Skip to main content

Full text of "Diachronic hierarchies in romance"

See other formats


For Reference 


NOT TO BE TAKEN FROM THIS ROOM 





- IIBRIS 
NIVERSUTAIS 
he RUAEDSIS 





BSOSOOSOOO OOD 
The University of Alberta 

Printing Department 

Edmonton, Alberta 











Digitized by the Internet Archive 
in 2020 with funding from 
University of Alberta Libraries 


https://archive.org/details/Lipski19 74 








eDereeuaNele yrs hee lol vor © FA bee Ret A 


RELEASE FORM 


JOHN MICHAEL LIPSKI 


NAME OF AUTHOR 


IBMR Oo WW TEESE SARE OS GARG Gs ao OOo GO mUDonaG OCD On DODD OOOO moose 


@eeeveeveevreeeeeeseeeereeeeeee eee eeee eee eee eee ee ee eee @ 


VEARSIHISSDMGREE, GRANTED UMM, tT 7 2ea0G et cdc. seeccesvcccdscoseecetes 
Permission is hereby granted to THE UNIVERSITY OF 
ALBERTA LIBRARY to reproduce single copies of this thesis 
and to lend or sell such copies for private, scholarly or 
scientific research purposes only. 
' The atic reserves other publication rights, and 
neither the thesis nor extensive extracts from it may be 


printed or otherwise reproduced without the author's 


written permission. 





THE UNIVERSITY OF ALBERTA 


DIACHRONIC HIERARCHIES IN ROMANCE 


BY 


JOHN MICHAEL LIPSKI 


A THESIS 
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES AND RESEARCH 
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE 
OF DOCTOR OF PHILOSOPHY 
IN 


ROMANCE LINGUISTICS 


DEPARTMENT OF ROMANCE LANGUAGES 


EDMONTON, ALBERTA 


FALL, 1974 











: : eiamr A a 
aa i eats abe. TR ST ee 
“pen se ao emer at =. RIE aa eT 
YHEOeRITHA tO HOTOOT FO eh 

“ui » a 

DOITATUOULS ‘SOMA 






THE UNIVERSITY OF ALBERTA 


FACULTY OF GRADUATE STUDIES AND RESEARCH 


The undersigned certify that they have read, and 
recommend to Pines Faculty of Graduate Studies and Research, 
for acceptance, a thesis entitled 

DIACHRONIC HIERARCHIES IN ROMANCE 
submitted by 
JOHN MICHAEL LIPSKI 
in partial fulfilment of the requirements for the degree of 


Doctor of Philosophy in Romance Linguistics. 





We do NOT know the past in chronological 
sequence. It may be convenient to lay it 
out anesthetized on the table with dates 
pasted on here and there, but what we know 
we know by ripples and spirals eddying out 
from us and from our own tine. 


Ezra Pound 


iv 





aes 


o range t 


awe 


worm QW 


brood sxsw 


® 
(enim 
J wet. 


. 
LOPE strip rub 3es 


OQ” wt ts pe Litenrice ry 


wor od 


of F 
3 Ww ee - 


rm 


AF. 


ad 


roti 


“he rity Aig art Ps ne 
2 RLAW ki gw hi ie 


ve hfe Settee is 
Ur ee YOO 2ibos Ett Ore 


oxid 


Yan 
od+ oo, ber Rearthaents tuo: 
ns exod TO Betasq 
agtaqix wi worl - a 
not? Bas a mort 


seared two 7% “IO f 


‘worn TOY ob < ae 
aT -S0fIBYBpA2 






ae | 
ae 7 
a 


_ 


ABSTRACT 


The hierarchical behavior of phonological elements and word 
positions has long been recognized in the Romance languages, but 
never studied in detail. As a first step toward a detailed analysis 
of diachronic phonological strength hierarchies, the behavior of 
atonic vowels in Italian, Spanish and Portuguese is examined on the 
basis of a numerical count of instances of loss and modification. The 
results indicate a strength scale of atonic position and one based on 
inherent vowel quality. Among word positions, initial pretonic vowels 
fared best, followed by the internal pretonic, posttonic penult, and 
intertonic syllables, respectively. Among the vowels, the following 
descending scale of diachronic strength was observed: /a/, /o/, /e/, 
Jaf, Si/. 

The resulting hierarchical structures are further discussed in 
terms of phonotactic compatibility of consonant clusters which result 
from loss of a vowel, in order to avoid spurious correlations of dia- 
chronic strength. The theory that language change tends to encourage 
open syllabicity is tested by comparison with the numerical data and 
the results, while not definitive, due to the small number of cases 
involved, show no conclusive trend in this direction. 

Elementary phonetic measurements suggest that the diachronic hier- 
archy of positional strength may be based on relative intensity, while 
inherent vowel length has little or nothing to do with such behavior. 
No clear acoustical correlates may be found for the inherent differences 
in diachronic strength among the various vowels. 

Word-final vowels generally resisted loss or modification to a 
greater degree than would be predicted by their physical properties 


Vv 


























Bas inpeost 


Set yeegewonsl sonwnet offs at 


. pkéyisos Baltsteb s frpwot Gade Jessi 6 2h of trade: rit be 


stokverled off .2zetrtotevoiaA roposrts ison 


eft oo bonimexe ei stsupudte?’ bas delnsge Reiss l me as 


an j . ee + => han meet J waeinres & 
off .f Sitibort brs. eeol jo aeonsteat Io twee feokismum 6 
; . 
= ° c t g . _ = re = Tata rs ‘s'4 
nD besad em bes nolgieda oinois to olsoe mipneste & susoiox 
" 7 , ae eg tweets tice vet a . > a 
sfowov otootosg Isitini ,.an6isieog brow poomA§ .\SLeup iowov 3 
" . a —— o ‘ ¢ rree 
bas .diurwe oimetdeod .ofmogsiq : Aeeni oct yo Hewallo® 
r y = e =x = r r. re > 
proiwal tot ort ,elowry orf pnomA .yievisoeqest ,asidsliye > 


: X ah 4A es - eee See 2 vee _ to ales 
La, « \O* \B\ Josvisdco Baw noprist as Op te ele: 


7 


- ¢£ ae, — r r +) rpc 
ni Sexe ri sexvispursS fa ee : v4 Le Ja (of5% ) Be PEG; —— wf 
a * » 7 Z _ wpe ls rie o~> — z 
“. $igesr rbirw exsteplo tasnoesoco Io yitlidiisa@noo Sfoosconorg 
Tw ” " : ~~ Ie : > * 


c 7 
: . ~ — —— = 
ae dvs , - > . Pon'e ty ' ee A ee ee ea a T ures 2 “.? on : 
“BID TO eno Lssiorico BuOLTINde DIOVS Cc fsrrO met isuovy 6B TO a . 
a 
eorcoone at abroad sormsrio spears. js:tt viooarlsy sont -titprens we oi DS: 
fas 62 insmn off siw noeiueqnco ya bojveae GE 


epee 26 xodmun [fame art ot cub ,ovitingeb don elite ae 

| A . 

-iimenib Bint 1 bastt svieulonoo on worle ,bs 

y , : - 

“tet oinowibsth off tsnit tasppue atfemeiuesam offenorg wastnemels “2 


Slisw ,vixrenstnr evicttalor no beesd oct yan Atpagete tenoisieg 


fon | : : ao P a) i 
‘tO.tvsiied dove iditiw ob od prirton xo ais z aah dtp raf Law 





esomsapitiih tnerscdni oft xoi Batot od yer aatel arxco lebis 
asa 
e lowsy asoitsy ort proms. Hires = om radials a 


“ene - 





* ae r ni — 


6 of pottse tihom 10 esol Beteleer yi iszenep elem | sinha 


a t egdeitinis isoisyng trent we Batoiberg | Biss S 
a : j # te 
““ ‘] yy © ba 


; t er _— : . 7 evs if ge . 
Wa AS. a aed i , : 7 _ aa ts i 
a ; ; 


alone. The role of morphological function of final vowels in the 
Romance languages, with special reference to Catalan, is examined in 
the light of more general considerations of information content, in 
an attempt to evaluate claims that morphological function inhibits 
phonological change. It is concluded that the overall rate of final 
vowel loss in the Romance languages studied is roughly proportional to 
the rate of word-internal syncopation, hindered by the morphological 
function of the final vowels. 

The hierarchy of positional strength is further supported by 
application to the problem of vocalic modifications in the future and 
conditional forms of Italian first conjugation verbs. Examination of 
alternative solutions yields the conclusion that the change may be most 
readily characterized if values of diachronic positional strength are 
included in the description. 

In answer to the question of the place of hierarchies in phonolo- 
gical theory, it is concluded that such hierarchical indices belong at 
the metatheoretical level, serving as interpretive devices acting on 
individual rules and feature specifications. Some elementary suggestions 
are offered concerning the manner by which hierarchies can be made 


compatible with diachronic rules, and with distinctive feature values. 


vi 





















ch doedimo nobiwrercint Io ekitersbkems bs 
esti anes iad tyoteliquan 0 tS AcE A 
Pend? 2o ‘ser Lisi ot Jar bebitons Ab IT paenpreer 


a | 
AEE oitscrcme Lamesdrektaoke 26 ‘ete ist : 
elowew Legit ay Yo nksoneR 


o 
se a 
Piss 


_ 


Ss 
| 
- 








a 
ie a 
oe 


_ = 


» i ae 
* 


Ye Hedroqque asrhial et iiteterte ‘Lenorttaog 36 goaeeld ott 









i a a 
: a bas suet of m2 aroitsottibon oiisoov to me Leloney ons oo noctlegs wa . 
+ | 20 notjsninisxs -wizev aakiseotae jai msrisdl te eorrot cecots BAC as : 
[ 
* geom sd vem somes sii tord poked Isabe sit eBLery anoitufoe ortiaiislds. : 








; - 

% ous thipnsrte Lanolt tao nicerntibsib to 2aglsv 1i besivedoered> yiiiese a a 
= —_.noitg hme ort ok bien 7 
ye — 

' -oftonoda i esisdetersid to sosiq ort to cobtequp. ort oF ccicniat tat ue : 

a ~ Ss paaied sen ltri isotroisis: id toue tect bohuisaro ak dd road fsokp. a, 

| ho pisos soniveb eriteietedni as pnivese fovel Ist torceettsichan Gets 4 a 


_ > 


erokisentve yisinemels ano® .enoisesitiosge crutpet Brie soli: Keublvtiat fe. 
abent @d nen saitoys¥elt Hottw yd rennee St pndreteones: bowtie iA ; 


aeniay euctsst witoniter div one (eelirs > iopateslb raw atdegiee. x] 
kia: a 
= oo, oy A 





a ( 





ACKNOWLEDGEMENTS 


Although bearing only my name, this thesis would not have been 
possible without the support and encouragement of my wife, Beverly; 
therefore, thanks go out from both of us to the many people who have 
aided in the realization of this project. 

A special debt of gratitude is owed to Jo Ann Creore, my super- 
visor, who carried out her task under arduous pressures of time and 
distance. Her constant help and patience brought the thesis together 
under trying conditions. 

Working under similar circumstances, James Anderson contributed 
his share of assistance, smoothing out many passages of snarled prose 
and helping me realize the true value of brevity. 

Thanks are also due to the remaining members of my examining 
committee, Tom Priestly, Edward Green, and the extemal examiner, 
Herbert J. Izzo, for agreeing to evaluate my work under less than 
ideal conditions. 

Help, encouragement and moral support were provided by friends, 
colleagues and students too numerous to cite individually. Most 
outstanding on the Canadian side of the border was the tremendous 
encouragement offered by George and Janice Patterson. Silvia Budde, 
Rob Cameron, Francisco Cerda, Kim McCalla and Francois Néve de 
Mevergnies helped me through long winters, short summers, and strenuous 
moments. The Interlibrary Loan department of the University of 
Alberta library provided invaluable bibliographic assistance, by 
helping me obtain numerous otherwise unavailable works. 

On the American end, I have benefitted greatly from the rational 


perspective of Dexter and Ellen Pease, and from the support of my 


vil 
































need aver ton biugw atest ait) end NE NERO 
pyizeved watiw ym 30 tnetepsiwaone as droqquenae 
sunsl only elqoea yrs SS ay 20 dod mote duo op sams 
Gwe tostosg eitt to noises cisss “ie. 

ot od Bow BF shiver to odae Letoage 6 > 
aycubts xsboa Azed tet tuo beieaso one xb | 
ated Jnoteaeo seit sense 
bedudixdnos nocishik comBl ,eeonsJemor > yeiiuie meen ontat 
easy heliene to accaezeq ye Jvo proce somptaiaes Xo onsite eid 
ere ¥> cytev art att esiieax em pated bers” ‘i ; ' 
ALEmSst csit of cub cals ous sam ie 
yiiesixa mor .sedt iene 


“ee gros) unt 


‘See anit To zouwsses74 





-gerttegos aieeds oft gripucttd sonsiteg bas 





onisetinmae an io = pr 
vTenimaxe Lepysdxs oid Das Naat Bnew 
aed) aeal tebe asow wi steuLevs oF pilots 20% ,o8sl -\ tredasH x 

| 20s thee, Leet: 

,ghreis yd bebivoig stow Trogyve {stom bos taemispetuconme: eh 
gem) -videwbiyibntk sth of svotemn oot stashute has seupeeliog, 

euchremsis 219 @BW tebiod exis to obia neihensd ode no SriBaesesUD 
“shbu siviie .noexadisT Sorast brs a faa vd bersiio FASTIN, Z 
a ove atognsst Ste Sl is OM mix , Bhs con toner’ A BE 
sucuneite bos ,aseuwe trode .erciniw pant npuoui an begted ee: 
$0 yttexevint) adt 30 Jneottsge asol hart: airs = 
yd .somdeiaas oifasxpoildid aidsntewnt sah x iw 


a 


ok a 7 iv, 


sees a ies td na om nen ore: 0 


colleagues Orlando Edreira, Israel Rodriguez, Cayetano Socarrés, 
and Charles Wendell. Rev. George Herbert, K. F. Montezuma, and 
Steven Strange helped in many intangible ways. A special word of 
thanks is due to the Princeton University Library, for granting me 


unlimited access to all their material. 


vill 


a _ 
ait 


os { enter Se OF The! as oS sxe, 6 
shite es ' % ae a 
7 7 anh 

de bes ness dora! | a ered * Lm * 


“e Brow  Enio soe A .2tw olclionadstt von nt Se 
am pat inere vod y yrerdit A ciasinaias porsontst add ct 


~— 








.ist ist y rier ifs a eas 








= . 
=e 
3 rv 
a F 
i 
be 
a? 
al 
14 = - 
& fi i + 
ae a a 
i 
i . : ’ 
1 Dem wae | o 7 a - 


CHAPTER ONE: 


le sd 
Je 
£2 
et 
Ba. 2 
NAS ee) 
1.4 
diye) 
1G 


Mod! 


CHAPTER ‘TWO: 
25 

rs BP 

cout 

PERE! 

Zaz. 3 
Dein. i 
Dstpoee 
PORE ES) 


2.2.4 
2. 3 


TABLE OF CONTENTS 


STRENGTH HIERARCHIES IN PHONOLOGICAL 


The lstucymoGeniewarenies. osnigis secre c sss se ecee 
Types of phonological hierarchies............. 
Hierarchies based on phonetic properties...... 
Ranisniguorevowe LITINDIE ce sicac ces ac cteicleis ws 0 cies 


HierercnieasoF .wordeposdit1.On . sc. \.\s siaie os es S's 


Thesposition Ofeanti—pionetism..........ss.0.ss 
Thespreblem of cause and effect............... 
D&GABtOlp WIG BENCOL Yami. oso os <ncts'e sess ss so els ee 
Scope: OL Lhe presen’ SEU .i< mice cles cicle Wiaie clave ce 


NO ESSE LOmChap US ta CRIC oats. isie's systeicis o slelcie oo cisverctol cs 


HIERARCHIES OF DIACHRONIC STRENGIH............ 
LTE OGRIGH LOM steele elets © eisieteleieisic\ «\s'eieley « cvlsssie so rs 
Hierarchies dnglital ane eantwwe ...05c- ses otek o eaht 
Tem GhOt CSLOLPLtol Lalvaeiiines.-.0c.> «oss sss sos 
SEQHeSORR tie LODE sad <Wrecterepe ols cis ew s'eiseis se 50 
SELECRINGRMSRCOLDUS .)fpe.s: ie eles sn oisis s 0.0.0\0 v10's vine 


Tx HeROCICTE 1 Ge Views pevaatenenere ase ecevavelelactls.c 6 6 6 ace's ee ee bes 


BROTH telsutce peal Cla Ge GEE. 0 Old Cee ae er ere ere 
Pred iminatveetesults:. .astyadaaste sas sce cesscenees 


Hierarchies AMSSpanLShie Bee soos ps aeeesces 


ix 


al 
30 
36 


38 


4h 
yy 
6 
46 
L6 
49 
kg 
50 
52 
af 
58 


7 | ae. et 
_ ‘ 
jane ee. saber 














i s£ ‘ : 
ae 
he a, 7 
1 - in a y + - yi in -. 
AS IDO IOMONY UT CST RORARSTE HTS 
~ A 
» N er 
, 7 Verra 
’ - i ye Siete a vid 40d. 6 biaseip' «ven, aa ine Dk eee ee 


ae 


ay a -» a - . 
ime Pog eT S —s ent 
>» -f£ ew eeeueeonrvreenet **eerree® am. geene= + A. A? VE Sa 
oP Z ss 
4 * > 
) . ta Tenivotonoda cto seer 
< see oe 2é 7 Lk Siri is |? OL OnOn ZO eggy? 
7) a, 

7 4 be . ‘ i rae 
. ee A | rey: oy 
 tekeds mq Diss : 38¢ friose2as8 

f ~“~—> = Pah tt 
Keen ee ccwcevvceucen es SICLLI Lewovr to pretoiish 
\ 
- ‘ » >. in 
2 - ; uy 7 = r tea @ 
a+ aeeestanes wees e seit iaod Sicw So LAOIBISLA 
» ow oe Wiic 
be ial eer e ew ene ee eee eune seteteoeweeen e-« “ee ee 8 {a 
1 
aE - ‘ P J * p ar) 
Ce ~ wise ».- Meizoanorg—c: noisvtesg Sat 
ro 4 TS oe ae sees ; ; 
eevee ee sacetw 20) Drs oe § Suit 2 SLGOIG orci 
a c 
Oc z wre T+ ori Oo” S268 
ws eeeee* bee eeweste eet * pido pe ge 6 © | oe Fa kt 
My 
t P Lt te “yat} 
ia “2 . * een eevee y eas y ~ IQ OSG 


CP 









bio ebiacaee og. ¢ Shite DIVORDAI.4O analy : 
7 > 
Me sted oe wate ee owe ; .- Moc togporsal 
j — _— = (ths | 
Oe! . Bhasin Pens oe bien orem . f5tissT at seithustem 
} . 
a it Py §. 
OB sad eavedecs sevcssvecevelOllet) 10 Solo Si 
>a - - P " ‘ > 
o a owe oe emis . ° malay ILI. TO xO? 
i os , 
_ = i. . 2 oe, 
CP gins peas 8Ge onsale acacce ess sn QCiGo Of Dirsareme ‘a 
j 7 
Oe ren, Pole ibowt +) "~ 
x “wee ewrn ee w en enewes ds 6's Sates) tae a ae ms 
: 2g _ 
a ye ~ = 
~ eeeee ‘ **, > *-* ** “* epee ee ee + ee : 
— a ae 
= - 
; 
GCP oa 
a, as eee ee ee ee ee ee .s«patod Deda. intia ° 
¥» 7, cy fy 
bh 7 Te ab & 7 4 * 
Tt ry a ebikibusnidee leds <neteaceg sane 2h ett 
Sy o, : oy 
j ; a! 2 ; baa 
utp 3: cle bts wale mee een oie ieee as aii 
7h. 7 a 
Z ; 
pate 5 
i: ; 
a 


" my '. yo's 


7 Caen fie 








Pind dlee. 
DistDiwe 
eS a 3 
2.3.4 
2.4 
2.4.1 
2.4.2 
2.4.3 


2.4.4 
2.0 


CHAPTER THREE: 
ek 

3.2 

fee 

B26 ol. 
So. 5.2 
Gc0.03 
3.4 
3.4.1 
3.4.2 
S425 
Bee! 


Beoeds 


TABLE OF CONTENTS (Con't) 


Page 
Mabmoduccich yews. vielen as. sss sw ese ess 38 
Spanishrasfaltes ti langquageee.. .cccccescccence 60 
Seleebiontot ENeCECOrDUS ae elite tie + c's es sje.0 0 wets 63 
Brcosee tyres) CS. fa)c:0)0'oeaske aie eae eiciete ne aie. stele’ 64, 
HEGPaAreneSiiin fPOLtUuGuese Brie. .sie ce cs wele ces se 66 
TinerOduclron wae. a. Mace Seis c.c.cic,e + ec cvdiae ce alder 66 
ThencholcskotaPor tuguese.skiia.. os. sek teense 67 
Selecting PSR CONUS Seis talents a:0 we cles cteisisissiaces 70 
Bre emi nia ryeresiil CS wees, 6, «104s <icle eso ciclo hel aie specs “Th 
SSUMIUMILIOY, tiscoterefede olere sic'e eoevalel ceive. bieielete et sce veretstere see Te) 
INDESSELOSCNODLE Er £1WOss <iecm os cterelsie co dtars ina erie ite scr ak 
ADDITIONAL PHONOLOGICAL FACTORS...........+4.. 83 
UO GUCELOMNS. | asi. Care eels: «5: o1c10,c10,0 bo/s « selene eceiace 83 
Consonant iclusterscomatibidditywe.ese.. 060... 3h 
Thefdevedopments#in yi talianss:. ccc eieiciee cc cee s 87 
Compatible! consonant; Clusters ....... 0. casicvccsss 88 
Behayronrot fatomic posteionseec. .. ces see.s se 89 
Behavionror mindiyidualGvowels ss... 0..c. esc es ne On 
The developments in Spanish......cccscsccncces 93 
Conpatiblemoonsonant .clusters ick ..:00s0 cece cn oe 93 
Behav Ors OiredtCOnLGtPOSLCIONS ce o's/0 001s «6 0j0/0)010 00/0 95 
Behavior Of individual vowelsy............<... 96 
The developments in Portuguese.........seeeeee 100 
Compatible consonant clusters..............ee. 100 


x 


senses eas er weer s os oe sae eren 
7% 


- ‘ _— w- ff Doc? 
oO oe 2 sis © eo « Fee © Seed oa Cc TiS he Sasd B 
Seal - 4 : 

a oe 

> ' © ~ ~~ 
£0 eee e¢@eaeetere «ee o tore ae 4 OTe 

3 


27:3 


a eG >, oo rort F, 5 
Vv iwee ee * » «ee | AW OL >. ~ ort 
7 ou pleot mae p ob + ve 
a i ’ s ? £3 lo> 
OF ase -* 4 eae eee + Cel tX) sai ME SOS, 
me or E sp tae trvery crore fl ott 
} eee er * “wee ** . | » X= p65) 9 me ES 8 he | 
“ = ee = 7 a 
ee e* aes oe esse aeweees ete * eeeee ¥- Th }} a fae 
+ c 
; La « * nt 
y 7 f 
‘ aiee ee wwe n en te ork os He Sa 1 © ) OS Searcy. 
a aor “yr TAS Tew ns > « t Trt K 
£5 en*+eeceeew eevee ewe i A uw w a a= Uwe aie?) 
re +r 
a) se*s#* ** peace e eee tae . + “eee mE = b op. ft 
Le ry ry) " > 
~ 7 oreee J. ens he Cok » JIbmivaika 
~a . 
of) = r —- om ip 
re anne ts-ew ap cinae whee fr atrengaiavss sat 


fe 
ti eeer ev Be © . . * 


exretecina tosnoen. 


. 
t 


iawov {sub.tv fae ga 


aeeewe et 


»* Stair pte a6 


eo " +eytorct= brats 
BA dew eee bieaveediende -QTSCRULD SISter 
ag srewi 4 toe ade 
4 4 ee © L 1LROG ais On Es 


ov 
7?) 


ms : e 
/* 
‘ 
. 
2. 
‘ 
a 
+ 
‘ 
‘ 


ue 8 ewe @ 


A 2 
, ic a ' 

‘ q y A 

ey Oy i 
~ | Ss 





to noldosise 


Spoiosborctal 


=i) 4 daknace 


*, 
(etry Ter cr teres 

> ee c + -o- a } 

oa seoeee eee” ove ws «SGPT fa pe AELE SD Ssi5Lh 


eld issanc 


4 ee eee oe 
aa) se 6 re ae, a ee . ANOCTIZO OLS 3 toiverist 

a 2 oy 
70 YOrvVsens 


no e aaa Re, adr 


30 sobbed 
ples — a “s 

-elanor Leth ro koed 205 nonhae 

ee een i spree att ‘e 


* 7 
a rye ; 




























= 
_ 





7 


hea 


ide = 
:GSaHT SPA 


an 
<% 


a ™ 
ae >t: 
oy — - 
”, *, Zt . 
—— 
; 7 
" £ 
— 
— ’ * & 
- VJ ee | 
: : a 
— 7 


_ 


wy 7 ty 


— 
a 
: + 


bY 
oie +0 
oe 


- 
Ma i 


cy? of 
. 





TABLE OF CONTENTS (Con't) 


Page 

oye? BehavaOreO te atOn1C  DOSLCLONS weleulcclsies sss 6 + «\s'e 107: 
2563 BehavlOLlOrelnalvidial: VOWS LS ss. cc's ss ces e's «> 102 
376 TMevcuestion of open sy llabicvty... 5 cee. as non 104 
SI sag WAC EOI CL OM ceistcts sens o's tee GVelegetstetetersts cities tials os e,61 1o4 
B.6.2 Qvpenesy Liaoveiey er nel Cali an ee ctesteis 6 cries crs 107 
3.O.5 Qoenksy Tabterty mane Spanish. sss stsvslsie 6 </sicin.are 108 
326.4, Open*sy liabicity in Portuguese. 6.0.66 esse css thee 
327 Possible physical correlates. .....<.sceeecees 114 
Notesm ton Ghaptemelnrece sr. ce. nese. see ese: 124 


CHAPTER FOUR: FINAL VOWELS AND THE EVIDENCE FROM CATALAN.... 126 


ae Imghas ibe: Weigh rcs oo ah FAO COON Ceo aCe 126 
4.2 Mievtateeor Tina le vowels... cece s+ess es chorareteretetore Ley, 
4.3 Aon CEVvOwel s ait Cate latices sais ee oeleieists sfeiale ie oe 128 
4.4 Preservation of morphological material........ 133 
4.5 Gender ands rumber sn tala. ...5.0« cle cc c's TS 
4.6 Gender and number in Spanish............eee. 136 
4.7 Gender and number in Portuguese............... 139 
4.8 Gander andsmmbentin: Catalanaenr<. ssi tees ss. 140 
4.9 Information theory and redundancy.........eee- 142 
4.10 (Wo ealivtcbliais Sige ene BOSS USO EIR IOS OO eG 149 

NOLES LOrenaDtel: POUL. vet sclcleisie civislerss se csislee LL 
CHAPTER FIVE: A MORPHOLOGICAL CHANGE IN ITALIAN............. ne 
ed TLC LOL lerettt utes eta ce a ne taverns ese a'r whe canis ss 6. 12 


LOE 
sor 


bor 


tot 





en vai 

x iss 
~~ 

Pi! acr 


Q co ~J 
f me | fm 
r - im 


Oo 
= 
Ee 





. Hot 


Cs ts Ud on a eee 
















sokee asin =v ALES NS TE 

coceeessue  Uipidelive inegs 20 noises act bd, 

aban aati gc cae -P 
. ike? 


.. WATADAO Mond SOVEIIVE. SHT GMA eTHNOY GAnTS ACO SAPARD 


oa 60 oe bse Qe ee Bee sor 2 60 tae ee oaanfolo iiosinl i i 


je aase ww Ciwes ohne ce wag sfsts) nt elowov oiricdAéA ) €.5 -, 
eve t§iueten Lsoipoioigaan to noiiayresars he 


tes > 


a Tees et np hist! cf sedan bas. tabs 


subu ate eves yonsbnuber brs daca coisa 
saainin eon iceoh ss ee ae ae ‘ah 
ie 


oo aren Lae 


ld 


Slay 
216 8} 
5.4 
See) 
2.0 
Sd) 
5.8 
See) 
Deel 
Did 


Sele 


CHAPTER SIX: 
6G. 


GeZ 


Giant 
Ge2e2 
Grees 
6.2.4 


6.3 


Sion 


Gree 


TABLE OF CONTENTS (Con: 


Page 
PER OLOU LEM rae Gia i aie erate aon rae Sane 156 
OeLGUO POL EUS SCUANGS , carclee prensa se s/s. ek aT 
Thegrals ingen? uence Ofelia. sarth. ctr. alley 
hes lowering inf iience Obst | leer se. ss 163 
Bilis erste Mares. Ole | Yo scree treeeate eerste reece sat 165 
Pe erO LESBO PE SCUCSS crce olen sie wists ocees ECE oi LTS 
Raising of unstressed vowels........e.ceeee. lie 
Tee Ole OTN LG DOs IL ELON csv s ce et tciss sies cee 17 i 
DOSE ena MoE MoumLliie..t. ot ccs aecmen tie loe 181 


The vowel scaley a speculative interlude.. 186 


CON CLUST ONG sree eer ieis te teralave, ss ohtnerortis o steucteee ene siete IES) 
NotesatoaChapter F1VG....c.cccrs ese einer eee 194 
BUN MCONCEUGION tet cemmieicrs sce ier ccs cette atsiere 198 
Sata? ie jaSsel ec ye oe oo ae ao 198 


The place of hierarchies in phonological 


TENS OF ANE Sr GO AI OO OCR OT as COD bs) 
URE EE OCH 1S CL CN es per meee Net rte’ to greater ee tercicre chearc = 199 
Hierarchilesp cca thiCOYy ss css siete ctcelc ele sie es se 200 
Hierarchies as metatheory.....s-.wassscen es 201 
BSVChologuce ls amolLiCaELONSinie «swe sss cts 6 ks 202 


Representing the hierarchies: diachronic 


Uiches Msle Mes So SI SCIRIES Saree oc ao een i hee 204 
204 

SSOMEM CLIO CS) AAA. ssc steer visie css sw alse vc 

Hierarchies as step functions.............. 20k 


vom 
| an 
Se er rer Tt sprit oi rigs 
ee i serdar [1] 20 sbeBittint eating Set 
[x) 2 orton fir pacbrawol sar 


Sh, Ss eer elt) so phe serie 


. 















rat 


eee Oa ecnnaese 


OTE ..cuanns---++--chowov Beesertans 29 pita ten 
TTL ccuceeccuveave vee ecbitieaq biow 20° ae sgt 1. on 
a De Vek eee ry <8 e davertt ‘ova Tengteteot 


alsoen Iswov sar 


Ww 
ce 
Ke 
E 
i 
a 
st 
i‘? 
Q 
0 
5 
Rel 
v 
a 
ve 
ie 


(ee es ae Br Po Por .  “proten iors 


daad Alp hha de <8 ee va +. BVI xedqedd OF Bagon 





see tadwetoaovanenevrvr ee cic eweccese MOLEULMOS VE 


am 
ae 
~~ 


ose eeen eee eer erweowe eee 


OP pawide tas cates 
DUS. .seswu events cess aes 


fos 


sos 


#05 





ig * at ay aa ; ; : 
4 7 = aa - - i 


Geo. 
6.3.4 
Gea ae) 


oes Sy a8) 


BIBLIOGRAPHY 


APPENDIX: 


TABLE OF CONTENTS (Con't) 


Page 
Hierarchies as continuous variables.......... 206 
MOATTAMLG a EU GS ae etets cial ates ote ert ccd wae nat Gaia e 207 
Hierarchies aS Variable rmiles..s.<sesecss sess oles 
Hierarchies as variable metarules............ 214 
NOLGS at Chapter aik assests as wate see eet ee cs 218 
Biche tchetelererecn cletecere Terence else ab bro esateleleceale eevee stone cclerets 220 
LATIN—ROMANCE ETYMOLOGIES...ccccccccccccccccs eu 


yatalal 








ere ee: ena 


sentinel EE 


all ahi Ma ov. tae Oe Fo ORR otibtay. 06 eal = 


So , e* coe 





eee Rew ewe eee |e eee eee oe a ee 


asta} bs tein ess «oe « Cobbea kd 





Table 


10 
i 
12 
13 
14 
15 
16 
iy 
18 
ik) 
20 
Zu 
Ze 
BN 
24 


Pe, 


LIST OF TABLES 


Description 


Italian atonic vowels 


Spanish 


atonic vowels 


Portuguese atonic vowels 


Italian 
Italian 
Italian 
Italian 
Italian 
Spanish 
Spanish 
Spanish 
Spanish 


Spanish 


Portuguese atonic vowels; permissable positions 


atonic vowels; permissable positions 
initial vowels; permissable positions 
/a/; permissable positions 

/0/-/u/; permmissable positions 
/e/-/i/; pemmissable positions 
atonic vowels; pemmissable positions 
initial vowels; permissable positions 
atonic /a/; permissable positions 
atonic /o/-/u/; pemmissable positions 


atonic /o/-/u/; permissable positions 


Portuguese atonic /a/; permissable positions 


-Page 


Dt 
65 


Te 
90 
91 
92 
92 
Os 
9 
96 
ox 
98 
sy) 
101 


103 


Portuguese atonic /e/-/i/; permissable positions 103 


Portuguese atonic /o/-/u/; permissable positions 104 


Italian 


atonic vowels; open syllables 


Italian /a/; open syllables 


Italian /o/-/u/; open syllables 


Italian /e/-/i/; open syllables 


Spanish atonic vowels; open syllables 


Spanish /a/; open syllables 


Spanish /e/-/i/; open syllables 


Spanish /o/-/u/; open syllables 


XLV 


107 
DOT 
108 
108 
iT) 
laa 
gale 


lala 





7 


a 
2 


7 


> a 
x i - 


is 


= 


oe 


7 _ 


7 


ae oe 
i 


a 





Ew 


tolsginoesd 
re afewoy obras setingt 
2 elenov otnads dekamee 


a 
ey a alow Dincds sessed 
Qe aroitiecn afdse2imeg. (2iawov sisots ns tisat 


Re aos S TBO eldszeimed :efowov Isitiai dsifsdt 
se anoijieoq eldszaimsg «\s\ melisst 
$e enoistivoq eftseaimeg i\w\-\o\ meiisat 


ee ” vaaodttsoq eldseeimisy s\E\=\e\ ote iteat 


560 «Cs pamkdicog sldsaeinneg icloww Sinods neinege 
oe | amet ieog oidsan.oprisg vefowov [sidint retinas 
ye anoitieoq oldsesimred 7\s\ cirots datisd 
§o -  appitisod sidszeinmed ;\u\-\o\ oimogs dainsge 
ee emiitieog sidseetmraq s\v\-\o\ oinots deinage 


fOL anpisteca sidsecinrtey yelowov dinods spsupusict 
€or anpistieng sldseeinrteg <\s\ ofnods oecupudaxo4 
G01 ag@otsiace aldseeinneg ;\t\-\o\ olmods szauposio4 
Sor snoitizoy sidsaziimed ;\w\-\o\ olsods sasopctiom 
yo. esldalive risqo jelewov o1negs mpi LetT 


yor . peltafive msgo \e\ asiistt 


GO. ; aaldelive nsqo +\c\-"\c\ "ap Elesl - 


BOL aeidaliye naga >\i\-\e\ ostfest 
Of - eofdelive neqo ;2flowov oitois Hennege 


“ eoldslive raga 3\s\ dete ee 


Ek _ asTdstitye 


























LIST OF TABLES (Con't) 


Table Description Page 
26 Portuguese atonic vowels; open syllables 13 
27 Portuguese /a/; open syllables 143 
28 Portuguese /e/-/i/; open syllables 114 
29 Portuguese /o/-/u/; open syllables 114 
30 Speaker A (Italian); vowel length in ns. 118 
31 Speaker A (Italian); vowel intensity in db. down 119 
a2 Speaker B (Italian); vowel length in ms. 19 
33 Speaker B (Italian); vowel intensity in db. down 119 
34 Speaker C (Spanish); vowel length in ms. 120 
35 Speaker C (Spanish); vowel intensity in db. down 120 
36 Speaker D (Spanish); vowel length in ms. 120 
Se Speaker D (Spanish); vowel intensity in db. down 121 
38 ita lieneacons Chasbefores r, 170 
39 itallansatonie ve before r ie 
40 Italian atonic e before r and nasals 171 


mwob .db ai yifansdat Iovov 4 (as EfeoT) A Jevbene: ) 
mob .db nk vite est Layo" ; (as cledT) & ionlsSge. 
Mob .db mt yiengiat Towov 4 (fetiege) 9 teAsage 


mob ab nii-viiensint Lowov + (rainsge) a xsileeds 





















sofa live nso yelewov phrads 
ys asl dal tye asa AX 
las Oy 
satda lye Bas ARAN 32a 


satdelive meq + WAAL sett 


eat ok dgnal Lion, smsiEs4t) A selec 
.am ni pael Lowe y (apts) a terisaae a 
vam ni ditenel Pawov 4 (detnsqe) 3 soneeme 

«am tk retposl Lewov <(datasce) 4 roxsoge 


+ sited s oincis nsEledt 


1 sxOted' 9 cinots miledT 





efsesn Bos x et@ied 9 oinots nerissl 


CHAPTER ONE | 
STRENGTH HIERARCHIES IN PHONOLOGICAL THEORY 


Tvl The study or hierarchies 


A linguist may approach the study of the sounds of human language 
on more than one level. For example, he may study the physical properties 
of the sounds themselves, and the vocal gestures required to produce 
them; this is the level of phonetics. Or, he may observe the manner in 
which the set of sounds found in any given language behaves as a formal 
or psychologically motivated pattern; this may be temmed the level of 
phonology. 2 Within either of these two levels, no matter what parameter 
is used as a measure, it will be found that some units behave differently 
from others. For example, on the phonetic level, some sounds are more 
easily perceived through noise than others, some involve greater artic- 
ulatory energy than others. On the phonological level, some units may 
represent a wider range of phonetic possibilities than others, and 
certain pairs of units are more likely to be confused by speakers or 
listeners than others. Having observed these facts, the linguist may 
ask whether within any given language there is an internal consistency 
in the differential behavior of sound-units, and if so, whether any 
patterns are discernible which are applicable to more than one language, 
perhaps. to all languages. 

Investigation of many languages has revealed some sort of cross- 
language consistency among sound systems and the possible existence of 
many universal or quasi-universal patterns. Such observations have led 
to the basic axioms governing the study of phonetics and phonology, and 
have provided motivation for the concept of distinctive features and 


broad classifications of sound change like assimilation, lenition, and 


ra 
_ 





Jas 


wilt 


_o_ 


a ees ee - oy scit bis .sevicemerit ebauoe 
elitr od betiupoet eaiwsesp Lenov ocit Bis , sevice 
rr ' 


Stic 


i 4 ats Cat ear | yu am rey be 
to [evel acd? bemet ed yan ais tiraetisg hedevitom yi isorpoL 


Si, 


bas ,ypolonorkg bos epitencriq 
LOO 


os 


- 7 
br T) moins af ,noltslimiess 


' D r . 
Ps Aco cena ito abruoe oats t. facts seid rosoxggs ysm tatuprkt A 


a i Terre 


oo 


ovr - 3 ‘t“aadt in ‘al olexratl cori > TS CS fitioty «¥ 
— SxX6Q Jethe 2 ah Le @ 1, on =) eee 9 Farm Bee me he a ’ c 
; 
wait auvedet oticn emoe tarts beet .od ib i (N2ear 6 RE 
Misitio sveanec Ssh Se Jie ta a LS ‘ , SA 


~ - = + 
r ¢ ast r las | re, “Try a 
M8 etrmoe spe ,isvau sitseaoia eft no ,sigmsxs WOU .cisto 


Sicm S3I65 six es 
= cc 
om bbe ae ae ons [ceri rie 2iot ito rari ecton Anco av iaoxe 
= Se ‘716 OCH YD SViCViis Stevo «Gd ii DS b ite 
"oe eee a ee ee a ee ee 
vem etinw ame ,ievel Isoipolomorg ait nO .t oO agdy yous 
= ’ 
j , 
-_ . ; ~ « 
"? 1 ry 4 ;.* -- ~ “ ~ 4 » -, r;* * ~~,» r Ld c SS 
oa i.3 we oa F- iroorywy * “¥¥ rx { MID Lo LW > 4 
{ fas ,erertto mst eeitiliorgeog SisokxKx > SOMBL IP. % 


: ; a on 3 a aid, 
teironil et .adost eee? beyresdo piivelH .sranto nem aveciadill: 


witeixe eldiesoq st fas expeteye Sivoe pnam youate 

























LT 
7 


. fenral ann + 
jnegoag Lnoteyre wis Yem od 42 signa 103 «evel aco asild 


{ . Cet =pry ft od 7 ‘ iat wel Dey Tenge agti+ oF oral -« 

i ory ¢ rssedo Ye TT 2, 4 10 oI Snic: iO isvs [ Ww Gt enn 
reciect > f yy f rit rey’ ehrurce to toe att 

as eovecied speupnsl movie yns aft fro ehavce to toe of 


\ 


i - ari 
seat Yo tari: 1 basi 


* ae = - 


rlesce yd i beestinoo ad o¢ ylextil exam sis sticw Yo etisg niet al 


e 


in 
noo Lensetai as et eaedt epsupast nevip yns ain tectirte 2S 
a 
ae 


, - . s: | fet my 
soritertw ,o@ ti brs ,echinw-baboe to solvered Isisnexetiib at Ae 


ano neadt sctom ot eldsoliggs ets doidw eldinreosth aus are 


; > 
ae ite se ce 

4 -2590800M5. i356 Go Bs ete yi 

7 - “ee 

. : af a ee eee 
io t1o2 saioe belssver ead eepsupisl yiem Io nottephi mare, i 


x 
e 


coco sean 


7 
evel anoitsvisede dove -erredtag Ls i exav inten: p x0 isexevirw yea 


. = aenctant svitonizeth 


oe 
ma) 
‘ 





“ 


| ro autmievop emotes 2 aed 


e 
at 


apocope. Various parameters may be chosen which can be quantitatively 
measured in each sound unit, whether in a given language or universally. 
It is generally possible to let a parameter define a scale or hierarchy 
along which may be placed all the units in question, according to their 
behavior with respect to this parameter. 

Empirical evidence pointing to the existence of hierarchies may 
be obtained in several ways. A hierarchy may be established on the basis 
of measurements or observations of the sound-units themselves; for 
example, a phonetic hierarchy of oral closure, or of amount of glottal 
vibration, or on the phonological level, of the number of correlations 
shared with other members of the phonemic system. It is also possible 
to define hierarchies by means of indirect observations, concerning the 
manner in which units behave with respect to some other phenomenon or 
process in the language. The most frequently observed hierarchies 
obtained by the latter method involve some variation on the notion of 
"strength'. 'Strength' used in this sense refers to the tendency of a 
given sound unit to resist inclusion in a particular phonological pro- 
cess to a greater or lesser extent than other members of the same sound 
system. Since 'strength' so defined is an abstract term, and since 
most investigations of strength hierarchies have been based on the 
behavior of sound units as part of an abstract system, strength values 
are generally assigned on the phonological level. 

The study of phonological strength hierarchies per se was not 
common in early linguistic works, although many of the fundamental 
notions were present implicitly, and occasionally explicit statements 
were offered. More recently, the theory of generative phonology has 
brought forth a renewed interest in formal patterns, the mathematical 


structure of language, and the discovery of language-universal 



























xloviselen & nO beasd emnisipivensianl Iszievinw to pdiviry 
nage Lint ext od noitnetts sis3il bisq oved son dite 2: ctemall 
fanigofonedy cial | Seeoreoah lemot to noitexoguoont vit ete ; as 
‘ed yino {liv elseievinw Lsotpolonody wniriet idistee to Lsep ad? i a i o 
; -igns to ite s aiv .modtod att mee ti pritdasowgqs yd besiisen vine % 7 
So? ,badon as ,xsvawoll ., aoe a seve ridiw amnivsoitewat tents i" - 
20 noltslumoos tesa sdf = (81 +f6e1) senect sopedet yd ,oignexe _ a 
no eben. ath enoitestiexesse eeeira ,sulsy stil 3o visme ak 20 LonsKo me % 
" Snwons nissiso s oftrw \basrt terte edt 00 .'aottsvieado jo atesd sckt 4 
perth wat pnifisnpie nt : sfdentay eyewls 2: noiteluosgqe jnspilisant ta se 
ed. taum febom Lemroi sit tatog emma ts , note ive 10 dowseas: 103 anokt — 
-\ ,g@eteeu od ifiw yross slaw edt 10 ,6¢8b Levsos xi dtiw a 


. 
‘ 


niesiss euolficden sit ests co tonetys 18 etasesiqes Youte snseesg oct : 

~soittxeg mo prikenoo? yd ,soiiglugeqe bas paixertep-stsb ~yino rooted 
[soitsxoserts Istenso sion 4 ot tuogque bao! rau cut co iriw sonnet a 
thencvbs asd (TTS :STer) ywhoiws .bae ehit of a 

~ a 


es 
a hy 





astoeni yino Insaltbapite er acirerustaid 90 aoieeuoaib A 
(gexovied tisit <0) —on uslimie to ome aft ee 
-tih wren to exemmste ait at csiog a Sa 


Ih 


ey ~asq 8 tortd Jnemous o& OL deci? vespeupasl Inga * 


~oldnorig at seipiecn sagt 8 tal ata br ai ato ) 
A 


Rie ieee iain Li 
ola Isaksson ia30 30 * 


i : 1 i - 


r 
: Dat os) é i \ se ‘Sx 
Td MaAGe ) Ev tee 


r. 





“ey 


an account of the relationships and interactions 
among the various hierarchies, and to be completely 
satisfying, the tenets of such a theory should have 
some external justification, for example, in terms 


of phonetics. 


The present investigation seeks to demonstrate the operation of phonolo- 
gical strength hierarchies by comparing a selection of data with other 
known facts or hypotheses about the workings of language. Since it deals 
with a highly specific set of data, drawn from the Romance languages, 
the problem of universality is approached only incidentally. By re- 
maining within a closely related family of languages, any theoretical 
discussion which may result from empirical observations will not stray 
too far from concrete reality. 

Until recently, remarks concerning the presence and activity of 
strength hierarchies have been almost exclusively confined to general 
statements describing the overall structure of a particular language, 
and consequently have never been applied in any detail to specific phe- 
nomena. The feasibility of employing the notion of phonological strength 
hierarchies not only as a descriptive convenience, but also as a tool 
to be used in characterizing other diachronic developments is explored 
in this study. Vocalic strength hierarchies, based on word position 
and intrinsic phonological content are established for the early stages 
of a number of Romance languages. These hierarchies, common to all the 
languages under consideration, are isolated on the basis of empirical 
statistical observations. Once the presence of these hierarchies has 
been determined, and their interaction with other important diachronic 
factors assessed, the hierarchies are applied to the description of a 


particular morphological change in Italian. 
























Lo)  spheeiiqnao “edt od? bans! «ere 
wei biuode yroad? & dave 1 ates a8 
amiat ni ,elonisxe tot ,noigsotiigesg & 


& 


ae P 
Dee vi 


-oLoiotg Xo noitessqa eit s ertanareb ot exes noiapiteownd 


testo ditiw stab to soltzalea 5 pristagmoo yd es Lrintexete seems / 
pris “to apaitrow edd atuotis esestogyl 30 Srset ewok 





‘eissb Si sonie .spbe 
,espsupast sonanu ont moxt owerth ,stab to 392 oft rose “yiriatd ss ee ms) a 
.Viteonebiont yfrto bediosoxqys ak vii iseseviag 10 mofdong orth i. a 5: 


net 20 ylimst beielex yleaafo's iste pine 7 


~ox Va 


feoktovoett yrs 29 paLR 


ern 


— yertta $on Lliw srioktev1eedo fsvittens mort tivest Yam fio Eriw ooiaaoaib ¥ = 
.Wilssr stexaaes ‘nowt 18 od : 


a 
oi 
to ae Bis sonsesrq oS piimisonco extiemet \ylansoer Leo 7 
letenen ot benitneo ylevrautoxs jeomlé nmsed svar eo ienetoil maya 
apauprs! xsiuo Ltd 20 gutta Lferevo, ont pridtisesb ctemetese ” 


~srq oftivege ot Listab ys nk heliggs stood steved oved - Scteeqpeere® ban. : 
—A st 


dpaesta Lsotpofon aks to notton ahs ontyolams 20; yotiidtasst oat SOS 
fort s as cels sud ,sonsiaavnes arLiqtnesh 5 as ylno Jon ala 


soittesd tow mo hessd ‘Behe dipnost2 obese » vba eins a 
espate yiase af? yot berdeiideses sxe dneinos LsoipeLorada, tenant 6 
ods Lhe ot Romod: ahirorsxis asset ne Ser | 

~ ‘feobrigne to &LP66 ait no. Badeteak O15 eo 
ond seitoremin . sesdt 20 eumsesig sal ep Peeree 


n 20 ear ext? otf — 8 ail : 
ape A a . er pe 


= esi ai 
cae mh): z* 
; 9 al belo A 
nl 4 a 


1.2 Types of phonological hierarchies 


In her study of consonantal dissimilation among the Romance langua- 
ges, Rebecca Posner (1961: 48) noted that a segment may be ranked ina 


number of different ways, considered in terms of its overall behavior: 


.-- the strength of a consonant can be of three types: 
intrinsic, i.e. according to its intrinsic phonetic and 
phonemic features; 
positional, i.e. according to its position in the word 
or syllable; 
historical, i.e. according to its historical tendency 


to be retained or changed. 


A closer examination of these remarks reveals that Posner has not iso- 
lated three independent measures of phonological strength, but rather 
three interlocking aspects of the search for a scale of phonological 
resistance. 

The fundamental dichotomy synchronic-diachronic is applicable to 
the investigation and description of phonological hierarchies. Segments 
may be arranged along a scale based on their degree of participation in 
a synchronic process or alternation, or the scale may be formed on the 
basis of diachronic developments which selectively affected certain seg- 
ment types in a manner indicative of a hierarchical ordering. In most 
instances, diachronic hierarchies are much easier to establish, but 
there are also synchronic processes of sufficient complexity to permit 
the inference of a strength scale 2 In a few rare cases, it is possible 
to make an instant comparison between synchronic and diachronic hierar- 
chies within the same set of data. For instance, Foley (MSa, 1970a, 


1970b, 1972a), on the basis of his study of Romance and Germanic data, 








foi aes at on eae ‘ely niet 
riG@ivetiod Llexsvo ett to amc ca wi sar al ,ayew. Saini: 


-Oei ton ae tanRod 
erties dud .fitonerite fsatpolonorig to esieson srabrodsiai sount Bedek 
stpolancig Jo. eisce 5 tot doisee el) to ejosqes pritoolzedmt seadt 


; = aa 
ot aldeciiqgs et oinosrpsib-oinosdosys yeotoroib (stash oat (pt 
a 
agnempes .us isoipa Londres to cottgiceeb has notsspiteownt edt a 
mt appeepoionsy to senpeb ahedt no beena eisoe 5 pAols bepaerat od yea a 7 
“= 7. 
: 


ed mo bearrat od yen alsoa ort 10 woisetatis to eesderg cinta s ae 
-per nistuen hedseilts vy lov itosice asta atasarp Iaveh ° staoissib Io. tan 
geom nl «spnineixo fsoiiorsyeid & 30 eviges teak TSnEM a sone pe 
gud: ,tietfdetes of totese coem o25 cutie slemruinal 


| aBOveL ee) yator | ease ao us 
+ edit oieiganis) bins ee a0'% Gude ab 


“eh 


¢ ‘ 


4 : . 
ast esWiy 20 sd nso Aratoence 6 to Lreaoste re re ae ie 5d 
ts oitenddy cianiigat adi oF i hea Sek . TS ‘ete 


_ Ertow : 
olds five 10 
7 
. ai 7 
_ * ; i 7 Wl 
yowstucd Isoftotaid esi od pritbxoons .o-t Lspisegatat a 

























7 per. 


-dentitieet cement 
és 
t noktieog ati ot paibtooss .9-1 lenoitkeog 


5 


.hepasio to herisie: acd-et ; Aa 


seed 2ieavet etiemor Sead to aoktanines aaah 
ma 


re 


athe 


a) 
- ear 


is a, 


Yemarked that during the process of intervocalic lenition of stops, 

velars will be lenited first, followed by dentals, and only then by 
labials, thus defining an implicational scale. These claims have been 
supported by Zwicky (1972) and to a certain extent by Lass (1971). Izzo 
(1972: 82-108), in the course of an unrelated study, found Tuscan dia- 
lects in which only intervocalic /k/ had been lenited, as well as dia- 
lects leniting only intervocalic /k/ and /t/, in addition to Tuscan 
dialects leniting /k/, /t/ and /p/; in all dialects, spirantization of /k/ 
occurred more often than spirantization of /t/ and /p/. 

The intrinsic phonetic or phonemic strength of a segment, and its 
positional strength within a word are important factors in both syn- 
chronic and diachronic processes, and must therefore be considered con- 
currently with the time element in any discussion of hierarchies. This 
investigation is primarily concerned with the role of strength hierar- 
chies in shaping diachronic developments, but given the interrelation- 
ship of synchronic and diachronic data, it is impossible to avoid 
implications for synchronic aspects of the languages under considera- 
tion. In order to interpret the historical data, reference will be made 
to strength hierarchies based both on the intrinsic nature of segments 
and on position within the word. In the last analysis, a hierarchy is 
merely a scale formed on the basis of observations of the workings of 
an entire language, and does not presuppose the exclusive use of a single 
set of defining characteristics. If a group of segments behaves his- 
torically in a manner suggesting the existence of a hierarchy, it is 
because a Similar hierarchy existed in the synchronic dimension of the 
language throughout the time period in question. ‘The converse, obviously, 
is not necessarily true, for a synchronic hierarchy is not always the 


result of a single historical development. 


ise tas = i 
Ob hes eh as, bene rai a 
se , 0) V Va ~ ES | J 
’ - wie a 
























a 


* al “beds ad ya one pee s soeld Beton. tab : 


Pe > ren + a 
satiated r lexsvo ‘aot Fo 5 amos ok berebie: ON q asin : a 
: . 7 ‘ . 


2 t 7 { : aS 


19 SO Tee ey — ‘ 


& 7 jue Pa m ~ of ah - yr oo? att . re. 
=) a a ‘ ae eft oe ts 132 ie | t= = i cur 2 = eae 7 
taeqys seis 19 trsric : 


——. ¥ = : : - anita 
ft ; ,; = Tone cere fD an r : 
_ - - bw ou isos xy of: ~ aeioae a. og paid Pee | awed — —_ » : 7 
| i De 


i. ‘poNutsst oenemorig, os 
. | ‘ i 
; im 


‘ies ; . inane o — fenorerec 
brow xit at noddieoc ar oF peiroNOoos .9,1 , eis 
7 : = - 
7 7 


Ww 
_ 


:eidelive 20 
~*~ a ; 
” i 


i : 
s + . a : - a rm ¢. airs % 
wimed Inotreseisd ett of ortibrooos .9.f ispiross eis ; 


. 2 Ree - 
bepaedibd 20 DaniLesst od Bits > 


Cae 
‘ 
aa ar 


. . 7 A ae — wait ‘ Bed a " 
-Cei gon esc tenect 2665 algevez ex soft to TOLtSsniiicxts Tes 


wotkg Yo asxussean Jtosbasqebat souits & 


: ‘a ne 

+ ef t + rino lf tadre é Qery ae 
3 40% dovsoe oft To atnedes BihivoLltad eins 
“ feotpofanaig 3a elaoa 6 10% Ao et JO Bfoedes & t tae 
j 
a 
aldsoit sipsif-oinoiionye. yendoroth IntaenaGnst 
~a 5B = y) * J. ‘ - a —_ } da s 
ot eldenifqgs ef sinosdosth ine LGTY. 

‘ - - 


4 e a P 7 - its de ; rs 
a aa rrreyierr® rj faoJ a ats x 3 tol Gs ae one) ©) Leis nor SY ne & avn 7 
Pee - = ~ 


A e TLI3 7 LE Ass ail 
_ 
act- ro 


> ‘cor . ; vo oe hago atfaenm ££. * enci4 tare! c (Eat 
NOLIBOID t~isxy TO GSIpSep “NEOTIS (10 DeGEG S4.boe & wu a Le. pS 34) r a oe 
{ a 4 . A ; 


. oa 

* S 

r a of a \ on 9 ee P a -_ _ ve? oS 
efi CHO Bes rror oxi Vw LBoe ois a JOLIET TS ti 1 2ao00ry sas — bi 


en) 


by 


rae i be trex) > + 733 wv Len rt ae foe fy criw 27 Sper fou oe wi aot to al Le me 


om al’ .pniueio Ian Meseueii 6 Jo evictsociet NECINEM 5 ek 2 


if Process. 


sifdetes co retese them ote csiriote tected Sane ting. 


‘timeeg of yi ign et iektite to reaaesor sigoliaene ATS cei — mt 


7 Fie cated * . S Leo Wier 
aidieeoq ei 31 ,asaso sust wet Bent ‘alee i snail & no hiatal 


~ Stseidd ofr niniosdh | pris ol see ee sass > sae oe Ls 

‘i ~ : ef .3 

Jans ere ae 7 
x «sorver , 9am) yoic 7 ounsger cal anbaahan eee sti utcit iw 





us 





 ~pdeb oarentist) bes: on aa ta 
et , “ie 


Fat ie’ Tee 





Yemarked that during the process of intervocalic lenition of stops, 

velars will be lenited first, followed by dentals, and only then by 
labials, thus defining an implicational scale. These claims have been 
supported by Zwicky (1972) and to a certain extent by Lass (1971). Izzo 
(1972: 82-108), in the course of an unrelated study, found Tuscan dia- 
lects in which only intervocalic /k/ had been lenited, as well as dia- 
lects leniting only intervocalic /k/ and /t/, in addition to Tuscan 
dialects leniting /k/, /t/ and /p/; in all dialects, spirantization of /k/ 
occurred more often than spirantization of /t/ and /p/. 

The intrinsic phonetic or phonemic strength of a segment, and its 
positional strength within a word are important factors in both syn- 
chronic and diachronic processes, and must therefore be considered con- 
currently with the time element in any discussion of hierarchies. This 
investigation is primarily concerned with the role of strength hierar- 
chies in shaping diachronic developments, but given the interrelation- 
ship of synchronic and diachronic data, it is impossible to avoid 
implications for synchronic aspects of the languages under considera- 
tion. In order to interpret the historical data, reference will be made 
to strength hierarchies based both on the intrinsic nature of segments 
and on position within the word. In the last analysis, a hierarchy is 
merely a scale formed on the basis of observations of the workings of 
an entire language, and does not presuppose the exclusive use of a single 
set of defining characteristics. If a group of segments behaves his- 
torically in a manner suggesting the existence of a hierarchy, it is 
because a similar hierarchy existed in the synchronic dimension of the 
language throughout the time period in question. ‘The converse, obviously, 
is not necessarily true, for a synchronic hierarchy is not always the 


result of a single historical development. 


f t i : 
keeles 

























ased evad amisto eed? . .9f6oa Isqaiseol font eeLiettt hex ‘ 


KLE ee Ys 


wee we 
— 


essr .(1Vel} seni yd trade misdxao 6 so Tris! ae Siw ry 


Haley Reel, Sees , Ware Siaiole te Mee iat 


-6ib as Llow es (adbiaal nocd bat. Aa, ot issovaesiut paper 
nanes? od poitifhs rt .\ F\ bas ot oilscovisdat yino paksinel int = : 


7 
WA, 26 apissstinetige etoetaih Dis si \a\ brs \o\, a vecisines avon 
Bos \g\ Zo soitesitnssige ask nedto uot berms 


\ : ba - 7 of 
et? bas .Jnommee 6 to diprerte ofmanodg 10 nijenodg obpaiitres ot r™ care 
“tye xitod nt exodoe? Jasdxogmt sis brow’ s aisthitw ditpnsate isnostieog i i 


a 
: : ; 2 5 ve.’ : 7 : 
“1100 bexabiencs & ed exotevads teun bus jeseaodorq SineuriaecS bas outro 1 
\y - 
abi? aetdoxsroid to noleepoeth yas ai triansts omit ont rit iw ciara ; he 


y 
a : 


wii dtorsxte To afew ant rdiw boerrveonco yLinemiag af noiteetteoume : 


- rT 
} 


oe 


re, es : 
“noiisleriaini ot neyip gud .etnemqofevebh ciaouiberb pnigeda ik aoito i 
Biovse od eldieaeogit ei.tt \steb Sinope ES \Bfis, oincuioaye to gids | ki si : 
tin - 5 
y oe of 


“sypbiancs wseint asparentl oft YO etseqes obnondorye x0t erate amt is 
ahem sd [fiw sosrster ,eisb isoiiqaid eit, dorguetnt ad sabto oi — 7 
5: Sueno 

esoampes to emdsn otentyint sit no ated bevaed ai inated cheb eat . | 


_ 


| ) 





ai ydoxersid 5 .aieylans yo6i ony fl | btow sad snisiw noi sO, exe 
—- =. 


one 








to apniztrow att to ancitwipsec to atesd -sde 10. “beamed alsoa “t 
eigita 5 to van swiauloxs oo seoqquesiq jon Bagh bets repavpast are 
-2it zewatled adtriempee Io quotp s iI obtataesastert & sade 5 _ 







e 


‘et ti yisusteii 5 to sonatelxe edt paiveeppue senna 6 nt y 
: “eit 40 anlage oknouterye edt at. hodeixe cmon ie 
weenie ido SexgTIOD art nists Ie 





1.3 Hierarchies based on phonetic properties 


Virtually every historical grammar contains some implicit refer- 
ence to the differential and quasi-hierarchical behavior of sounds or 
word-posSitions across time. Remarks concerning vocalic hierarchies of 
positional and intrinsic strength, both synchronic and diachronic, 
appear in most treatises on Romance linguistics, and in numerous minor 
works as well. Very few scholars have made explicit reference to hier- 
archies; the most common practice has been to scatter, throughout the 
text, a number of comments concerning the relative strength or resistance 
of certain segment-types or positions. All such remarks are, by virtue 
of their ‘sporadic and incidental nature, purely descriptive in value, 
and make no attempt to transcend the description of individual pheno- 
mena. Nonetheless, a survey of major works of Romance linguistics reveals 


a high degree of awareness of the differential behavior of sound elements. 


1.3.1. Ranking of vowel timbre. From earliest times, poets, philoso- 
phers, and others with a keen ear for language have noted, in the syn- 
chronic dimension, differences in 'strength', 'sonority', ‘weight', and 
other impressionistic criteria, among the vowels of the Romance languages. 
Voltaire, in his Elémens de la Philosophie de Neuton (1738), compared 
the seven colors of the visible spectrum with the length of vocal cord 
vibrations, thence to the seven intervals of the (Western) major scale, 
starting with violet at the low end, and progressing to red at the high 
end; in this fashion, he established an inverse relation between the 
wavelength of light and the wavelength of sound. 2 

Feliciano de Castillo, in his Tratado de Metrificag&o Portuguesa 
(1850), offers a mumber of interesting impressions of the intrinsic 
characteristics of the Portuguese vowels. Speaking of [a], he notes: 


"Esta letra 6 de todas a mais franca'. The vowel [o] is described as 
































2 os 7 e 
‘to afence to soiveried ‘iste brs ile | 
to satdosaeisl nitsooy puirctsonos ashemat sents | eras & 


.sinorres.ib Sts, oinouinery2 ritod tprete nirertssent = 


ov. 


+ontm eucxemm ni bie  epidatiupalt sonst ao noe iteext jecm nb se ae a 
‘- 
anki ot soneietar tinkigxe sbsan over, eislodee wei isV . Liens eal AD 7 
ds 


‘ 
ext juroriguowtt .is3se52 ot msed esl eoisos7q norms teon a reoltoas a 





Suveteimex ro titpasite ovitsiox ant pn Lerxsonioo inceiee to xedaun 6 dint oe 


‘ 
; eusibe yd .ous eixeesa dove IIA sanort£E0g 1O zequs-trenpes cistaee 20° 

y 
9. elev ai subset yvlering . 2x13 61 Lsdnebtoai Bas. DL bRIOge tied 20_ - 
a? , .. nt 
Z onary taghiviiat To naar. eds basvensit at jomadts on ookesn es a 


eisaver enolic tempat anno io attow rofsn to yayviue Ss zeslodtondt svat 


waste fe anne to soivevisd laisasiett if elt Jo eesnousws To gems) fais a 


: —daclitkg .ateoq ,eenis gesilses moxd . Sues i+ Sowow 30 peblnee opi 
of ~tye ot ai Hadoa eved. sperms 10? 269 road 5 ritre eterko bees ese 
: nea .'drioiew' .'ydhronos’ ,'ttpnerte’ ct esonerett i sioieramib sinersrio 
-eepaupasl eons edd 20 alewov ois pcone Sivediro olteinotesssgnk wedIo 

hersgnas , (BET 1) nedusit ab abigoeol itt 5b eb ensmats aie eet 
bioo Lsoov xo dchoreal ows riche mrioeue eldieiv edt to =xoloo. aswes sett 2 
valsoe totem (irsaseeM), act to ateyrtsdn! newse edt at somackt — a 
dpid oft 26 Bex ot patezexporg bis ‘Eno wol edd de dato sti @ 
er somatod eniinies sevornt: ne tedanten st vaattee’. kia wy 

€ sewoe 30 dipaalevew 2d Bae Steh Ro xt ws 


- gemupeanon tproittats ob chagent ait nk voltae 96 cms 
. otankefins sett to eee tor eure 1073 = 
a 


gedon ach fs] to wnideraa ree 
" ; 4) vy: \ 
es as fo} few ae | ens ap? ‘ 
: | a _ ee *) f AY: oe 
; an ; ce “srg Tee 


ay 7 _ i> . , foun ar. 













follows: 'O o é na segunda escala das vogais o que o A 6 na primeira: 
som franco, rasgado, enérgico, € como uma explosi40 da alma'. For [el], 
Castillo notes: 'Com menor volume, com menor explosao e resonancia do 
que o A'. The vowel [u] is 'um som abafado, que se emite com a boca ja 
quase de todo cerrada'. He continues: 'Se a vogal A que nos abriu a 
primeira escala dos sons, expressa a grandeza e a alegria, 0 I,.em que a 
mesma escala termina, parece convird com as ideias de pequenez e tris- 
eza'. Finally, the entire vowel scale is summarized: 'O A é brilhante 
e arrojado; o E tenue e incerto; o I subtil e triste; o O animoso e forte; 
o U carrancudo e turvo.' 

More recently, Claudine Hunting (1973) has offered some impressions 


concerning the French vowels:4 


Visuellement la bouche prononcant le son A se présente a son 
point d'aperture le plus grand, n'opposant aucun obstacle a 
l'émission de la voix qui conserve ainsi toute sa puissance 
Originelle ... la voyelle o est la plus sonore de toutes ... 
les 1l€vres se prolonguent alors démesurément en un instrument 
de musique des plus sonores: le clairon. la bouche, ainsi 
gigantesquement agrandie, résonne bruyamment ... en oppo- 
sition a la voyelle é6éclatante qu'est le A, le son E, bouche 
plus fermée, lévres lég@rement avancées, est a4 peine per- 
ceptible. I1 est plus un souffle qu'un cri ... l'atonie 

du son E correspond bien a celle de la coleur blarche qui 
est, en fait, absence du toute couleur ... aprés 1l'accalmie 
passagére du E éclate le rire du I... la forme allongée 

des lévres trés rapprochées suscite tout d'abord 1'ouver- 


ture 6tirée d'une blessure ... 




















Lagth 


bait y cacd co occa wenn é 
“02 © sermupsg ot) esiabi as mo Bxtva03 soxreg sre 
Rasixewine ei olsoa fewoy axis era at g ie 





_ 


_ Séoedlizd > A 0 








asic = cecmins O o wetait se [hte Io ‘ce-mbak =) chad a0 im 


- 

i" ‘ ‘ay 

! ; ot sr | : i i 
Be B atncctay oe A noe of Yosgnory stood of 4RamLisuane | ea a: 
3 ; 4 
' i ) 

a 4 slostedo niuays triseoggo'n DAS Ip euiq ef stutssgée'b srtod eas a 
| aa 
i * aa sorsaBing se otunt tenis svisehde ‘bep xtov si oh notecind' I 7 - wv wi 
a : : 3 . A a reed 
_ --- aaduay sh sronce atti sl dee o siiseyev bE Ms al foniginie” ah ae 
- 7 = — sie o <* ba aa 
| Soevemseni ony ne Snes e1ofs Inspr roloxq 88 aonvel, 

i - 


jeris, ,etisuod al Louis ef teotoue eulg eb 


| z: “GO 1 «2. sneumeyrd annesdy ,siiderps! cumaresttig 
i ertoiad sak 108 of ol de8' Up sunstalos eLiayor ‘al st. 
=w8q anisy & é-des .eobonevs: somone £ oo i 
eines h 42. i vay epee ra wulg $e 

typ Srintia tc swloo Mv ‘ebcaSfs8 6 nikal 


simisson'{ eran ... swelucs ait? hs on ds 


spenols maui al .:. F a seit: 
~ernie! Bass" tua = | mi 
| mee ra 


aes 
cn 3 









Finally Hunting arrives at the vowel u [y]: 'C'est la voyelle dent 
la sonorité est la moins forte, mais la puissance de souffle la plus 
grande, car son énergie se concentre et vient buter contre 1l'obstacle 
des lévres presque fermées'. 

Such remarks are prevalent in works on the Romance languages, and 
form the basis for much of the literature on phonetic symbolism, a do- 
main which also represents a ranking of sounds, in the synchronic di- 
mension. It was not until the Romance languages were subjected to com- 
prehensive linguistic descriptions, however, that any attempt was made 
to characterize the diachronic behavior of the various vowels. 

Mendeloff (1969: 11) neatly summed up the evolution of the atonic 
vowels, from Vulgar Latin to the Romance languages, in the following 


statement: 


The evolution of the Vulgar Latin unstressed vowels was in- 
fluenced by one or more of the following factors: (1) the 
intensity of stress in the language concerned, (2) the persev- 
erance of the vowel, (3) the presence of contiguous conso- 
nants, principally nasals and palatals, (4) the fact that the 
vowel, through the loss of an intervocalic consonant, found 
itself in hiatus, (5) assimilation to the stressed vowel, 

(6) dissimilation from the stressed vowel, (7) position 
within the word, (8) confusion of initial vowels with vo- 

wels of articles and prefixes, (9) the fact that the syl- 


lable concerned was checked or unchecked. 


Menéndez Pidal (1966: 67) describes the high degree of diachronic 
Strength of the vowel a: 'La vocal a es tan resistente, que aun inacen- 


tuada, se conserva en todas las partes de la palabra en que se halla... 


ye coved vale a) hile a 7 
sb atlayor st ta! 2 Hh Be tao 3 4 do ak sibiakte rt 

a i 
ore Pe er ses itd arabe 


7! : oo) 
ra | ellsectado’ f euitios teiud tneiv ts exneonto: oa —_ 


7 
















, 
> 
fl 


{ n é 4" ; i - 
h, Bas seeaaedin’s sored <1} no exxow ni troleverg Sis ties 


oo 





. eminent en 
J : 
i} 


an ib oiacrdomye of at ,abrvoe 20 paisnsr 5 adisesitaar cates 
—Mbo' oF Hacipatdve stew saperpnsl sonenof orlt Ligou too B50 a 


be ee cheek ated ‘ee ee OT ae Ge ‘ oa : 
ebam ssw Jometts yis stadt  »weveworl >Etqin2esb rere Besar or edie 
: i 


i 7 * 4 
a Fetes. acihiteotes: AA a wen rere allele a on ee — _ 
-CiSwoV Sviorisy & | r JO x B50 DINOW sip Si ; OR bis Wash) SiN co 
77 t _—- ae eo 


Le 
: : ® 
vr att ie ma a . " aa » 7 f 2 
Saws . Sit Je. no w= St qy bomme yltson (if r0e@l) Bakes 
"re eo a ~ 
Y rwoiic A be; 8 ih {lz I S&S giz : + - 
7 Si 
~ 
“Li arw 2! Y DS ore? oie an aw} : Lov ve he 
$-) ory , 
art | » 6 “eben ~ re af fA > > — 
+. ak a 7 OVE s TOL J : JION. XO Sf 
' 
Pa iu 
op ra 4 ‘ft _ 
] . aT Oe he rags vO 7} le Ai dime re jer aan ed J a 4 . 
YSEI9q Of o> OST SoToS SOK azs 6 n 2egiste IO 
~O2rt A Se yy to owreoperer: ets ff) cz Pe 
Os Cp Scareie 2: : Svea <U = 9 it ,  AIOYV LI 
, 
wd 


teri3 Jos wit (6) \2istsleo Bos aleasr Yi legs Pea ig wid: 
Drmro} . . dren © oi fscovasini ns 2o.eaof sit fecomts Seales 


eeera? Re skavto. x pk. rey oh Slee , 
4 WOY Hoe sayte a oo motte Dbimeas el 2ha8i0 fi ° Bast. 


1 
; 


i ie . an a om, 
ov Siw efowoy [sitin® to nolavtno (8) , Bitow ott ork i 
“hye oct tadd doe? ot) (2) (eemtioyve ore 2s _— 


—_. a 


-Dexosdaan xo Beximars apw berrrearton 


ve 








a aes ee 
a ce — 
| re %o serpesb Seid oclt eediaBeb 0 ager 0) tebit se 


U 





>? 





2” ale sites 52 sup Fst) wads leg Bs sb 
Us 

js : : 

- 7 Yi ' o hp 

OF re i . Pw 7 7 . a 


1O 


la suerte de las otras vocales atonas esta determinada por la resultante 
de dos condiciones: primera: su colocacidon respecto del acento; segun- 
da, su colocacion en el comienzo, medio, o fin de la palabra’. He also 
notes (p. 44) that 'en sflaba dtona las siete vocales se redujeron a 
cinco cuando son iniciales de palabra y a tres cuando son finales', thus 
in effect classifying a, 0, and e as the three 'strongest', vowels. 
Bourciez (1967: 144) also makes implicit reference to the diachronic 


strength of a: 


L'opposition n'est pas d'ailleurs absolue, car, en Espagne 
notamment, il est resté un certain nombre de proparoxytons, 
ainsi ceux dont la voyelle penultieme était una... les 
voyelles qui, sans étre dans la syllabe initiale, se trouv- 


aient devant l'accent, ont eu aussi tendance a s'@ffacer, a 


moins qu'un groupe de consonnes ne les protegeat ... leur 

chute (exception faite pour a... ) a été normale au N. de 

la Gaule. 

Grandgent (1934: 98) remarks, of the intertonic vowels: 'In some 


regions, they began to fall regularly before the close of the Vulgar 
Latin period, but a was generally kept'. Speaking of final vowels, he 
notes (p. 102): 'The vowels regularly remained throughout the Vulgar 
Latin period. Later, about the eighth century, they generally fell, 
except 4 and i, in Celtic, Aquitanian, and Ligurian territory'. 


Guarnerio (1918: 301) describes word-final a as follows: 


E la pil resistente delle vocali finale e si conserva in 
tutti i domini romanzi, meno che nel francese, in qual- 


che zona ladina, e in qualche dialetto italiano. 


; ; A 



























- : sinikees eyF Aaplesbios $1\.sb nit o vi, 4 O38 s19.kmco abd 
| & nore tubes 22 eelpooy atoie- asl snioed sdeSte net Sada 
ors il « Saobepndd roe cbasun ast ¥ elas wb snetss totes aoe — ni 
” elssov Swieiatebin asris ody a6 5 Bis ,O \5 nom 


- obmosdbsst® aft ‘of. sorerotsx 3452 Cont aeaser Oo Ls: (pid Feel) madi: ig 


—, zz ‘4 everf nt . as ie or } 
snpsq2t ns 160 ,Sufoeds ewoellts'h 2sq tasle dors racks 
4 


wd puab re ~~ ~ = ta ~~ 4 7c - 
,anesyNoIsoc sh sidan. atesyoo mi Stas 4Je5 f 


o 
Jb 
. 
| 
t; 
‘ 
a 
as 
i 
t 
rt 
ip 
; 
f 
~~ 
om 
bang 
7 
ro 
| “a 
Pa 
§ 


UIT 3b eatettrar ~aelit¢o «< > ak Ato arr re rr — 
VIIO heed we ‘ oe © Pt5 z bs Lt Yo 24 < a Ww. = ad SS Bri a ia 
; 7 ; r 
£ epee soshast feeus ve tao «dr 

1 yseosiIS*a : aigois? eg 99 tno .3drSoos’ £ : tnaveb ‘jnois | : 


. , ; 7a 7 
‘ “rt rea | +c —wte bid { <a ONTO rt : 5 ~“"1T ae be 
SL wis WISI Bol SY aomiToentoo sp SOTp fits SLT. 


7) 
P 6) .W us olemitn $i6 5-( ... 5 wood sdist; notjigsoxs) saat 


” m prt? : Picea ¥ nel < . a 
amoe mi - ei Sworn Todysinit ofit to .2:remex (Be <be 
~s al * ~- 





aa 


of een nf ered a> fare Ff 
; aty to s20lo Sd e1oted ylislimey si ct raped ee 

; : a : = - : ia by 
on ,eiowoy Jsnit to priisisega. .'3qed a aie & std ois 

’ 
ven{rtl arix tresetrrteydr4 enw treme cy een free 7 
SPIOV SS JUCot Dex BTS yivs fore aioww aft r (SOL g oan aT 
X 


,fisit yilertsrep yor Yisdiso ddripis sri 12. SOR teted botseg nse 4 
/ _ m_ Z . ss, iS 
, Iodrixnes nshumid bas ii hg sien yn E Sew: < tcpsooc 


a 
S 


:2wolLo3 e6 6 Isnii-brow aadtincesd (sae, s8feL) oinsn rice 
7 a ra 


2 


ok svisenco ie 9 aienet £ Heoow oLlab ssasdni ail te 


ine it we vd 


: a ey 
ie orted Ls: dio obtelett oot as Pry ts 
-y ? : te - A hee 7 mn A q ; ; 


~Ssyp mi ,sesonsxt ieee at xt \ 


- 





For initial syllable a, Guarnerio notes (p. 336): ‘Anche nella sillaba 
iniziale atona, 1'A e la pit resistente in tutti i domini, e subisce in 
una misura relativamente debole 1'influsso delle consonanti antiche'. 
The author also remarks on the resistance of a in the posttonic syllable 
(pp. 319-20): 'L'A ... mostra maggior resistenza, cosi nello spagnuolo 
dove E cade davanti aN, 1'A rimane ...' 

Lausberg (1956: 96-7) begins one of the most comprehensive state- 


ments regarding diachronic ranking of segments by noting that: 


Spricht man alle Laute mit (ungefahr) dem gleichen Atemdruck, 
so besitzen doch nicht alle den gleichen Schallftllegrad. In 
diesem Fall ndmlich tritt der Eigen-Schallfullegrad eines 
jeden Lautes (d.h. der jedem Laut durch seine spezifischen 


Eigenschaften zukommende Schallftillegrad) in Erscheinung. 


Lausberg then goes on to classify all the vowels and consonants of 
the Romance languages along a numerical scale of 'perceptibility'; the 


vowels i and u are placed lowest among the vowels, e and o on the second 


highest step, and the vowel a has the highest degree of 'perceptibility'. 


A similar statement is offered for Spanish by Navarro Tomas (1971: 27), 


who ranks the Spanish vowels along a descending scale of 'perceptibility' 


a, O, @, i, u. Among the consonants, in Navarro Tomas' study, as in 
Lausberg's, vibrants are classified as the most perceptible, followed by 


laterais, nasals, fricatives, and occlusives, in that order. 


1.3.2. Hierarchies of word position. Generally speaking, among works 
of Romance linguistics, there has been a higher degree of awareness of 
hierarchies based on position within the word than of hierarchies of 
vowel timbre. Nearly all discussions of hierarchies start off by noting 


that, in Vulgar Latin, the penultimate and intertonic syllables were 


ae 

























sldeiive sinetteog add az o Yo tomate at 0 
olcunpade allen iaeo \sprode teox sotipan avjeom « tee ere 


by ‘oo. etemix BL) Hs Loeb ie me 
F ey 
-stsce dele iarsdtinrstetiis sjeom ait toe ane aénipsd (+e + s3e@L) pradawed 
: rdedt prison vd etnsmpse to paivnss oinotirios£5 slams ean 





veerbiesh nedtoiste mab (ad8topms) Jim eduel sLisyiee trotage 

Pe en ae ells sdoin rloob rasdiee@e 
_ abetks betes Litt Pedoasnapist ob tt coi Deter fish meat | 
sitet isons esiee ds rab suryl aatet web .d.b) estusd nebst 
oiunistoesd ot toartpeLtietherne shaamolus oettectoenspia 


10 wimanoanno Ins afowov ed¢ Lis yiteesis ot no ssop cad bredecmd er f 

~ orks | Pwd Lictitaentod’ to sisoe ([soiromin s pnois copscpns L Somnemanl Set < a 

brqoee ont no 6 ins 5 alowov srt paces seswol Beosig sus u Beas £ alee is 
. "yiliditgsotsg’ to seupeb teordpid_ ord asd s fawoy'sd3 Bas .qade Seerdoid i 
V@S 2LVeL) easier oxisyat veh sainede t0t bars to 2. dhenstst3 ree: 


:"wilidisgsoxsq' to olsoe oA ienneh 6 paols alswoy Heide ort stir one the 
ar I: a 


ee 


, a 


ai 2s . yee ‘atm omisveli nf ,Sinsnoaton sat pac ay a eae 
yd bewollot patentee jeom oft as bait teesio 15 aieicdiv 2a hig 
ae 
-usitio Sant ai .: agviauloos mis venison  GIRSBN y | 


4 
~e2eow paoms ,prciasqe ities oft eo aaa 
Se sential voi a eatin ea es 204 


i 


highly susceptible to loss. Some investigators attempted a classifica- 
tion of atonic positions in greater detail, also taking into account the 
early stages of the original Romance languages. 

Bourciez (1967: 36) notes that 'on peut admettre (avec certaines 
réserves, surtout pour 1'Orient et 1'Italie du Sud) qu'au cours de la 
période impériale le peuple, en parlant, ne faisait plus que rarement 


entendre la pénultiéme bréve'. He further notes (p. 143) that: 


La force d'expiration donnée 4 la syllable accentuée devait 
entrainer une autre conséquence que les fréquentes diph- 
tongaisons: elle ne pouvait en effet se faire sentir qu'au 
détriment des syllables voisines, dont les voyelles ris- 
quaient de disparaitre dans la pronunciation. Déjd, a 
1'époque latine, c'est pour une raison de ce genre (due, 
surtout, il est vrai, a la nettété particuliére de la syl- 
lable initiale), que certains proparoxytons étaient 


dévenues paroxytons. 


Grandgent (1934: 91) remarks that 'among stressed vowels, those 
of the first syllable had most resistance, possibly through a lingering 
influence of the Old Latin accent'. Of the intertonic, Grandgent notes 
(p. 98): 'Vowels so situated probably became more and more indistinct 
towards the end of the empire, and occasionally disappeared'. He also 
states (p. 99) that "the Vulgar Latin rhythmic principle tended to ob- 
literate one of the two posttonic syllables of proparoxytones. The 
penult, being next to the accent, was weaker and more exposed to syncope’. 
Grandgent (1927: 46) describes the fate of the intertonic syllable: 
'In the rhythm of the phrase [the intertonic syllable] has a position 


not unlike that of a final syllable, and is exposed to similar influences, 





W 54° . My 
Wa) a il Mt oer - 


fs . panting 

stele aA pring 
| hi pee 

aaniarico 24vs) eae yniers - -_ | 

Ran: yr ob Se ae eat e mere 

sasness otp wit Sdinwtet on ‘doakisg ne volquey sf s } 2 

tert (ERE .q) eacton sathien sydd omsiit for aN 


ee 


jieved sdutnsoos oldeiiye sl 6 ory wcseeipiias oor? at 
yi ib acdiisupont est Map sonseIpBEnoo oxaue ony soniaeine 
silica esas 3 gtint-ce dette n5 sidgnaky all REE ‘enceispamd 
tas eolisyov zat nob Tomas te aaldebiys cab name. 
/ £ <BR. fobtsionundsg sf enh adbbisqakb 8b Inseam ri = 
(Sub)  s1nep eoceh ndeiel sm Ie te3/9 vets t supedeht a 7 sed 


-fya 6L ob oxStivcitisq Sfatton sf & , isi 329 a ae 




















erage 
i i 
















nisi en us eas ge 
aeis oH .‘borsogapeih yLtsnoiassoo bas ovis ie 
-do 0% febast olgioaig odes gett. niset Lech ws 
ect paRaePEN oe eo [ 
creel Rane eens oe meee. a 
nets whe ou? 

| oti wi —- bath i» trie ett] 
a | 


see ed 
a 





but, having no inflectional value, its tendency to syncopation is 
stronger’. He also notes (1927: 56), with regard to the penultimate 
syllable, that 'here, as in the intertonic syllable, there was a differ- 
ence between popular and cultivated speech, although the incentive to 
omission was not so strong in the penult as in the syllable between 
accents'. Finally, turning to the matter of the secondary stress, Grand- 
gent (1927: 14) remarks that 'when the secondary stress remained, it was 
apparently stronger in words in which it preceded the main accent than 
in those in which it followed'. 

Elcock (1960: 41) remarks that 'with reference to the syllable 
structure of the words, it may be observed that nearly all these shifts 
of accent bear witness to a paroxytonizing tendency in Vulgar Latin'. 
Essentially the same view is offered by Lapesa (1968: 61): 'En los ro- 
mances occidentales el ritmo del lenguaje tiende a concentrar la fuerza 
expiratoria en la vocal acentuada, detras de la cual no suelen tolerar 
mAs de una silaba'. Speaking of the pretonic vowel, Lapesa remarks (p. 
56) that 'en menor grado se debilit6 también la vocal protOnica, que en 
algunas regiones, sobre todo en Galia, llegaba a elidirse'. 

Menéndez Pidal (1966: 35) offers a good characterization of the 
inherent weakness of the posttonic penultimate syllable, by noting the 


behavior of this position with respect to assonance: 


La escasa sonoridad de las vocales post6nicas relajadas 

se aprecia en los asonantes del verso, donde la vocal 
posté6énica no cuenta para nada, cualquiera que sea ... la 
perceptibilidad asondntica de la vocal final es algo mayor, 
pues s6lo son equivalentes las dos vocales palatales entre 
si, y las dos velares ... la a final no admite ninguna 


otra vocal equivalente. 


13 




























¥ 


ak cokdacggms- ct sondan at ne oir: Dia 


etanitionsg oft od Busper dtiw:. (22 2 eta — 
~aathib 8 asw exerts fcativ olaatstnk 0 he ae 
of avisnsont edd dquoitis dasa Bataleon Bas xed penile 


rk 


ssessid attaibes sus ni = He ee w 
sce mon yt i ee a * 
gew ti \bentemex seeute yuebrousa eft cedw' tert avtneinae GL +! 1s Fy 
neat trscos cian at aoa ithe eee ar 
. pewobiet 3¥ dodtiw sib sacra mi | 
edsiive a od sonersiet diiw" pst eAIBTOT ULB: :08BE) bdo . is 
ettirid ogedd {fs yLised dail! hevysadio af yen 3k ‘inte laa 
.'atted wepioV oi yornsbnsd PeEsinios yg eG £ of seamiv sad Sag008 20 
—ox wot a" : (1a :B80L) seeqed yd Beistie at weiv pe ais yiisiomedea - 
: esxcut ‘si *eadreonco 5 epaait Sstenpiel lah oni fs colainshicce agoaam 
ssysicd nefare of Isto sl sb ebtieb \shesgnecs fsvov sf nS stores 
vq) extvemex sesqet \lowov oinoiexg sit to peritsege dette ane ob aie 
m6 sup ,BOindio1g Isoov si aStdmet csilideb se cbexp xonem ne" Sant (ae 
| : ‘emtibile 5s edspell .siisd m9 chess S1doz. vesnoieet samuel 
i edt i0 noigssitatneisro boop s exsito (ce :33ef) sbi “sobniaett ; 
| 4 7 oft pation yd ,sidelive stems hat néncdaog oft siempre 


1 roonienoses OF Josqest Hate a ae 











witioios es.incdsog eolnoov asiosb beninonoe 8 


= ery 


fssov aL ebnab ,ORTSY. feb Cs US C 26 ropa 


af. . Bee sup sisinpieus abs xe Bt 


zoysn opis 25 Isat i mc | 


i i i Maan: 


a ols 
oe 





14 


Lausberg (1956: 107) offers the following general remarks: 


Die nichthaupttonigen Silben (und ihre Gipfel) werden mit 
geringerem Atemdruck ausgesprochen als die Haupttonsilbe. 
Es gibt dabei verschiedene Druckabstufungsmdglichkeiten 
---Je nach der Stellung zum Hauptton unterscheidet man 


zunachst Vorton- und Nachtonsilben. 


He notes (pp. 107-8) that 'die erste (worteinlietende) Vortonsilbe tragt 
den Nebenton, d.h. sie wird mit dem nachstgrossten Atemdruck gesprochen'. 
Lausberg adds (p. 108) that 'die Nachtonsilben haben geringen Atemdruck 
("unbetonte Silben"), jedoch ist---im Falle des Vorliegens zweier Nach- 
tonsilben---selbst hier noch eine gewisse Abstufung festzustellen'. 

In studies in which the differential behavior of atonic posi- 
tions was described in greater detail, diachronic strength came to be 
equated with acoustic intensity. Thus, for example, Pope (1934: 112) 


states: 


Atonic syllables all varied slightly in the amount of stress 
they received, the weakest being the unstressed penultimate: 


dor-mi-to-ri-um . The final syllables of proparoxytones, 
26S 3t 51 ess 


which were separated from the tonic syllables by the penul- 
timate, received more stress than the final syllables of 
paroxytones, which followed immediately on the tonic. The 
valiue of the intertonic syllables approximated so nearly to 
that of the final unstressed syllable of paroxytones that 


they followed, broadly speaking, the same development. 


Lausberg (1956: 96) begins his discussion of the fate of various 


atonic positions with the following statement: 






















a & 
tim nabrew (fetqio » mute a) nsdlte cma 


« 
: 





fed iedro | Socrepa mseents mrtg canta: exV' 1odiB ah i 
. : : At :? "e . ' 

2 berine x: fpt i + 7 St ol ‘ 7 
rasa sebsiaetoptionre nottquel ms rus. (iste xeb — Le - 
a 


> , |, fischl renedripsy iow ~nodr0V dartoe 


’ wel ers , j LT <a =e | * r ry ak at 
testi sadiianosrov (aieateriniedzow) ster sib’ tar (8-TOL Jaa) 
ra E — 

a ae To. rr Pe eee ee Se joat. iy pe al nit © 
- nadoorazep ou Bese TS neaGipsen tt ED Jin CIIW Sfe -f1.5 


Joirsoreté mepatico asdad meditanatrins 9i5’ tartt (801 .q) ee 


oe 2a : : p57 ents » 4 4 nt a. * i ae — 
“(et yorsys -enspoi lov, esb sills mi— tel reeber ,("Aedite eine 
seins: 
. asllstepstes?  oreluteds seen op Sais doom *6id sed£ 


~teca inate Yo xoiveried Isisners?tib sd¥ Aoiw pt salts At 7 


me. ee a ae < iy hizveeeiciaee PEER yy el t Red 
20 GO Sm6o Arpnsite oinomiosib , [tatab tedesrp at pedixseeb | as 


7 
7 
= 


cor ' =, ery? * + N+ ay PT! teh re ater = ‘ ~ 
if :O6¢1l) eqot ,sfamexs tot sun) .vitenstni sigsnoss ‘ah 


2asz3e tO thveme a at yitrpils bobtey Ils eeidsliye omen 


~Senitivonsq bepesytenry acts pried tesxsissw ari: + , Bevisosr: ok 


a Re ae a ee see: : pel = . na ae 
A Sh (S =240h..Ve SLO 27s Cit cect ABE. Sifw 

, : j ,. =. rs oe 
cx abe Ls ha 


r bee i: b) ¢ Ay Po en 7 , , 
S10ds6itve tsatl of nsdy 2asrcte evan Bovisoss 4 
| | | topes , cited 


< PF - i 


ed? .ofned adF nto y Spsiham Bewollot dotrion aadterhy eres 
ot ybusexr ce. batamix qa ay sidslive geese Sit 30 


ad 


esis senesyxotsg to sidsi tye boeasaterar seats oxi 3 


+ sone sietsntich emtse ‘ens ceetblagte, ybdoud Bonnet 
oe | 7 eS ~ p> fs He r bn i o bal 


) 


_eteiztey to osst axis 320 fc taeda! aid entapedt (ae ve) 


eit Lal 
Line See eo : ad 





Bf 

; 

7 
> 
> 
> 


i 


Das Schallfiillegrad eines Lautes hangt zunachst ab vom 
Atemdruck, der zu seiner Artikulation aufgeboten wird. Es 
ist klar, dass ein mit maximalen Atemdruck ausgesprochenes 
[a] einen grosseren Schallfullegrad hat als ein mit schwachem 
Atemdruck hervorgebrachtes oder gar ein geflustertes [a]. 

Die Verteilung grosseren oder geringeren Atemdrucks spielt 


die Hauptrolle in der Wortbetonung. 


He further notes (p. 97) that 'die Abfolge der einzelnen Schallfulle- 
grade im Wort gleicht nun einer Berg-und-Tal-Fahrt; es gibt Schall- 
fullegipfel und Schallfiiletaler'. Finally, Lausberg arrives at his 


most general statement regarding the fate of the atonic vowels (p. 108): 


Die Verteilung des Atemdrucks auf die einzelnen Silben des 
Wortes kann sehr extrem sein, d.h. der Hauptton kann dusserst 
stark hervorgehoben werden, die unbetonten Silben dagegen 
kOénnen dafiir mit ausserst schwachem Atemdruck gesprochen 
werden. Schwachung des Atemdrucks ist Schwachung der SciHall- 
fulle, die unbetonten Vokale werden nur noch gehaucht, ge- 
flistert oder angedeutet und schliesslich ganzlich vernach- 
lassigt. Diese Entwicklung ist besonders im Westromanischen 


verbreitet und besonders frih im Fr. eingetreten ... 


1.3.3. Summary. The remarks which have been quoted in this section 
are by no means exhaustive. Nearly every author of a work on Romance 
philology has had occasion to mention the topic. The statements chosen 
for inclusion in this discussion are, however, among the most exemplary, 
and state in the simplest possible terms the manner in which diachronic 
strength hierarchies have been viewed by Romance linguists. Within the 


traditional framework of Romance linguistics, such differential behavior 


4, 


oe fart : oF | J Sb 
ays mow le edoscess ae ald 


; t, a m4 ; if hah 
ig , -- eg Bret fe ose ls paeiniene sith > ore - 


i). oe +a 


7 - - 
a 5dr Sachi sou reianectA nie Leniism tim nite aeab 












































 fedoswiog tim nie els ted Bexpelin tere cate 
ike 5 | . at 
oa VD  « [8] Seteotenitep io xsp) 16hS Bod (oesiouior ae Hou riba: 
iF a + x se cP 
Pant | atilge etoorbista cosep parisp tho A etseadap ‘ased lisdys Y ia 
7 - boned sdtte Ww 39D me. ae 3 ib 
G 3 | . - 7 
. a 7 >t. Ae 
ste wel liiitsring contesite 19h spioidA sih' stadt (Te <q) — a 


~{fetin2 tdip 29 ine )-Let-bar-pied sears OSC tilolsip _ int 
eid ts Bevirre prexvtausd <vitenta f {frre at +Lfacse | 
eid ts asv! pisdavsd ,yitsatt .'yolete liter sate Bru .63 


120 a tareee «4 ’ i oe at ; 
(801 .q) efswor 2incts oft to sisi sit poibiecst Jeamede ie 


5 tus atouxbrectA esh pre [ied 34 


s 
Ses 





7 
SSIS8E0R Miss sodas 1b .dvb .atse meytee aise all —— J 
}- 


WORED NS0lla fesncssdmy sib .nebySs maxtor, provEsri sage. ae 
- - 

ak MSTIOTWIe9S) 20TomesA medoowibe pseyenauvs tim “usteE far at ee it 
= % ic O% ns ¥ 4 im \& é¢ r +» vJear WY} f, cal 

SLEIE IS) 2b 24 , J2f B@XnibnNeatéA |esb fe: inioswrios sie 


Sosp (food aun neiiew siskov netaaedae sth (9H of 


+o xy fon > ai 7 © ~ Hy , e+% ~¥ Scheie _ 
Y io ilessisioe bay ts US0SDTS TSano Systane 

: - — 
‘ > _— 


oi. owed i mi mae ~ yah TOe5G TEE estan’ | two i at Sac fa” tei 
Naat ternta (~6riet aes , 
Jsitspnis is mi Ate eyeb mowed Brus tes 26% 
> Sa 
} 


fh aloes = > As 


nikiner sidt mt Bbatauyp nes over tiotdw aan: od? oy 

_ ~SOMSIO mo stow 5 to xodsius reve yittsavt +Sviseusdxe eri mom yd o 
aseods atagnesste sit .oigad stit. aobinen on sien ra 
one $200t ent’ proms Sd sword, Sus —— + mt a 
a. a 


vi 
t 
(Pa 


athontiosib noteiw out orien bas te nies 
mA ody ori ie ade tn a Sonne 

aoe e ei liad 
“solvated, istinbtst ib) tou; sea 
7 ie 


- ts 7 _ Tron os 
7 vito De 2 ane a oF cos bh 


16 


has always been acknowledged, and some investigators have even attempted 
to establish a causal relationship between diachronic strength and 
acoustic intensity. All discussions of hierarchical behavior have stopped 
at this point, however, and no further investigation has been undertaken 
within the descriptive methodology of Romance linguistics. 

The authors cited above have all agreed, either implicitly or ex- 
plicitly, that certain patterns of diachronic strength may be observed 
within the Romance languages. Both individual vowels and atonic posi- 
tions have been tentatively ranked along approximate scales of diachronic 
resistance. Among the individual vowels, a has been observed to be the 
'strongest', followed by o, e, u, and i, respectively. Initial atonic 
syllables generally appear to have fared best, apparently followed by 
word-internal pretonic syllables, posttonic penultimate syllables, and 
the intertonic syllables. In the present study, these scales are ampli- 
fied and justified by means of comparative statistical data. Many of 


the conclusions substantiate impressions stated as long as a century ago. 


1.4 The position of anti-phonetism 


Fundamental: to the investigation of phonological hierarchies is the 
amount of phonetic data to be incorporated into the description. The 
investigator must determine whether to arrange the segments themselves 
along a scale defined not in terms of their phonetic properties but in 
terms of their behavior with respect to phonological processes, or whether 
to define a metatheoretical index which replaces the actual segments in 
question. This question may be reformulated as whether phonological 
hierarchies may be considered properly as distinctive features of phono- 
logical theory, or whether they are a system of interpretive devices 


which act upon, but do not replace, phonetically-defined phonological 


stags et sci a eB 





_ 
i 











sili wel ot Kenrssads cued Ga atone SerSevattt ane di peers 
einots Isitinl “dev Emeqag: it Bok au YS ov PLS hile | 


yd Bowol fot yitrieiagas teed Bers? avext ot 1ASdgs ‘vierrecee: penta: 7 20 


' Bes ,aaidelive at enistor: I SInodsts5q eeide hte” oiaarsa Keorsscnd Bacher 7 Ne 
c . : 4 


ies eas zolsze overt tate Gneessq sty al ealds live oitnodsreeet edt 7 
70 yneM = .sdéeb feoitsitsje svitexsamoo jo ensam yd beiiveut Bas Balt 
“@P8 urineo 6 as pol us hotste anolecsxgnt sts linstedve. enolate att f 


a grag elie 
e at —_ 
—- = 

« 














~~ 


ji SS, i 


_ 


a | 


= 


a aligs 


_— 


me Deaeriqy iiae’ ae bio} — dL 


any ei estrousteid {sotpolonotig to nolspidzeuni Poe op 


ey We 





“ae 


ed? .aottghtoesh act ont betsyoroont se oe stab 2 te 


— 
1 


Bevisemedt etnempee ort opnsyxs Go <oihtoriae cri 
ai tud seidragqosy 2itanoriq tistt 30 eared mi- sor, bertt: . 







vr 
“Te 





texiteriw xo segeescong Lsorpolonarg at t4sqaer Sioa . 
fue atrsmpee leat efit esosigan piste xebnt = oon 


_ 

Pes ne 
_ : : 

eae iva 7 






Ly 


units. As seen in the preceding section, 'classical' diachronic linguis- 
tics, particularly Romance linguistics, has always regarded the ranking 
of phonological elements in terms of a descriptive convenience, while 
the elements themselves were characterized by phonetically-motivated 
distinctive features. Recently, a different interpretation has been 
offered by Foley, who is of the opinion that strength scales should 
replace phonetically-based phonological features and hence dominate 


linguistic theory. He stated (1970a: 88): 


Phonology is not concemed with the physical structure of 
sounds, but with the relationships that exist between entities 
that manifest themselves as sounds. ‘To attempt the establish- 
ment of a system of phonology on acoustic data is a contra- 
vention of this principle ... The distinctive features of pho- 
nological systems must be based on phonological, not phonetic 
data, if that system is to be viable. Acoustics has only 
peripheral relevance to phonology ... It is possible to con- 
struct a system of distinctive features based not on phonetic 
data but rather om phonological relationships, establishable 


by observations of phonological change. 


Elsewhere (1970b: 11), Foley remarked that phonology conceived in this 
fashion has no 'physical reality' but only 'systematic reality'; l.e., 
reality relative to the phonological system which Foley advocates. 

An interpretation such as Foley's represents an extreme position, 
which must be supported by specific information before it can be accep- 
ted. In the various papers in which he has presented his theory, Foley 
has succeeded in demonstrating that segments can be represented by 


distinctive features based on abstract phonological hierarchies, but 






) d 
it 7 — , : ) Ve ry 
‘7 - on tt ene “fst BPALS' «ID, 


; 7 id a te 
- pttiainws ants babzse t syewls asri a Peper 















a nved a a ee L drents 
©. 
= 
7 - bf pode asisoe dbtos exte coach poinigqo aad 26 -er obi: vs oa 
7 - a , : 
¥ : - e a af, 
eteninoh soma bes aesiwtsesl feotpoloriodq Ssesd<v ne fs Ke 
id ‘ oF nat ne lget 
7 i * 
we, - : (BB. :K0TEL) palmde oft ayoent 2 aa 
7 7 “4 , i ; ) a) 


to suudmuiie fsoleytiq oro rhiw bormgsonmoo ton ei. ypoioaads 
asititne neowiad sakes tend agidenoisteiex sat dmbw dod ,ebar — 

3 ; : a. - a > A 
a 


“téiideses att toqnetis dt abnor as eovisemedt testi iielbvsice = 


"Sith 6 Ef, SIF ~iJeOOsS 10 vooLonarig TO eS ava 
> a ie r 2 
4 AO »* Pexane? ro vite ts ad alorarercy 
1 « ass J ms < c Se RA **-e mt! Neh it > 
. 
5 FAN - i > bossa ect tec ons 
£5SFIOMNS = i 4 ad rf OSabna Sc Seu Sik 
, ne . i a. 37 + ‘< a ; 
a rs 3G ad oF ef asdeye sed 2: 
nd : os) 


rr 
‘ tb of eldteecg ai JI .., ypolonmrig et sonsvelss Lsxorig tea a0 


Ma? 


















Sitenmrig 1o Jon bsabd eeuést evitonissetS zo meseye 6 < te 
ae 
ool mer i . ong sing Se glfion se rae Pet ra Zz rs - a Ce ee 
eidsrisifdssas ,aqgidanortsle: fsohpolonoig ao hima sud sash 
: ar ms b 7 7? = 
-spasri. bso inet Garg Io el ssyrseds. we 
f a aa. 
ok ee | 
iia 
od ew eat eile es xe -. ‘ a . _ 
Bie i. BDOoVLSOMm“s> WAAL. Sits. Daas ih? § yalot : Aer 20089) 8 weals 





i 


+L ( Yotiset so a Se Bk ye’ Yloo apd “ya fast sro) on as 


ves , 


'_» ytiekt2eog Siete ne ainseoige: e yal. oe | v2 dotiss qed 
5 Pere 


a 


Hy 


7 
tS _ ~ - J : 7 

—“ocs sf nso Si sxoted noitenmotat ofiiecce . x af t+ 

cits ut oe. vege yd bad aogqne, 3 
. = wha tee 2 a: ‘ aw | 5 7 ear "7 
Naka  toort afd bagnsesaq ssrh oe fotiw of aE Sod « 
<— a 6 fa : ; ee bg te - aa 
a ; vd Redaoesrgen 5 stent Tis) oy 


e 


igo & a S| pti pat 


tee j * 


i” i i 
wel 7 Jud \Boisdoasotd featgolororiq, 


18 


he has provided no arguments as to why they should be so represented, 
aside from pronouncements of the sort quoted above. 

In a critique of the anti-phonetic position represented by Foley's 
analyses, Victor Cohen (1971: 316) noted that both syntactic and phone- 
tic facts must be taken into consideration when constructing a level 
of phonology, which in effect constitutes an intermediate level of 
representation. By remaining within Foley's system, Cohen notes, 
"similarity of phonological elements' could only be determined via a 
circular procedure referring back to the needs of the phonological system. 

In a direct reply, Foley (1972b: 458) remarked that Cohen's 
criticisms are only applicable to generative phonology, but not to his 
proposed theory. He further comments that while circularity is a valid 
criterion for evaluating mathematical proofs, it may not be used to 
invalidate linguistic theories. 

It is obvious from observing developments in the Romance languages 
and elsewhere that Foley has succeeded in formalizing some actually 
existing strength hierarchies, but his own arguments designed to show 
the predictive power contained in such scales must be considered largely 
circular, since they refer back to the original processes which suggested 
the existence of a scale in the first place. Whether or not circularity 
in the system renders it invalid is largely a matter of personal 
taste, and depends upon the goals which the individual investigator 
has set for himself. 

Foley's main concern in the studies referred to above is theo- 
retical self consistency, together with a maximally general (and hence 
maximally abstract) formulation; Cohen, in contrast, wants to know how 
Foley's theory may be equated with specific language data. Cohen voices 


no objections to the internal consistency of Foley's theories; he does, 




























) ~ 


; «OH axis ails 02 whee RSMO —. — 

eysidt yi badrossrqet fois tedd obssnorg= ins “aa aa <3 
~snorky bas otsuateye cited $s badass (OLE =1¥ ef) ork POPOL Ye es ; i r 
fovel s ae nieces noketebianes cont satiae on cen 
te. fevel pre eee om aedytttanco tostis Ai doit epotonbig Ye ae : 
veson merkD ymateye 2'yslot nivgty Tei Lema ya nbitesneasgs: -- 
‘ eiv benbrtatah ad vin Dinco ‘adnamsle «Leo. alos lo wheelie’ wee 
wmeseye Isotpolonoity sit to ehear ent of cosd prizrstoy auubsoorg usivoxto 
a'nedcO tc baskismer (BEA :dSTOL) ‘yore (fees tooxis 5 at a 
Bid OF toa gud .ypolonndg sy ist sisp ot sidentiogs yiao sexs aminitio : 
bitev 5 ci Wisslistis aitriw Fert SSMS tort oH .yuosrt besogiag * 7 
od ‘beau ed ton yen 3} ,Aioog [sottanotitan ‘pistecisys 10% realebigiitety 


a5 proactd Diteinpnil sfsbiiswak — 


eepsupnel sonmnmesi o's ni etnengofsevab paiviseds meat ayobvdd ab 31 - 
ine irre 
Ylissdos ame prisifammot nt bebessous esd yolod gent ocertge he Beis me ' 


wos of banpiest Stn Sneipis nwo ait sud \PSiibisisin citpnsxe pebtetes - 
Yiswusl berebiencs od tum eslese foye nk bentetheo iswog svisorenq att 
bedeegoue mhirw essescerg [Anipise s+ ot ged tster vec? se vteluotis - 
Winshoxut> ton 10 rode -2o5lq tenit old af slepe's to sondage all ; 
fenomieq to vetsen 5 vise! 2i Silevat 3i sxeprsr mame at = 

wotspiteoviti Ledbtviin edt cobdw alsop srt Bi li “ vf 
-tfsentt rf aa 


~oset abi evods af Baretay’ sothutte oct cub, erisono in beanie a He 





sorte Bris) Lstonop, yf eeaseien cs abe ak sak WOnsd2iBA, tie 2% ae 
worl wort od aegis aesitooo mi. rteric ri. ae ( jon | 
ese oii vookroeds Waren. 108 Pie ~ 













A) 


however, demonstrate some cases which serve as apparent counterexamples 
to Foley's predictions. Foley (1972b: 460), while recognizing the 
existence of counterexamples, states that correspondence with data, 
together with empirical testability and circularity, are not the 'most 
important' criteria for evaluating a linguistic theory, and asserts 
that 'a theory cannot be disproved by data, but only by another theory, 
pr attall™ 

A word should be added at this point concerning the 'philosophy of 
science’ implicit in Foley's work om phonological hierarchies, which, 
despite the author's claims to the contrary (1972b: 460), bears little 
resemblance to other statements with the same title. Foley has claimed 
that a theory may not be invalidated by a counterexample. This is the 
converse of the generally accepted scientific methodology, which assumes 
that no theory can ever be proved (merely successively confirmed) but 
that a theory must be modified or rejected when faced with a counter- 


6 If, as Foley claims (1972b: 460), a counterexample cannot 


example. 
invalidate a theory, then presumably nothing can; certainly not another 
theory, Since this theory is also uwnfalsifiable by counterexample, but 
only by another theory, and so on, ad infinitum. As to the possibility, 
mentioned by Foley (1972b: 460), that the data containing a putative 
counterexample may be incorrect, if these data meet the same standards 
which the investigator has imposed on the data which confirm his theory, 
they must be accepted; otherwise no theory will be falsifiable. 

Equally incongruous is the notion that circularity does not invali- 
date a scientific theory. While it is true that the charge of circularity 
does not disprove the theory, it is also true that a circular theory 
is empirically useless. The circularity inherent in any particular 


theoretical proposal may be overcome only if the theory is extended 






















colemen cru STREET A vase Hk 
itt prisiopads: alidw . Comb: ste) ote 
weseb tiw. eonabioqeenr109 16H “eerste a nol taken nde = 29 
- toon erie garners vehinslvopio bis erticaseey “eee 
2tuoees bre a Sale yisatupiel s 4 pabisutaes 0F abredine nie” a 
yyaoeret teBons yd yin tod -stsbh vd bevorqeib ei sonmso roads aaah 4 
te et ie 


ea 
ne | 


- 20 wigneoliriq' aris en irrtSonieg Sniog att te betbs ext se tuorda iow Av 7 





,Holtiw (eeitrowsrsin ee ee oo dxow a wolot nk tiob ict ‘eorsioe 7 
a eiaght exesd (0a) idiVel) quetine vsrit ios : emginto e'saiec, oft ‘oni Le 

a) 
ore bemislo esd yaicot .oalsit ome oh rtiw anes eta settta Pa) ooveteaaet " 


a Sip-ai eiiff .olgisxerattice s yd.bojebiiavil at Joo, yen \noart & test 7 
i Bar 


ie semgeek tibtriw . ypofoboritem oft bineioe Hatageos yilsrenep ont Te orem 
x : s ‘ Ton | 
fs Sud (bawritnco viaviessooue vieran) bevoxqg sd cove neo yrosild Gia 7 a5 


7 “Sednnte & ddiw beost roswbatnofer uo beitiban sd yea ytosnkiie Bae 


|) m Fe 
a Jonmo signsxerstouoo 6 , (09h sdSlel) antefo yolod gs. ,21 
“ | oo y 
| ted4oms ton yintetves ten pniirtion yicemresiy mont , possi eb hhevat wa 
: , 
de? .elqnexstedmien yo oldeiiieisiq cals at yroadt 209 sone yaa > 


Waaitdiiedog edt od eA’ .mitinitot 56 Jno 08 bos ,yrosds secon yd ea f 
Witaing 6 priaisinoo acto. ott $57 , (a8 rfTOL} yelot yd bartodtrem co 
eiraineste ems eis joan stsb searS tf .joswiconl sa yan sSosmacea8r008 
Voit eid mito rbidw bi8b ods no baeognil ebri ae | 
oldettiete’ ed [liw yxoerkf-on Seiwsetto, shadgecos ee 
-Eisvai tort seo Votarnisrorn io gant aoiton a eb pucumpneonL yg 
‘et Duakotio to aaa ert Jett art ef te olin ttt ‘i 14. mes 
moa selessis 6 dant sexy’ dale ab i oe 
Jblooiireg Oe ek Snertonté pease ea a | oa fe 


_ Rehsense ‘ek oat 24 yfoo emeanayy, sf on 


oh neu 


ps inal i 











ST 


xs , : 4 ; 7 i 
y - ' Soe 7 ‘. 


“¥ “a, - } Pa! 





ne - 


eee a wider domain, so that predictions may be hazarded, and the results 
of these predictions compared with the claims of the theory. 

In order to justify an abstract phonological system such as the 
one proposed by Foley, one would have to demonstrate a set of empirical 
correlates for the abstract features which could be given some sort of 
psychological or perceptual reality, and which would hold up to testing 
better than the presently employed physically determined distinctive 
features. It is impossible to accept merely on faith claims that features 
ferived from phonological hierarchies should replace ordinary phonetically- 
based distinctive features. Indeed, some current work in psycholinguis- 
tics, for example the experiment reported in Derwing (1973: 318-20), 
indicates that, while some of the presently used distinctive features 
may not in fact be perceived as such by the untrained native speaker, 
the perception of segments is accomplished by superficial processing, 
and does not necessarily take into account any more abstract phonological 


patterning than is necessary to effect identification. / 


During a dis- 
cussion of the feasibility of representing one of Foley's specific 
examples, Spanish intervocalic lenition, solely by means of a numerical 
scale, Anttila (1972: 112) remarked that 'even though it is quite possible 
to use such tricks for presenting shifts as simultaneous jumps, their 
value as historical indices is highly questionable'. Similarly, the 
remarks of Zwicky cited earlier show a desire to bolster the theoretical 
notion of phonological hierarchies with empirical data. Foley, on the 
other hand, has constructed abstract schemata which may be mapped onto 
the present theory of phonological structure only by his own interpretive 
conventions. 

Another alternative to Foley's conception of abstract phonological 


features is to represent the observable strength hierarchies not in the 
distinctive feature system itself, but in the phonological metatheory. 


20 


- i 



























Nos edt. 30 enisis ih tin Bese : 
ott 2s stwe mateye {sotpoforerg sopiteds| cued od xs i Pe 
fenisigns Yo.dtee 2 atert crema ot ever Blsow spo velo we 





to Seer amoa asvip - bios Ao it eutite3t sop teds oct tot 


(ye 


pnitast of qu Blo bidcw do irk Bre yt ifaex Lendqnousy 30 tosigatotneg: | 
efitoniszelb benimisdsh yi Lsoteyrg “Beyolois vignisesrig ofit nipirth, wetted 
@eutest gads anisip tts? ono beeen - igssos of sidipeoomi ak a2 ea 


” 





: - a 

“Yinoiionnds| yuanibro sosiaqs: aos ss triotersit Taohpolocota mat berkpab _ 

-atrroni lc winyeg il 22OW Jeo embe .psehnl .eieunteed a Son Ete kB | P c 7 
ont 


- 
,(OS-BLE 2éTCf) priwred pi Dediogst tuenitrsyes) sit Signs coe Rane 





gemrctset evitornitats boay Visiieesica ar Jo anna afiriw tems ashen Lenk . ; 
) sEevieoge ovicad hetissio: acteyd dove as bevisquey od toet unk jon yen 
voruteecsona cioljuaqe yd Setetionnos ab 2taaipes io noi qsoten, at "y 
fsolpoloumig -scetteds sion vos taticods Got Stes vitiseesoon ton, ean Ems 
bi on! bioete iy 


Bib im ‘socio itnsbr Jostie of yaseasost af - cart ee 
siiioasge s'yslot to sm palidsessasx Fo. yWilidtesar ei 26; sostemro 


i. 


fsoizomin s to. arieem vd yletos Sotdinal silasovrenit Helis? ealqmene 

el{diser: sdiisp ae $k devors neve”. gens ixenst- (SED sSTO@t) SLissAR ,@igoe — “4 
PO - 

Sree, ,aqwt avoenstinnie 26 eri mb aoesne Sct drloiis cue see ot u 

et, ,vinekinig ."aldednitecup ylantd ak asoibak Jromosetth ae stew - 

fsaisexoent sit rs2eld6d mi grtseb « wots xoiftss Batin woiws: Leenwzall 


@ 
as ‘ es - 
ois a .yalod, ..sdeb fan ixigns iE cinbzexatd, IsD.Tpe i om 0 


nse 4 oo 


q Taeaed 
i rt 
svisesquasni mw eid yd yl corto! (ss inal onedi 30. p act Secale adie 


: aan Ys sa. 


r 
a~ 





, Bns begqgsh' sa vam Hodetw Ec ke joe dace Paden extort 





an 






el 


Viewed in this fashion, such hierarchies would serve as additional in- 
terpretations of certain configurations of features, rather than actually 
replacing them. This is in effect the position adopted by Zwicky (1972) 
in his discussion of a particular phonological hierarchy in English, and 
by Lass (1971), who deals with Old English data. For example, Lass (1971: 


22) notes: 


It looks as if the feature [+ coronal] is crucial to the re- 
sistance of dentals to intervocalic lenition; but it is 
easier to say this than to explain it. At the moment, I am 
at a loss for a really motivated explanation, but it may be 
at-least possible that coronal articulations have a greater 
amount of articulatory energy associated with them, due to 
the raising of the tongue blade above neutral position, and 
this may make them more resistant to certain kinds of leni- 
tion than non-coronals of the same positional class. But 


this is still an open question. 


Lass is clearly searching for an explanation of the correlation between 
a given hierarchy and the current usage of distinctive. features; he is 
not attempting to substitute the points of the hierarchy for the dis- 


tinctive features themselves .8 


1.5 The problem of cause and *Srrect 


For the most part, strength hierarchies are discovered by observa- 
tion of historical events, or by synchronic alternations which either 
result from, or may lead to, historical events. If the events in question 
indicate the existence of a hierarchy, one may ask what effect the hier- 


archy itself may have had on the ensuing events. If, for example, the 























WE iia: oe 
Yilsotos ned waite ;eourisot. to snares 


(STOEL) wlodwe yt Basqons nals teog ert. oo a 
bas \dekipal ni videusteicd inoipe tect sean re 8 30 a 
sIf@L) seed olgnexs 1a ated lai lpo® BFO <jtw'atseby ort 


ony 

o 
me i 
ae 
eaten 48s - f : 
y " 





ne 
wey edd od Iptouto at Usaoieo 2 *) sutset orks ti 26 tool JE 





pi +i qd ynottine£ olfsoovaaint of a tase to coca ae 
ms I denon oft: +A di mis ine at cat eae NBS ot we wees 
’ a * 7 ee 4 
. 7 jee | 
) ad yen Ji judd’ ,cottsneligxe hetsvilon yitser | 8 103 eeot s sal : rae 
| isisore Ss aved noits iors ye Leaceco Jsirt ofditesoy ee ee wT; 
oh: . ae oe ae 
i ‘od sof \mevit do iw hotaipoess VeisiS ynosalps Ge to Suen was 
a ‘ves 14 ae 
7 fies jmots¢téoq Teasusn evods sbeld sipict ont to. piteioy ont 
ps | : oa a : 
— --nol to ehyic nissi1s< of drial bers stom merit solsiti aa eee : 
ny] 7 
fi $id .easio lsiottieog aise si? to 2fsnoteo-rion reacts wee . 
ee 
ce. ; woLrasup eepi ns tiija ef eiag we = 
i>, == FE - 


neowied nottsieriod edt Yo noltsnelqes mB. 362 pridertece eo wh ean . : 
ef of saoutsaot ovitormieib to Ssoseu TASTE e ‘bas yeoustedt aap 
~ath ott sor veovsaein ont Bo etntog ait gus itedue oF pulisquedta ton a 

* coviaemel comida eral a 

-_ a 


7 a. 





Mei” < 






__—— 


ae eee a 





acideaupint Sonar a 5X 
i ib Je as 


ce Re ata 


'weakest' member of a posited strength scale is observed to undergo 
lenition or loss, can one attribute the cause of the lenition to the 
low position on the hierarchy? Such hierarchies are established by ob- 
serving historical processes, and hence an attribution of causal powers 
to hierarchies is in this sense circular. On the other hand, there has 
been a great deal of recent investigation, particularly within the domain 
of generative phonology, which hints at the possible causal action of 
formal structures; for example, ‘rule simplifia@tion' by means of alpha 
variables, feature suppression, and so forth.” Such studies often give 
the impression of having put the cart before the horse, since the formal 
conventions upon which they are predicated have generally been derived 
from observations of similar historical developments. In the particular 
case of strength hierarchies, knowledge of their existence presupposes 
previous knowledge of a great number of synchronic and diachronic pro- 
cesses within the same language. It is therefore impossible to set up 


sucha hierarchy as an explanation of a historical process, unless perhaps 


"explanation" is defined as ‘accounting for particular events by reference 


to general laws ... of accounting for laws by reference to principles 
still more general' (Caws 1965: 91). Popper (1959: 59) offers a similar 
definition: 'To give a causal explanation of an event means to deduce a 
statement which describes it, using as premises of the deduction one or 
more universal laws, together with certain singular statements, the 
initial conditions'. Within the domain of linguistic and literary theory, 


the same interpretation has been proposed by Tanaka (1972: 41-2): 


Traditionally, the two most important functions of a theory are 
to provide explanations and make predictions covering particular 


phenomena ... let us say that when we explain something we 


2a 


























ait of noktine! oy 20 etre | 





aes aia maid hod gy 
wt yarns os, cine Hoa Bt igelesaits aie 





'? ‘ra 
- prowad Dseuss Jo nolediss36 116 | sonet Bae pees a ies 


a & 


Gel cult aed serio Sad 00 eats ono i sae ey 


Gye i iL 


fitmmed sts aiddiw yirsivotrisa sotispicteoval $053 to tes seme apes a 
» ae 


Jormpiias isauso sicdteazog ett de again ciation vqpeTonodiy, ovrksenene 20 
aries Yo amsen yd “noktmriienie slvr’ .olegnexa sort cecutowsse Sema 





—— avip mavio eatiie forse, * ptirnct 2 eet woleesngee. sate. validate 

a 
f - Iemct ois conte .caxcd aig sidied dibo sm a svat SG ieenaialll i 
on Beviee® riesd yilsisrsp, sven Selon aT yor tioLtiw: age sooksnemco 


by aes 
teiemiseeg odd nl satin stveb Lechiotats wale te eons at 
. soacugesiy sonaivixs sist To epbelword. adiriouetete ‘hiocaater 20 gas0 
“tq oinoriosib bas olnomianye 20 seam Isoip 5s ki. sphelworal HOR, 
e qy tee of gidieaegnt sictersdt af 51. opsupnsl suse oe ris ow ‘wammaD 


- Z 


t ii . i p ; ee ) > 
¥ adertseq eeolti: .erqocrg [nottorerd & to notjenn igxe me 25 tous =e 
a 7 d ; ? 


gongictes yd ehrare WwlLoitisq 107- pines” 26 bert 358 ai ‘ okeneigee' - 
@elgioning ad sonsisitsy yd Swell 168 peitevodos 95, ton sunk Semanep ot 

‘ talimie 5 exc¥ie (02 2020s) ssqgod .{J@ s20GL —o \ exons sion Aheae ae 
: & sqybeb of anaan tHsvo ns to. noitsnsions thane —— 5 sii ott smokin 
: 16 snc saitoubed ort to sseimeig &s pate +E eadixpest rote semeaete 
a ord Sarianed site xs iartre aisds105 dtiw xofiepos, as 


A “preearis usta it fers ottetrpeatl Io oR atts neisaW - "ga 








: (S-2R :S090L) salons? yd bepoqorg seed eu . 


ne on 52 sr on sn 





wr . - = “~~ y Be 
ie + uD) hme ht be 


23 


show how the statement describing the particular phenomenon 
follows from a more general law-like statement. For ex- 
ample, we would explain why a particular sentence can 
passivize by showing how it meets the input conditions for 


the passive-transformation ... 


The term 'explanation'’ as used above does not entail the determination 
of the underlying causes of a given phenomenon, which lie outside the 

theory, but only a determination of general statements or laws. Witt- 
genstein (The Brown Book, p. 88) offers a distinction between 'cause' 


and 'reason' in seeking theoretical explanations: 


If you ask 'why', do you ask for the cause or for the reason? 
If for the cause, it is easy enough to think up a physiolo- 
gical or psychological hypothesis which explains this choice 
under the given conditions. It is the task of the experi- 
mental sciences to test such hypotheses. If on the other 
hand you ask for a reason the answer is 'There need not have 
been a reason for the choice. A reason is a step preceding 
the step of the choice. But why should every step be 


preceded by another one?’ 


As a further remark along these lines, we may consider the statements of 


Rapoport (1972) concerning theoretical explanations. Rapoport defines 


two terms, theoretical explanatory power and explanatory appeal: 


The greater the increase in the a posteriori probability of 
an event (given the assumptions embodied in the theory) rela- 
tive to the a priori probability, the greater the explanatory 


power of the theory. (p. 322) 








fi enctstinee + ct wi 















bed siden ene (ratas ton neck svode bea as "nohisnsiges® 
eit sbieio sif doiiiw .nonbaaeestig. feevip 6 0 apadEo oni soba a 
—#iW .awel 1O ieee ‘Sette zo noltsHineedob 6 vine Sanh 4 , 


7 
1 - 


‘geuso' moowiad noiipaiteib s,eretio (88 J AOE swore eth ott) 


ere sae sioue fon ioapert paiviose ni ‘sone o2 
ies 
x i) em 


Snoesor ot sol 16 Sepss exit 10% 286 voy athe *yriw' alee or aed ae C7 
Ve 


at ; : a : = ' i 

_sofoteysc 6 au And of miperoris EEO, Bl +8 savas odd wok BE ba, aoe 
—a q ts we 
. 2: 

esfodo aicd anisioxe coiciw sagixiat fsorpolortoyag x0 facies rey 7 


“% 


-hisaxe aft to asd si ef 31 “enotsibacs neyrp add ebay. SH, _ a 


yorito att mo 11 .segaerbtocyd fous Jess oF 290n2 boi een bial? ae 
; ; ve au 










' 7 
; ; d : 4 7 
oad gon bead sxeft' 2i xswerns or nogbex 8 Tot tas oey ewe. gall ae 
7 


a aT > 





pribassiq gate 6) 21 netsex Ah .ooldris rit <o8 nOgRES & ali = a 
efi desde yisve bfvede yw jut ~eaten> edd to ate ay aii 7 


‘con0 oe ae sn = ae 


.< 
rs 


to. egasmststa oct ushienco vant Sw 4é gs angry tls ae 


testis woxsatase tne +s 


2h 


It is not necessarily the predictive power of a theory in 
a Single instance which makes the theory acceptable, but 
rather its ‘integrative potential', the extent to which 

Iany apparently unrelated events are seen in the light of 


the theory to be related. We shall call this aspect of 


a theory its explanatory appeal. (p. 324) 


As defined by Rapoport, explanatory power is an objective criterion, 
arrived at by the ability to offer empirical predictions, while expla- 
natory appeal is a subjective criterion, relating a number of phenomena 
according to the investigator's personal point of view. Both criteria 
have a place within linguistic investigation. 

It is useful to consider the following lament of Bernard Cohen 


(1972: 401): 


It is, perhaps, paradoxical that despite the abundance of ex- 
planatory efforts in sociology, and despite the similarity 
in the goals of explanation between sociology and other 
sciences, there are very few satisfactory sociological ex- 
planations. Our explanations are unsatisfactory because 
they do not lead anywhere. They are not useful in the 
pursuit of further understanding of social phenomena. Most 
often sociological explanations are ad hoc ideas applied 

to one particular situation and rarely, if ever, used 
again. Typically, these interpretations are plausible, 
certainly possible, and often interesting, but the striking 
thing is that the ideas which one author uses to explain 
his observations are rarely encountered again, either in 


the same author's later work or in the works of other 









































aud bite — ‘st nn 

Hoidw od tnetxe oth Habeas 5 
1 anne onic “TTC 

to toeqes ets tio Ciecte ott Baislor ed meme 

YY" (BSE sq) issgeis casein. ast oa 


smotred fr sy ictnotct $0) 15: 2i 3swoo yaotsas lox Frodoaes wid 
~siqxe siirw . eet He q Tedataite xatio a, vittids ort yd 36 bowie 
sesmone cic to sedan 5 pn tjslox sfolresix evidoageue 5 at ce eetias | ee 
siyetit> fom .veiv to dmiog [sncereq e*sedspetsevat eit a ‘en tbrxo008 a , 
| -noigsoiseaynt oitetupnil nidtiw sre. s:e0nd 


gexicO Bxanyos to tcomsl ptwolfot ort rbienso. ot [vise ef #1 ) - 


+ (08 2k @L) 


~xa to -Sonchmids sad+ atiqeab tett Lsotxohsisg tiga cae By a 7 ; 
winslinte sit stiqesb fas ,ypolotoos nt. airoite’ rode ao a 
ractto bas yoo! toi (332 coaveted noitens.igxs to aisop ai) nt oe te 

~-y¥s laotpolotoce yrodosteltée wet yiev S15 axedt estat a © 4 
squened yiciosiatiseny oye enolvtensiges 100 -enobtanslg MG 

ent ni Lise tor S26 alt ees ae. ob: york 

tao .anemonong {slooz to priibris azebas roigut To seg 
be Figgas ena cod hs ‘ois enoitsnelaxs tsoipaioigos. ae 

hens adds ti ,yferex fas dotdeosta wwii 0 


eidtevsty ens anoijstoxqodni a2eilt 


a. 


a ee 


ney! oo . 
Bae: 
Awe 


eo 


sociologists. Furthermore, these explanations seldom 
generate new research, for they are often untestable, or if 
they are testable, they are presented without any guide- 
lines which would allow other investigators to apply them 


to some other situation, study, or phenomenon. 


This statement provides a convenient summary of the problems involved in 
considering theories with only explanatory appeal. While not directed 
specifically at linguistics, Cohen's remarks find wide applicability in 
various areas of linguistic research, particularly diachronic problems. 


As a partial solution to this problem, Cohen suggests (p. 409): 


To make our explanations more useful does not demand that 
we immediately achieve strictly deductive explanations; 
it does demand that we strive for them. Even the gross- 
est approximations to strictly deductive explanation 
which result from striving would immensely enhance the 


utility of sociological explanations. 


These remarks also find a ready place in the methodology of linguistic 
investigation, and especially as regards the problem of cause and effect 
in historical linguistics. In the field of diachronic linguistics, it 
is necessary to consider both formal deductive theories and more tradi- 


tional approaches, along the lines suggested by Rapoport (1972: 337): 


Ideally, speculative concept-generating theories and rigorous 
technically competent hypothesis-testing theories should 
complement each other in the development of social science. 

I do not join the positivists who demand that every theory 


must be immediately translatable into testable hypotheses, 


ies 


a © 


a we 


alts 








Sa 





ee 
















, 





a aon ee. 


"pic ise- anois sacit \< 
ee ae ey i] ease e 
af to tam ot Hoxsea : 
cae ; pare Mae a tie ; % oe 
pea ish oer ; i 
; nctienotiodg 10 \vause oli ae xt ‘ nae Bs 


CA “s a 
if Bevidurit emeidexq or) 10 Viamma Instteyacs & gene aes PRoncrere= ise, y 
‘es 


i 
beuerth ton oliriW ante a qasaaslges vine ddiw ‘sotto? pabBieme —- a : 

$5 a are ray: 
fi peaiatine sfitw Bika eXamen a' ceil spobfabspait: Je yitacithonge ‘ 


EG 


Apbidery oinoxiosib vlgols 369 iosse2ex diteiypatl to ones ania i oa 


#(@08 .0) edesppue mended \moidore ail of noitnios bie sak — oa 4 
> | 
at. ue ands 


; id on 
- ; 


eed Dremsb ton goed’ Ilsa. Siem ciate Pactcla FER hi 7 


= ey 
is ft aes 


ne iy , ed Pr ‘oe nt 
a? e 


~~ 















a 





7 















-anottsasiqxe svijouBab yitoiide evsiios Viessibamt Sw sa - 
nd ‘ : FUE O 
“220xp ons asva .mereh) tor av itta ow teett (Bramad 2sob ai 
gottsnslaxs-svittoubsb yitoiste ot aan termbxougas fee 
eit sonefins yloenanmi Silvow privinte mort divest dokelw. 


.ennitsnsigxs Inolpofotooe to ELE 


Bee to. ypoLoborism sks te spat ybsex Ps bait ovis | 
tgostis bus sauso to a ee at abispes 2s: yListosaes fas. r : | 
$i ,a0tteivgonit ebaorie ts to bier2 eds al .colsetnoaht a 
oo 


~thexs.otom bas 2otxoartt ay Lrwwbeb ae fied Sora ah 





and that every hypothesis is formulated only to be tested 
and disregarded if found wanting ... concept-generating 
theories are essential in social science, because social 
science simply does not have a catalogue of ready-made 
concepts with which the physical scientist operates in 
full confidence that his concepts reflect the essential 


features of the world with which he is concerned. 


To a field as full of variables and indeterminacies as diachronic 
linguistics, these remarks are particularly appropriate. More specifi- 
cally, when attempting to discuss a notion such as phonological hier- 
archies, it is important to accept the relevance of both the deductive 
aspect of the theory, established by rigorous observation, and what 
Rapoport calls the ‘integrative potential' or apparent scope of the 
theory. While a series of historical events may be shown to be consis- 
tent with a proposed hierarchy, this hierarchy may not be said to 
explain the events in question in the sense of accounting for their pri- 
mary motivation. Instead, the hierarchy must be regarded as an interpre- 
tation of the process in a fashion which might be related to other 
developments or phenomena in the same language, or in related languages. 
Speaking of the problem of cause and effect in formalized theo- 


retical linguistics, Anttila (1972: 127-8) noted: 


The point of view that changes would first occur in the 
program, in competence, and only later would change the 
utterance without having anything to do with performance 
is clearly inadequate, although popular. This notion is 
a result of the scoring mechanism for changes. Language 


is learned and hearers interpret utterances largely from 


26 



























stan a no Sati ala 
. PASEO IqEOMOS re shsenijs 
feipoe seusoed .sariog L6iooe at tnhecen ieathe 
barca 20 sigoteino 6 sre ton een Vigne 


ri BatGASqo teisisine Lscidyda farts Ho ietwr nioiw 
a 


laitneeas Sas stoeltsx etdsouco ¢ — teat caret tn Ph 


-harysonco afi Sd fpidw rit ws Bice ott to nonueey id ; 


al aa. 
be nrg 


Ofnmotiribsih 2s. esicaninnatshri bas sefde tev 20 Sait 2s pists 
-fiose s1oM .aisingqowes yiusiinmitisy ois ettemer seadd ,200el 
“tsi Isofoolonoriq: as dove sotton 6 aeimelS of prittorstds mare ee 


i a 
witcher sit cod to Ssornsvelisx ait iqsoos o¢ Jnsttogni ef-ti ,eetd 
$6 bos ,noijsvisedo eyoropit yd badeifdetes ,yroadd edt Yo 
aft 2o sqgoo2 tnexsaqgs to ‘Isitnetog svitsxpetni' srt eiié saoeee 
~elenoo sd at owore sd yem etneys Inofuotett to asinee 6 sftaw wer 
Brow 
Dbise od ton yam Yrorsxeid efd+ \yrisisisic beacqorgq Ss Asiw 4 $n 


: ; 
ne 


cr + _ 
| J a 
~isg Tisds tol pristmuecos to sanse Sit mi nokasip ni etasve arid 5 LELqKS 


- - 4 > r P iy } 
“Sigessat ne cs babispe: sd tec eeots1e ii ait »beetenl  .notsyidon T, 
cat > ‘ 


1030. oF Bassler od anos foidw neidest s at e2500%g only 20 sokted 


ae 
ad a2-g 
& - o 


~eepsupnst bedslox ni 10 ,9pstronel amee sot ni patie cet 10 Hnengofeves 


: 


“cart besilemro? at jostts bas sauso 20 maicoig orkt to ena - 


hetom (B-FEL5;S0el) stigina! eolselimnit 4 ite1 
€ | 7 
SU Mr wW990 t=1i2 bivow eopnens tats wety io —— all ri 


edd! Spretiy b lick sel tie brs 2 IED oh i. : 
SENSMETOT tog ddiw ob at pautdtyas paver’ tts ; ste 
et “Rotem sicT beer 3 ay hie re 


performance, that is, real concrete speech situations. 
Causes have to be sought in the totality of language and 
the relation of its use with the total culture, and indi- 


vidual speech acts. 


On the other hand, Kiparsky (1972: 191-2) has characterized the current 


State of affairs as follows: 


The bases of explanation in current generative phonology 
are formal constraints given jointly by the notational 
conventions and the evaluation measure. These formal con- 
straints may be absolute or relative. Absolute constraints 
are given by the interplay of the notational devices and 


the evaluation measure. 


Given such a procedure, it is virtually impossible to formulate the cause 


of a historical process or event in terms of formal notational devices, 
since these devices are themselves outside the empirical limits of pho- 
nology, and belong instead to the domain of mathematical metatheory. 

As visualized in the present investigation, the study of hierar- 
chies is not the study of causes, but of relations. It is impossible to 
determine with any degree of certainty the cause of any sound change. 
One can hope to accomplish an enumeration and systematization of fac- 
tors which aided, or equivalently, which impeded, a given sound change. 
While it seems unfeasible to seek causes for sound changes, the inves- 
tigator may at least hope to gain a further insight into the structure 
of the factors surrounding such changes, and ° to ultimately arrive 
at a theory of the conduction of sound changes. In the long run, this 


latter line of inquiry may turn out to be the more productive, and of 


el 


mee Seok 
if ara 


-anoitveciie dosage stesonco Iss \ai gadd sometime «= 
brs. sosuprsl to wiilsto? ert nk Udpuee, of of ved some 
“that frig voutivo intod ott riviw ear att 20 noitslst ert 
.etos risseqe tenbiv 


Inerivo aid bes riesidexedo esd (S-[el :STel) \wlexsait .bned ‘tarizo ens nO 
awollot es eiistie to atede 


yoolonaiq avitsiator Fnexuo mi noltsesiqxs to asead Sar 
fenotiston at vd yitnior asvip ainisitenco Leme? eas 
“100 Lamro} 32047 mmzesemn noftsulsve sit Bas ermotinevaae 
Sirisywienso odtoeds -evissisor 10 ovuloeds od yam ainisize 


Bas esp itvab fanoitsjon sdt to yslouetht any vd neve ome 


.giveseam foitsisve. att 


omtIS Sid edeluaxcl at oldizecqmt ylisurxiv et si .embsooxg s faba wee 


a 
a 


yaemiveh [enortsta: Lamrrot to 2m + at treve 10 aescoo1g issizotdesd @ 36 
“tig to ejimil fecixtam sli sbietiuo eaylsamsdt ors BSoftvsb Saane Samke 
yacerbeten iscitsnsisient to nism edd od beatent paofed Bas , Yeofon 
“asieii Zo yirte sit. .moitepitesynt gnseen ty] a nt besiiauaty BA 
om Sititeaaoat ee awmissist to tud ,eseueo Jo yvbitte sri con ap eee 
cere bruos ye to savso srt yiaistres To seypsb vas Titiw Saemeee 
“O62 70 noises ctemeteye fins noOwternysie. is datiqnooss oft sqect reo SAO. 
-Semist> Suse movin s .bebsont toidw .vitnslseviens xo babies finiciw exod 
“awn of ,asposrio Srvoe xot espn dose ot oldissoton apsen ak Shed 
cuusowise of) atai triptani redtwi 6 aisp at sood geesl se yam tedepit 


9 


wirie Yladamitin od Bris ,2epasity finwe. poibavorsue anos 38 ond to 


and ,qux pmol et al .aspnst boos to ee ony 40 oat 8 a8 a) 
: 4 r, 


to bes \svitouboxrg .sxum eft od at Jo nxt yam. rdopat 20s 3 ia ee 





28 


greater value to the study of linguistic evolution, since while the 
actual inception of any sound change is probably sporadic and random, 
the course eventually taken by the change may be accounted for by a 

theoretical model. By looking to the past, in formulating a theory of 
| historical phonology, one may also hope to look to the future; for, 
given sufficient ses upon which to base a theory, it should be possible 
to hazard predictions as to the direction of future developments in a 
given language, with better than random probability. 

From a more epistemological point of view, the introduction of 

the notion of phonological hierarchies may be regarded as a form of 
reduction of part of the theory of sound change. The following char- 


acterization of reduction is offered by Causey (1972b: 176) ae 


Occasionally, one scientific theory T, is explained by 
another theory T,. When such an explanation satisfies 
certain conditions, it is usually said to provide a re- 
duction of TS to T, ... I use the tem "theory' here in 

a broad sense to denote a set of laws or putative laws 
which may be either well-confirmed or merely hypothesized. 
In addition, I will assume that T) and T, are concerned 
with objects in the domains Dom, and Dom, respectively 
...- I will say that a microreduction of T, to T, isa 
reduction of T, to T, in which the elements of Dom, are 


2 1 
identified with certain elements of Dom,. It is 
further assumed that Dom, always contains a set of basic 
elements, which from the point of view of T) are not 
structured wholes. In addition, Dom, may contain com- 
pound elements which are structured wholes whose parts 


are basic elements. Thus, depending on the particular 




























ext alinw eonia nots Love, cibinliriet a 
“aban bas vibsioge vidsdoxrg 2k. opasite sot bit 

B Ya 102 Raxmucoas at yam epaiario otf ye codec eee 
oath ea s paiteiemrot ni ,jasq aid of pattool ya febon teoenoeds . 

tot (outtiod ait of Acol of ogo dete yet 5A0 yeolonadg feotuateid 1m i 





aldizaog edi binorle 1 doo 5 saad oF satan noay steb Jasiok toe nove 
@ nt 2inemoolisveh swiut to noisoench oft oes eno Holiness bussed od 
Viilbdsdoxg mobast mscd3 sedge dtiw .spespost ssvip 
Yo noistauhowtei ati .waty to dntogq Iso tpolomstebae sian 6 mod e 
to amot s as babusper sd-yen esinoigsetd Lsoipofanodg 20. noigen ot a 
~xatts’ piivoliot aif .spnsdo Savoa Zo rrcert soit to. t15q te 2c sobigubes | ; D 


rani 


¢ (ONL <dSTCL), gets yd borvstio at noiogbet to aokteeisedos ~~ 


j : : : 
=, 
ya bertisloxe ei .T yroodt oftcinioe eno ,yiisnoiaasgd - 
antietiise nocisibigxs 15 soue aedw. ct ios serio va 


-1 6 sbivorg od bise yilsuey el ti ,enostsbnoo mfeiaee : 


ni sisd ‘yroors' mrad edt ce I~... .T OF EP ao on ca 
awel sviijstuq 10 ewel to tse 5 sdonsb od sense beaut = a 
e P S . ‘ ‘ ‘ 
-besiasrtoqyni -yforen 10 bam iines-Lisw xentte ed ‘yen dotaiw- 
- a 
herxsonco sis .T hos 7? tedt aweas [ftw “nossa me 54 ae 
f - s c * 5 . : 7 
(lev Hoses . mod bis jmod antsmob edj nt etosido Haw. 
set (T a  T to nobophercm ins slat yee Lliw Eva . 
_ e 4 7 
is .mxt to ainenefe erit riokriw mt re Dot - to ell al aT 


ai 31. med to airamis fLiadtrss: pi nie Detate 


“igo enischnice yen imo colds ar - 


- 


 ehaed s2adw eeloriw paar ci ik: 
Salut Fieger 18D, | 


2g 


microreduction and the particular elements, an element in 
Dom, may be identified with either a basic or with a com 
pound element of Dom, . 


In order to apply these remarks to linguistic theories, it is necessary 
to construe the term ‘explanation’ in the light of the preceding obser- 
vations. While it is beyond the scope of this study to determine the 
full epistemological significance of Causey's remarks to linguistic 
theories, it is easy to see that, at least from a heuristic point of 
view, a theory of phonological hierarchies may conceivably serve as a 
reduction (and even a microreduction) of a more general theory of phono- 
logical change. The theory of hierarchies deals with a domain consisting 
of metatheoretical statements about phonological structure. The elements 
of the domain of a theory of phonological change may consequently be ul- 
timately reduced to a metatheoretical level stated in terms of hierarchies. 
Use of the term 'reduction' in this sense is not meant to imply that the 
phonological strength scales are to replace the actual phonological 
values; in fact, the opposite conclusion is implicit in the preceding pa- 
ragraphs. It is possible, however, to place the sort of methodology 
involved in the postulation of phonological hierarchies in a somewhat 
broader perspective, by viewing the entire process as a theoretical reduc- 
tion. Overriding any theory is a level of metatheory, and the extent to 
which the metatheory intervenes in the application of the theory varies 
considerably. The precise relationship between theory and metatheory 
also varies, and it is the failure to clarify this relationship which 
has led to a great deal of criticism of certain linguistic theories .1! 

It is the intent of this investigation to maintain a clear boundary 


between the metatheoretical level, exemplified by the hierarchical indices, 
























7 eon ‘a ah Pee | 
ni-taemale as cesemee | 
Pe Se pete Pd i 

“OD 6. fot wo vid site ib 


" 
Aq A, - Loita 7 
: a a - . 

Te is ; « St 


J 
* 


~iswedo paliaosag arth 30 aeiebt tlt fie bend ae 
eft ahimmatsh G vbude abt to aqooe ant paced ae Sad oli 
' Site tupriil of extent 2 y9e8o 
%o aniog Sitefwend 6 “most eesl 36 saat 992 ot yess. sae pa 
& 26 ovise Y idgy.Loonicn « ean asidorsroiri Inotpotonnda to ros - voi 


-onory te yroals [srensp szan 5 to Gant eine & nove Bris) roieabes 





paitaienoo itiemob s titiw elses Boirbrsisid to ye ocr -opnsio. . 
Lan re yr Ce 
etnemelo.eca? .sxujourte feo tgatonedt duets aynemosate Leo isoxosdsate 30 
~[9 sd-yitnewpesanco yam sonsdo Lsoipofonadg ie reed B to skeet exit 20 





 asidousssid to -emtes ni betste lovel isoiteroeritsien 6 oF booubss yhechemit 





ants Sect yout ot tnsem Jon 2k sense aicdt ni ‘no fsoubss" mist ort al aac < 


wai 
ae 


i. fsoipofonoda Lejos oft sosiqa: at Sis eolsse ditpasite teoigatenoea 
i 
’, —aq piiheosiq sit of tiotigni at cs ad Feoqqo ont ost Ae weouiey 


ypoiobordtan to jxce sc sosiq a visvewod \aldiaveg ek FT. a. AL 


i 
ue rv 


Jadweme s nt eviriousyoin fsotpolororig<2o noite leteoq ods mk. sean on 





tortsey ynosdt odt to nortsoilags ody i sonovaoani 4 . 
yioettstan-bas yroddt neswad digembisiex id 7 . ve 
ridtdw ditfenstsdlies atts Whisks tis Seger as 


ates <a 
Lt corres oitaisoni ail > aed Se feeb | 
iebmnod sala 6 mtisnisan a ne epidaevar: 
| ivi i My es Re ce nN 
; at ae od. gaa 















‘ee a 
- an _ ie 
- , i} 7 vn aa’ 2 i 
ma ; es ee 5 slo Nake 
. ‘ -- - fe ae y* 
1 a = Pa a. *,) - 


30 


and the level of linguistic structures. As mentioned previously, 
hierarchies must be regarded as performing only an interpretive function; 
in this sense, then, one may speak of a theoretical reduction. 

In conclusion, this study is not directed toward the search for 
causal explanations in diachronic phonology; what is sought after instead 
is an explicit portrayal of the structure inherent in certain types of 
change. An accumulation of such data may then be utilized to further 


extend the search for explanation in linguistics. 


1.6 Data for the theory 


Phonological hierarchies, by definition, may be either language- 
specific or universal in nature. Often, a hierarchy will emerge upon 
consideration of a single historical event, and is not applicable beyond 
the study of this or similar events; thus, for example, the caution of 


Lass (1971: 26), following his proposed strength hierarchy: 


... the schema I am proposing here is intended to be a 
possible universal characterization only for intervocalic 
lenition, not other kinds. Each basic assimilatory type 


will presumably have a specific schema associated with it. 


On the other hand, observations have been made which suggest that 
certain phonological hierarchies are universal or quasi-universal in 
nature; that is, they are a function of the human species, and do not 
result from the peculiarities of any particular language. The search for 
universal hierarchies must encompass a vast amount of data of every sort, 
and to this end, the problem may be approached from a variety of angles. 
First of all, it is possible to undertake a comparative examination of 


the phonological systems of all known languages, and search for any 


= 


5 ie; 





( OT 2 















wh Siva i hihi tt Sea ‘i 
Heetant xetis scspciell ~~ saci : ye fonorig oiacuioe 2b nt Bh " 
fo equ nisdieo nk ‘Grenoria sutuite adi jo HE ek | 

soctHint of besilisy ec. ribs yen. ate5 dose to colt a Lumsct FA spre 

p  eoitetopntt ni nottansiqns xo rictsse srt Bred 


Lite 
al ee \ aad 
yeens ari a o "2 
~epeuppnust tertia ed vem obtinttes yd. eotdoeteid Issigolonodtt * a 
fogs eprsms [fiw yrioxsisid 6 edo omen fe isetovios lia 
broyed sfdegifags tom si bos. neve Tsolicsetd ofpate 596 cotts2ebtenco "s 
to noitues oft \signsxe x07 ,aurlt ‘adnevs xslimie xo eit to ane oct ae 


syroiese trl ritpneste pascaend air wiwoliee {8S 2fPCR) east 


oe sae cre) 
6 od ot Sobootni ai srad pateogorg ms I nneiie ot . 


ron Wee : 
oilscovysini 16% vino ‘notes ixatosrado. Ssetey icy stom 7 ; 
_ (we _) 
sy wots limtess ofraed ees + ectnat sito) tog ft ad _ 
Ae 


ti ctw bedstsoozes smerise olinsees 5 ovad visio LE. “ng 


tadd tespoua foidw abem cased aan ‘anoitevasado en ae aa 


ft iseisvinu~tasup x0 Inexoview S15 ssioaiioht sp ipolon 


q "il 
at iz Te 
: 


ton Ob bie ,esicaqe nanud sit to nolan} 5 StS 
we we Am 


pans ue 4 * 


S13 


Significant recurring tendencies. The nearest this goal has ever come 

to realization is the study of Trubetzkoy (1939), where it may been seen 
that, while certain phonological system-types are more common than others, 
it is nearly impossible to offer any non-trivial generalization which 
will hold for all, or even the majority, of the world's languages. Among 
the trivial observations is the fact that all languages exhibit the vowel- 
consonant dichotomy, thus suggesting this as the fundamental phonological 
axis 12 

A study of putative universal change-types is also fraught with 
many of the same difficulties encountered in the search for universal 
phonological systems, and no truly comprehensive attempt has ever been 
undertaken. Referring back to synchronic observations of phonological 
systems, the theory of 'markedness' has been proposed to characterize 
phonological change. Utilizing these theoretical proposals, a number of 
attempts have been undertaken to sort out the most 'typical' or 'uni- 
versal' types of sound change. Some investigators have even sought after 
physiological correlates of certain change-types; for example, Chin-Wu 
Kim (1971: 26-38) presents some interesting neurological data which sug- 
gest that intervocalic lenition and terminal devoicing may result from 
optimal neuronal firing sequences. 

Observing phonological hierarchies via historical processes is 
generally a straightforward enough task, but searching for universal 
hierarchies on the same basis has so far been undertaken only on a very 
restricted scale. Lausberg (1956: 96-7) and Navarro Tomas (1971: 27) 
have offered a scale of 'perceptibility' in which all the vowels and 
consonants of the Romance languages were ranked, with results similar 
to those to be reported in the following chapters. A similar attempt 


was offered by Chin-Wu Kim (1966, 1968, 1970), who attempts to define a 


4 


ee 






ems iad wed Ssop eidt Jeouson sd? eaimiabriad . 
7 
ss ses-tan ui us totes) sedi Aen i: 


exerito aed Ademco s1on sxs asr-iedeye- Isotpofonnda riistieo obbriw \dedkt 


doitiw nolistilexarsp fsivint-nonr ys Istie ot eldieeognt yiesan af 2 
Baan .sepetpes! ='Bixow oi lo ,ywirofam eft aeve 10: qils tem ibed ainw 
| : . 


( Pe ol 


~fowen odd Jide eeverpnei [ls tart dost of? 2: anobteviesdo Isiebehed? « 


feoipofenedy lainemsbnyt orf as ait pabesppue eure victors ih tnaneenco 
| SL. ins 

fitiw srpusit ceils ai yys-sonstio Iserevint evissing to youde & 
feetovion x01 ioe oft at betetnudons esitivoitiia smpe sit 20 yaa 
need atave asd tumatis grlenaSagneo spheres on brs ,emedeye iso tpatonardg 
isvtpolonctg Io enoitevieede/>tnoitbays at sbsd pa Peien _ egesiebrm 
esizedosiaiio of hedogoxy ased Bsr. ‘eeanbexisn’ to oad ort vansdaye 
to sede 5 2fseogorq Iso itoseers “seorit pans Lat» sepsis Isodpotomody 
-irm* yo.‘ Isotqyt' Jteon, add. jue 310s OF nedieisbar msed eved etqrasts 
Yedts jdedos cave svar TIQT SPL AWNL SMO Sones Bree TO esq ' fpersy 
oW-nirl) \afonexe 162 teacys-apnero ‘nisties to eexelerieo inoipoloteyrigq 
gue mbirw etseb Isorpofomesn ottissusoni amoe aineeoiq (St-d8. 21088) mot 
moxt tives: vem prioroved Leninxss bos noitinel ohisoovisdet tan 329P 
Leapeeepen prick Lsrarcren Lamt3go 

ei aeeeecoig [ssinoteld siv esiroxsyeid Ispipofonaig patvrsedO— 

Jeatsvins ii poiriovsee stud (aaa gsions biswiotddpiacta 5 Esnensp 
yiev 5 15 yino solsiishm nos ash Oo edt elesd emse eit nO neisiodeepabet 
: | yt 
(VS :LV@Ll) a8mol ormevel bas (f-a® 7acel) gradaus ans nin 
Bes elowov ont fis ri tet ot. ‘wii Lidctqsored! 20 ane & 







telimie es Lasers Hiw bolts: oraw eapeupis! santana at 20 8 
36 xefimte A exe 
sqres sari priwetioa onl os 


ae aya $ 
& ania ot somedss ov (OTEL Bart 982h) i hea 


Bag 
eae. Py 7 


& yi : Ges 


Scalar feature of aperture, with results essentially the same as Lausberg 


=e attempted to define 


and Navarro Tomas. Foley has, in several places, 
a scale of resonance, with vowels being the most resonant, followed by 
glides, liquids, nasals, fricatives, affricates, and stops, in that 
order. 

Some of the more salient difficulties with Kim's proposals may be 
seen in the paper of Lass (1971: 17), who has noted some discrepancies 
with his Old English data. Lass defines a scale of 'strength' (i.e. 
resistance to lenition) which is identical to those proposed by Lausberg, 
Kim, Navarro Tomas, and Foley. Later (p. 24), considering the aberrant 


results of Old English lenition, Lass ‘notes: 


This suggests that the simple hierarchy ... is not really 
adequate; at least the movement of segment types down the 
scale is not regular, since voiced stops drop two positions 
(from strength 3 to strength 1) while voiceless fricatives 
drop only one (strength 2 to strength 1) and voiceless stops 
do not drop at all ... Now let us see if we can formalize 
these conditions more accurately, specifically in terms of 
strength hierarchies. It seems, on the basis of the evi- 
dence, that it is incorrect to say that aoett down the 
scale is simply linear; it looks as if what we want is some- 
thing to the effect that the first stage of lenition is 
voicing, so that a segment goes to the nearest voiced 
segment type below it in the hierarchy. I suggest that 

the process of intervocalic lenition is best handled by 
positing two hierarchies, one for voiceless segments and 


one for voiced ones, which cross-classify. 





me oid 
r 


- 





pssdanct ax empe eds a eaapancisi. ol? 


A. 


j 


— 


pusdews.! yd barogarg saat ‘od ispitnabr et sisiriw (noitinel od sonatatest pai 


a ee ee _ 


ers =~ 





























: won 
eafteb ct besamets ©! zeosiq feteves nr ey boii a 
g ata i 


xd bewollck ,toanceex teom ont pried See ae npest TO . iat 
jady ni yegete bos ,ecdsoig%is | seeetioagie ev amaieic nif, 
te stsimo 1 
sd yen ninaceant pt ox rit iw. esit. aoidine rihib tastise orem sit to qk» om 7 | 
@eiouscena th and2 baton asd oot {00 - Fer) 2eeI id seqeq ar at ages 


+9.1) ‘dtpnecte’ to sisos p-eanitebD azal xae feifpall 10 ais gin a 


Sasrreds oft ptizebiaac> ,(d8.g) astsl - yell bas \eime? onmeved get ; 
:gedon Based ,noltinel feifpnt BLO to aiivesz re: 


; ; ml ne ee 

~yiissx jon 2i-... Yiotsrsin slquta odd sents ataspuce are , y 

: ss - - 

oct nwob esqyt tnampse Fo dremsven er? taesl 3s yecsopeie e 
ot 
: - 

enoitteeg owl qorb soote Baoiov seaie .teitpes Jom ai Siese | | 

aovitsotzt eeolsotoy albiw ({ tipnsate att dipnste ag) 

y* 5 a : 7 
agot2 eesisotoy brs (I rtprerte od S rtpasxte) sao vino gaa ~ 1s 
aSifsmiot ms ow Ti sea ey tol wo ... [is ts gob ton ob ve 

<b 
16 eared nk ylfen Piasae WVistemnos sian eros tho seer | 
a 


~ivs orld to atesd sit no .emses ST esinomsasid dposwte 
oft woh Jasnesh tacy yse a+ tosrioonk at di gerd 008 
~amps at tnaw ow jedw ti 26 bebe $5 visarnet vai er tn 
2i nottias! to eaeis setit scid “té0d tog | 
Beotey sestson oft od 29% yi 8s Ss: wi 

sats Aasppe 1 -vibusxoiid ort va di 

ae bo Sneed dead at rots tne i 

bee ehaamgoe | 


- * 
%, 


Lass then attempts to bolster these remarks by noting possible explana- 
tions based on articulatory phonetics. His statements indicate the degree 
of difficulty which is involved in any search for universal hierarchies; 
Lass is caught in a dilemma between an intuitively satisfying hierarchy, 
which has been observed to function in other languages, and a set of data 
which do not conform to this hierarchy. In the above quote, Lass was 
forced to recommend a modification of the proposed hierarchical structure. 
Elsewhere (1971: 22-3), when discussing Old English data which do not 
Support the hypotheses of Foley and Zwicky regarding hierarchization 
according to point of articulation, Lass proposes that lenition and 
deletion be considered as completely distinct processes, with the strength 
hierarchies applying only to the former. 14 
More recently, the problem of phonological hierarchies has been 
approached by Vennemann and Ladefoged (1973), who offer a distinction 
between phonological primitives and phonological cover features: '... an 
adequate phonological description of a language must be expressed in terms 
of two kinds of features. Any empirical theory has to have a number of 
primitives which are definable in terms of concepts which belong outside 
the theory. In the case of phonological theory, these are the prime 
features, which are definable in terms of the acoustic or physiological 
properties of sounds. Each of these features consists of a single mea- 
surable property of a kind such that sounds can be said to have this 
property to a greater or lesser degree. In addition, there are phonolo- 
gical features that are not themselves prime features but are disjunc- 
tions of values of prime features; we will call these features cover 
features ...' (p. 61). The authors then consider the manner in which 
this distinction may be incorporated into current generative phonologi- 


cal theory. ‘Two logical possibilities present themselves. First, one 


33 

































“ania qx wfttaoy liseet Wel aye 
germpab corks agsoibat airreamdsis aie 4 
yesirotamein Isetevin x0 doreee. nt a pines 
wyibusteld! patyteitse vlovitiviat as ond serail 


a) 
a 
stab 7o te2 5 brie , 2ogpHnAs! retto o£ sotrorne oe tevseedo nnd ane Abt * 


’ 


a ol 


esw gaat .Sdarp syods elt nil .ydtrexsid als ‘ce macRsiee Jott ob dic * 
eudcarate froidorsyetd bezoqoxq et jo soitsostibon 5 _ Enron ot Beosot 
gen 06 Asiriw aie dei fond biG pateasoatb ascw ees +. £84) crockeoe f8 
neitesifesseretd po sealed wioiwS Bas yoicd 20 sonalieged orkt droge 

bes noisinel jst essoqorg gael noteleo ive to $niog at Hite 
feotexte odd rGiw ,esecsoacng soniseth yistelqnao 25 Sersblanee sd, moiteleb >. 


: MM emo? off at vine eniytogs soisimewsit ~ | 
rss seul asirowe pit fen cpetenanig to sweidora Sit \itnecst. S10 in 
noltodiseih s 1otte odw ,(€Tet) Bepotebst Bas stemenne¥ yd Belibeonags 

fis...’ :29%M4se? 1sv0o Lsoipofonadg iis asvidiming fecbooicnedg aiesd 9) 
cals fi Beessuepes od Jam spsipaat & Xo nobtqhrozeb Lsbtpolonodg ‘eteupabs | ) 
to xednn 5 .aved od asrf eae isoinigms ynA .ssaiset Jo eleLiat ow to ca 


sbhiesuo pacled tiinw aigsoncs To enriat at oldsaitsh aed ottiw covisimitag 
emtia sit ore sean . roar Lestppiortade 19 seo ort nr. <vaoees add 
fan tpoloiayda xo Siawoos sft To sims az oldenitsb. ws riot some 
“sem slpaite s Yio interes evuutsst easdt to dosd ‘aad er: 
airtt sve o¢ Hise sd msg abrmoe Ssrd doue bait 6 10 regen: 
“olornoig, sits sxodt nob} tbbs AI .a9xpsb ssaeal = ad 
+ enieib-sre 3p asurtpot omiag gov teanody tor! oan th a 
tayco somiss? ‘seard, Liso Lite aw een y esi 
tS tn ti i 
ae ean ovizeronsp InerINS © | [cael fie 
ot, ai ens — eters 
x Sui’ , 


é 6 


<i 


3h" 


may 'treat both the prime and cover features indiscriminately as "dis- 
tinctive features", by cross-classifying all segments containing prime 
features which enter into the definition of specific cover features' (p. 


67). The disadvantages of this approach are (at least) twofold: 


(1) By erasing the distinction between prime and cover 
phonological features, it would obscure a differentiation 
based on a well-defined phonetic criterion,namely the 
presence vs absence of a uniform measurable phonetic pro- 
perty. (2) By treating prime and cover phonological 
features as primitive concepts of phonological theory, 

it would increase the number of primes of this theory, 
which is undesirable on general methodological grounds. 


(p. 68) 


The alternative solution is to define cover phonological features in 

tems of prime phonological features. The authors opt for the latter al- 
ternative since it avoids the methodological disadvantages signalled above. 
In particular, maintaining a distinction between these two types of 
phonological features permits distinguishing between a level of phonology 
in which units are determined by phonetically-motivated features and a 
derived or metaphonological level characterized by the use of cover fea- 
tures such as phonological strength. Ladefoged and Vennemann (1973: 68- 
69) specifically note the applicability of their distinction to the 
problem of strength hierarchies, adding that 'the situation with respect 
to a cover feature such as Strength is slightly more complicated, both 
because this is a multi-valued cover feature, and because different 
languages employ different strength hierarchies'. Rather than elaborating 


on this problem, the authors offer as an example a reformulation of the 






































omixg phintatnco etmemge: Lis gniytiarsin-eaoto ye." 
«q) ladwedest davon ol iosqe to moliginiteb Sy otek xeon vith an 
‘hiotest (tasef 35) con Hoss eit Yo sept co 


1400 ee eaten noewied ‘nakjoniteth off paiders ya wo 
‘ 
es we 
noHisitwersttib s sribedo ‘filvow tt eens isslpotéoniie 
A 
ety yCensn, noixed ins nitenodg eaten mall 5m bowed = i 


hes TH USF 
-oxrg oitsnota sidsammassai mxoting 6 45 Sonseds: sv aonses%q 


; Lie 
{56> Ppoioaorig 13703 bas smiagq piitsext Yi (S) .ywusg oe) 


mK i 
Reta leolpotomeda to. atusonies svijiming as eourtnot 14 


: 7 ga fab : 
rosrit eitt to atmitg to tein ott suspaat Binow +i ° = 


v ¢ 

.abevonpe Iso tpot ohetigean Tstense no, Sidetieehan ai fohdw 
(23 .@) he 
. : ae : 
ni asuttsst Isoipofonnda ssv00, enigsd otjer soity ipa ov ctsred is ont 7 
“is xetisl aid x1oi too etedtve sat - BSCE iso ipo fonarig gms 20 emsed® 
.audds Sollenpia 2 aledimeypeaits iso ipolobortsm ae ph hy +i omnie ovitsmaisd ‘ 
to zeqys one Sacxit ae noitonctalb 6 - paiisanten viskmisysg al 
yeoloncdd to [evel s neswisd gnineinpnitelb atime” eeuied Lsoipafonadg 


5 bis esmdsst hotevitom-yilsobienddg ¢d bon mate sts eivihadiie., 










“st x9700 to vay add yd Bes iedpexsio devel teot . a 


“85 :EV2L) routes brs. bepotshrt gnats 





: ae ' ; => a 7 7 
eit o¢ rieisonidsath tisk. to biumintng ss odd. ea iin. f [ih 
i 7 


soehes1 dtiw noigaudia exis" teat potbbs , *) 


poate: Mig 
ded ,badsoiigneo . sion yisrpife ai cou e ante Me 
Peas rene we. 
gnexstiin seusoad bas sete eves soudey oa 16 ei é 
a oe gh _s oe 


SD 


consonantal strength hierarchy of stops, fricatives, and approximants, 
utilizing a set of redundancy rules. 

As a final observation preliminary to the detailed study of phono- 
logical hierarchies, one may note the work done in language ontogeny and 
language pathology with an eye toward establishing 'universal' hierar- 
chies of language development. The first linguist to coherently suggest 
the study of these two domains as a source of information on language 
universals was Roman Jakobson (1941) who, based largely on the findings 
of Grégoire (1937) and various Russian observers, including the neuro- 
logist Luria, suggested an intimate link between child language, aphasic 
disorders, and universal aspects of phonological theory. He noted, for 
example, that children tended to learn the phonemes of their language in 
a particular order, suggestive of a universal hierarchy of 'acquisitional 
difficulty', and proposed that these observations be linked to other 
investigations of phonological universals in the hopes that significant 
correlations might be found. Jakobson also suggested that the dissolu- 
tion of speech in aphasia followed the mirror-image course of language 
acquisition, with the phonemes which were acquired last being the first 
to disappear, and so forth.1° Jakobson's original observations and 
speculations have subsequently been subjected to a great deal of serious 
inquiry, carried out within a number of different disciplines, and the 
emergent results have been no less diverse. In general, it appears that 
there do in fact exist hierarchies governing the acquisition and loss of 
phonological systems, but these hierarchies are not as rigid and unyielding 


aS waS Originally proposed by Jakobson .1® 


7 ia’ @ 


g 
~ 


vetnaminorays bas aevitertst a oes: | na 
ston eae ae ist =i i 
~cretig o youre peliesb ert od yreniniiesa noktevisedo Latins eae | 
tne yrogerco spavueet at anak sow orth ston en ane \aRiniomeal Leola 
—rBisid ‘isenovinw’, piifeiidetss biswod sys ms sitiw ypaloritag eR 


tespoue vitroterioo at Fhivonil Sexii sat .tnemgofoveb spsuonet to esirp 
- J ; : 5 - = Fl 

; ‘ - my 

avecpes! ao moitsmrolnt to some ses antamob owt seedd Yo yeaa ort ~~ 


























oe 


7 





A | geetthai? add ao ylopas! ‘beeed jodw (THE) aoedoast nema sew elemmewtnns 

. | Guyer: 4 patbeloni ,eteyvreado melee auoiusy bas (VOL) éetenn to » 
oiesrigs \spsupas! Alirio nosed iil stantial ns Padepeaue uatmal depot 
tot .badon cH =. yroertt teniepafonorig to Bite Isetsvinu fas sexabroe ib 

fit spsupisi utedt to esmpnorig srit otss!l at bebnat neibite> tert “ee ’ 

paidietsenspes' to yiousteil ipsioyisu 6 1o,sviiespene . sito Isivobtisg 6 “ 

+1), ; rwitiFo od beoitil sd anchtsevysedo seadt tert beeeqouq fas , io isa 4 

h | ttepitingts tsnct eaqor eis ni elsexrevinus [so tpolonmarigq to ano itepiteevnt 

‘etHoestS sig tari: hed espaue oe Le spades . bavor sc tip it one tislersco : 7 
Seeuprel to smujoo spami-xoryin ‘age bowoliot siesdgs ni ddsege do AOR? 

sexkt ond pried Jasl bevinehs “as doiiw demenioda) Sis atiw .noddiekepos 


; , fs annitevisedo Ladipixo e'noadotet. ©) stot oe Bas , Ts8qqs2 ib- ot 4 
‘: auortes te fash tsesrp 6 ot bespsfdue nosed yitnsupsedge ares arnitsinosgs. 

y ort fim ,ecbiginaib Jsrexttib io aschun 5 mittiw juo Beitiep <yntupett 

£ tent axresqas at earth nl .sexsvib zest on fised. aysri ad Iueon Srepiens 

a 


to gaol bow noitietupos arit Reh RniNe "RE eae 2ONe Oy ¢ f : 
[Bitoni bo nts non sam SER meni int ee See 


enema Oe 


7 : te, : 








36 


1.7 Scope of the present study 


The remainder of this investigation represents an attempt to 
demonstrate the possible role of diachronic strength hierarchies in 
characterizing sound change. It is divided into two major parts. The 
first part, consisting of Chapters Two through Four, contains the data 
establishing the existence of hierarchies of vocalic strength and word 
position in the early stages of three Romance languages. A selected 
corpus of etymologies is used’ to yield data regarding the rate of loss 
of vowels in the various atonic positions characterizing the languages 
being studied. The computations of rate of loss result in the establish- 
ment Of a phonological hierarchy of atonic positions and a hierarchy of 
intrinsic vocalic strength for each language. A comparison of the 
hierarchies isolated for the languages is then used to provide informa- 
tion about more general strength hierarchies common to a larger segment 
of the Romance languages. 

Chapter Two opens the discussion of the search for a rigorous 
formulation of phonological hierarchies, and presents the results of 
statistical analyses of Italian, Spanish and Portuguese. In Chapter 
Three, the phonological strength hierarchies are discussed in terms of 
the phonotactic compatability of resultant consonant clusters, in order 
to view the interaction between the two major diachronic forces of pho- 
nological strength and phonotactic admissibility. Chapter Four discusses 
the fate of final atonic vowels, with special reference to Catalan, where 
loss of final vowels was much more frequent than in Spanish, Italian or 
Portuguese. The fate of final vowels is evaluated in the light of their 
posited morphological function as the marker of number and/or gender. 


Chapter Five contains the essence of the second major part of this 
































Od Sqned3s ms einessxqe soisneizeoen 30 
nt eattorsusiri dtensise oinowiosib to abet & fire _ al eee 
eat aie tofan owt at i Bsbivis ar ee ae ~ oY 

edeb rit artinanits ery ripyvowit Ow iets to gntsezenco sae sc 

bmow Gre itpneaste oiiscov to aeftdotsye tt to constene sat vaiattdtee ; 

ferioaioa A  eapacoris i Sram 9c1 ot vines aft) mb ete a: 

azol to stay ait paibrspor stsh Bioiy boats at coivolomgte 20 mia , ae 
egprerene f edt piisixsdopreris anoitiecg ninots avoissy ott o£ aloo 30 
~faildssas ext ai tivess ezol sag sji5x Io ano bs BSuqmoO, edt -betbects ented 


to yioiezeid 5 Ban ancis +ieog otnas to ydoisisid ‘eoteotorerg 5 20 om 1 






es : 






od? to moeiisqmeo A .Sptumcist ross 102 Mpasive obfssorr otaniint ) 
“soot? shivorq of Bee: oof al aopeopast orit rot. hedeloar zoishouseist al 
trempse ssprsl s at nonmoo esidoysroili rtpasize Isrensp sia pours ants | 
.eepaupasl sonata add $n af 


oe” 


avoxropia 6 10? douse orit Io noresimerb ond aneqo’ ow? ae y 


iy 

to atinvesx ait atneesia ns. asidoisretid Isoipofonorig to otic @ 
oe 

tadgsD al ..sesuputied fis deinsgé ,asileeI do seeyiens inpitebbase 


; 1% ars Wy » 
to ems ai bogaumeth ei5 setrorsxcid dipnsrte Le>ipolonotiqg hig eunett 
<0Bx0 i \exedaulo trancenco taedlues: 30 WiLidedeqnes 5 i 
we 


-wfig to saox02 aisoutonib totem ont ort nesatad nobton sent na ot 
seamicath wo vatqsd> ..v tl iditeeimis. obtasionadg tie Henonte & * 
eitedw \nbisieD of soneaste: Isinaqe Hsw elawov oinads fen ve 
| x0 nei Led eine? ni csc Susuess3 sora rx eaw ) 


el r 


:h pe y 
sacheeene lands otiens-nice 3 


ane ee Stes 2 ee 
sae i HAR ‘an 


>, ¥ ‘ ’ : 
_# id oa! 


37 


investigation. The strength hierarchies which have been isolated are 
applied to a particular change in the Italian verbal system, in order to 
demonstrate the feasibility of employing hierarchies as a tool in dia- 
chronic investigations. The results are summarized in Chapter Six, where 
several suggestions are offered concerning the manner in which strength 


hierarchies may be incorporated into diachronic linguistics. 


OF AWSDIO 1 


“eib a 


Jofoat asad eved doirw gsifowexetd mtpaste str  -.obepaeamee 


i maxteve Lecrov cetisttl ert ri spasdo we lveisrag 6 oF heilogs 
loot B 86 eatdoistoH oritvelgue 20 ytiiitdiass? ont eoprieacnab 
sodgedD oc! b3siveame,sus ef lyesr Sat _anobispictadend 2 tnouris 
finrtw mi uectmen of onimponce Dexsiio ets enol eepede Istevee 


ypnil oinordbsto offi hetsroguoonr ed ‘yeni 2s iiyraieicd 





38 


Notes to Chapter One 





1 Among linguists there is no unanimous agreement as to precisely 
What constitutes a 'phoneme', and consequently the terms 'phonetics' 
and 'phonology' taken out of context are vague and ambiguous. Strictly 
speaking, any abstraction away from physically measurable properties 
may be considered to represent a level of phonology. In the present 
study, the term 'phonetic' is reserved for such physically measurable 
properties of sound units, while any classificatory property serving 
to include a sound unit as part of the overall sound system of a 
language is qualified as 'phonological'. In particular, the idea of 
a 'phonological hierarchy' refers to the observed tendency for sound 
unit tokens to consistently behave in a hierarchical manner, while the 
actual motivation for such behavior is presumably to be found in the 


"phonetic' properties of the individual tokens. 


2 For example, the study of Zwicky (1972), in which moder English data 
are used to postulate a hierarchy of consonantal strength, and the 
Study of Thelin (1971), establishing a synchronic vowel seale for 


modern Russian. 


3 For further remarks, cf. Jakobson (1941: 82-4), Reichard, Jakobson and 
Werth (1949), Osgood (1960) and Coenen (1965). Some interesting, if 
methodologically worthless, remarks concerning the relationship between 
the vowel scale and the visible spectrum are offered by Schmidt (1947, 
1955). Schmidt's works., despite their nonsensical character, indicate 
the potential relationships which may be established by the untrained 


observer. 


4 Subsequently, Braun (1974) pointed out the relationship between 


Hunting's remarks and the French 'vowel triangle'. 


eee 


ond teas of Sayan 


yiseine:q oft es tnarnsscips ariel ieee: or elsxett  etacupnel parca { 

tesitdenock:' amst add vicnarmsanco bas ,“anehorig’ 's) Seder ieage sere 
Yitots2 .evoupidms bos supsv Sis txetioo Io fuO meas J vpofonery? Tims 
anita sidevuesés.+/ isoieydq moxt yswe soltositeds yas yeriiseqe 
ti1se ate “al .veolonodd to [Svel s dnséstdqex of Beaeiienss ef ya 
Sldexuessn yileoreyia cove tol Bavisast ei ‘otjsaaig filmed ot ,yiecte 
mnivise yireqciq viotsoitreesio yos siiviw .etiou Bayoe To sentzegarg 

6 te meceve bavoe {isyevo att to t15q es tity Bouoe 6 Sierioak oF 
20 6ebr art ,xelim ited ai ‘Tsnipolodorq' es beitileyp ef spespaat 
bratoe xo iebned £ sido ort o¢ sister ‘ydousyelid Lsofpolonody’ 6 
aft sitriw’ .zerwiem [soidcistSin 6 Mf Svsned yidaed 2banca at zastiot tiny 


ett sti bres 2 Losi ma 2 ~orvsetied dogs tot cbtey Gan fepdsos 


ensvios Lsubrvioni sit to asch SOO ‘oe sanorig' 


eoeb faifondt sxsabon fi criw ML 4 Cres vaiotwS to yvoscte edt ,siqnexes 30% ¢ 
si¢ bos ,dtorerte [stnsnoenco towrorsissén 4 etslcteod oF beay os 
107 eleoe Lows Oxubirye ; piineridstes ,({IVYGl) alia? 26 yaa 
feieeut crshom 


bis moedoxct \fausdo ts? (6-88 :is@I) nosdovxsl .to ,atrsemsr sede acm € 
ti \pnidaeroini ance ede!) meatscO brs (0d€L) BoopeO , (ERGL) ciaew 
Mewied ginenoitsici ot prinrsonc aissex ,sesiristow Vileoipolomairan 
\YROL) thine vd beasito sre musts rege cidietv att Das sisoe Towary ane 


etsoiivic ,tedorrsflo [sorensenon wisds otigqeseb aha e'sbomisk fees 


bents«tma) att yd berlelldesee sd yam dotriw eqidencitefer Is tineaa ae 





39 


5 The literature on phonetic symbolism is enormous, and much of the 
relevant material has been collected by Fénagy (1963). A sample of 
other relevant studies includes the following, where the potential 
ranking of vowels along a subjective scale is mentioned: Jespersen 
(1922: 402-3), Sapir (1929), Newman (1933), Jakobson (1941), Bally 
(1951: 54-5), Maltzman et. al. (1956), Chatman (1957, 1967), Hockett 
(1958: 295-6), Wertheimer (1958), Cressot (1959: 19), Marchand (1959), 
Miron (1961), Taylor (1963, 1967), Johnson et yaal. (1964). Weiss, (19- 
63, 1964, 1966, 1968), Gebels (1969), Chatman (1957, 1967), Brown (1958), 
Brown et. al. (1955), Brown and Nuttall (1959), Critchley (1970: 101), 
Wescott (1971), Michelena (1972). Further references are given in 


Robinson (1972: 137, fn.) and Melhem (1973). 


6 See, for example, Popper (1959: 76), Caws (1965: 182). A more lengthy 
discussion is found in Derwing (1973: 225-47). Consider, for example, 
the following remarks of Rapoport (1972) a& propos the role of counter- 
examples in formal scientific theories: 


The phenomenal success of physical science has been attributed 
(justly, I believe) to the physicist's preference of explanatory 
power over explanatory appeal as a criterion for accepting a 
theory. In the physical sciences verification of predictions 

is the 'final court of appeal’, as it were. Whenever the derived 
consequences of a theory result in predictions that consistently 
fail to be corroborated, the theory is modified or discarded. 
Thus, a 'feedback circuit' is established between theory and 
experiment, between logical deduction and observed fact. The 
process propels theories toward greater generality and precision. 
(p. 327) 


If observations fail to corroborate the consequences, the model 
is thereby refuted. But if the consequences are corroborated, 
the model is not thereby 'proved' to be a representation of 
reality; it only acquires more credence. We can then contrive 
to use it, drawing additional consequences from it (if we can) 
thus putting it to more and more severe tests, perhaps modify- 
ing or generalizing it in the process. This is the 'feedback 
circuit' mentioned earlier. (p. 329) 


Rapoport's statement represents a succinct appraisal of the role of 


counterexamples in scientific theories; their disconfirmatory action is 


seit 30 ccm Sah \avormiors et mebfodmye cisenolg A suseweel ed? @ 
to lasses A (é > tT) ynsetd yd Betoalioo need ear isivagem jJosvelox 
Paismetog ad? smarty .patwol tod oft ashylont eolbete tasvelex serito 
iwereqesl rberoitnan et oslsoe sytostdve 5 profs, aloyov to priistwes 


VI LSG (fet monctowiel , (LtHl) asmwed , (ecer) xiagse + {£-S0Ob €Sef) 


Sjoteow (Teel .teee) nment)erD \(02@l) .In 135 comehiem , (e-82 seh) 


GACT) Brarioxem .(@[ 18208 naesx) .(6cPD) szencerdtre , (gnaee :B2@L) 
SproW . (hat) [s .ts qoanint ,{Toel ,bder) zalyer , (fee) nowIM 

(SGGi) meotes , PPT) nemteand ,(@3€L) elisded , (eel aes bael .£8 
» CEO s0VOLl) - voli rm) .(&cel) Listsu Bas maord .(cecerly” woe J8 Swouk 
mM meyip sis esone + rearoauva .(STCi) snoferioim .(IVGL) sgtepas# 


Cf) oewife4’ bos {int .VEL <SVOl) nommidod 


’ com S{S8f =2a@l) eeast (AT: ;@e@l) szequol ,okomers <8 ~ase 3 
MEAS LOT Sn ienoo b-2CS :EYCL) oniwratl af Bavot et ‘OLeeupe rb 
mires to slox < 2OgOTG sV@f) Jxogegsd to extramer patwotict ery 
29 frost oLiitisine Lemxet at 29) Quisxo 
Peto ittts Ased = oneins fsofeytiq to eesooue isremoosrig aaP 
icKo to suieysteyq e'’teioféewiy sat of (euenied 7 :viteut) 


LJts0o6 Tol noOLratix eS LESS YICTSENBiGksS ISsV0 Isawog 


oihetg Io molisoltiney 2zsonetoe Isoievde act, ab =. yer 


: Sviseh xis isvansit, ,syoaw ti es ,"iseqqs "io Suto Tanke" ere et 
Yocsteieng) tait enottotisrq nt tivest yaoedts fo ZAOMSUPSGHOS 
£ b xo Bel i neat ei ,bedsxodvsuos se ed eee 


2716 yacerl neces herlatidstas at "o.biortn sosdaaet’ & ,aeair 

ort o6i Devteedo bas mitoubeb issipol asewtad Jtianizeqns 
moretosig ons yiilstensp, 1-¢seip brswd estrosd? aleqoig easocng 
(TSE .g) 


+3058 SiS ,asoreupsento ey sisicdortes.c¢ List “eneideviseme: ak 

DETEIOUCIICN ONS esofSsupstm@s siti ii ta |. besirter ycleredts ai 

iO TOlMfaAsmneesids ys 5 od Qt ‘havorcr’ yderartt 7O0 47 iieeg Ps 
Srrismn nett so SW ISAS via dstiopon wit - 
ib aw ti) ti test esoncroeseics |i isttiots. ibbs oneal tt — 2 
“7 Bom acsrite SJ ,eteeat sxoves sas bas Sion oF pe bar ete) cet 
tascbest' ait ai girl .eaaodiq-end mh sbvpeiebiorages col 

‘i | (QS .q .zelizss hanokinen’ ' ik 


Yo efor oft 2 iseteunys 4 fortinoga s aT@eatgeT jrenistete 2" xcopgea y aa) 


ai aoiios seencamat Stowe i thot .egtoert oALnoige stk molemmecntatmuce pai 
faa 





¢) 


clearly indicated. In his reply to Cohen, Faley claimed that his 

own opinions mirror those commonly associated with the philosophy of 
science. Foley's position, however, is radically different from that 
to be found in the empirical sciences when it comes to the question of 
the role of counterexamples. As a further example, Popper (1959: 113) 
notes that: 'Theoretical science aims, precisely, at obtaining theories 
which are easily falsifiable ... it aims at restricting the range of 
permitted events to a minimum; and, if this can be done at all, to 

such a degree that any further restriction would lead to an actual 


empirical falsification of the theory'. 


Cf. also Ramasubramanian and Thosar (1973). 


Consider, on the other hand, the statement of Foley (1972a: 97), after 
having established a lenition 'strength scale' with velars being 
weakest, followed by dentals, and then by labials: 
It is important to emphasize that this smle does not refer to 
the phonetic properties of the segments but to abstract phono- 
logical relationships. Since our goal is the study of the 
human psyche of which language is one manifestation (along with 
behavior and mythopoesis), we surely want to construct our 
linguistic system in a manner which will allow correspondences 
to be established with other psychic systems. A phonological 
theory which includes a set of distinctive features based on 
phonetic data, because of its parochialism, its inapplicability 
to other fields, will not allow us to reach this goal. 
Once again, as in the passages cited earlier, it appears that Foley 
has lost sight of precisely what the 'distinctive' of ‘distinctive 
features' is supposed to signify. The very choice of terms indicates 
the fundamental desire to seek out the basic psychological correlates 
of phonological theory, and this search must in turn be guided by 
those aspects of language which can be demonstrated to enjoy psycho- 
logical prominence in the minds of speakers. It is quite wlikely, 


for example, that a speaker will consider the difference between two 


. ; 
Sid tort) Bemiaio yale ~ned@eay glqdy amt cl -bedeottak yiaseld 
Yo YWignsalitiq srit Aa tw he stnoges vinomico seordy toreim enpiigorame 
harit moyt tnevetih vilsoibsa ei “aevewor \soitiets e*yeiet . 90s ipa 
to Rottesep att at sanco jf asdw esonsice (soizigm sd? me Bric ad as 
(ELL 2O8OL) zeae paignsns Tesi. at asiqnmexaracnues Io sicor er 
eeircens paintstdo 3: yiastostg ,amts sonsine Isoivsioed?’ tie eaor 
to serex att pnigoiuteou 3s omis 3h... olderitels? yliess axe ae 
od §f6 45 2005 od-cem eit 34 bas wma m o} atrevs badd Inmegq 
sud7s, m5 ct beol birow noitoisteer tecthwt ys dedd sexpe 6 me 
. 'vroartt+ oft tO noitenitreict Lsoretquy 
‘efeOl) aseodT bas nsinemsidvesme eeis .2) 1 
15976 . " ‘sS\Cl) yalot Io toappgete ot ,bnsn spate ij no .teiiemO 8 


Miisd sie! richtw isos fipreite’ noitias! 5 bederideses prLived 


~ phd 


at * ton Be05 eirt tedt ssfesdqms of snsttogm ef 31 
“onoig Jogi oj tud etrempoe art to eérdiegqeta oo Srodg arts 
ait to vbete ort ai [Sop mio sonia .aqmienoiisiexn Jssmoet 
itiw pools) cottscestinem ead et spsvonsl doldw to steyeq cama 
Iwo tourdeno> oF tnew ylexua sv , (eizsougdty bee tolveried 
asonmebncgsarico wolis Lliw rinigiw sanmgn 6 Gi Meda Siamese 
leotpofoncia -£ eye ofipyeq szaite sidiw bedatidatas ou Go 
mo Reasd aouct: tormitecth to t6e 4 zebolont roinw vroem 
(tilidsotinnsai asi meliisibtoorsg ati-to sausosd “ets Srysaemg 
rity ross co wolis ton [itw ,eblem sede co 


i 
(y 
S 


2epaensg a ni 26 \flisps eomd 


i 
) 
ro 
— 
oy 
ty 
b 


yeio! tads exseqas 3 Si [6S 
svismniszetS' to -'svisoniierb' ada dartw yiseiverg to dipia: seek aac 
eadeoibai amieat to sororh yisv aff .ytiepl= of beesague ef “eae 
estaiorios Lantipeforoyeq oieed off tuo Asee OF Sekasb Tet temmbnie oct 

ye puestie tt mms ri teum rivtssd id bos, prods iso rpoLonarg 3o | 
“odyeg was,  betsctenomeb ad ae Ainiiive spsupnet to aioeges aeodg 

 eiartitoy 8: Lup ei.tl .ermiseqe to ehaim edt mt sonsaimeng fepkpet 


ret 4. 0) 0 


wt rieewted sonerst?ib arid taiienco acid terisege 5 tecd 302, m 


@ 
& 





hy 


phonemes to be a reflection of their relative resistence to lenition 
Over a period of time; rather, such strength scales are generally 
observable only by the trained linguist, whose job it is to determine 
the correlation between such hierarchies and the psychological and 
physiological characteristics of language. To define a language in 
terms of the abstract hierarchies which arise from the language is to 
evade the task of determining the abstract structure of language as 
separate from the concrete. As a consequence, unless subsequent 
testing affirms that abstract features of the sort suggested by Foley 
are in fact 'distinctive' features in the narrowest sense of the 
term, any investigation of phonological hierarchies should work on the 
premise that such hierarchies are merely a metatheoretical interpre- 
tation placed upon phonological data. In the case of hierarchies 
based upon position within the word, this conclusion is hard to avoid, 
Since at present there is no indication that speakers respond to word- 
internal environments in a manner suggestive of distinctive features. 
The following remarks of Martinet (1966: 13) are of great relevance 
to the case at hand, albeit in an indirect fashion: 

The model is not the structure, for the structure is always in 

the object, latent as it were but only if latent is not opposed 

to real. The best that can be expected of a model is that it 

represent the structure exactly, and it will do so if the scholar 

has succeeded in correctly disentangling the latencies involved 

and has not tried to force them into a prefabricated model 

founded on the set of a priori ideas currently in fashion. 
For an example of this type of interpretation, see King (1969a). 
The value of this analysis is questioned by Newton (1972b), among 
others. For further examples and discussion, see Harms (1967), Chafe 
(1968), Wang (1968), Cairns (1969), Davison (1971), Naro (1971b), 
Andersen (1972), Haiman (1972), Shapiro (1972), Vennemann (1972), 


and Chen (1973). Some counterarguments to this line of approach are 


r 2emertoriq 


5 tTeavo 


ss i == @) 
t » at? 
ode 





‘eo 





10 


ll 


ye 


3 


14 


15 


16 


ho 


offered by Creore (1971). 


Cf£. also Causey (1972a) for further elaboration. 


Perhaps the most obvious metatheoretical bone of contention is the 
competence-performance distinction. For a thorough discussion of 
this issue, see Derwing (1973: Chap. 8). Among the numerous other 
works treating this same theme, some interesting remarks are offered 
by Reichling (1961), Hammarstrom (1971), and Pylyshyn (1973). Other 
metatheoretical questions, especially the notion of 'simplicity' in 
phonological theory, emerge from the debate between Householder (1965, 


1966) and Chomsky and Halle (1965). 


Even this statement has been disputed. For example, Kuipers (1968), 
based on data from Indo-European, has concluded that consonants take 
phonological precedence over vowels, from the standpoint of a wmi- 


versal hierarchization. 


For example, Foley (1972a: 97). 


T. Priestly informs me that contemporary (standard) Russian exhibits 
an analogous phenomenon in fast colloquial speech, where deletion of 


consonants is quite frequent, while lenition is ‘rare. 


The correlation between child language and aphasic speech has been 
experimentally disconfirmed in several cases reported by Fry (1959). 
On the other hand, Alajouanine (1956) cited other data in favor of 


Jakobson's proposals. 


For data pertaining to the phonological hierarchization of child 
language, one may consult Messer (1967), Menyuk (1968, 1971: 54-91), 


McNeill (1970: 130-41), and Ferguson and Farwell (1973). Studies on 


(Fel) sxosexD yi beretio 
[is yards * ssvel) yoeuus) oelis .iD 


+aqm eit aqertxe! 
ffonttes rb soterzotisq-Sseresegnco 


MLIWISX Ss ejeec ai 


OL 


MiiD 20 sithsuroisioid [sofpolonoiq at ah printedseg saab woe 


-, MONS awl} r2eesM t3ivenco yen sao ,Shmepnel 


SO 0F% teva) ; WISH Bis nosupis% bres (Ldb-O8S sO9EL) Li teiiomM 


a 
a Tf 
Pe | 


di 





43 


the phonology of aphasic language are given in Shankweiler and Harris 
(1966), Blumstein (1970), Pilch and Hemmer (1970) and Schnitzer 

(1973). Further evidence from experimental psycholinguistics, although 
restricted to English, is reported in Wickelgren (1965, 1966). The 
problem is also discussed, from a somewhat different perspective, by 


Bromein. (1971). 


my 


» 
c : wi orig on 


ry Wie ao. . ‘ 2] 
rt 7 a 4 4 
{Tt it - ¢ 3 } 


St q ¥ 
ctx (ever) 
: 


* 


eliuaorg 


(LLARCr? 





yy 


CHAPTER TWO 
HIERARCHIES OF DIACHRONIC STRENGTH 


2.1 Introduction 


When postulating the action of a historical process, the 
linguist may look at successive stages of the same form, and write 
an equation which summarizes the essential changes which have occurred. 
Fundamental to such an equation is the assumption that the events in 
question proceeded uniformly, with no intervening reverse develop- 
ments, unless otherwise indicated.t For instance, given a formA 
ending in a consonant, and given the reflex A' of the same form 
several centuries later, in which the final consonant is no longer 
present, one must assume that the consonant was simply dropped, per- 
haps after first being phonetically weakened in some plausible fashion. 
It is not feasible to assume, for example, that between the time of 
form A and the time of its reflex A' the final consonant was succes- 
Sively geminated, aspirated, nasalized, and finally lost, since no such 
intermediate stages are attested, nor may they be posited through ~ 
parallel developments in the language. If in fact any or all of these 
changes did affect the original form, the results posited by the lin- 
guist are inexact. However, this basic indeterminacy must remain, 
and may only be attenuated by amassing a large amount of data concern- 
ing the nature of plausible phonological processes occurring in the 
particular language under discussion, as well as more nearly universal 
phonological change-types; the indeterminacy may then be qualified 
by a statement of probability or likelihood. 


The present investigation treat$ only the observable data from 


the history of a subset of the Romance languages, qualified by the 


















ef eze007 fesisccreid 5 to noes ae 
c, Se 


eeiinw bas nfo? ance edt Yo espsi2 ar ipaeoye 36 en 

























r sberiss00 eyed doiriw esprisia faitnoees Sat tia eis soso a8 
& ni eifitove act? Sacti noidguees eit} ei notsaups os dove of Leinamebeet 
F | —pleveb apraven pouinsvietnt on ntiw ,yimotiag bebasoorg spiteeup 
A mo? & msvip sonscteni A | “bags tbr seiwtertto zest seine 7 
mot ame srt To - xoLior. eit navip Bas  dnsmoanes 6 mt exits . “si 
+ xeerol on ai tremosngo Isnit alt roitwnt pistel Aarne Is19ve0 ‘ bas 
. _ Renmoab yiquie aew Jnendento Srit, t5c3 emees tenm sno - Sreseng, | . a 
(1 piesa’ oidiauslg amaz st bonsasew yLiso.ctonory. phied Jest eas eget | - 
a, Zo amit aft doowisd tet ,aiqnexs rot ,amees OF efaresst jou 2 3T 
: widens ‘eine Semioetce isnit edt "A xeltsy ati to ont at bas 4 seit : 
} fie on conie abit yifeni? bas {Bes iiesen ‘bovexiges <badsrtes -yheteda 4 
fipuorrit basieoq ed yard: | son \botesdt5 ots eepsia sdetbememnt 
. -. geet} % [fe 10 yns fost ni 21 -svsvonst oi nt e2nomaoleysb Geokbeag= 
5 “thi aft yd batieoq atiuzs1 srt jm10t Senipiao = soe: tb empmada 7 
rt vniemex jeum yosalarcetebni steed. atrit sare: foment as see : ; : 4 
; “rdonco ia To snare spxal 5 pilgesms wi Betsunedte of Ylne oy ‘ ioe 
ert ni privunoo RSassoordg Lssigosomts sldtausig 10 stn on Co 


at fier 
| Esereviny yinsen san 2s fiw as soleeibalb sow 6 at ez ] 


paltitasp od start vem yosnbibsatin ot — 


vr) i 


ws re | a ry 
»~, ¢ 7 
‘on a 
é ; 


15 


general methodology of historical Romance linguistics, and hence all 
reported results must be weighed in the light of the caution stated 
above. This limitation is common to all of historical linguistics; 
nonetheless, since it is a common practice to base theoretical claims 
on the results of such historical records, the shortcomings of these 
theoretical foundations must be admitted. 

A further methodological assumption which is the common property 
of most historical linguistic studies is that whenever two groups of 
forms are related in a manner which may be expressed as a single dia- 
chronic equation, a Single diachronic process must be posited, by ap- 
plication of 'Occam's razor'. This assumption leads to an even 
greater indeterminacy than does the preceding one, since it is not 
unheard of for a language to arrive at the sane results through dif- 
ferent processes, either in an apparently principled fashion or as 
the result of the fortuitous overlapping of diverse Bee sccate Here 
again, however, the indeterminacy is unavoidable, for unless it can 
be demonstrated that more than a single process was at work to produce 
a given result, ‘factoring through' by the conmon phonological develop- 
ment is the only non-arbitrary method of assessing diachronic data. 

In the results reported below, as well as in the following chapters, 
all the data were evaluated in terms of the methodological assumptions 
just stated. Since the events in question occurred many centuries 
ago, it is nearly impossible to sharpen the results any finer, due to 
the lack of adequate documentary evidence. On the other hand, by con- 
verging on the problem from a number of ldifferent directions, a 


plausible case may be constructed. 






























tts sonset bre eokse usp hl $0) 


basade coisveo att Yo dripil ont 
viteiopntt teoixonaii io (is at ‘nemo ai soiasgimct oil Pe 
eabslo fsotteocerlt assed ci alee abana 
seed 20° eprimadtione ved’ (ebxooer fephrovelfl ‘idee Suid 
treet sown ssi) ai cbiriw noki@aees Lsorpotoborttent sects die 
Zo ayer owl Iovarsriw sent ei caibote >italvontl ee 
~sif sIpvie 5 25 boars ed ‘yam doidw tondian s ak feales ove emmtt 
aah Ve (beriscy ad taum ezsco1q Dinobsib aipite 5 nod taspe inex 
fave ns od 2bse! noxsqaes sin? .'tosex &’mepao* to nosganilg 
$n at 3i sonte ,ano piihsosig alt asob asd Pe | 
Ahh devout stiness sns2 ot Js avis of SePEUDBL % Brasd Io Bieerde 
| 6 1¢ moiesi bafcioning ylineisqus. as ni isis ,Zogesnctg Inert 


¢ F ' 
ors estes saxev ih fo lak a anoc hot att to sigan ‘St 











miso ti weal 10i .oldshiovarw) ai yosciertatebat art visyewor yitisgs 
seers od atitw +5 dew eascor ofpaie & nsf gicm tertt bexettenanab ed 
“qplorab Tedipolonoriq scm anit ye ‘avout priiasost! Jivest devip 5 
ysis oinoribsib pnitezse2s to Sortan ywertices-sag vino ort et Satem 
ersdqetis piivolict att ni’es Low as wold Bodxodee: erivesr ein 
srorkgiees Isoipoichoriian wht io maid mt boreaievs sxew' ish st Lis | 
seiuinte yasm beri300 nolseaup at sane ort sont ae 

ot sub end wis abiverr cei nieamald + etateonand Crs ay mer 
“and ed brisrh xarito artt no rig gal 
5 RIALS i 2 








—— 
7 





Nat > # 
ir @ 7 
os OC - 

: 5 -: 


| ee ome aa ‘aiid 
‘ 
haces ie, ee 
ae ya “— 


é ) é. 
ee > id ; a ~ - 
- ro 
7 : 4 ’ 





tw 


46 


2.2)\Hierarchiessineltalian 


2.2.1. The choice of Italian. Italian was chosen as one of the target 
languages for this investigation for several important methodological 
reasons. First of all, Italian is sufficiently documented to warrant 
some claims as to the nature of its phonological history. Moreover, 
Italian has not evolved as far from Latin as have most of the other 
Romance languages, with the result that observations arrived at on 

the basis of older written records may often be compared with modern 
Italian, with a reasonable assurance of convergent results. In addi- 
tion to a more restricted phonological evolution, Italian exhibits a 
relatively small amount of external influence, such as borrowing, sub- 
strata, or other processes which blur the traces of past events.” For 
these reasons, an investigation of phonological hierarchies in Italian 
will hopefully provide a solid foundation from which other similar 


studies in other languages may be launched. 


2.2.2. Scope of the problem. Throughout the history of Italian, as 
in the other Romance languages, there has been a differential treat- 
ment of vowels, based upon position within the word. By and large, the 
Vulgar Latin Stressed vowel survived intact, unless modified by a 
process such as diphthongization or metaphony. Substantially the 
same result holds true for vowels bearing the secondary stress, for 
these too remained unmodified in nearly all cases.4 In the case of 
the atonic vowels, however, a variety of events occurred, including 
raising, reduction, replacement, and loss. Furthermore, scholars in 
the Romance field have traditionally recognized that, even among the 
class of unstressed vowels, some positions were more prone to modi- 


fication or effacement than others. Nearly every manual of Romance 










| Japrss ait to ae 2B nsearh aot Asa sitet 
iso tpoloborten soscogni “ne Estey 
 gaexiaw of pathname’ Uisnsh fits et nett ith ‘ re 






















a | 

Savess «| ynavelifise ipolonoria 251 30 > abe’ et edias 

-_ tee ai to teom saved 26 met mort 162 ws tein ot 
: 

| 





ao se bev iris an.sy ieee jars 3 iseor: on is bv . 263 ; 
ferebom itiw boxcar sd natio vem ahxoost neddixw <a o* ee a. 


| -ibde al | tives: tnspisvnod Io sonewezes aldanorss 5 ris kw Pr a ’ 
8 adidirks isifeti .no thi Tove. iso teolonoigq peisinctest saan 6 eee ‘ag 
—the \eniworod a6 fue sonises Simi lenistxs 30 Japons Them visite led 7 
a ce ’ -aittev= tesg to esosext ong ae fiotsiw esrasoo sg 1e¥o 10 eee =a 


exes? Ai sebriniservotd [scipoloamia to noltspigesvit ns \aronset seats 
tefimia xerito doidw mond obitebrol biloss abivoig (iitaed Bebe 


: -berbausl ed yen espeupnsl x~edtéo ag esibute : 
gta 


es \nsiletl to yrotei oi duced? .meidoxg oft Fo sqgeg SS 
~jeent Ieistrarcitib 6 need 2eH Siot ,sspsupaet outed tate: ott nt 
4 vopusl bas y8 -brow rit aintiw noitiedg nog bsesd velswov 29 Jnem 
& vd beitithen eesfru: ,jostnut bavivave Low peseside ie tee 
ert yifeitastadse .ynorgadamn xo noldss.benodtdgl 2s foe sasoorg 

x64 ,2z9nte Yesbiapee edt pribusecd alamo 6% ert. chlor siuaee me 

2o ee6o edt nl * cons Lin yMtsen ai ae 


_ Peiioat Seruiden eiears 20 yiabee 5 eapaniep: 
at wale ,sroteint BE Ear ve: 








i ™“ 


. 











7 


philology contains an account of the 'strength' or 'weakness' of 

vowels in different positions within the word. In addition to the 
general statements quoted in the preceding chapter, a description 
directed specifically at Italian was offered by Meyer-Ltbhke (1967: 55f.) 


who proposed the following set of terms to cover the various positions: 


canto - protonica atona 
canterd = protonica semiatona 
canta = postonica atona 
cantano = postonica semiatona 
canterd = semipostonica 
cantano = semiprotonica 


A similar, although less complete statement, may be found in Grand- 
gent (1927). 

The manner in which atonic environments have been classified in 
itself implies the existence of intrinsic differences, hinting that 
certain positions are somehow 'stronger' than others. Meyer-Liibke 


(1967: 55)=states further: 


Le vocali atone sono di differenti specie. Anzitutto 
alcune si trovano prima dell'accento: canto, altre dopo 
Waccento ss Cantos Maul "a— dilcantosnon)é identica a 
quella di canterd, perche la prima é affatto senz'accento, 


e la seconda porta veramente un accento secondario. 


This and analogous statements which may be found throughout the 
literature suggest the possibility of arranging the atonic environ- 
ments of Italian along a hierarchical scale of 'strength', ‘accentua- 


tion', or some similar parameter. In order to postulate the existence 


a 


a pate: 
o ‘ezemissw’ x0 ‘dipisnie’ Sad Fo 


od? of noitiths nT. - brow ttt eis ter ec oo yaa abe 


inlkitgises 5 revatio pribsosiq srt m ‘aes pa le mica 






















cM , pony eerrrt wal 
(322 <TRCI) sdent-asyert yal Doxs¥o | eaw nbi lest je ine ce - | 
| vanottisod evoisv seb eves ad Binds XO Yae cniwohlo®’ 2 HO 4 a 
: Ons - 
. : Snos6 soinctotg - éimines * ar : 
ot \b . soots imBe soipasoni - Grech ~ ne ) < 
7 / Bots Bo imeadaog - SIRES: ie rat 
7 _ Sftedsims: soinateeq — - ones ne: a0 _ 
. sodnoteoimeg “> Sustness | 
soinatorcgiase » onstasa z 
. “bnss) i Ewe oc ee Jhembtste sfalquce. easi fous cekinie & ; 
a Mi 1 | - (PSRE)) reo rs | 
4 nt Beitiieess maed svar ainanMnox ris ofnots doinw af 2enem of? Lae 
“ Set pattotd esonsisiiib sian fisni to sonseixe aS esiiqnés vise i 2 a 
a eoldiut—soyat .<2termdoc mott “teprnotte' wriemoe sts annisiemeg asaiiel 
: sei? escava d@2 steer) 
a | ‘ 
a | GtaictisnA .sinsqe iinersitib if enoe — ifsoov al : 


5a, 
ogob satis ,Ginso Dotnecos Mba eniig onstoss ieomols + 
“Se ag 
Fs sobinsbi & nor 6taso ib 4s aes SM .oghBo = «6sodnemas*£ 6 
| a oar a 
obnsons 'sse ots ; fue 
: Se OF as 5 iss apie ae a * 
-olushnonse ojmseos mu stemisisy stig sbasses sf 
i ' “<4 7 
i oe 


mt surah i emi ik 


















of a well-defined diachronic hierarchy, however, it is necessary to 
demonstrate not only a differential treatment of atonic environments 
in Italian, but also a set of behavior patterns which are consistent 
enough to warrant inclusion in a step-by-step hierarchical arrange- 
ment. Precisely this second demonstration has been lacking in tradi- 
tional historical grammars of Italian, where one finds instead a 
random array of anecdotal evidence, by means of which the authors 
then attempt to infer the existence of a diachronic hierarchy. 

Even considerations from a narrow perspective have not yielded 
unanimity of results, for variation still remains after the common 
core of observations has been collected. For example, the category 
names chosen by Meyer-Lubke imply a differential treatment for initial 
and final atonic vowels depending upon distance from the main word 
stress, an implication which has not been justified or substantiated 
either in Meyer-Liibke's works or elsewhere. Furthermore, the degree 
of constancy suggested by the remarks of Meyer-Ltibke, Grandgent, Laus- 
berg, and others has never been empirically justified. In order to 
remedy these methodological deficiencies, a comprehensive survey of 
a large sample of phonological processes affecting the Italian atonic 
vowels was undertaken, to convert the notion of hierarchical behavior 
from a series of scattered remarks to a more coherent statement. 

In view of the available documentation on the Romance languages, 
measuring the phonological strength in terms of weakening of vowels 
is the only feasible method which presents itself. In general, loss 
or raising of vowels may be considered to be evidence of phonological 
weakening. The replacement of one vowel by another vowel, not a simple 
raised variant, is sometimes also related to phonological strength, 


indicating the hierarchical interplay of distinctive features; for 


48 


od yranasnan @i $i: revewct! 
axinarostent since to semana G 
Sustetento sus cite acini eae ial | 
~sonsy1s Iscistrusreint qede“yd-gste 5 Scahianetionae! 
-ihiet at paisosl negd pest cobteetencts aces St 5 ) 





5 bsotant dont“ "3m iia 
. ites ii oinourob ib 6 26 conetaixe elt teat 0 tyme th " 
a babisiy ton wel ov Loeciened WORCEREL 6 eet enoLsersolenco oe 

feemoo odd iafhs entemer Li tse Rope, tot ,etivess to io ihrwees 
eiopedao stis betoal fo eed end sbiteracteo Yo eo 
{eisinst ‘ tosittssxt [stiass: ath 5 yioiak cok cae ya Alea ‘ant 
sonmat2eib og emtbrisqab efswov oinode Leni Bas 


,siqusxs OU 


Buow nism sat mort 


| i Sets tyosnjackic “top> = he tt Barr nedd ae 9 epfi fo.sriw noise Efe. Reale eg 
ie 
esipeb aij ,cxiwusiiwi .sieriwe2sis to exiow 2 ‘ vid -xSyom nb xedtis 


; ; ve oo 
| “isi .taSpbasxD .oldid—roye! Io eofiraner ont yd beseappue arts aD BO ’ 


-bativest ylisoitians mesd yeven esl atotte Bas ypred 

; is a | 

Io vovwwe aviensieignao 6 ,esistainizs isoipoloborten seat ybemen 
Simetn melistt ari? pridvetts Kahan. fextpoloonig i ieee ante 


of xeerro al 


.Jnematad 2 FUSIAOS atom 6 OF SALTS | OS 


,eensupisl sonst bd ise retreat ano tai 


alawoy 20 pritrtones nie) eta tt srptente isoipe sells page A 


_ peel Istadey oi -isett atneesiq rtd otis 
h teotpolonarig Io ¢73Give, od oe sie dle ie 
‘ vase: ~ieok aa 
‘¥ signis 5 Jon see, ee gr a iy _ 
vr he A i‘ 


‘ sls os 


Jittgnsese (Bipotonoda ot bath “Pa he 
ee Ne Ean 


1 


es fi 4 a : \ 


ae 





4 : ’ 
P r 
ue. a we 6 


7 


satis to denies incall aaa 7 
eee be 
euotties eft co icw to einen eee 1 


pan 
vabwediod leciteseisthi do aoiton Sn +197009 oe APT ot alin ab) 


























_ 


—_ 
a 
“ 


/ 


ci 


‘ 


1: an 
7 


example, labialization before labial consonants (devere > dovere) . 
More often, such replacements are the result of analogical or morpho- 
logical contamination, and hence do not bear directly on the matter 
Of phonological hierarchies. In order, therefore, to determine the 
extent to which phonological hierarchies have operated throughout 
the history of Italian, a statistical survey of the evolution of 
Italian forms from Latin was conducted, which was then compared with 
more indirect observations from later stages of the language. ‘The 
observations of key importance in this regard are the instances of 
loss or raising of unstressed vowels, aS compared with cases of their 
retention in unmodified form. 

Since final atonic vowels in Italian have generally been preser- 
ved intact, no useful results may be obtained by studying their beha- 
vior; from a phonological point of view they must be considered as 
essentially of the same 'strength' as vowels bearing the primary and 
secondary stress. However, a more detailed study of the behavior of 
final vowels may reveal the relationship between their morphological 
function and the general process of unstressed vowel weakening (cf. 


Chapter Four) . 


eee ee LeCt Lng etie (Corpus 


2.2.3.1. Introduction. In order to attempt a statistical analysis 
of vowel weakening in the languages under discussion, the data must 
be drawn from an adequate corpus. Such a corpus must be representa- 
tive of the entire language, without giving preference to any parti- 
cular forms or developments. At the same time, it must be of manage- 
able proportions, consisting of an acceptable sample of the entirety 


of the phonological history of the languages being studied. In 


49 

























a 
= 


} | Nensiecby < ofee8) eg nenoendo’ sn ee 
/ eiisnesbinbosas 25 ee SL 


win it 1. Linch ee Sore sen A) midi scart 


eri ortintiadeh od (ostierot, <eeb30 aT coitesald tigate 


Spotipuoxs bedex3go ever asisotsxein isofdolonedy stoi et Sead 
; eee 


. noisutov: y ait to yer tue Isaittaistata fig kiSST 30 ysodedal! ects 
ate bersgmee neds esw doidw ,bedoniico - sew kde mort enti cles 


— 


— wa 





ett? -aesdpnst srit io “eapede ‘yodei mort saotteviseds sosctthnt on 

te saoteteni sft sie Bisset ate? of sonstzogal yea 30 anoitsvisedo 

sis} Yo woeko ft iw beteqnan es, ,zloww beaescteny To piieiex wo saat : 
| amex feltibonw at noltnetex 

~3983575 nsec vllsvaesp svat cBbisil ci elewov oiscds Ten aon é 

‘ais sie pityarta yd bentstio Sd (YAR atluest iiutsey on ,doatak ba 

25 an acd taum vers watv to.tntod Lsomelononas mork: yroiv 
bes Yiamig at ertisacd “aiswoy es ‘ptonege! ance et 75 Vi iainmaes 

. 16 aolverded oi to youte balisieb stam 4 ,IsvewoH «=. BaSwss ‘\gueba003a : 

Ssiphyotodcroon tistth neswctsd giffenoiisies sm Peevey yan aiswov ent | 
An) painevisew foucy hoehorisms Bo sasnoig fet4isp of bas mojo 

. Cet  weeqet 

Se mes ae : 

zidyisns isoitaitese & toncdits od sch ai -soistouborda oSs€ S45. 

Jeum sisb sedge ,noteeuperb 125nu Bopsupiisl ot si selenite 

_-sedneeei9e2 Sd teur auction 6 v2. ,auqred ssieiomemiatet? a 

Sag si osc ne Naa aa ne 


7 








20 


addition, all data regarding specific phonological developments 


must be well documented and as uncontroversial as possible. 


2.2.3.2. The REW. Since the required corpus should be relatively 
small, easily obtainable, and fairly representative of the whole 
language, it seemed most feasible to utilize some sort of etymolo- 
gical dictionary. As the direction of evolution proceeded from Latin 
to Italian, a dictionary giving primary emphasis to the Latin base 
forms was regarded as most Suited to the task of establishing the evo- 
lutionary patterns. The volume most appropriate for such an inves- 
tigation is Meyer-Ltibke's Romanisches etymologisches Wéxrterbuch .> 
The limitations of the REW are obvious to any investigator who 
has ever utilized the volume. Perhaps the most penetrating analysis 


of the shortcomings of Meyer-Lubke's scholarship in general, and of 


the REW in particular, is offered by Iordan-Orr (1970: 22-3): 


The strength of Meyer-Lubke, for the period between 1885 and 
1905, and, be it said, his weakness for the present genera- 
tion, lies in the very fact that he espoused with such ardour 
and conviction the neo-grammarian doctrine, applying it 
regularly and consistently, not only at the beginning of his 
Scientific work, but right to the end. It was characteris- 
tic of his attitude with regard to linguistic problems that 
he should identify himself with the theories of Grdéber 
regarding the reconstruction of Vulgar Latin by a compari- 
son of the Romance languages ... Meyer-Liibke attributes, 

to all sources of our knowledge of popular Latin, other 

than the comparison of the Romance languages themselves, 


very scanty significance, and maintains with conviction 


















aa er 
sfetiw s:tt to aria eae etiee tes brs 5 aleniaiety wim ee 


sdlomyte 2o zoe Gin os ilisy od ofelisant Jeon bemsee ti epee a . 
mitel mest bebosoorg so itufove 36 noises ert 2A sponse Jeoke 
6eed nijel oit o etesrigns wetlrg priviv yisnoisoib 6 ate _ 
We ei prninlaridsies Io dass efit at betiere . igon es Bebispst cow eich . i 
7 ~ sign és rose x02 etaitgotiags tact eat L OW aT vatraiseg vamos 
| sibechist OF 2 ee = earn iaaona 2' svictil-reyom ef mlieett 





we notepitcevni yas ot auolvco sis WHA oft 30 esolted baat ot i 


genplats pritextsisq téeu ory eqariss .ampfov ees basil Ge eve sat a, 


Ae 


to brs S , lexsaisp at ifeteforos =‘ sidit-rsyat Zo epiiihood ode se 0 aa 


ee 
= a 
' 

4 


: :(€-S8 :O0EL) x-nsbHOL Yd bexsiio, at .saliro bitsg fi ven ats —_ 


“Bas S681 mecwtad Boirseg ort om wad ~1eyoM to aerense ont - F a / 
-<Terep nseaniaa edd x0? ezendeow eit bisa tt od bas | S0eLs” > 7 
woobis rowe ritw bseucges aff | ents 4osi, Yuov at mt ert i. 3) > a = 3 
$i pnivicg: .adiioob (is PLSMIBTE-OSH eft so boivnoo bas mete OP 

airi to prinniped ait jn yfno tom .yiinotetenos bas elmsiuoo sig 
~eiteiosiaie enw 3t .bas ont ol ons tui Jw pietaaioe. Wik 
Sack} ame ldo, Dictyinmendt os psspes thin sinititae: ais 90-08 i 


‘18x Yo eelxoers oft thew een pane 










_ 


hoe 
j se ie 
Sarath saldI—rey +=* € 2 Upris. z ia rma se is <1 + te 


, pet 
sors sexaa sft was 


eee ine ites 8, Bi hee 4 


wie 
' 


; Sez ° 7 ate i 
= aa aay. , —_ SC eet @.5) a me 





that in cases of conflicting evidence, credence should be 
given only to the latter ... His works ... are full of 
linguistic material obtained by this method. As an example, 
Wwe may quote, primarily, his Romanisches etymologisches 


Worterbuch ... 


As a consequence of Meyer-Libke's methodology, the volume contains 
a number of dubious etymologies which may be contested on various 
grounds. Moreover, Meyer-Libke was not interested in listing the time 
period in which the cited evolutions ensued; he merely listed the 
before-after correspondences, generally without stating his sources. 
Most of the inherent drawbacks of the REW are generic in nature 
and may not be categorically eliminated. Instead, individual cases 
must be scrutinized, and a composite statement compiled. On the 
other hand, any tendency towards biasing the results in the direc 
tion of particular developments or word-classes can in principle be 
eliminated categorically, by establishing a procedure of random selec- 
tivity. In gathering the data from Italian, one word was selected 
from each of the pages of the REW, thus yielding a sample of 814 
words. The randomness was achieved by first establishing a set of 


categories which would have to be excluded on methodological grounds, 


a 


and then picking the first word on each page which met all the criteria 


of acceptability. In the event that a page contained no acceptable 
words, two words were drawn from the following page. In such cases, 
the first word was obtained by starting at the top of the page, while 
the second word was obtained by working up from the bottom, thus mini- 
mizing any bias which might be introduced as the result of including 


derived forms occurring in close succession in the REW. If three or 


od odie nebo 
20. Fiat sie . ‘oe fa 
afqnecs aa 2a cbarlten eid vd 


























enisiron auiov orld aypalonoiiien 2 ' old -eyaM to < SOUPS aa 
: oO 

Wolter no Sefasinco Sd. Yen silt as.ipofomyte : eectdb 30 sedan 6 

' 


anti aris ptvett ni batdexstiat Jom asw oldiniarsyat tqyosto ebowse 


< 
aris betel yiovem ed ;hevene enottulove batto a dbldw i Bobreq - 
~monmioa ais ent tate socttiw yilaiase ,Soonsbtoqeenxa xette-omied. : i 











gousten at pjisedes ois oa axts 35 asbedwesb trex ea jo 3a0M > J 
esas fexbivebni \bectent .betsnimiic vilsobopetes od jon veot bas 

eis nm .befrgrnco jas vatete atieognuco 6 Das bes itase od Jeu 
erih ett of esiues: aft pnirasid abrrswad yonsbned yas beset aeifo 

x ed sigitoting al aso eceasin-Diow 20 einengofeveab asivo.iiaag Io roid 
P -gefee mobs to subsoorq 5 poinderidetes ya ,yiiscoricpsts° eee 
| bajoaise esw brow ono ,raiigit mort sisb ait piitatte: al ives + 2 
B18 o sigma s priblsiy suds .WaH edt fo sopeg as ie rose moe a ake 
f 3o doa « pnirletidsizo Jari? yd bevaitios Som svormcbnibs oat noe 
le aru Lecieniobarisem 0 perros odo hig Sivan dotitw connate, in 


sixetixo ots fis 24m Adtriw SRE bas no baw ees ae beeen 


Sidsdqsoos on Banietrco spag 6 tert ana aial - ek 2 
#028 roye aT -Spsg miwoltoa orig ped Fai eas et 


‘okie oes ot 20. go it 35 sate Baad ae 


t 


pueus eats - a ae nowt mH oa itow Ye me ehh teow Koa k aitei 
ne - pe) : Ate 4 _— / 





ark see = ‘ y:.* 


ea eee n.d 
wo aia ar vm a ie = =i Sa 


* "a 
> 


more words had to be drawn from a single page, the sampling alter- 


nated between the top and the bottom of the page, thus ensuring 


maximum diversity of entries. This procedure necessitated inspec 


tion of most of the REW's entries, resulting in a reasonably compre- 


hensive overview of the phonological development of Italian as 


recorded in this volume. 


le Pa Be 


following 


dl) 
2) 
3%) 
4.) 
ot) 


6.) 


1”) 


8.) 


om) 


10.) 


ely) 


12%) 


Eliminated forms. In drawing the sample from the REW, the 


classes of forms were excluded: 


Forms showing no reflexes in standard Italian (Tuscan). 
Non-Latin words. 

Words formed with productive suffixes or prefixes. 

Vowels occurring as part of grammatical endings. 

Words of fewer than three syllables. 

Vowels in hiatus, unless the hiatus was removed at an early 
date by a process such as metathesis (nucléus  nocciolo) . 
Words in which a vowel was absorbed through palatalization, 
labialization, or similar assimilatory processes. 
Onomatopoeic words. 

Words whose cited etymologies show a high probability of 
morphological or analogical contamination. 

Vowels in verbal stems whose position relative to the pri- 
Iary accent changes within the verbal paradigm. 

Proper names. 


Demonstrably late borrowings from Latin. 


(1) Meyer-Liibke often gave the reflexes of Latin forms in a number 


of Italian dialects. Since the results reported in the following 


De 

























. : - as 7 
: 7 : = ite 
Pp A im Weg 
iy . ba : nee a> 
a4 és 


| _ “weds eaniiteiaas REG aula | 

‘pabtians eukt .opeq oft to. masdod. mpage 
cegani batstiessse subeootg eid | eixhng Yo tn: emt 
or ee fal 


-gsgmo yidsnoesar 6 mi oe isi frrcerr, asians os wt , 0 7 
ee as 
arte 


: 7 


as miietI to IN Is. Paclbeneet 33 30% 
Gee 


4 emi Lov eists a | 


= 


7 


ott. Wat art mort ofonee silt piteiD a ec} eetcd, DeemimetA mt 
pay ' sBeholoxs eyow peer Jo aseesio pa 


,(asoevT) meiiail Bisbaste mi eaxo itor on piiwade amof (f° 


: | | abioy sited (S he 
/ . 4 . . / } Ph 7 
ive a) itetg to eaxxitive svisauborg djiw hamot abiow (.€. i ‘tS 
- —  ,@eoiins iso cisameip 10 Tis 7 2s PEERS aly (bh - 


/ .eeidsliye saadd neds sewer Zo aia (.€ 


: yltse ne Js bevomes Bs cuss i eft eesims ,2tripid om alow (.2 - 
7 3 (@loison < erie) éteottstan es doa oxg B Yd a be oe 


= - 
se. 


.~ Weiten tints eq fowot badrosds asw fowov.6 do irtw ck atest bit: 
sé 
,eonassoig wroiclinraan “relimte 16 oiaastttiat Li 


ae 7 
_ -2b1ow oieceednse® gn 
; To ‘hilidedora rp li 6 wore. esipolamygts beds sea 2baoW i.e “4 a 
7 ks , ° (ics riinptnoo re cs 10 BAG a 
| | = 


7 


- 
“iq ect ot oviteiox'noit teoq s2orhw means Ledzov, at 






° 


; et 7 
: m@plbssagq fscre, sad nicthiw ee cecal i‘. a. 
pea ae ae 

 rittsl mot soaluaerd <a 


tian 0 aati attec 20 emer nie 
emia ox) 





' 










wa 


as inaiaty ? 
ae 


se 















oy 





Sections were drawn from entries for standard Italian, derived 
largely from the Tuscan dialect (but see footnote 3), no citation 
from the REW was used unless it contained an entry for "Italian" 

or "Tuscan". 

(2) Words of non-Latin origin were excluded, since it is impossible 
to determine with any certainty exactly what form they took when 
adapted to Latin or early Italian. This is especially true of Arabic 
and Germanic words, and somewhat less so in the case of Greek, which 
is better documented but still retains areas of considerable 
uncertainty. 

(3) Words formed by the addition of productive prefixes or suffixes, 
such as dis-, ad-, were not included, since in such forms the presence 
of the morpheme boundary generally alters the stress patterns. In 
the case of prefixes such as ex-, which became firmly affixed at an 
early date, such exclusion was unnecessary. 

(4) Grammatical endings were not used to provide data on the evolu- 
tion of unstressed vowels, since they were nearly always preserved, 
with a greater frequency than would be expected from the phonetic 
environments alone. This is trve not only of final vowels and vowels 
of verbal endings, but also of the theme vowels of infinitives, which 
indicate the grammatical class to which the verb belongs. It has 
gererally been claimed that the preservation of final vowels is the 
result of their morphological function; this matter will be returned 
to in Chapter Four. 

(5) Since only the evolution of relatively complex stress patterns 
will provide useful data regarding atonic positional hierarchies, 
only words of three or more syllables were included. Contrary to the 


remarks of Meyer-Liibke cited above, there appeared to be no difference 


26 
























boriveb \naiisdT barbrese\46e Be 
nottetia' On. (€ stomoo? ee 
Maples? a tine one memset soneriey : 


sidisecqnt ei ti sone habeas bea niset-oan 30 ance MSY 








hs tay aes tat yidSexs Sates yas die snimmeebre ie af 
a SidexA Io arxt yllsineges st etd .nstisdl votes x0 abiead at nae Pix q 
. “po fond 0 vec = ni tr scene Be i ma i ay 
' SStisrebianco to esos aristet Lise stud badaamyeach zetted at nS 


eaxitiwe io eenile:¢ svtophosg, 1° roltifis of yd Beceog. ae {€) x : 
2Of182SiG ort wre? roe oi sonte .bebyisai ton sraw “be \~Bib a6 dove. ‘ q 

, fl -.amadteq eesie ot eigtia yi feicasp isbrved arsvignen ad to : : ‘ 

: | fe te becitts ylert? amesed roiriw .-—x5 es dove eacttong to saad enti 

| .yiserscemar caw nofeufoxs rove deb yise t 

; “wuhovs oxtt 00 Bish ebivorg ot beau Joa suv, epeibite inotsenmena. (8) > 

; Wee wwestg nn yitsen sxrow yes sonte velowo beeeexs anu 30 mR . 

pitenorig oct mn heilidieine 20 bluow asc \onscbeid adsorp i aii oe 

- eiswov bus elowov Isnt? 30 vino jon sro el enn’ .soois atnemmosivne- a 

, fbitw ,eavidinitai te elewy -smeit ait. to oes sud vepnitine Lacie Yo 

r ean 31 -epnpied coy ort doédw ct aeni> feoktemer ott sapitiak ii 


ms 0 sine Slt Yo eottarames eo ns a, 
Hacrustsx sd Lice asta arnt ES an ae 


amattea eaort? xakgneo viovise ter to moist 9 | 
| hectic e tit ad “a . > J ee 
: bag - see a 

| ont os vse ssiarapesdes sais oe om oe ae a i ar igeh 
spnaesthib on oxi ot be s eed overs thi~tayeM Io 
oe. foe , waa: 


Cee » eer, 





















a a 


oe ‘# SRY Lae ‘oe ee ee as 


54, 


in the evolution of pretonic and posttonic 'peripheral' vowels 
correlated with the number of syllables intervening between them 
and the main stressed syllable. By including only words of three or 
more syllables, a much greater number of individual cases of parti- 
cular environments could be presented within the stated limitations. 
(6) Vowels in hiatus generally do not behave like vowels in other 
positions, Since hiatus combinations often become diphthongs or re- 


duce in other ways .© 


AS a consequence, an unstressed vowel in 
hiatus was not included as an example of an atonic vowel, although 

a word exhibiting a hiatus might be included if it contained other 
vowels in allowable positions. 

(7) Vowels which were subsequently absorbed through palatalization, 
labialization, or similar processes were not regarded in the tabula- 
tion of unstressed vowels, since these vowels fell prey to other 
processes. / For example, in facio > faccio >[ fatto] the i would not 
be included as an example of an atonic vowel. In most cases, hiatus 
combinations were involved, so that this category may be considered 
aS a special case of (6) above. 

(8) Onomatopoeic words, including words denoting the sounds made by 
animals, were not considered. Sometimes such words undergo the nor- 
mal phonetic Pe ries niene in a language, but :it may be that their 
evolution is retarded, due to the special function carried out by 
their phonetic pes 

(9) In certain cases, the etyma posited by the REW were too far from 
the end results to have suffered only the normal phonetic evolution. 
In some cases, the path of analogical or morphological contamination 


9 


is obvious,” while in other cases no ready answer presents itself. 


All such words of doubtful etymology or containing obvious morpho- 




















alow, ‘Saradgieey” sinmssety 
mers raeneted etic: pean 

x0 sett Jo auow vin pitts font ya -oidetiye & 
rss ces tune | 
anoictesimil basete gett vee mel 
tedio of elaww agli t ountiad Jon ab yl isxedap acetate (Bh 
were to epaortidgib ameced asst aroman.cano scl eonre 
nidlowo heawontett 2 tis, Soasipsanoe =. 2A. agen tio ab a 
feuottis fswo vinois os To elgnexs: ns 25 babetiont tpn asin easel : . 4 






sito bentstacs +i Vi beboloni ed Jripim eats s pctsicccee Baow . _ : 
| - anoisiecs eldewobls: ni alow . % 
F Moijesiinisleg devours bedroads, ylisipaadue ston doivine elowN (tT) af = : 
f | ~phudst i ad babrisysx jon sxew 2ee2eoo1g 3slomte 2 ooitks Ustdsl . 4 
| saiio of yorq Liet afewoy seed sonce. .Alowov boaeansenis Ye oaks 7 q 
a Jon Biaeoe i oid [ SAT disat < inst mi yalgnexs: sot Y cogesoorg | i. 
| ~wenéte tel Beas gecm nt . Lew Sinots ms: 20 olgisxs AB as balaalopt.ga | 7 : 
| berasbienco sd vem yiopaiso aft ser o2 pevionni ore nae 





7 


; | petits (8) $0 sso Istzage 6 86 
_, 
‘i yd stan — ot priszaneS ebacw Peniras tense bey ooaaban, 7 


. ~on ot ogsebia show rhu2 eenitawe Jpoxchioncs Jor sxsw vale 


te 
: Lies sent ed yom ti tod ~sptopant & es cross tee oisenndg tam _ My, 


YE ito bain sottontt fekseqa oi 6) eB me - 
se 


+ 
ee 


he D akt a 


\ pert eR cot Giaw WH oct a Bhiiecy amos ate esaao it id 
ioe 
; Po 3 2 7a) 
meni 


| pied aa po ie ae v4 a ie py: | 


- 
i © 


7 : 7 : ’ , 
: age on | iS + ewe 6, 3 i 


# 





logical contamination were eliminated outright, since they suffered 
intermediate changes which are not fully documented. 

(10) Since the point of the statistical analysis was to study the 
evolution of atonic vowels in well-defined and consistent positions, 
it was not possible to include instances of vowels in verbal stems 
which altermate between stressed and unstressed position within the 


10 This often resulted in the exclusion of verbal forms 


paradigm. 
except the so-called 'strong' infinitives of the second conjugation 
in which the stress falls on the stem rather than on the theme vowel 
of the infinitival ending. In other cases, however, available evi- 
dence indicated an early shift of stress away from the there vovel, 
and such forms, if meeting the other requirements, were included in 
the calculations. 

(11) Proper names were not considered in compiling the statistical 
data, for they too appear to form a special class with respect to 
phonological modification. ‘Tbponyms are frequently the result of 
derived or inflected forms, combined with cases of borrowing, sub- 


strata, superstrata, and dialect mixture. 


While of great interest 
to historical dialectology, toponyms often present insoluable ety- 
mological problems. Names of individuals, perhaps in view of their 
clearly referential function, seem to enjoy a certain prominence, 
and their original pronunciation may conceivably be retained, at 
least for a while, conflicting with the course of an otherwise regu- 
lar sound change. For these reasons, names were omitted altogether, 
rather than incur further indeterminacy. ‘The number of Such cases 
is small and offers no serious impairment of the overall results. 


(12) The most difficult question encountered in any historical 


study is the separation of 'learned' from 'popular' forms. The 


22 






















baad Yortta ot 2aw afaytatis 
vention itetatarncs bras suntan ai atone dinade 
errata Leciroy ni. i 0 seonsd ition: ot oictimeog. 2 
orks ekets iw qpbrivsa.b SomaoTT Ra aie beaeene nested oe 
aor isdrev 30 optenloxs ‘ertient. bothiresn medio aid? of, 
nokIagutim patoea ot Io navisinBint "proe’ Bitte a Se 7 
lew anad ent no cart xoriders mot] ae no ells | anende ona seiaabitk a b 
“ive sfdsiieve .xzsvowol \esee0 terbo Al - DIL EIS fovis.tn ties ate ae 


7 a 


Pa 


a 


law seit ot mor? yews gesrte to sabes Yites os ec 
ai betufodi sew ,ctnensxigpsr radio oft paiseen ii ,amet rue bas 
~ - 

_ = , . 7 > ¥ 

PS). es 


) »s ' 
. : Pi =e 
lesttelista act oniiiqnon ni bedsbilenco. tar stew eames regord ay ae - 


‘e yu - 
a Av » 
oo Josgeex ritiw eesfo Ls ivege 6 moi’ ot issags ocd vers ot ees —— 
> 
etn a 


30 Sivas otf yCtinsupes? sts emyrogd?’ moctesftinan ive ieetemedy 
~tue .pritworyod to eseso miw bearidmos yeMragt bedoediint amis. 


s2eqatt> tasrp to olny ii .dguudiim toeied DAs cea 


oat Vie Ssfdeufoerté SOS REAT medio ae peSaeee, sen 


shedt to woiv. at egsrtteg atiesbegeeie to emit 





i Sonscnimoig misses 5 YyornS o¢ mace sortoeust sf 


sins 
: 


‘3s ,barnisse: = cd ytclev fsonco: yen ae - bse 20 
as oy 4% = 
“upset seiwierio nn Ie seine ‘sat tid cus Ras Lor 1 





- toritapod is ‘Boctiino. sxow eemen seroassr ae j 


Bess> fou is recimun sxft' ‘Wosna Irs: 0) ri : 
ey, eo 


lasts om 


generally employed method is hopelessly circular, since leamed 
forms are regarded as precisely those words which have not undergone 
the "normal' evolution expected of words of the language in question. 
In the REW, Meyer-Liibke noted the evolutions which he regarded as 
Buchworten, and in the majority of cases, these are forms which have 
not undergone the 'typical' Italian evolutionary pattern. He noted, 


for example (1967: 14-5): 


Sono d'origine letteraria le voci colle seguenti formole 
imitatte: BN. elteeprotoniica; *perche “dall'e- lat si-aspet= 
ta i. Reprimire, repubblica, declinare, regalo e simili 
sono d'origine letteraria, midollo< MEDULLA ecc. d'origine 


popolare. 


Meyer-Ltibke also brands as learned '1'-o- ai =0107-ola\ in such) toms 
as cupola, discipolo, etc. Grandgent (1927: 37) also lists as 
'Latinisms' developments like sebaceus >sebaceo, secare > secare, 
micantem>micante, etc. When considering such statements, it is 
tempting to regard the appraisals of learned versus popular status 
at face value, since much more consistent results may be obtained. 
HOwever, the circularity inherent in such a procedure is unavoidable, 
and consequently, in the data reported below, no attention was given 
to Meyer-Liibke's classification. The only 'learned' forms which were 
rejected were those which may be attested, either through Meyer-Libke's 
writings or elsewhere, as relatively recent borrowings into Italian 
of Latin terms, generally of a specialized nature. The result is 
that the true percentage of learned words in Italian probably falls 
between the figures cited by Meyer-Liibke, which are undoubtedly too 


high, and the figures reported below, in which no 'learned' words 


56 











a re oti ie cs 
7 - ole 


oz 
as iy ; 1s “air es 
led Miche tenors" 


a wil 
ties 1 sempre! a Yo son i Rao an 


ws bins et nin eS. Bane aoa 
ges! dottw atriot Sap seats congo Se Yetsopan sad 
Gatos SH orraddsc vaso oti Love rissa ‘Isolayt” « 


so 


slomiot itnetpsa ef foo Sor na a 
“dagen ia .tel =o! Lieb, Stiorsg tenis at os: ea” | eis Pe 


Hi inte 2, QLSgss ee ecg cm fr ‘eo | 
etiipiro'b .o0s AdTUCGM > OLInb in stteredtel _—— ae . 

‘eh oh “es 

“HO bua od “elo- Qlo- 1b o-M1" Sartor 2 beeke oats Silene | | 
) es asail oats (ff :TSOr) aneahasyd ods oe ae 


‘ SIBQIE < ; gsose wanndse< eusosds2 ott! 


a eee 
E adtite teluond auanoy bersal 36 alseisrggs' 1a teen alpine 7 


a: ».beniside. sd ven 2+ fess Fist aleqos' scm samt Gonie Sete 2988S : 


Gili benaeietis et suibecorq s pee ai Srsctorint ¥ ' 




























Tey ip -esw roisnedts on ,woled amie 2 i685 ‘add ab 

grow rhtiw amok ‘hsertsel' vino ofr oi eS 
ee 
ee fisiisdt edn Spniwomod tsa, uieaaes a 


et tives sn etasan ene 


— ee 


nae + 


have been excluded in principle. 


2.2.4. Preliminary results. By applying the sampling method des- 
cribed above, a corpus of 814 forms was arrived at. Since final 
vowels were not studied, in view of their potential function as 
morphological markers, and their almost universal retention in un- 
modified form, word-final vowels were not tabulated. Also, since 
conjugated verb forms were excluded from the discussion, the number 
of different positions which could be studied was further limitea.¥ 
All together, sufficient data were collected for the following pos- 
itions: initial pretonic, initial pre-pretonic, second pretonic, 
posttonic penult, and intertonic. While this list by no means ex- 
hausts the possibilities exhibited by Italian words, it provided a 
sufficient amount of variation to permit a more detailed study of 
positional hierarchies. 

The preliminary results arrived at by tallying the vowel devel- 


opments of the test words are displayed in Table 1: 


second Ist 
pre- pre-pre- 
tonic tonic 


120 lasyil 


# of 14 
example 

36 

6 


altered 


Table 1: Italian atonic vowels 





a 















7 
“J 
’ 


a “A sly : ad : 
~eab boritem paiignss st pay ae ed tueen 
Lit hs 


font aonté .te bovis. fas rot BEB Xe 
me 


—.) 
as noktorait icicoiclig Phot soyeiy Ht cit bout y: si 
“ras ni poltneder Cinesov ing tears ‘stot ee exonuer 


sonie ,oala Lbedelieded Jor eiew elewov (f os 


Techn ati .soleenouih ott mort’ bebu loxe or set , Ame 
ST becimis weit? eew Belixte od bitmo ahead ‘ " 
“80g priiweil fed ott xot mare ECS Siow 6385 noite yaaittoget tin an 


7 a _— 
solmaterig bronee \oinessigrarg [eitiat ines sig 


tek temo es 
“49 aiisaq on yd isi! airtd, atic .abnottednt Bas ~~ * 
a bebivorg i \ePaow Bile yd Bediditke cots cts in| 
+ Yiucte balists> ooh s +imreg of moltstsy 20 races Skate 





a 
ie 





There is no a priori test of statistical significance which may be 
applied to these results to determine whether or not any meaningful 
figures have been obtained, but some preliminary observations may 

be offered. In keeping with traditional accounts, initial syllables 
appear to be consistently 'stronger' in terms of resistance to syn- 
copation or loss, than intertonic and posttonic penult syllables. In 
addition, the word-internal pretonic syllable appears to occupy a 
weak poSition comparable to that of the posttonic penult and the 
intertonic. The percentage of vowels retained unmodified is more 
Significant than the percentage of vowels simply retained, since modi- 
fication of atonic vowels is, like loss, another measure of diachronic 
weakness. 

There seems to be no Significant difference between the initial 
pretonic syllables and the initial pre-pretonic syllables, while the 
second pretonic is considerably weaker than the initial syllable, thus 
failing to confirm Meyer-Liibke's statements about these positions. 
Further discussion of these results must, however, be postponed until 
the following chapter, where the data are broken down into finer 


categories. 


2.3 Hierarchies in Spanish 


2.3.1. Introduction. The preceding section has sketched the struc- 
ture of the diachronic phonological hierarchy which may be inferred 
fram an examination of the developments occurring in early Italian. 
Based upon position within the word, atonic positions may be clearly 
ranked according to their intrinsic degree of phonological strength, 
measured by resistance to syncopation and modification. A strength 


hierarchy of this nature may be utilized to lay the groundwork for 


58 














noideilye ipitias SaCTIOR 
—ftye.at ontstelaen, to. aia at: emeet 


é es 

i .aeldsifye tfune; Sroodtteog fas shen & 
& YRIDOO ob 3 (SOqUE ofdel lye sihateny 

edt bas jitreq nititidtaeg eft to Jar of atom 


exem ei beiiifemw hettister efewov to coated os 


5 





_ 


+ihom sonte ,banisis: yiqnie elewov to opsirisomee oe rats 


wets 
Simorineib to sywesam todtons . 320! attl ,2t 2iswov cine, te a 


* » oad °y@ a> alt 
‘beiste sis msemtod conersiiib jnsoi! inpie on sd a3 S nese sesutt =e 















edd efirw ,asidsaffye simodsiq-sig [sttini sft Bas as 
gimt jaidsfive isisini sect astit texhow yidszebtemo ak. 
-an0ks 20g sesdés dwods einanedst= <‘eAdiimisyst sick 


figou Mircesiilan a!  Yevewod .tanm ativest seont to woisare 
santt odni owobh assord ote Bt6b oct oradw' wedgado: eri 2! 





i ) ’ 7 7 / a 


sounds ‘ortt Bertatside earl AS Gi tah 


para ts i eines Sa 
ps 






os 





29 


characterization of later sound changes affecting the Romance lan- 
guages, and as such may represent a theoretical proposal with far- 
reaching consequences. Because of the importance to the incorpor- 
ation of a rigidly formulated theory of phonological hierarchies into 
diachronic linguistics, further evidence must be adduced which points 
to the existence and viability of such hierarchical behavior. In the 
case of the Romance languages, the phonotactic heritage stems from 
Latin, and hence it is most natural to search for comparable configu- 
rations among the other Romance languages, in order to disqualify 

the null hypothesis, namely that the observed hierarchy in Italian is 
the result of mere coincidence. 

The utilization of data from other languages can serve as a test 
of the validity of the results obtained in Italian, and may also 
function independently to provide additional information on these 
languages. In particular, one may hope to discover analogous, if 
not identical, hierarchical behavior in same of the other Romance 
languages, Since several of the languages bear a marked resemblance 
to Italian in terms of phonological development, and any such dis- 
coveries will thereby strengthen the arguments used to document the 
hierarchy which has already been isolated. In the remainder of this 
chapter, two additional Romance languages are surveyed, with the 
goal of providing a more comprehensive view of phonological beha- 
vior among the western Romance languages, and in order to strengthen 
claims that values of diachronic strength be included in historical 
descriptions of early Romance, and perhaps in synchronic descriptions 
of at least some of the moder Romance languages. Insofar as the 
data converge in certain directions, the parallel evolution of the 


three languages may be traced, and the possibility of encouraging 








cn srt eta 6 tes 
aetog Hoist benubbe, off dann somebive Hedy ai »eoiseivpnki © 
ot nit sotenited Teokdonsielt iiwe do yiitidniv bes eonateiae ed 
apes anate spatiied oisvesaadg ‘rt seg i i ‘ 
isis efdstsamsc 162 ‘doxege Ge fsraten feon ef 3i tee Ra A 
PasdeipetG-os tele th ee ee 
@f-msifesz al wivbtelt beriakiia off saci yLemart ,aieactouyt pe 


iu he 
sonebinaies saan Se << nts : / 







aa 










i. : 


32ed 6 56 sree NES eepeupirs! teddo: gus atsbh to moijesitiiow sam 
cals yar hos  msifey ni fonletdo etiveer edt io yibiher edt 0 i 
ceois mo Aitearotet Ieavitifits shiveng of ySnshaeqebat colton’ 7 





ti \exteedens sevoorif at sged yan eno “telooimegq al -eepeppaed 


ote _ ae 


consol ‘aerito eft t¢ sme fri. soitversd isoifbasis hi io ip 7 





sonsidmses: hexzem 5 sed sSeaEAL eds 30 tereven: sunte ,eepsupanl 
‘ 
-e86 Aove'yns Bos diame fsotpofenoda’ io enmsd wi nei fesT og as 
“, 
ort Snemoo of hse aa oct nemiprsise re Tite a i i 


a 


aitiy 20 2s8niemer ort oT .bassioci asad ybsexte aed ibiedw 4 oul 
ort dtiw .bevennme ois aapevertl conus Sana on ga 
-rited kepipoioncig to wiv oie baiaiataaie Siail 6 Liha is: a 


neriphests ot iabro ni bets ere eeaee eB solv 





; F : — 7 : WA O34 s 
4 —_ . ——— a sic oammed 
7 - 7 ag : , 7. 
7 15. ¢ ; 7 4 : 7 a ~~ 
: a oe r oe ay 
&s an rs ri} eh bic pes ; 
: 7 ' —e = see les 
c a = 7 
é : 
s , y > 
= cor o > : pr; . . 
p ve.) 
> ab 


60 


the search for mutually shared phonological characteristics may 


be viewed in a somewhat expanded perspective. 


2.3.2. Spanish as a test language. The major Romance language 
bearing the closest superficial resemblance to Italian is Spanish, 

Or more appropriately, early Castilian. Spanish, like Italian, exhi- 
bits phonotactic characteristics suggestive of hierarchical behavior, 
both with respect to word position, and with respect to the individual 
vowels; several quotations attesting to this fact were cited in 
Chapter One. Among other investigators, Lapesa (1968: 61) has char- 
acterized the relative position of Spanish within the Romance family 


as follows: 


Dentro de la Romania occidental, unas lenguas se muestran 
mas revolucionarias y otras m4s conservadoras. El francés 
ha llevado hasta el Gltimo extremo las tendencias generales. 
No se ha contentado con suprimir la acentuaci6én esdrGjula, 
sino que, debilitando toda vocal posterior al acento, ha 
generalizado el ritmo agudo ... En cambio, el espafiol es 

mas lento en su evoluci6n. Domina en 61 el acento llano 

o trocaico, intermedio entre los abundantes proparoxftonos 
del Oriente y el ritmo oxitono del francés; incluso conserva 


la vocal postonica con relativa frecuencia. 


The modification and syncope of unstressed vowels in Spanish was 
largely confined to the earlier stages of the language, representing 
the gradual emergence of a well-defined vernacular from the spoken 
Latin of the Iberian Peninsula. In gereral, word-internal vowels 


fared rather poorly in early Spanish, tending to drastically 













spear l EDM stOfso onl 





Abineqe ef ast ia + See seaants = 
sists Med Lett earl t ade je ferclitasd yixse ;yiat 
\Woivarled feotioxsretd fo sv itesppve ex ctebisdosterb oitoat toni q atid 
7 Leabity tbr i oa ad Josgesy dgiw bas ,norticog brow os sooper site 
@ ’ 2 be ak 5 J a Le oe | “= ~ 4 al 


mk Betis stew tosi eid of phitesids eacttaiéaup fsrevse 


—Yems eefi ({[9 :800I) seeged .erojaprtesvar rerio paron, sho ™ ee) 


yiime? sonsmesx! ond oindtiw deiasqe to nmoitieeg evitaler ont 7 


\ 


¢ ee AP ee ‘ 
22 esuprel eso ,, istneitooo sins ef Pn oe 


de AD 

Am asixjo Y¥ asitsookoulover aim we 
ae 

-Geletoiep eotoishast esl onsitys anitlo fo stead obavall ag 3 
at zi ; rs. | ea): 
oe Slu(isbes sOlosutricos sl timtique feo obstretinds. af 68 eit) |» 













' id 7 
6 20i1s)20q Lscov’ abet obmatilideb aD onde me 
wusgqes fs ,ofdmss nd“... obune omis fo = cece 


ous.1l atnsos Io £8 n9 brian aignlove ue. ns ane al E 
' ‘ 7 . _ 2 pied 
: ~ ete a ; - - ' v he. s rat 7 
SONS EXOTBGOIG esunsbnitis gol sins oLbenxed ne iso 


avisenioo cadioat + esonsit feb atottxo, ont fo ¥ eres J 


er 
5 fonrxmenk, ayitelen. nen roinitocs . 


— i é 


S6W deitsaqe ni eisvov bovnataity” ‘a Sqcory e* Bis noises ae 
2 » — . aw Ve 
Er iwiagatqes sspeteaal xiz tO or isis ett a) beanies 


2 


nein. srtt moxk tad cal iW Sone en e 3 . 





« Pe - 
»9 2 iT 
Vee 7 o _ 
' f* 
7 ’ ° ' 
7 of 7 a 
a _ 2 
4 ~ in 7 Zl 


61 


diminish the number of proparoxytones found in the language. Lapesa 


(1968: 61) has noted, to this effect: 


En los romances occidentales el ritmo del lenguaje tiende 
a concentrar la fuerza expiratoria en la vocal acentuada, 
detrés de la cual no suelen tolerar m4s de una sflaba. 

En consecuencia, ha desaparecido o se ha reducido mucho 
la acentuaci6n dactilica. En cambio, los romances orien- 


tales conservan gran n&mero de esdrijulas. 


Spanish appears to be a promising candidate for an analysis of 
the sort carried out for Italian, since the language has consistently 
exhibited a differential treatment of stressed and unstressed vowels. 
In addition, Spanish shares with Italian the advantage of not having 
strayed too far from Latin in its phonotactic evolution, an advantage 
not enjoyed by French. Moreover, Spanish is well-documented in the 
early stages, a critical requirement when attempting to deal with 
posited etymological developments on a large scale, and the patterns 
of evolution have been fairly well established. By comparing the 
early Spanish forms with their respective etyma in Latin, it is 
possible in most cases to trace the route of evolution, with the 
result that tabulations may be attempted which will not seriously 
suffer in accuracy or completeness. 

Focusing now on the specific vocalic evolutions of Spanish, 
one may reconsider the remarks of Menéndez Pidal (1966: 67), summing 


up the history of the Spanish atonic vowels: 


La vocal a es tan resistente que, atin inacentuada, se 


conserva en todas las partes de la palabra en que se 























pals sternal [Sb auth £9 saletaeBloss ssorater Sat # 
vebectrete Isnoy si cs eixateziaw si14adt sl asrineom si 

sod sn alee eee shoo in aa - 

cobain chinehex af seo obtoersqnasb af a 

1 agixo esonsmex 0k ,ofdtiso nc .sotlisosb miinarsiteos aft? ?- df . 


% 


.aslwiiabes ob orann nesp MeviIeeAcS aoiet al 
? 


vishedtatanco ase a eit Gonte eiletl wt 20 boise Se ae 
salowy hoaoonjeany be beeesiie To taantacet isitnexstRib & mannan: 
ptivad jon 2 sestusvbe af astisst pihiw esunie dainee eottiibs nt ‘a 
spednevie ms .nonhilovs 2isostonorg ati al nite. mort ist cog boysrda | 

ad} ai bedesnumeb-Llew ai dainsgé isvooxoM .poneri yd heyorms gon ¥ ¥en 


_ 


dtiw Isai od piivgnedts noiw tnariivpe: iscitino s 2opete ims - 
; 


re | 
gimetieg sit bos ,oisoe agref 610 agnangqo! sab ino rpolompte bactieog cea 


oi pairsgmo YS -bedetidetes Licw ylrist asst ever names to a 
- gk ot ,abtel ot emyets WLIO ES asians dt iw enor deieeqa — a 
aris nites casei fo aduct ony sos of esano tacu mt Lc * . ae : 
‘pend biee hei fii 1 shy achopiethe: o8 vibe" atpiseidst sets 3 . ax i” 


: -eaeateiameo 70 ‘oeaioos mt 8 ren 

ye ; 

; Heine to anshisiove ob isoor oftineqe srt Wt Chae ned * 
pate . ene a ee mn & 


et | laa Aca cu xk¥ 


wi 
=| 


se » dt up edmec te ie ode | 
x a My : Bs es 


- a) eo 
pe eee ae 
28.2 m8 Ca ze GA. a = ss 
7; oe 


im 






| 

J 

a4", Pay 
_ rae 





oa 













7 
Satie in 
. 7 


halla. La suerte de las otras vocales 4tonas est& de- 
temminada por la resultante de dos condiciones: primera, 
su colocaci6n respecto del acento; segunda, su colocaci6n 
en el comienzo, medio, o fin de la palabra. La posici6én 
inicial es la mfs firme, la qv da mds resistencia a las 
vocales, la que m&s asemeja a la acentuada; sigue luego 
la final; la vocal menos resistente es la medial, que se 
pierde frecuentemente, lo cual se explica bien por su 


cualidad de relajada ... 


Tuming his attention to more specific cases, Menéndez Pidal notes 
(p. 75) that 'la post6nica interna desaparece en general, debido a 
ser vocal relajada'. This however, is qualified (p. 77) by the ad- 
dition that !la A, que se perdia en latin vulgar, ... dej6 de perder- 


se en romance'. Additional information is added (pp. 77-8): 


Se conserva la I posténica en romance cuando se pierde 
la consonante oclusiva sonora ... Fuera de los dos casos 
anteriores, las otraS excepciones se dan (aparte las 


voces cultas) en voces semicultas. 


Speaking of word-internal pretonic vowels, Menéndez Pidal adds the 
following remark: 'La vocal A se conserva siempre ... las otras 
vocales desaparecen por efecto de su car&cter relajado. Ya en latin 
vulgar se perdia la prot6nica después de r'. 

The above remarks give an indication of the extent to which the 
phonotactic evolution of the Spanish atonic vowels paralleled that 
of unstressed vowels in Italian, for many of Menéndez Pidal's 


remarks quite accurately pinpoint the results reported previously for 






























abioracg si peer res ‘dale pear" il a 
: as! 6 cisoaigider elm sb Suprel amr alu of 20 tolnk 
opserd amic iabarinens af & steiéae atm aup af vestsooe 
se aip ,Jexben al no shaviaiean lous Teste ef {Lent a4 7 
a hg notd cok lege ee fam of couemednermant Pon, op . 
. ... SBpisler ob ebifaww 


a 
aaton Ishi sabnnstt «29860 bi inege ssom a noitnedds aid. were os - 
: ie 7 


5 chide \Istansp ne: s0s1sgsEeb BLESS eoiqieeeg si' sot 48T @). iar it cat 
. he ott yd (US .q) Seeileup ai \tovowod aif? = ‘abstefer Lo = a oe 


. 18breG ob Ors6 ... .teplny nidsi no slhisy ce ap A sl’ Sets mbib 
2(8-TT .qg) Sabbs. 2k abitamotni [acisheA . ‘Sonam agen” 


) oy 
ehisiq se obrisim sono as soinéseog T sf syissnco * . 
‘ 4 > oe, 
goes eob aol ab siau3 ... fA1ome BULAdEIO SHHENOROD 4 a 
aur. “ iV 


esl edt6as) ABb of agopingsoxs emito iki eee a 

.aedionimss eecov me (ascii <r ws 

| .—: > ee 

od} ship Lebit xsonans! .efewoy cinoJonq Decrdei-brew 20 paileage — 

aestooaal ... sgmeie svxsdnes 92 & issav al! annie | 

Alisl no s¥ .obsinier 19103160 we Sb aimsts senna: 

2 86 adogesh sichioweeie 

ot} doiiw cd deta ort. Io soiagothab 1% wi hae y 
jet polalisxeg plewov sFaDts relepie ORY Ap £1 oe 

; er ishlt nme? 260 ie rape ie | 


¥, : 


pearance 


ny > ae 


7 a + oT hs, ¥ 
of) eee ore a eer 


63 


Italian. This degree of correspondence between Spanish and Italian, 
added to the methodological advantages signalled above, facilitates 
a more thorough analysis of Spanish. ‘The groundwork is laid for a 
comprehensive statistical survey, designed to put to empirical test 
the anecdotal remarks found in historical grammars of the Spanish 


language. 


2.3.3. Selection of the corpus. To ensure maximum standardization 
of results, recourse was again made to Meyer-Ltibke's Romanisches 
etymologisches Wérterbuch to obtain a sample of Spanish evolutionary 
developments which could form the basis for a statistical analysis. 
The class of forms to be categorically excluded from the Italian 
data is also pertinent to the study of Spanish, and consequently 
the same eliminations were made from the Spanish data. This is es- 
pecially necessary when considering words borrowed from Arabic, }3 
of which Spanish contains a great number, for these words, while 
eventually attaining a configuration indistinguishable from that 
exhibited by words of Latin origin, underwent a series of complex and 
often untraced phonetic modifications before being adopted into 
Spanish. A good survey of the scope of such phonetic mutations is 
found, for example, in Entwistle (1962: 126-34) and Lapesa (1968: 
95-110) ,-? where the unfeasibility of including such forms in the 
following analysis is vividly demonstrated. Greek words fared much 
better in Spanish, since far fewer modifications had to be effected 


15 but in the interest of accuracy, these words 


to regularize them, 
were also eliminated from consideration. In the case of Spanish, 
more verbal infinitives were eliminated, since Spanish merged the 


Latin second and third conjugations, destroying the independent status 







ey 


Z g. mee aaah oo sitar bx. 





ants Baw Hae aed a 
amiaritinnt eveds bo! tsnpte 29 aint 
5 132 bist et PronbeRse OAT ot 0 BE yi 

















” 


Jest Sent tiget os tug ot erpieab «yas sorrert 


¥ ¢ bw #0 
dining eit to - aia teobatath re — 
























yxsnoitiuiove deinsge to Slgnse & misido ot 7 
wieviens leniseltese 5 1ct etesd oft miot Sieoo ‘ce acide Pea) 
SePIAST adf eG bebtioxe yi feolsopstes sd Gt: anti Go east air 
vlinenpoence Drs “ie bres to vinde xt at trent oels ei eh ts a 

See at eid? etch sebeeqe off mov? eban a6 encidehimite same ei a 


Sp 





£ o 
OL nidexA mock Beeorrod ebuow palsobieros mere {BRASH qlisioey J 


_ 


_ - - r q 

A elinw ,ebitw seeds sot .xsdmun Jserp 6 antietago” fleLnegs small tl og 
u : ’ = 7 
ts jadi} mort Sldereiupntteihbai softsipiingo & primeeiss Lisieneve a 


Boe xalquoo 3o seizes s tnewxsbay \nkpino ated to abrow vd beciditrs 
ogni bedaoks puted sicisd encitacitiban o Domai bsosisag asdio a “a 


St atoitesum oitsnotg cove ic ia sit 3o yarawe Boop A deine 
G0CL) sesqet Eris (te asf ser) eltei i Signage: si on 
sf ‘ni anmot shai paiiuiort 26 erse tie ames ont ote *.¢ aan 7 

thin died AABG Jood hesitate WIRE ak ora cadedln 


7 


paca co bettie 





of the 'strong' verbs still found in Italian. 

When dealing with Spanish, Meyer-Libke's REW becomes an even 
less ideal research tool than for Italian, which was one of Meyer- 
Libke's primary areas of competence. The volume contains a number of 
errors with respect to the etymological sources of modern Spanish. 

In those cases where an error was suspected, the form in question was 
compared with the results to be found in other works of Spanish his- 
torical grammar, such as Entwistle (1962), Menéndez Pidal (1966), 
Lapesa (1968), etc. The corrected etyma were used in the tabulation 
of the data. Although all the entries were scrutinized, the total 
number of apparent discrepancies was relatively small, and conse- 
quently the overall accuracy of the results has probably not been sig- 
nificantly altered by these manipulations. In a handful of cases 
where the available resources were insufficient to resolve a problem 
of etymology, the form in question was bypassed in favor of a more 
straightforward development. 

Once the categories to be excluded had been established, the 
same random sampling method utilized to gather data on Italian was 
employed; namely, the selection of the first acceptable word on each 
page of the REW. This sampling procedure yielded a corpus of 814 


words, which was subjected to further analysis. 


2.304.) Preliminary results.) As in the case of Italian, the only 


atonic environments studied were intertonic, posttonic penult, second 


syllable pretonic, initial pre-pretonic, and initial syllable pretonic. 


Final vowels were not considered, since their role as morphological 
markers may have prevented their loss or effacement in Spanish (cf. 


Chapter Four). The uncorrected data for the five atonic positions 


are given in Table 2: 


64 

























neve 6 B4moood WH <'oAtiI—rg 
~1syal 20 900 sew doisiw nd enenst seahascill 

90 nednun & enkaince steior of? ssonstesgngo 30° aoa 
dekaage «riche 4p) saosice thatsolsinge sit oF tosqaan iw ome 

asw moitesup ai moi ott padosqane esw tor9 9s este obimo Simi a 
“pis deinsgé to extsmw isto Mm boot ed os ativesr ‘oat iy Beeman 

. (aaer) Eabis ssbb (6895) slsetwtath es rises vistiissp Deo biad teasbite 
moitsinds+ oft ai boew eipw sate bajosrzco Sar eth (Baek) pro 
lgetos ort becinitnrsoe exsw eatisaS ort fis spurt TA stab si 20 
~eeico Bre \ifeme yisviselor asw asisnsqsralé sais to ers 
~pie aged ton yidedoug est et ivesx arit to Yosiioas [isxsyo seit ‘trae aoe 





BS685 to ipthesti s al .2noiteluginem saat yd barat Is 


7 


moldeng 8 sviowes at tnsioftigant srew esoiosss alds{teve oc eapacee Bi 
7 1a is 
| sfon 6 Yo tovsi ai bseesayd asw nobteayp at mc ort  polaings 2o 


, QP | : 3 . obenreb snevactiplaate : —_ 

orls parla fededeo rised Beri Rebisloxs sd ot eobxopsdso oat —_ vd 

iy aes 

eew asiissi nm stab silsee. od “Ess tlity borgen clients mobesx ase 
| 

des cp brow sldssouioe testi ‘4% to to notineiss any’ viet eretae 

. $18 Yo eagne © bebe? ewibboous ent es hi te aie "iy aaa sa 


s eieylens seta Baa at A BO ae 


vino ai scLas) to ses ort ai eA a lvegs vig tm lord oe 


baense tinned ptepesece: \pinedaed at exo baibede) bdncmmntume % 

‘pinging ofdsitve feisint Bre, alibi Sey ‘isBind « | Liye 
isotyniodduom 2s’ sfox thal sprite’ y asa H 

te) si «ava a | 


I IO ee 


Si 
* 
we) 





65 





Table 2: Spanish atonic vowels 


In this table, the atonic positions of Spanish may be clearly ranked 
along a hierarchy of diachronic strength, based both on overall ~ 

rate of retention and om rate of modification; from weakest to strong- 
est: intertonic, posttonic penult, second syllable pretonic, initial 
pre-pretonic, initial pretonic. Furthermore, the two initial posi- 
tions are of nearly equal diachronic strength. Both the posttonic 
penult and intertonic syllables demonstrate a much greater weakness 
with respect to resistance to loss or modification, while the second 
syllable pretonic (i.e. word-internal pretonic) vowels occupy a posi- 
tion nearly halfway between the posttonic penult and the intertonic 
on the one hand, and the initial atonic syllables on the other. Taken 
as a preliminary indication, therefore, the results reported in Table 
2 strongly suggest an analogous hierarchical behavior pattern for the 
Spanish atonic vowels, a pattern which, however, must be further 


refined before any conclusions may be drawn. 

















my = oe 
tei jal itrese 


“aq l-—aig-erq | Ig 











P| Lot 
Me 1 \ ‘ 
j } —-— 





- _ 
” £ 
r o Lae 
sane + 
: 
afs 
° fe 
E bs * fa2inbac.to enc 
+ e's nary : i Worse 
ee : ~ fn Be 
: a es 3 os ta -& at 
—_ ° fw 
SSxnBow Hk. + OST Sores Ps}! 
. 
ee be >¢ ete} Pie NCOs 5 
moteiq sicdsiJ bneose tin 
: ; ~ np rt bape Ail 
Imes a ent ,STOmaiI Ts 
7 } - 
«> a~te aed ——s 
tA wt - - as P< —— i ml xt 
i 
a 
i ra 
° - ™ ok “ 5 = - crixc 7 pee Zz 
13 - 2 wo LS TC x cs {five od a5 92 a ‘cae Pete ta 





= 
alicw ,.aoaitseoiliban +o eeol of Soneseiasi oO} Joagear cet be 


. “ pa r si Fs ii a 
simon afewov (oirtotora Lantetoi-biow .s.5) ataaten afds Liye: 


f 
ci eH brs tforieq olpxiteor xi ie taxi yswiled yitsen noLt 















to of op esidieiive sieote Isstini om Bos Geen amp-of 7S 
styogex stines: oft ,syotexzsid ,nolsothar yrenimiferq 6 
- ‘ ~ re 
2 pat ae 
STIEQ. OL iveded [so llors15 oH BSOROLEAS sesppes yipaosde 
; 7 a 
7 \ ; ¥ ® + 


iniwt sd team ,12srswod fbirh a iin tae Sw pinot oS f date 


¥ 


owen ad yer’ sroliions wie = a : 


CY 


2.4 Hierarchies in Portuguese 


2.4.1. Introduction. In order to lend additional substance to the 
argumentation presented in the preceding sections, data from one 
additional Romance language will be presented for discussion and sine 
lysis. By comparing the data from the three languages, a more 
general formulation regarding the role of phonological hierarchies 

in diachronic developments may be arrived at, and more substantiated 
statements may be offered regarding the incorporation of such concepts 
into the theory of historical linguistics. The choice of an additional 
language is not an easy one, for all the remaining Romance languages 
present methodological obstacles. The various non-Tuscan Italian 
dialects, while providing many useful observations, are insufficiently 
documented and exhibit a great deal of intermixture nich would render 
untenable any investigation trying to separate them from data concern- 
ing the Tuscan dialect. Rumanian is also unfeasible for considera- 
tion, for several reasons. In addition to the almost nonexistent 
documentation for earlier stages of the language, Rumanian has been 
strongly influenced by the neighboring Slavic languages, particularly 
Old Bulgarian, with the result that it is impossible to consider 
diachronic developments in Rumanian exclusively within the context of 


16 French, on the other hand, eroded atonic 


the Romance languages. 
vowels at a very early stage, and presents such a complex and inter- 
mixed set of developments as to render any accurate etymological tra- 
cing nearly impossible. Rheto-Romance shares with Rumanian a nearly 
complete lack of documentation, aS well as a great deal of influence 


by neighboring languages. This process of elimination therefore 


leaves for further consideration only Portuguese and Catalan, Romance 


66 
























«it at sonmiedve Iisnoitibbs Emad oo seb at eae 
ano - mom sish ,emisoe7 piibidae att ac batnsesmq anisace 
“S05 ons ankeagoeih a bstiseerqg ad iw seaumnst sores 
erat 6 er sexy) Sig. murat Ag? ads prltisguco ya 3 ner 
esirixstoi | snipolonxiq to1esor, ort phibaes coiegtirmsct fexease : a 


_ & 
ne 


batettossedre sucm fos .ts baverss ad yan asnenol aveb oinoe Eb r 





 etqsom=o dose Io noliszoproonb et vintbnisoe beientd Sd yam @ 





fenoisibbs os to sotem edt .acLttainoentl fecizadeid to \proatia att cack ‘ 

aepstone! soem printans: off) [is tot\ Seo yasovis tom ak spaagest 

| ‘ neifistl apoeuT—-non ehoriev aif .aslostedo Les tpolebormiant snsesig 

x Yisesioittven: sis ,anitsvicedo Inteay yan palfiverq slime jeri ; 

ve tebnex Siu? coflw-omcxioreint to [seb tserp & tidtsike Sas Dacrsmuoab “a 
(0) =ifteneoo stab mort moti scsxsase of privat soliserisovnt vets sidsnaiay 


f -, “tebiance yol eldiassjqu ocis ef msinamst, .josisib macau aa Rot 2 


jnadabrnas jeans att od moivibbs ai ~.enpessr [erayee ot fold 

7 @ esd netinescs! .sostmnisti 10 sSoe7e Ff~re f : ; 

7 need sed neinees! .spsupnsi att to = ~sifass 16h no Dhaccemupab 

yiuskoitieg ,aspsvonusl oivai2 pntsodipisa scot yd Ssonec iia vipat 
tehiemo oF sidiazeami ei +t tad¢ tivesy aif dow he frepiet BiO 

to gxeonoo St nimtiw yioevre: fos laa rg ca Snenyoleveb > ingastb 


‘ obaate Beboro (fami zortio edt no Pa of -eoprupas l Bite sas 2 BS 


i o7 
: her "a 
oan | 


Yiases 5 _nBiosadt driw esisde Si hah ech “oldieeogt ) 
soneulini to Isab soot 5 BS Low 26 Sausemeit fe Ral | 


isin. bas xeicnos 5 dove etnsesiq Bos \spaie yiiss yisv B 38 


“ext Isoitpolomeo Steumos yas tSinet os 25 oe ito 3 


is 





67 


languages closely related to Spanish, but which also enjoy a well- 
documented linguistic history of their own. Due to the methodological 
difficulties associated with establishing a large corpus of Catalan 
data, described in Chapter Four, Portuguese was chosen as the third 
language to be surveyed in this study, subject to the limitations 


discussed below. 


2.4.2. The choice of Portuguese. Portuguese derives from the same 
Ibero-Romance proto language as the Spanish dialects, but from the 

earliest times, Portuguese has exhibited developments which justify 
its classification as a separate Romance language. ‘The primary rea- 
son which has been ascribed to this differentiation, aside from geo- 
graphical isolation, is the relative scarcity of Germanic influence 


in Portugal. ‘Thus Williams (1962: 11-2) states: 


Probably the most important cause of differentiation [of the 
Romance languages] was the intensified stress accent, super- 
imposed, as it were, upon the Vulgar Latin of Italy, Gaul, 
and the Iberian peninsula in varying degrees by the invading 
Germanic races ... Additional Germanic invasions ... brought 
about further intensification of the stress accent, and with 
the rise of the Romance languages, syncope of the vowel of 
the posttonic penult and the intertonic syllable became a 
general phenomenon no longer limited to the special posi- 
tions in which it occurred in Vulgar Latin. But the addi- 
tional Germanic invasions did not reach the territory where 
Portuguese was to develop. Aside from Visigoths and Suevi, 
no Germanic tribes ever settled in this territory, and Visi- 


goths and Suevi left but slight traces of their stay. ‘The 











aroitss imil oct soeidue ve me 20 ror 
pampe 7 fe; 


ar 
<a 







deme oct mont aovixeh sastromxot seeped 39, spies sae pes F 
edit moxt sud edooisib deinece ot! os Gonopist otena sonemd-cade a 
@isaut dirty esranqolsveb bed idirxe asd suarudigt semis elke 

ser Yisming arf? csupna! sntieil atexidie’s és cpiaholaieaelin eae 
~cop ment abies agtisttnerettib etch od bediwes, nsed asst seb AB 7 
praxtens Trt ohne) So vtiotsoc ovidsiox ort ai .nobtedoet feoiagem bs 
rowtme (Sql :S800) eimstbh iw apar oe | a 


) 


























“ 
= 


ey) 


ed3 20) aottsitnsrettih Io sauso tnssaog team ort vided q a 


Fi | | 3 : 
a ~teque .Ins098 eeotte boitteneini oft esw [eepaspnal sonamaAn ‘ s 


‘ ius) .yistI to ala aul oft noqy , Stew $f 26 _becognt — 

peiisynt ort yd aserpeb gaia ci aloeninsd aghred! eit bne = . 
a Ytipuord ... anoiveyni oinamicd fenoitiopA . _ eeOar Sirened 7 

> tte bas .tneovs sesxte orit 30 sy ieee an ae 

ay | Sc wal a 


= amoed ofdative olnstiaini oft Bos stinsy obstinacy Slt 9) 
; nage bt ao. Si, i 
~ibbs orld Jus Beebe ih eusc see wi _ 


in 


ania variate ee oe a mcd Lenoit 
on @) ae. i : 4 
vse ia i * id 0 is eee weer 
wind) bee yecogbmed eich 72 eis Ot 
ai 2 ee oe fare r : 
oi? .yste stadt 30.260 O8: fa tud 322! “ie 4 


en ree 
e) : - 





68 


linguistic result was that there was less stress accent 
than in other Romance territory, and accordingly, less 
syncope ... The separate Romance which developed in the 
south among the Mozarabs was entirely free of Germanic 
influence: hence the especial fondness of the people 
of the south even today for proparoxytones. ‘Thus while 
certain characteristics of Old Portuguese, such as the 
fall of intervocalic 1 and n, arose in the north, the 
resistance to syncope, a far more distinctive charac— 


teristic, was stronger in the south. 


Entwistle (1962: 278) remarks that ‘Roman Lusitania extended so far 
into Spain as to include Avila. This dialect was eminently conser- 
vative ... proparoxytones appear to have been unusually tolerated’. 
Thus, perhaps as a result of decreased Germanic influence and also 
of political independence from Spain, Portuguese developed into a 
unique language, which in some ways resisted certain sound changes 
affecting Spanish, and in other ways underwent changes to a more 
complete extent than Spanish. 

As noted above, Portuguese frequently tolerated atonic vowels 
in positions where the vowels were lost in the other western Romance 
languages. This is particularly true with respect to the posttonic 
penult. Williams (1962: 52-3) has characterized the conditions 


under which the posttonic penult was lost: 


If the posttonic penult was e or i (Cl. L. € or i), preceded 
by 1, m, n, or xr, or preceded by c and followed by t, it 
fell in the late Vulgar Latin or early Portuguese period ... 


this change took place before the time of the fall of inter- 





















jneoos Reoxte zeal anw By a : 
ae 5 a ceaniaal vce onamA ion al . 

~ ine ae ee 
otasme to sss3 yLortins | eat pansy > ae 
~ eh gaey og 20 beer act fstosges att coned 
4 . slidw enrl ‘sahcomotneend 302 ysbos nave innate 
’ of en cone AERTS BIO io mideiratzersto erases: 7 


ore  fitton ait ni s20i5 1 5ns £ ‘pi leovzedit 20 List -sail 
: bah v> En ) ; 






v —3fisis svisoniseib sion 1615 SpOMYe: ot sonsjeteo late i. 
~iygo2 orft mi bee ‘a5 oiemet : 

i i 

| 4s} ce bobastxs sinstims! cembf' tac: ottems (80S :SeGk) Se 07 

, - ameands ylissaime esw josisib aint .slhvA sbulont of) as)qeege Game: . 

, * thajexatod yitscvetnn nsed sven od 189098 aonodyNcISgong -.. Witsv | 

eS eels bas soneu tint oinemso beassrosh io titeer 5 oa ageitieg seu 

j ~ & Ginl teiiades seaupitsed ,fiage mori eon sbrsqebrt ino ttOg to an 


—_ 


sgensiio Boude cisiiso hatatacx avew ombe sk doiriw , spsinrish Supt ¥ 
giom 6 ot espns tnewrebey eysw tedto oi hie debacle 2 enisootis 
i einww oinods batarslot iene sesipusiod ,Syods ee 
bi 
, cnet resbeee the oft nk Sect, Bee eats ers aad engi booq at 
' 
 pinatyeog att ac s3egest isiw act yltstio bred te so ae 
7 ae 


anoliibrss sii. poh apts ni. {é-2 a LS a 
ae 





tf i = 


69 


vocalic 1 and n, but after the time of the voicing of inter- 
vocalic [k] and t ... if the e was preceded by m but followed 
by a short n, it did not fall ... the nasalization of the 
penult by the adjacent nasal consonants may have increased 
its resistance to syncope. All posttonic penults which did 
not fall in Vulgar Latin and in which the [above] condi- 


tions ... did not obtain, remained in Portuguese. 


Essentially the same conditions are offered to account for the 


behavior of the intertonic syllable (Williams 1962: 55-7): 


If the intertonic vowel was e (Cl. &, é or I), or i (Cl. L. 
i) preceded by 1, m, or r or preceded by c and followed by 
pet cerelWeineche lates Vulgar Tatin Or carly Portuguese 
period ... with few exceptions this change took place 
before the time of the woicing of intervocalic [k] and 

t ... in some words syncope of the intertonic vowel was 
avoided because of the unprounceable combination that would 
have resulted from it ... if intertonic € was preceded by 
m buts followed by a short n, 1t did not fall 22. 16 in 
tertonic e was preceded by n, it did not fall as the post- 
tonic penult did ... this seems to indicate either that 
syncope of the intertonic took place later than syncope 

of the posttonic penult or that intervocalic n fell 
earlier before the intertonic than before the posttonic 
penult ... all intertonic vowels which did not fall in 
Vulgar Latin and in which the conditions [above] did not 


obtain, remained in Portuguese. 
































. a) Sra y 
= ee 


aint tis 
es pany - 


e wt 


. th - 





7 


- cert 
ind AN Per 7 7 4, © 
cast Sagas ieee aad | 
aap apenas fo 
oft 20 noftasiisenr ori} ogame 
sxc ant on Sees ean eas a 
Bib rite axiuideg cin sa0q ELA aro raat ah 
: sites [evods] ort bide at Bre nited Isplwy nf iistaea” 
-seaaipatien ok Begtanrs <i EN on ja 


= 








| 
gs 





edt 2O% 31tiCoDb ot bertetto ois 2nort longo ans? odd yitsisreaed 
:(v-22 :8a0f emeiff{iw) oideflye pinatisink ortt 20 ~wobverlod 


a 2D) B “<0 ; (E x0 5 15 »19) = e5W fowov Sioodisdnh ont 3 4 J 
: . ; 
yal bowolict fas > yd Dabsoesg 30s FO it LE \e Bebeosag ¢- | a 

sequmatics vitss to nits! stepiov: etsl oxit me ‘Let * ~. ve 


“- -, 


eosig xoot spnsdo aint enditqsoxs wet Motw ... botisgq ae i 
ia . 


Bas [ot] 2iiscovraini to paiibtoy art Io smbt one crotad ; “ 
asw fewov sinotrcynt adt 20 sqoonye ebiGw Smoe atk od - 
bitow tac fisen ich BIdacororae si3 io sound tows _ 
yd babeossq es a binnirredtt 2 ... +i mort baslveex ovat a 
“at th ... Dist sé bib 4} \n Juote's et batch au 
“tog ors es List Jon bib +t 20 ee ae a 


aa weittis aisofhni at nese eit - Ms sheers 
+30 2 ea “i 
sqsemye nsrit astal eosiq Mood nenopiednk ot 
fist n oilsoviednt stsit 16 FLU age es 00 ) 
_stenctt204 oft sxoted ners oekalaie pie a a , 
a oe See ; , - : 
ak fa ton bib cin steal 1 
en a fevedsj pepe ae 1 


Le 


= a i 


q 
’ 
: 7 


2 


Williams' observations, representing one of the most compre- 
hensive treatments of the fate of Portuguese atonic vowels ,1/ pro- 
pose a rigid categorization of environments characterizing the loss 
or retention of atonic vowels in Portuguese. The present study 
sought to refine these observations by offering a statistical 
analysis of the evolution of the Portuguese atonic vowels, which 


could then be compared with the data representing Italian and Spanish. 


2.4.3. Selecting the corpus. As when dealing with Spanish and Ital- 
ian, use was made, for considerations of consistency and faute de 
mieux, Of Meyer-Liibke's REW to obtain the basic data. To a much 
greater extent than was the case when dealing with Spanish and Ital- 
ian, the REW presents obstacles to the study of Portuguese, which 
must be discussed before an examination of the results may be under- 
taken. 

The same categories of forms which were eliminated in principle 
from the Spanish and Italian data were also eliminated from the data 
of Portuguese, for the reasons given in the preceding sections. In 
addition, Portuguese presents special problems which required the 
elimination of various other categories. 

The most obvious, and also most tenacious problem concerms the 
loss of intervocalic 1 and n, a process occurring in the earliest 
stages of the language. Both 1 and n were lost somewhere around 
the beginning of the tenth century , 7° leaving, in the former case, 
two oral vowels in hiatus, and in the latter case, a nasalized vo- 
wel in hiatus with a following oral (but progressively nasalized) 
vowel. Since the date of these processes nearly coincides with the 


date of syncopation of the posttonic penult and intertonic syllables 





- wyobeus od vem etivest oft to: noitsnimsxs m6 eicnad pecaioatb od team 7 


-leliqtonixg ni bsisnimile sxow cOitw amok 2o eo.ciogatso mee aT a 

















> Enoblelist2 & prietto 
feidw ,alewv oinots en ee ap otovs wth to sbeeiae 
srleknwae Mie meiiedl pritnsssrqex sib ont slaw Set cok eet: Seat 


“faa bos rains? itiw oniigeb carl eA cose mat fib ; 

ab stust bie wasteteacs 46 enoitsyebienco Yor Sam aay oe oat a 

kum, 5 OT, 6tsb otned eg nistde ot WER afeiiht-—auh 20 san . 

—KeiT Bon deidene ritiw hits’ naxiw. 5365 edt asia asst snedxe zedso78 = 
rin fw . see eypusiol to viede ody ot Sa ~roneg WL net 


= 
- Mevlet  ' 


- =a) ° 





ate! sta) vit myst Pale sees cna oeis STA Sts asi feo bas carwiet ott not : “<1 





fk .anbiteee pithsosig sdt pr nevi enoessr silt Btoid seeuetso 30 | : 
artt £ ee foxrdw eneidorq fstoeqe ainsesiq. mea . 
+pobtopsdss verte eiokxsy 20 miles 

sit amraoroo mafdorq eyoiosaedg zom aes bas ‘ 
seokixss art mi palicusoe aesoorg & wt bos I 
Ewens storwanoe teol axew nt bas L diod ewe 


269 temic? ors nt ypaiveer 8 undies dame St 40. 


ree 


in late Latin or early Portuguese, serious problems of relative 
chronology arise when attempting to attribute a form either to the 
Vulgar Latin epoch or strictly to the early Portuguese era.!? In 
general, if the Portuguese reflex of a Latin form shows evidence of 
a creation of a hiatus or of a hiatus reduction (e.g. nebtila > n€voa; 
poptilus 5 p6voo> povo), one may safely attribute the form in question 
to the early Portuguese period. However, in forms giving evidence 
of syncope (e.g. *caracXlum s caralho), one runs the risk of total 
circularity by ascribing them to the Latin period, since one is 
then defining the period by means of the definition itself. Ina 
few cases, Such as words listed in the Appendix Probi or developments 
listed in reference works such as Grandgent (1934), Kent (1945) or 
Palmer (1954), evidence exists that these words had undergone syncope 
in the Vulgar Latin period, but in most other cases such assurance 
is impossible. For reasons of methodological soundness, it seems 
preferable to view the situation in terms of two conflicting proces- 
ses, yielding mixed results, rather than as two successive processes . 77 
Consequently, no attempt was made, when collecting the data, to dif- 
ferentiate between words of supposedly Vulgar Latin provenance, and 
words undergoing modification in later stages of Portuguese. ‘The 
end result of this methodological assumption is a much higher appar- 
ent rate of syncope in certain atonic environments than is probably 
to be accurately ascribed to old Portuguese itself; however, since 
the topic of interest is the overall evolution of Portuguese, these 
figures at least provide a complete picture of the diachronic strength 
of atonic positions over an extended time period. 

Another serious obstacle concerns the presence of nasalized 


vowels in Portuguese. ‘The exact date of vowel nasalization in 




























gvidste: to amefdosg suites en 

ante o¢ xeitis mot s atudinite a bok , es 

a 
at © ere exceptions yiass et OF iolagadel > mitel c 
EN pO ee 
| fe 

7 

ysov8e <aithden .p .9} cotroubst aitsid & too adel s Io nokia a 
One 

aolseeup ni ane? ody odvdistis yletse Yam sno , (eyog < cows <suigog 
sousbive piivip wmol ni ,revawoH =-balweg seeped yitss Ont Od 
fetod Yo seit odt emer ano » (dfs <mbsBoene PS) eqcomye ‘to A 
ris +26 
‘ er ‘ano soure (be hie f= 8 | nigsd ort at mor priditoes dimes vd vt hrefuosio 
7 Sal .tigeti noit inress onit to ereeam xd bo risa. ort prinitab mand i 
ae. 

assent ov ob 10 idexd xifasqah odt at hetell ebuow ee doe, 20680 wal ee 
+0 (ape) jxok ,(RECL) trophasw Bs foug etrew: caaeisten ni Begait aa 
| sqhonye.sucpisia baa chow seert tery. eteixe soasbive , (sen weeded 
is sovsuuess Goue aseao sorito deom af tud ,botiseq acsl spall wale eck = 
: >: ; 
ansse Ji ,azenbnuee [saicofobodian Jo enossst 107 atdiceogal at a 
. ~2e0cng piisS exe BL ioo-owt to amet ni nottsictie xit waiv at oldegetenn 
OS _esaeanomg sviereoave owl 25 eas xorpits1 ,ativest bexim ots ese 7 

; 4 
“2ib al ATSB Si tuctitosl foo pee \Sbem esw sqras35 on vibes ; ; 


Ss bas .Sonsnevorq mitai tepisv Utoaaque 20 ebxow nat aden ve 
| ea? .sxaugurtyed 25 aapade saa ne a cote oan gatamaha sao ‘sbrsow a 
| Ateqge radpit row 6 at. soligmess (eb ieotaborten akay te Sestak bas 
| yletscioae ot edd, aioertion bus ones misstiag AL, a in 20 aS a _ 
aa ata. reveldat yafeasi sesieig10% Blo: at nos) visas sy of 1 _ 


Seaett eecsirt Sophos 5 saras 


as 


r 

C BRA | -: ee 

Bie ritpneste eat: ais to omsany, oy ome — “e 

wl ; iicibwoy opshiy fu 
| ‘a 


‘hore . 


Bpepis to a ott a 
eter bones Neier 












co oe a 


i © 


go hi 





ee 


Portuguese is impossible to determine, but it is certain that the 
process was initiated prior to the loss of intervocalic n. As in 
French, Portuguese nasal vowels represent a separate subclass, and 
never undergo abe This is especially true in cases where 
the nasalization is probably of very early origin, but the point of 
disappearance of the following consonant remains an unanswered ques- 
eee Prior to the date of the loss of the nasal consonant, con- 
siderations of phonotactic compatability, to be discussed in the 
following chapter, precluded loss of the vowel preceding the nasal; 
once the consonant was effectively lost, however, leaving behind 
only a nasal vowel, this vowel has consistently resisted syncopation. 
For this reason, all environments potentially leading to nasalized 
vowels were eliminated from consideration, since no useful results 
could be obtained through their Periciocie 

Another issue serving to complicate the process of selection 
of a corpus concerns the raising of unstressed vowels. In providing 
figures for Spanish and Italian, raising of unstressed vowels was 
taken to be one measure of the phonological weakness of a particular 
position, and also of individual vowels. In most dialects of Portu- 
guese, however, all unstressed vowels were eventually raised in 
speech, to a greater or lesser extent depending upon position and 
also upon the dialect; such raising is not always represented ortho- 
graphically. Early Portuguese had no graphy with which to represent 
a raised variant of a, and consequently one cannot determine the 
date at which the general raising of atonic a to [te] took place. 
Some investigators have speculated that this change dates back to 
the earliest stages of the language, while others insist that it was 


only later that atonic a was raised to its present centralized 


eimneadanuatiees:' 


jaan 
1 (oetocan aimtinn » oie ie 


oF 


30 Gntog srit such ani ylase yrsv to yi eisdlorsey ak lommmanian?s 





TKD SAB MENOD isesn aft to. eeol ot Zo see wn od 10fra te 


my 


ect mt bereuneih sd e vi EL idsseqnoo staseheiiilig 30 enoitexshie 


| tisean lortt paitbaossq Iswy ort to Geol bebulbarg aradgpito pattwol fot | 
a Paided privesl ates ied ,t2o0k ala esy tosnoenco si? sono 


moitequoiye beteisex yOnstetenco ead Iswov aid+ \lowow Ieeem 6 vito 


| bes iisaen our prise! Vv riisttns3og etosmacs brits {ts  foRser ait wi 
7 ative Gfiew on sonia \noitsiebience mort hedecimets saw alswov 
. | eS ons ior qpuorit beatside odiidieep 
mitosiee to sesn014 aid steotiignos cat paiviss avess ee 


x “~ Sibivorg nl -elswoy bseaerteny to priaier edt arson eexqtap & 20 

cow eens hearsijen to pataisy ,nsiisst brs feicade sot semuprt 
~tefuoisisg 5s to sanibiesw Ino tpolonotiy ont to sxvesam sno sd Gt aszied 
‘ <atuct io einetalt, joom mI .alewov [subbvibni to osis Bae  iokeieog 


gi beebes vLlisucinove siew efswov boseortasm iis ,tevewost .sesup 


—oeting beipernest eyewis toa et mrieis: rhe ;dosisib Sd, aedat? 
ort sniuisate Jon sao yLinaspeatics Ban is 26. = 

sig toot [¢) ot 5 Bear AiR vane 
@? sad id spre eirit jefe renee 





great oso at amd vilsipeqes ak ais? is snoktnseonee Sebi Sra 


bas soitieeq poqu pnibneqeb tasdixs Jeeael xo 1adasTp & OF a 


treesrqs: ad sbidw igiw yigerp on hed sssuputros yixed. a 


Pe 74 
oy 


eid 

























peat 


» & 


a 


73 


position. a 


In the case of unstressed e and o, matters proceed somewhat 
more smoothly. In the majority of modern Portuguese dialects, atonic 
o is raised to [u] in most environments, and in certain derivational 
paradigms atonic [2] may be raised to [o] a Atonic e, initially 
raised to [i], has now become [2] in most of Portugal, the Cape 
Verde Islands, and the Azores, but remains as [i] in most areas of 
Brazil, although in Brazil the overall rate of raising is substantially 
less than in Continental Portuguese. In certain derivation paradigms, 
atonic [ce] also becomes raised to [e]. The raising process which 
accounts for the raised variants of e and o in the modern dialects 
may be traced with a fair degree of certainty to the sixteenth cen- 


26 and hence is of no direct relevance here. On the other hand, 


tury, 
Since modern Portuguese exhibits raised variants of atonic vowels in 
a number of positions, together with various exceptions and a high 
degree of inter-dialectal variability, the task of sorting out those 
vowels which became modified during the earliest periods of the lan- 
guage becomes somewhat more formidable. The problem, however is 
not aS unresolvable as it seems on first inspection, for, except in 
the case of the still untraced evolution of atonic a, atonic vowels 
which were raised in the early stages of Portuguese were reflected 
orthographically. This change, while recorded in numerous forms, was 
comparatively less common than in Italian, and apparently even less 
common than in Spanish. The raising process occurring around the 
Sixteenth century has, in the overwhelming majority of cases, left 
no orthographic traces of its existence in the language, except in 


isolated cases of Portuguese words' being borrowed into other lan- 


guages. 27 In consequence, a reasonable etymological accuracy may 

























ntocts 2toaisib seanares soar te wixofian at an agli 
faomitsvineb alist i frie \Stiamosivas Seam mi ty] oF & 





ylisisiat .o Dinega. © * fol od Baeiéx odiyem (c) civets amples 
eqs oft \fapiis6% Jo team nt [6] ae , "a , 
- a 
Io esors yeom oi [i] 2s, anton od ,zc1OsA Sc Sn vebnsiel eb = 


> 


Yilelinstedve ei perrainy to ati Efezavo acts iret ne’ dovorit.is ited a7 
yerpibeisg Inottevitst oietis at ,oearmuchio’ fetnanitre® at csctt eaoi i 
ADitw eaanox1g pietee ait, ...([9},0g basr6s eanopad oie [3] oiais 
efoaisifb mteham sii at oO Bris 9 to, einsixsy beeis: eft tet eimmocs 


“80 iasetx te oft of vtiistiso to esrpsb wisi s rfiwi bees sd ba 


7 a ef a 


fried tadto ar nO) . oie! naeeeiee tosrib oa to at sone as - 3s a le 
& A a 
a elewow ainods 2o einsitsy Sseisy atidinks seatpedao! meebo ene : 


dpiti 5 bis anoisqsoxs evoissv dt+iw tadtepod amidiseg 20 Sas Ua 
; } 

rary _ 7 

geevit $i pridsos to dash odd ,yilidsimsy fasoet Sib-raint to esupeb a 


ie otit to eboiisq tectisss odd paiswb battibar sms ped fbidw efewo 7 
@t sasveworl wsidoig off. aldabimset enon tsesmoe aamoned apene 
® Sqeoxs ,to? ,solineaent teri? no emses SF es aldsvloesirm Bs Jor. 
alewov cimeya 6 ints Io sotiviow| BeosrSatr ieige! ‘add 20 ‘eano ett 19 
Deine! tet ori sty Seaupusios Bio) i yinse oft me beaker sw ite ie - 
eaw ,amict aso ni betrcost ar bebe: vopasdo eke baencie a 
Beef move yisnsisqgs bas wositetT nc somo ene ss a 
> | ge bauets primo secooig priteter off , Sen 
#ief \sanso Yo etrobsn eaiatertaravo st} snk si Ef ¥ 
nt +ogEKB seers ort ni sien we a at Sh 
erm Lsnigatangserl acc 6 


‘ 


be achieved by considering the spelling of the words in question. 
Portuguese also underwent a number of isolated but frequently 
quite regular changes which must be taken into consideration when 
tabulating the data on the evolution of atonic vowels. Most common 
are cases of the vocalization of a consonant, which generally created 
a diphthong, which was then reduced to a single vowel of timbre dif- 
ferent from the original. The process of vocalization of syllable- 
final 1, begun in Latin, has continued to the present day in many 
dialects of Portuguese, especially in Brazil, but it is generally 
not orthographically represented after the earliest stages of the 
language . 78 In general, any cases apparently involving vocalization 
or metathesis resulting in a change of vowel timbre were discarded 
from the data, so that the tabulated results reflect only the evolu- 


tion of atonic vowels uninhibited by additional conflicting changes. 


2.4.4. Preliminary results. Once the list of classes to be elimina- 
ted from the data had been established, the random sampling procedure 
employed to gather data for the preceding sections was again uti- 
lized to obtain a sample of Portuguese forms. The only environments 
available for consideration were the posttonic penult, intertonic, 
internal pretonic, and initial pretonic and pre-pretonic syllables. 
The uncorrected results of the tabulations are given in Table 3. 

In this table, the appearances of hierarchical behavior of atonic 
positions are seen, along the same scale elicited for Spanish and 
Italian. This comes as no surprise, especially considering the high 
degree of similarity between Portuguese and Spanish. In fact, it 
may be argued that, in view of the large influence which Spanish has 


exerted on Portuguese at various points in the past, the data reported 


74 
































, mls at ce ati ia 
yLirexpest sud haintoes ee ee 
she satnonhsire ini se 
ee eigwoy Did to robwuiovs or} Ao mie oot § 
ipnate yi letiney soit ,Josnod2npo 5 eee 
Aih grdmi> 20 ftw Sfpaies ct beouher ast mew doit’ sna 
ae Qo nmitssilscov to easooig scr -isnipito cs) coed Sek ‘ 
Yass ai yeb Jnsesig add ot Baanitnoo esa cided eb auped Sed va 


ywilexonsp ek 3i tucd .Lisexd nt yits images .s2eupus10F 2° eivaliath © ae 
. AES ae a ca he 
ort to copsate test fuss ocd astis baineasigst yilsotigsrpertaa gon. 


moitesifaoo priviowi isnsisaqs asaso yous , sxsase.ot 8S epeupasl - 


~ 


babisoaib ase Sxdatt Teyoy. to annsto 5 itt paitiness zandttstam 20 
—Wlove ii yirm joss: utivaor Sataludsi ot tan oe {seb alt mort 


. Reprerin piigol ties s [smisibhs yd betidcdains alewov sims 20 aokt 


spiimife ai of eseenio to t2il ent nO easiness sagan feed bs 7 es aa 





-  gewheoeng paiifamse mobasr adj \berietIdetes need bed Bt6b ot moe bas a) 
oo 
¥ -iju nisps asw anorjose piibsoord otit r6i sdeh usrvsp od er 


; atnemmoxivns yiro eff amor seaputs0d jo siqme s steida ad beck 
A Winsywetat ,tluneq sinoiseog ais stew aoitexebieaco i102 ssson | 


Ve aside llye isad enig-s1q Bak pinoteng isidint bas. (binotsxq Semeednt 


if | Shen 7 ; 
iu | ££ iden ai marke ane snobte tudes exit o nie berooesoony = a 
a oe a 

, oinaxis to sotvared leogriorexaid 49 esonsesgis oft sed eit s “a 
pai! - 


| has dated 10) Satizils sisoe ompe, sat onols pene S16 anoiti 
; “dell ads pninxebiere yilstoseés SeHigae on 26 eelio eit y 

uy he viost al jek «shag ceca a 
| wt ier ‘a 96 ans ery 





in Table 3 are merely a reflection of the events occurring simul- 
taneously in Spanish. Such an allegation can of course not be 
entirely dismissed, but in collecting the examples representing Por- 
tuguese, every effort was made to weed out those forms which, either 
through explicit citations or through highly unusual phonetic de- 
velopment, appear to have been borrowed through Spanish. ‘The redeem- 
ing virtue of the Portuguese data is the scope of the survey, taking 
into account a large segment of the words derived from Latin, a 
scope hopefully sufficiently wide as to eliminate gross inadequacies 


caused by the effects of borrowing. 


inter- post- second{f lst 
tonic tonic pre- Diese re=| aa pics 
penult | tonic tonic tonic 
i of 
amp les 










# ‘retained 


peep at =p at 
% retained 
odi fied 


Table 3: Portuquese atonic vowels 







2.5 Summary 


The historical behavior characterizing the evolution of Portu- | 
guese appears to be consistent with the data representing the corres- 
ponding time periods in Spanish and Italian, which therefore further 


supports the validity of the posited hierarchies within the overall 














=Iimte enbracos €3svs oft 

ed Jon Sewco 26 mE statin 

-yod priitnsearge1 eelqmaxo ott peaLtsie 

wets \doite anit seo! Jro Boow 03 pels 

okiaendy ican yids eipeicrtrt +0 arobie i) 

cesabet ott  itetdege dence coed aed ove ob eae Re ae | 

eiiwes .yovme eit To scooe ont ai ets sosupustiot art So rab pr 

| | 6 nite? mort hesy crab Brot ait} 3o shawese opts & trmcops ott 
e eeinsupabsni Bacto stenimifie ot 25 abiw yiinoioetwe yLimeged ‘aapoe: 

it | vniwotod 26 efoetis adit Bema 7 


ess 





ahah . 
ner 9 enemas ie St ate jo 
Kieu - Hak ae a pore 
"iia 9 erie or pedi oi eee stew rit 
é 





76 


diachronic perspective of the western Romance languages. Portuguese, 
like the other two languages surveyed, may be described in terms of 
a metatheoretical statement, based on the facts of diachronic hier- 
archies. Such hierarchical indices, further bolstered by parallel 
developments in Spanish and Italian, provide an additional dimension 
to the synchronic description of Portuguese, and help lay the sup- 
porting foundations for a more inclusive theory of phonological hier- 
archies, applicable to a wider range of languages, and potentially 
add a new element to the search for links between synchronic and 
diachronic descriptions. It seems that the similarity between the 
Portuguese data and the Spanish data further supports the proposal 
of diachronic hierarchical processes in both languages. Further 
discussion of these results must await the inclusion of additional 


phonotactic factors. 


> — it e 
ge AA RE 

a 4 1 A 

= Pi | Ae iP 

\Sesute 108s = pO AIS 
a a ie 7 i 
te amrist ok bediisoenb oc ‘yam Bi oi cael 
\ t 


mais 32 oimourtyatb to wins? off no’ Pate ,inemstete fs 
: U) ne a 
Isiisteg wd bexstiesod york ,aecibat —— 
























_ eokenemib Leonie, 1 re gamer peti ah ahees 


= 


’ aa 
: a “edt ysl afc Bes 2 eeeQUsIOT 30 noktarsaeb site 


“teid LIsoipofosoric to yost teutoni excrs xt anoitebwct ot 2 utdrog 


ry € x A : as 7 
= : 1* —- 
.  giistinetog bis ,2Spet a | a6 SOnsT’ 18iWA 8 ot obdenkigas oh. sai dois 
| | 4 a ee a 
bos oiftrrfionye naew si oo if rol tidzese2 sit at casemieed BBS 
f " . 


edie ceenried \- 1565 mie ort. jadt emese $7 ' eneisgiztoseb > Die nore 


" 
: oa } Issoqurny eft asxooqque et birt steb delineate Sit pas Bish & il 


J : i. 


a 


sorifinys =. eopsrpast tod gi aczessord: ras 1 Ss oli 2 sib to 
a ; ‘ 
: ‘ ’ 7 ‘- a aie a bel 
Seepisshbs io noreiuloni sit tisws Jeu asia ast beset 30 2 sexsoalh 
an : _ 
.exodos? oi ian 
“+ 
f 
“4 
( 
. 
es 
= - 
g 
— - 
; be 
7 - 
ae Se 7 
- = 
i 
teow . 
- ” 
oF: 


Notes to Chapter Two 


1 This operating principle has been named the 'uniformitarian 
principle’ by Labov (1971: 422-3). ‘The matter is briefly returned 


to in Chapter Six. 


2 When the same results are arrived at by the apparently principled 
interaction of two or more diverse processes, the interaction is 
often termed a 'rule conspiracy'. See Chin Wu Kim (1970), Kisse- 
berth (1970), and the literature which has arisen from the latter's 
work. For a study on the fortuitous interaction of divergent 
processes, see Malkiel (1963-4). The possibility of multiple 
causation as a methodological desideratum is noted by Malkiel 


(19682327);. 


3 On the other hand, dialect mixture was extensive during this 
time period. This fact has frequently been used to discard as 
"non-Tuscan' any form not undergoing the hypothetical 'regular' 
Italian developments (cf. Parodi 1907), and has led to circular 
results. In this study, no attempt was made to discard forms 
on the basis of dialect, except where no reflex in Tuscan 
could be discovered. ‘The investigation was directed toward the 
loss and modification of atonic vowels, a process common to all 
the Italian dialects. Indeed, loss of posttonic penult and in- 
tertonic vowels appears to have been significantly higher in Sici- 
lian and the southern Italian dialects. Thus, failure to make 
provision for dialect mixture may have changed the absolute figures 
to be reported below, but probably did not significantly alter 


the relative proportions. Moreover, comparison with the data of 


TF 










balqronix Sanaa lgtinis vd 38 bevitxs oxe ativeot ouse af a 
ai poltomatnt ol lisa oeppegmnetee ‘ 
“paals , (0T8r) mK oh abt Be . ypstigens oltrs* 5 beams? nego 7 
e'xestel att moxt noebss earl Aotiw omutexeti{ ais Bas < (0TeL) dst 
tnsprsvib Io nbitoeieint antod fackrot ons no vate 6 Act ae 
sigitium 20 Wilidieasy oft (ener [otileM ope 129883001 rer 
fobtieh vd Geom ei martstebiesb Lan bpolobardtan 6 25 solisagso ~ ee 

«(TS « Ae 7 


hod, e. 
eit poixeb sviansdxes 2ew smuttim tooisib \bned xedso a ab E 

















a6 Breoaib of beav need yitneupsrt eal fost aid? .bolreq emis — 3 
‘wefpex' Lenitorisoqyd ott puiopraiay joa mot ys ‘neoedT=aan’ 





weivosio of bel ead brs (OGL fborsa .t5) esaampleves nak lest 7 ~ 






amo} brsceib of shen aéw-Iqnsdts on iviude eleld ot -etivesr 






msoavT ni xefter on sissw tosoxs ,tosisib Io Biased at mo 





- ; 


ott baswod betvenib asw notieplieewnt ont -beroveoatb sd Bigeo, a) 
ffs od noemo eEsvond 5 alewov pinods 2 noksestinitan bs-aeak 


f 


' tii bas jivmsq sinmdteog to aeol ,basbal 2dosietb nstisdl 
| “iit at oat yLinsoitinpiz eee over Keer yer 


a 







78 


Spanish and Portuguese, where the problem of dialect mixture is 
not nearly so severe, serves to verify the feasibility of uti- 


lizing the Italian data. 


C£. Grandgent (1927). 


The use of an etymological dictionary of Italian was deemed un- 
satisfactory for three major reasons. First, since the starting 
point would be Italian rather than Latin, it would be necessary to 
sort through a large number of derivative forms, foreign borrow- 
ings, compound words, etc., which would increase the amount of 
time required for the investigation without appreciably improving 
the results. Second, since only the posited derivation from Latin 
to Italian would be given, it would be impossible to make an in- 
stant comparison with the reflexes among the remaining Romance 
languages of a given Latin form; thus it would be harder to double 
check the accuracy of the proposed etyma. Finally, use of a 
Italian dictionary would eliminate the possibility of using 

the same source of data for all three languages, thus severely 
impairing the standardization of data. The only other available 
Romance etymological dictionary was that of Diez, which is smaller 
and less complete than the REW, and was therefore felt to be in- 
ferior as a research tool. For the present study, the fourth 
edition of the REW, of 1968, was utilized. This book, in addition 
to numerous indices and introductory sections, contains 9,721 


etymological entries spread over 814 pages. 


Cf. Mendeloff (1969: 4, 11). In no case was a vowel in hiatus 


lost, although in a few cases two similar or identical vowels 










aos 
a7 a: 





_— 





_ 





at ourteim toelakb to midorg edt stedw \sesipudeey bas etme 

| Se ARIS 
-iu to Wilidiesst aij yitrov'oy esvte2: ¥ ak, 

t | | r Be Recht Shes K... as > ine : 

ae osm — om nae ae -_ 


os 
ru 
8 

















- ii te 


- 

ania ce lacae hs ideaog eee 
—— edt sonie ,detit .andesex 10fam oourit xo? odstebiee : oe 
oo ytskecoer ect Biow Fi obted msc wertdox sb dt od Binew tniiog ; 
“wormed moisuot ,aurotsvitevineb to asd epist 5B ciguowt Joe 
9o gavams ort sesssont Blyow nbiriw ate \eb1ow “anoggnoo > ae 4 





: 
; -. ‘BAbyougnl yidsioexcqs sjuorttiy op Sepeers ott x02 berchupen emit 
; titel. merit nokta ia bet beog. sid yino sonke .brcoe a ieae ate * i 
Pa “ik. slam at aidiesogni od bivow ti .covkp “ott blow mrt iain ee 4 
: sunentcn aniniemex oft poaom eoxelier ait rtiw a a*g 
: Sidobh ot tebted od biyow tr aon Sate ited Htovtp 5 20 sépeumaek a 
7 . 6 to sew .yifentt .amyto boeogorg of To i a aad toate 7 
pritex % Wilicdizeog a3 stsnimifs bijow \RIBHOLIOND & | 


o > . 


j yisravee sucks , aopanpnst osu {fs xt stab Jo censos s Sane 
efidsl isvs vadto yino off .s3sb. to mitecibubate of prisisand 


st6ifeme ei doidw .ssid to jad =n ine ak 


“ai sd o¢ fet sicieist esw bas .WaH ert cert 
rigxsot |cti vote sesoma aft 36h_ eat cae oi 
neiibbs mi good aittr PT Se 


a et), 


San 





| : e . LOnsaAs 
on a ae _) a | ne Po er) <6 


9 


might coalesce into a single vowel. In other cases, depending 
upon the stress configurations, one of the (non-low) vowels was 
raised to i or u, and then changed to the corresponding glide 

[J] and [w], respectively. This happened in all atonic positions, 
with no apparent connection with the hierarchical behavior of 


atonic positions. 


Cf£. Mendeloff (1969: 11). In all such cases examined for the 
Italian data, absorbtion of the vowel appeared to be a purely 
phonetic process of assimilation, in no way connected with accen- 
tuation, and hence not connected with the problem of hierarchical 
structure. As in the preceding case, these instances were comr- 


paratively rare. 


For the special treatment of onomatopoeic words, see, for example, 
Sa& Nogueira (1936), Malkiel (1963-4), Wescott (1971), and Morin 


CLS PZ): 


For example, the paradigm of the verb fare, whose forms altermate 
with those of the proto-verb facere. ‘Thus, in the present indica- 
tive, fare is conjugated as: faccio (old fo), fai, fa, facciamo, 
fate, and fanno. Some variant of the stem fac- appears in all the 
other conjugations except the future and conditional which, formed 
from the infinitive, exhibit the stem far-. The imperative forms 

of fare are fa', faccia, facciamo, fate, and facciano. Other common 
examples of morphological mixing include the third conjugation 
verbs formed with the inceptive infix -isc- which appears in cer- 
tain forms of the present indicative and subjunctive; for example, 


the present indicative of capire: capisco, capisci, capisce, 
























er e! 


ee er 


| Bile eatmceero 9 of Hage oat ne 90-0 Om i a 
venoisieog oinats its ak Sedeqged eke -farisoogeen «(w] bas 131 "e 
2 wind ie jeouit at wae Sema +i 


«. 


_ 





ort xt bealmake aeeso, doe Lis or. + (i nonEE) Rofebuat 





- | 
: snes + Derssq7s few os To roids 606 a 
¥ “T1805 didi badsanmtoo . YAW On +i “iobet Be6 ‘fo 3 
P int eal oe - 
ie feoirioisisid to meidorg a ri4iw badseieatece, Jon sorts! Bas woe a 
7 ) 
+ 90 cow agonstant stot {sass palgsoc™ oti ni aM Resins ooh er 
; ; Sa 
uF : Ser ovis ia 
: aiqmacs TOI ,952 .ebsow ofsoqedeun@ to, seamaascet se shi 8 i : 
a fitoM brs .(IVCL) sscoasW , (b-£30L) Tenia » (9€2L) siisupo 82 | _ 
‘4 7 
is ; 

| Sf Wee 

al 


edsmiejis emo? seorw ,sis? drev srlt Jo meitiexee' ae ii coal e@ | 


“soiiai tneasig act ni .aurit ~-FiECGt ‘cored ol Bo oat i a “ 
: eves 8 
Sere OB vind . (Gt Did) CEE Bi oust 





ansot evitexagui SAT 2s mate ob sitio « 


2 
peeEDO rato ee ee ,atst Sa Bi 


mers 
presente Se zor tee 
oP eo ha 


lig et sole i 
singe» a 
a ee 
: ae | wee 

isch eal 
a ar 










? oh ete 


< e ; x 7 Soi) , eee eee 





10 


Ail 


12 


13 


14 


80 
Ccapiamo, capite, capiscono. 


Cf. the methodology of Clark (1905: 66), when faced with a 


Similar set of data involving Italian verbs. 


Cf., for example, Eloock (1960: 221). 


Since the greatest diversity of accentual pattems, in Italian, 


Spanish, and Portuguese, is found among the verbal paradigms. 


Patterson (1973) reports on a statistical study of the genealo- 
gical characteristics of the Spanish lexicon. Of the 5,000 most 
common words, representing 95% of a typical Spanish text, 23.50% 
are inherited directly from Latin, 35.24% are created forms, 

and 41.26% are borrowed fran other languages, including Classi- 

cal Latin. These relative proportions change significantly when 
considered in terms of estimated text frequency: 81.3% inherited 
directly from Vulgar Latin, 8.07% created forms, and 10.62% for- 
eign borrowings. As based on text frequency, the relative pro- 
portions of borrowed forms are as follows: Classical Latin: 
84.66%, other Romance languages: 9.06%, Greek: 3.27%, Arabic: 2.33%, 
other languages: 0.68%. These figures demonstrate the feasibility 
of utilizing only words inherited directly from Vulgar Latin as 

the basis for a representative characterization of the Spanish 
language, Since such words comprise more than 80% of the effective 


vocabulary. 


Other examples of the development of Arabic words may be found 
in Elcock (1960: 272-96), and Giese (1931, 1964). Politzer (1954) 


offers some additional phonetic possibilities conceming develop- 
















by 


Ti ast - / . 
| ite Tip es Ct te Wier ad 
ritiw beast nodw , (8a +2001) ohsi2 
; ‘ Pe i ok Fane : 
| . | ah 
>. ro 
: . (ISS 20801) abepkt .aiqpexs wt 4D L 


(neilett mi yanrstied mein x0 \ieravib sesdssrp ors Ba: 
embiosisg Lactay ort. prone rae et vsemcetrsos fis tana 





~olssisp oct io vite Ieoitaisste 6 ro atroger (eveL) wna ef : 





: $acn 000,2 srt 30 oni demise eds 20 itabxatomt® Saabe xq 

#08.£S ,jxat deinsg® feotqyt & Io #e pabinssesion. 208 ome ) - 

emmac? hassor ‘sis 28.28 ,nited ‘moxt (ites | - pa | 2 yi 

‘ | ~teast9 eet font (espeupisl tedto moat bawrnied Sx pee tle 2 

fer yitrsottiinpte sprsna enoitsogorq ov itefex eeod? .ttidet ie : i 
betiverini ¥6.18 :yonarpert txed badamides to amet ab bexehiene _ - 

: | “~ -ge8 @S3.00 bas yenmot betsesio #V0.8 cited sepiny mom yioetb : " 

| “OxKj evitisiet at ,yonsepert txet mo beesd eA Fe acer tae 

: sisal fsoiaest5 sean biice 25 SIS emit pewcintod 20 ~ 

D . pBEELS sobdexA \ATS.€ sxoc1D 820.2 : eepeupnal” soe atl . 

: Wilidfase? ort steiiecomeb acnupit caoct .#83.0 sespel | 


a6 fisted ispit.moxt ylioa id betfvetias ane 
. pee 
A feimsge ait 20 Mniesitxedosseto otigeas 1: 7 


qitedias eds 2) 808 nady eran sere do tam anit ss omar ial 7 
Ai oa - 
: ; a 


; ‘ws 5 









= 
~s 
7 


i % 7 


beuro? sc! yam ebrOw oldexA 20 InSilowss 59°30 /aatgnecs GBD BEY 





UES) 


16 


17 


18 


19 


20 


21 


22 


ments in Romance with possible parallels among Arabic 


borrowings. 


See, for example, Lapesa (1968: 44-7), Grandgent (1934: 64-6), 
Meyer-Libke (189.0a:35), Seelmann (1885: 42-9), Entwistle (1962: 
43-5), and Elcock (1960: 196-203) for data concerning the incor- 
poration of Greek words into the phonetic pattern of Latin and 


the early stages of the various Romance languages. 


For an idea of the scope and nature of the Slavic influence on 


Rumanian, see Petrovici (1957) and Hadlich (1965). 


For other reasonably comprehensive treatments of Portuguese vo- 
calic evolution, of which Williams' work represents a survey and 
summary, see Cornu (1888), Leite de Vasconcellos (1901), Huber 


(1933), Nunes (1945), Moffatt (1948), and Silveiro Bueno (1958). 


For data concerning the loss of intervocalic 1 and n, in Portu- 


guese, the most thorough treatment is offered by Sletsjge (1959). 


Additional information may be found in Entwistle (1962: 288-9), 


Williams (1962) and most general manuals of Romance philology. 


For some further elaboration of these problems of relative chro- 


nology, see Louro (1952) and Otero (1971: 58-9). 


The problem of evaluating the results of conflicting or competing 
sound changes is a significant one, and has recently been discus- 


sed at length by Wang (1968), Chen and Hsieh (1971) and Chen (1972). 


Cf. Lausberg (1956: 96-7). 


In most dialects of modern Portuguese, a word-internal nasal vowel 


81 







; ite an as 


is » a 


Tt if 
7 is 


(Oba =bEeL) snspeaci , (HES meted = ie 


















SO0L) oftatwins , (e-d :288f) rsa «(GE :s008d) ft shes a “age 
hein ie: ie oa 
niet wit painrads stab 18 (e0S-3er see foorla bas (2 a 
a 
bic mite] to matin ssatgihteealisdiaeama le an a a 





-expsupasl. comeneh einktey ot 20: 22938 RIMM, = 





| x 
nm siesuftni civel2 ot to, supten bas sgobe oct 30 soi ms WT aL ey 
; ~ AT - 
(2024) tokIbsH Bas (Feet) inbvorte? sea S aaiane! : 
7 
. 
| —w sesopertictl to at nartperct Sv Fenarergns> yicenosest setlo 30 w i. 
> bis Verte 5 sireesrasr show ‘ems ELLE rite 20 eoksntove obs ‘ou 
7 ' s <3 , is 
. sds ~(f08L) solfasnasesy > etial , (8881) ummoD ese .yRome 
; 
) (6201) onsea oxioviic hus - tahed) $3e330M , (cdf) egant qeeen) T 
i ie | a 
. sirtiol ai yo-bhs 1 vilsocviaini to geol sm RELL EES stab 305 a 
| (0201) -sepaset2 vd baratio bt snantsss4 Apron? Jeont eA yale 
7 


(R-B8S <S8@L) olteiwit? nf beaiot! Sd yam pot dined Lsapi3aBbA 
- <pololing eoneiri 20 olsen Lexenop $2om fr mae 


ee 





ours ovisslen 0 ene idoxg overt to nottszrodsis\ salted eno 


ig (ee 210 @f) onat0 bas (s2er) =uet es - 








be a i 

. Bnictegeme 30 pnttokitace to atleast ARES oe 
Me ieee Senta as Rae aS ea 7 

- (SPH) aed bas {ITEL) deiak bas rsd a 


i ah 


; i 


ag 






























me Bees 3 @ :08e 
: aa é we pha v 
a cy roe ‘7 he at ee eee > va fp 
7 ‘ a OA es y, i ‘ Ni 
= ‘ “ q ra i 4 of 7 ® 
.# Sa Qa): ve wy - wa ra 7 w 
: : “p Bd os : 
“t a aa faa, Se rs Ae 7 : 


23 


24 


25 


26 


2d 


28 


82 


followed by an obstruent generally prenasalizes the following 
consonant, to produce a fleeting transition sound homorganic 
with the following consonant. It appears, however, that the 
actual consonant originally giving. the source of vowel nasaliza- 


tion eventually disappeared in the earlier stages of the language. 


For further information on the formation of the Portuguese nasal 
vowels, and of nasalized vowel systems in general, see Nobiling 
(1903), J. Bourciez (1949), Rochet (1970, MS), Lightner (1970), 


Saciuk (1970), Foley (1970, MSb) and Schourup (1973). 


See, for example, Vasconcellos Abreu (1887), d'Azevedo (1900), 
Goncalves Vianna (1906), Rohner (1948), Herculano de Carvalho 


(1962), Williams (1962: 40) and Moura Santos (1964). 
Cf., for example, Jucdé (1950) and Earl Thomas (1969: 347). 


The most complete survey of the raising process is provided by 
Naro (197lb). In addition, see Herculano de Carvalho (1962), 
Moura Santos (1964), Prado Coelho (1946), d'Azevedo (1900), and 


Révah (1958, 1959). 


Some examples are discussed in Naro (1971b) and in Coates and 


de Silva (1960). 


Mattoso Camara (1957), in reviewing phonetically-based spelling 
errors of Brazilian school children has signalled several instances 
where syllable-final 1 (phonetically [w] in Rio de Janeiro, where 
his observations were carried out), has been spelled as though it 


were /w/, and vice versa. 








ees . si . i 
" ie A ‘ oy 
ae ak 
entwortal ot 5 elacridie tt 2 : = ; 
" oiewsentamort bao asiitegeene 
ort sect one »EIssqas 3 ‘“ ie Bnoert Rahs 
~esiisesa Iowov to so1w0e oe pute vane aL 


spaupabl sit Zo paige ose ae 


One > 
a, hate 
faéen seeogers107 alt 20 noitamict sit -n0 sisi adh 
cee 
gailiccn sce , Inxaiep nk emadeye foun be ie 70 fs tee : a 
{OVOL) sontrckt (es \OTRLY Saran ; eken, elk rae ie Sa 
(EV @L) quiver? Bas (dem et vaio on sues 


gj 











. 
' 
> 





ye a 
ene 


| ni 
(0081) -AbawesA'b , ATESI) trexds Ucienaeene < olanet 1a pay: 
orffsvis? s5 ofte.Lim 30H q (2beF) serio 08D) sesntV aL QOD ) 
.(b3@f) eotned ete! bos’ (0d -aet) | anit ise) 


- 


| we = 


aa 


Ai * 


7 
— 














ae 
~~) 


(VBE 2®9@L) asmorr Inem bos (0c@f) Bowl , Shanes: 10% iD = 





yd bebivexg 2¢ Sesoorq paieis: oft te yore sieigno team on? “ 1 
»(S8@L) Ofisvasd ob onsioxeH ase 80.C} BBS at AGEL) oe | a. 
- a 5 

: Bes , (00@f) obavesA'B , (aber) ori act chert s(baer) aotnse scm ey? 
=~ a¢ ~ (eet AR & 
) / | ay) pr 

‘Dee <etiaod ni bes (4{T¢l) os at bseauatb Sas colqmte ome: s a 


| ae hecsated ‘aa 


pailicage cic et est cp auectrs at ey aD 
esofedeni Lerysx belismpie: 6s aenbf iro. Saas le i? 
gisiw \oiiosl ‘eb of vi [¥} ylfsolsedotg) 1 


(pe 


5 






SBP annette en eee eet ed ho atm 
Fon 
2% heals » fj) - : 
ant aa : a ' 


83 


CHAPTER THREE 
ADDITIONAL PHONOLOGICAL FACTORS 


3.1 Introduction 


A hierarchy of diachronic phonological strength reflects the 
intrinsic characteristics of phonological elements across time. The 
values of strength represent the behavioral patterns characterizing 
each individual element as part of a total phonological system, which 
is in turn part of the sociolinguistic matrix in which the language 
is embedded. Phonological elements do not operate in a vacuum, and 
hence cannot be studied in isolation, free from considerations in- 
volving the totality of the linguistic system of which they form a 
part. An enumeration of all the factors which may have influenced 
a particular phonological development is virtually endless, and each 
specific investigation must select for special emphasis only those 
areas which appear to have exerted the greatest influence. 

One area which can and must be explored in connection with the 
hierarchy posited in the preceding chapter is the relationship between 
the vowels in question and the surrounding consonants. Phonotactic 
congruity plays an important and independent role in phonological 
change, and the action of phonotactic factors must be clearly separa- 
ted out of the data before the action of diachronic strength hierar- 
chies can be observed. The remainder of this chapter contains an 
attempt at separating indications of phonological strength from the 
diverting forces of phonotactic structure. In addition, the hierar- 
chical behavior of the individual vowels is studied, and some obser- 
vations are offered concerning the possible physical basis for 


phonological hierarchies. 



















- “ft _ a 
ett} etoolte: reeaese a coca as nS “ig 


ag? omit aeoxos ainansis Inptpatoncrig: to paler, oieatsmns : 


pais iiesoe ise armadtag issolvertod orks iinet eS 
' (ity \medaye Isoipol snontg fasos 5 to Saas es SoBSIS rhB9 
be is 
eonupist ott roirhy ni xittom alveiconitotooe a So. asa eat ak a ane 
Ses .muosy 6 af stsxsqo don ob atnensico Issipofonods re at ~ on 
=a? enoiserebiencs mixi. sett. .noltsfoat nt othuisa ec ronma ‘e8aoet ; 
8 west gett iskvtor to madeya-oitdinpnil sit 10 WoEtesoe ont pntviov 


Bbeotas(tai eved yan roide erodost a Lis to ne a xsrs aa 5q f 





iv) 
7 , = i] : 
iq foes bon ,caelbae yilkutiiv ef tnomqoleveb Isolpafonoig selEETS 7 
= gtr - - ’ 
esoarit yino steniiqus {sinsege soi Josise Jeu aoiispizaowni ofRipegs id 
i - “a 
: ,sortsyTint testssrp ond bevrexe svaros xBsags dokriw @6or6) 


emt ciiw nobseadeno es Perolcos od aeumn Bais os doitiw S238 iad 
qeeatted aitencitsfe: ota 2i scp pribsossty etit c not toog yiomwtatd 
bitestonaff .etnaseerion pribauorme edt Bas noitseup te 2LSHw ett “ 


fenipollonndg ni ofc: tnsbasdshat bos Insite AE i aa wine, % 


+stsqee yluseio sd tei s1otosi 9: Rosters to, naidos) ins Co 
i. 

—1staki fispnstte: otromikeib To noitiee aid shoted, Sadbiits me 

fi anissaco isiqsr> aint tor sshe Later. ont bariae SG TES Ee a 


eg mond ditprerte ao 





3.2 Consonant cluster compatability 


The figures shown in Tables 1-3 (pp. 60-78) only re- 
present a gross approximation to the phonological development of 
unstressed vowels among the three languages being studied, showing 
the overall rate of loss and retention in the various positions iso- 
lated for study. These figures, however, do not present a complete 
picture of the hierarchical arrangement of environments, for the data 
in these tables are independent of the syllabic structure of the 
three languages. The ultimate result of a process of phonetic weaken- 
ing is the disappearance of the segment in question. In the case of 
the weakening of unstressed vowels, loss of the vowel juxtaposes two 
Or more consonants which have previously been kept apart, or places 
consonants or groups of consonants in new positions with respect to 
the entire word. No known language permits all sequences of consonants 
in all positions, and in particular the distribution of consonant 
groups in Latin and the various Romance languages is very severely 
constrained. In view of these limitations, the syllabic structure of 
Latin and the Romance languages must be taken into consideration when 
evaluating the results of the statistical analysis of vowel weakening, 
so that spurious correlations do not result from failure to distinguish 
between inherent positional strength and the phonotactic resistance 
of a particular sequence of vowels and consonants. 

Modern structural linguistics has made significant advances in 
the direction of characterizing the exact nature of consonant sequen- 
ces in natural languages. The key notion in this regard is phonologi- 
cal compatability of clusters and potential clusters; i.e., the 


degree to which the distribution of consonant groups follows the 





* 


ated eft sot ,einoiner cone ‘Jo Jnenapnexts teotiirsessid os vo susoig 








js a 
a ae 


: al an 
“st yino (8f-08 2a) -E- est ak 





















oo 
£ 





= an 
es vin 7 
to. inamgoleveb Isofpolonorq ait OF MOLIEMXO: 


pekworle oe | 


~oni enol s i903 suoitey oct nt nolanete: bas 280! 20 stiex ett <a 
apatee B gusesuq ton ob evened zee sean? bait 





sactees| 





eit to ondoutte otal ie exit to snehrsqsbeti Par ‘paldet - 
“fetbaw pitanoda to eeapor¢ 5 To Piross stamisin ort » RapaLpRBL souls Ja 
7 4 - 7 


36 seen oft nl .nottearp at. dasmpse oct to sonsisogieai® exit ees - 
F ' : : , . =e : 
cor pre Lowou oj to-eeol velar bsezertanu 29 pole eet om 


pene!g © \tupas J$ged nssd yladeiverq svar ss aah orom 36 ; 
og Josopst dt iw anoitieod wor mi Binpnoenos Io aquexp ‘10 stoneanan we 4 
ainasioencs to aeomsypee IIs 23 imisq spsuprst fworl of  .Balaw a a 
¥@ 


$nempenco to sotindixteib sit tel itiseg at Bas enoitivog Lis wed . 
Viersvse Yisv 2i 2 eniiget sonemoi euolssy ody bas sotted ag aquoxp 
es : 
90 euourte oideliye awit ,andissdimil sasdé to way ai -benker3anoo 
re, 
cisatiey potssrsbiencs ojnt asbs sd tein cular sone ee 
ypainsascw feeoy to etaylens ing teltats ait 20 ae: ameeteshc 
demoniselb of gulist mort tiesx tort ob erobtelrm 
eonatatest oitostornity oft bas dtpdente fen 


” Jednencenco Bris eloway 30 & 


ft esomvbs tosoitinpia shen aati Eee . 


basic Structural laws of the language in question. Anderson 


(1965: 75) offers the following definition:+ 


Intervocalic consonant clusters are considered structurally 
compatible to the phonological system if they are 'dis- 
solvable’. They are dissolvable if they are composed of 

an initial sequence cc-, a final sequence -cc, or if the 
second member occurs in word initial position and the 

first member occurs in word final position. Word final 
position means before a pause or phrase final, i.e. sare- 
thing which can occur in absolute final position. On 

this basis the final consonants of prepositions are 


excluded. 


From this definition, structural compatibility is a function of 

the juxtaposition of consonant sequences in the chain of speech. 

A logical consequence is that, for any given language, the set of 
all possible consonant sequences may be divided into four categories: 
occurring clusters which are dissolvable; non-occurring potential 
clusters which are dissolvable; occurring clusters which are non- 
dissolvable; and non-occurring potential clusters which are non-dis- 
solvable. Anderson and other investigators who have dealt with the 
problem have claimed that during a process of syncopation, the first 
environments in which syncope will occur are those which will pro- 
duce occurrent dissolvable clusters, followed by those which will 
produce hitherto non-occurrent but dissolvable clusters, followed 


by those which will produce occurrent non-dissolvable clusters, and 


only then followed by environments which would result in a previously 


non-existant non-dissolvable cluster. If the structural characteri- 


85 





Pree 


== 


Wer se: ed 77 = ae i : 


| 


_ 


-_ 


lp are 


= picuncn snl: Savings ‘ 



























i 7 a @ 
R ae a ae iy 
I. uw rice ty ie: ‘* yEf 4 


aay so oe 

inst shine iis sna 
cei’ axe yattd 2 nctéye Teptedtoncig ait cd ol 

to beesepro Sax yectd +1 ofideinidnele fue Yeett «eke ert] © a 7 

” ett 2 to .po~ sonsupes Eehiirs 45 sonbupee Isidinbre jon | 7 ne : 

, _ ety Ses notdizoa {sitint ‘brow ni aunee sade BABB 4 i” 4 

font? brow + .aoitieeg Tenia. tecdat etinoe) taco Jack at, ,) | 

amos :5.i ,fsnit Seeing 4o Seusq a sxoted eres naksewor = 7 

©  .ferrisog fenit siefosds) ar ccemotinn ey ene one. 

ets encitiacgerq 40 ajasmoenco Tad ot ated abds a 


em " 


a ae 






So moktomG 6 af viilicitsqmoo (sustouate eee cud eae / 'f 4 
-fioesge to misrb sit ai aconcitps2 $hisepen00 20 anit teogedest: ect” ny . 
90 452 Sa}. Spa sf nevin yns xt tere el somsupedios iki iP 
fasiropedss wet odal Bebrvib od yen espaape toanc@cico nicthdnige Eis 2 : | 
{slinatog aabsicoco-—on -ofdsi/ioaalb ents ride + exaute waermnee | : a 
=fon 918 foidw exedeut> praitusaso isidevioes 15 aw ise emanauto ‘2 


ee a Tweet 


~atb-non ete tobriv erateuio Spitnatoq pruisaisoo-on |B 4 Le b o 

" Ut tS 7 i -% Ae 7 - 

ould rttow shied evar otf, axatsptd re dO: ia foe 

$aci® oft wioktegoonye 30 22so001G 6 ye sid F 
ra 


7 - 
ee are <" 


eww: by aia ag ae 


ce Hino a nl 


ae. 4 


86 


zation Of consonant clusters given above is correct, it follows 
that the implicational statement regarding the formation of new 
clusters through syncopation should also be correct. While Ander- 
son's remarks deal only with cases of intervocalic consonant clus- 
ters, the same structural formulae may be applied to cases which 
would yield word-initial or word-final consonant clusters. 

It was not the intent of the present study to test the struc- 
tural hypotheses regarding the formation of consonant clusters, and 
indeed, in its strong fom, giving an implicational statement for the 
formation of new clusters, it is probably impossible to verify the 
hypothesis for the early stages of the three languages under consi- 
deration. Nevertheless, since the goal of the statistical analysis 
was to isolate the action of a hierarchy of positional strength, 
the notion of intrinsic hierarchization must be separated from that 
of phonotactic compatibility of individual clusters. In order, 
therefore, to ensure that the results are not skewed by the introduc- 
tion of factors of structural compatibility, only those vocalic 
environments which, in the event of loss of the vowel, would have 
yielded a compatible cluster along the grounds sketched above can 
be considered. This requires an ennumeration of the consonant cluster 
patterns of Vulgar Latin and those of the Romance languages under 
consideration. 

As listed by Anderson (1965: 76-7) and Tai Whan Kim (1965: 56-7), 
Latin exhibited the following initial and final consonants and clus- 


ters, where the groups in parentheses cannot occur phrase-finally: 


wor 26 nolan? eat pd ioe sail 







—rebrd ol iv pe) 0 on as a <b 
—enlo tecioana nb fecoriasnt Se ac r2 oe 

na 
foie as 2 gales a hm lena Li bent 
Sadeaio tpeioense te eeee ae: RR a ay 


aside ort? teat of vende treeenq arit Io sqatok exis a0 aew a 































* Bas ,exstevfo ¢nsnoeny Io nolsemtet ont pcbspert 
; ede x08 joumedate [snoiteciiqni na pakvio was pao est nk tak 1 
sie Viirey od éiareowugts yvideder ef 31 eratents von te reatseonc i | 
-tenco ishoy eoenosupisi ssutt ait Io ee vines ott so} ameetzoqy 7 
aleyiene [eotteitate atit to [sos odd sonte \seotadixevalt pokssreb | fl 
q ; ~Atensite Ienoitieoo to yiorsteisd 5 Fo mois ott oteload ad esw = 
¢ | tedt moct batexsuse od ten soljssiroreid olenintat Ge meee _ 
abso el .etetaulo levhivibat Io yd i { idistsqmo oitnssonarig 20 20 
—yubontnt oat wd bewedle son ozs etivesr ont tsi syens ad sexoteradt ‘ 
, oiisooy aod yinm ,vitiiditsane Is nuidcinte Rie) eictost 2 soit 
eved biuow ,iawav scit to sees “to tneve ‘edt As vibithe etnemngnives a 
a) fae wvods herbseie eabmiots ay brie Pospaien oldtgramdo 8 bebloty > | 
néedaylo erenoanco elt io nolttetensnad. As vociow eft vbesobitance eb | 
wehess e2peupdsi sotimmn alt Fo pact, bas vise mei Yo seating 


= 
7 _ 


c oS © 


. (0-02. 22th) AX nary Ger bos (-OF Zee!) hanes 


ria be ameccancn Janik bas Eddi. etiobtled B 
ED: 


maaan deditmiati sii anal 
aoe 


‘ ‘ie eis an 
tye ~~ 
= : : ; . + aT ; 
i | | spies 
ee ’ 7 a a iy 


; | ae ik : ay 
 ¢ Q ; yet i 7 


7 ; ¥. : i - " 
: A 3p j * ° i : + 7 - Sisk ies 





87 


INITIAL _ FINAL 
sp st sk pl pr t (mp) st 
spl stl sk] rea kr d (1p) (1k) 
Ve spr str skr be Die S (rp) ps 
af gl gr k mps ks 
i; EY oie aL rps ms 
d dr le nks ns 
s 6 a m lks 
k 
g m 
n 
r ¥. 


3.3 The developments in Italian 


3.3.1. Compatible consonant clusters. Italian, while generally 
allowing only the consonants n, 1, and r in word-final position, 


exhibits the following initial clusters, where clusters in parenthesis 


cannot occur phrase-initially: 2 
sp spl spr pl pr 
st str GGh) eve 
sk sk1l skr kl kr 
sb sbl sbr bl br 
sd sdr dar 
sf sfl sfr val 1 € 
SV vl (vr) 


sg sgl sgr gl gr 











yilssensp olidte nististi axstauis Eis 1 
,aoitiaoq Lestat ni bas aD ei nwscioen 


1q {g 19% 


we 


‘att 7 oa 
aa ) wei 
— e re 


oe wir 


88 


sm 


sn 


sl 


sr 


sg 


While virtually having eliminated final consonant clusters, Italian 
has greatly extended the range of possibilities for word-initial 
clusters, with the result that, from the point of view of structural 
compatibility, the phonotactic possibilities of Italian are not 
radically different from those of Latin. Anderson (1965: 83) posited, 
on the basis of the structural distribution of Latin consonant clus- 
ters, ‘maximum potential holes' for the clusters sb, sf, sg, Gd, Gv, 
and lr. Italian, in tum, developed the initial clusters sb, sf, sg, 
and the medial cluster lr. The clusters Gd and Gv, with G represen- 
ting the archiphoneme of /k/ and /g/, are no longer structurally com 
patible, since /G/ does not occur in final position. Most of the 
Italian clusters beginning in s which are not direct reflexes of 
Latin clusters, are the result of the prefixing of the morpheme ex- 
to verbal stems, with the initial e being lost at a very early date.* 
The clusters s§ (sgelare), sfi (Sgnaulio) and sr (sradicare) are later 
creations, formed by prefixing s- to an already existent form, fol- 
lowing a pattern carried down from early stages of the language. 

By intersecting, therefore, the posited consonant clusters for 
early Italian with those hypothesized for Vulgar Latin, only a hand- 
ful of clusters remains which may not be fitted into both phonolo- 


gical systems; namely: ps, ks, ms, mps, nks, lks, and rks. ‘The 























as 
; Ove 

OOP ye 7 

gbifesT vereten is tascoened fant: Deteninils oaived yLisuriiv ef tow _ 
3 
Ist ini-beow o6t 2siilidteedq 20 Spnist edt hebasdee yRsemp aed - 

| ve i‘ 

feunioirrte to watyv io ti tog arty most, faedt Sinaor! it eihiow mana mars 


Be tot sis milstT to 20k Saas Site sonOdg: si jegitiigegno. | 
\betisog {£8 <tatr) Aneetaben etal to sport, are ingot yiisolbex ry) 


ao) 


‘elo tesa nitst to noitudhrsety tides: ort 30 altel ects = Cra 


a soe 32 «de crstevin Intiini oft beqolevab ams al tsb Let ah hee 
-\mppesioes Dd dtiw ,.vO bos bo eratenfo sit Re 86 istauin inthe al 4 


“ite Yilewtaase wepaof on sxe ,\p\ Bas oN te nonoriaiians =f 
7 ay 


et3 Jo tecM  .ncitieoq [anki af 10006 sh 2906 P\ eqn 


to esmeftsx tos1ib Jot sus ridw 2 nk ‘prinntoed eamtanto 
7 


: “ne amerigrom scit to pnixtterg alt 15 4 ybax act 8 aiatites 


Je € etab \yomss yaw s 35 teol pricds ee art aie <a 


I. taiel S16 (oxsoibsie) 1e Boe (otlusaps) tie ee Seat Saas 
a: “foi act tasdolxs yossrls ob ot -2 ia 
-setumnbl ant Yo eapsie ylasS mowt rmwab 6 


wot statenio tasnmoanmn bedeeog srt scar 


ps oe SN ee 


ar debit a i ee oe 


WO 50 pe ,2e ,de ecratenio ed sot 'aafod lait ered a | 


J 


ein 
be asin 


atl 


7 


pas rN ae 





es aie 
24 


cluster tl appears to have been converted to cl before the transition 
to Italian, and it is doubtful whether such final clusters as -mps, 
-rps, -lps, etc. were pronounced as such during the period in ques- 
tion. It is therefore possible to arrive at a structural charac- 
terization of consonant clusters valid both for Vulgar Latin and 
for early Italian, which may then be applied to the corpus of data. 
It is impossible to verify the order in which new consonant clus- 
ters were formed. However, in view of the above remarks, all those 
environments that would have yielded a non-dissolvable cluster may 
be weeded out of the data, and only those environments which would 
have produced a structurally compatible cluster, whether or not oc- 
curring in Vulgar Latin, may be retained for study. In this way, 
another variable will be removed, and the role of positional hierar- 


chization may be seen in a more unobstructed fashion. 


3.3.2. Behavior of atonic positions. Considering, therefore, only 
vowels in 'permissable environments', i.e. environments in which 
loss would yield a compatible cluster, the results given in Table 4 
emerge from the data. A comparison of Table 1 and Table 4 reveals 
that almost no vowels were lost in cases where a non-dissolvable 
cluster would have been produced, thus strongly supporting the struc- 
tural characterization and classification of consonant clusters which 
has been utilized. The hierarchical behavior of atonic positions 
emerges more clearly after the non-compatible environments have been 
removed; the intertonic and posttonic penult syllables seem to be 

the weakest, followed closely by the second syllable pretonic, with 
the initial syllables being considerably stronger, perhaps with a 


Slight advantage being afforded to the initial pre-pretonic. 


89 

























Be ih ee ad bi 
5 va <A. ie | in 
- ; a ie 

tT 


HA y i of ; 
: pe = a, ; 
| stares) et exoted fo a baie ste bile 


Ses = es Lah de sa ke 3 ee 
~seop nt bokteq aio grits dove 6 Bebroreug ee . ote saad 


Shana [eaydoent] 5 je Seis of eidisadg estoredd pei fot 
Sue 
bas abies xeplith wt od bilay sxesaul> 3nsnoanco to ac ished. 


.Sé8h 3o apqioo eff oF bailggs ed net? van. rhirw \nsiisdT ¥ x0 


~ - 


“alo tnsapenco wen dbidw ai aabrxo act Witev o¢ sidizeagn « ‘: 
ser iis ,atuemes: svctia: cit! So: Waku ni revewoH amo? otew © 
| 3 bo ' 

yee teteiio eideviceuib-non s bebfery vad bluow ject einemnox 


bivow sti atnemaoxivae saat yin fs. 6365 sit to a be oe 
~@ jon XO nasizonty vrayeriio slditemxo yilsudousse 6 eaemehe 76 ad ye 


Weweid ol .vice 16t Genisto1 ed yam tits! uspiuV at § a a, 
a i 
ors 
ae, 
) — a 
sims. se [1D io 


_“tereid? Inqoislecg Ic slot stig Bas ,fevawer od Li iw efdsirsy 





ottias? batoeriseden gran 5 mi mese od yer 


° vino .Sxofered} .palwebianc> .angiticog oinods to sobvatish $e 


— 


; 
: ‘* 
foitw at séneatairivus .¢.i ,'atranniorives ofdeeeionsg’ nb af i 


’ sldeT oat nevip eiivess art’  xatteto sidtinamcs 6 bisry blocw 2 gis | 
ioe, 
7 


ae 


eisever t aide? bas I side? =Car e .20eb is moxt 












— 


aldavinesif-nom s oxethe asaso ni teal sxew efewov an Jeomis Sane. 
ete off pritreqqe yipnoste euris \becubowg need evad Binow we 
foftia arsteylo Jranoencd to mnttscttivesio bas raises hatoetaip f q 
accittend ofmste 0 wiveded Ksoiibsmustt sett Desifknr t 2 
os ot eee amano A ae vas 
od of mece ealdniive thug oknaiseby bis ointseitt ond 5 bev 
“iw oinotene ofdeLiys bacose ocd Yel ybeaolS Rewollet tek 
panei eect nae: 


ae -_ 7 


ba, 








Table 4: Italian atonic vowels; permissable positions 


Another interesting observation which may be added at this 
point concerns the behavior of word-initial syllables. In general, 
initial syllables exhibit a strikingly high retention rate when com 
pared with other atonic positions. And yet, in the case of the 
forms beginning with the beri ex-, the rate of loss is almost 100%. 
The basis for this differential behavior follows from more general 
observations regarding the position of the initial vowel with res- 
pect to the word boundary. Word-internal vowels of the initial syl- 
lable remain intact, even when loss would create a structurally 
compatible cluster, although, as noted by Grandgent (1927: 39), ‘in 
a few cases, the medial vowel of an initial syllable disappears, being 
swallowed up in the roll of an r or the hiss of an s'. When the 
vowel is also word-initial, however, the rate of its loss is much 
higher, not only in the case of the prefix ex-, but in all cases where 
a compatible cluster would result. The results compiled from the 


data collected for the present study are given in Table 5. Here the 


90 


00 an 
. ae ks apy. Bod os 
ghar 
| a iets 
ae en ae 




















i| ‘ 

i 3 

| 8 e 
} 

| i 
4 

had i 

| 


/ anolt isos otcientinssd rere ine te ce 
~ ‘a : “ ae: 
_, 8h 36 bebbs od yen doidw ooitsvraedS mtiseonsnt 2 

,leretiep nI .esfdslive [sit ini-fiow Zo Tabu aE fo 3 oa a 
“fo notw ste1 roltnats: deid yipnitihse «6 sidlises oleate weit 
edt Yo anno 2 ott ni ,tey Bak ecole 22s so dk ee Ae 
,#001 2eomis af secl 30 aoe Soft «3 xitosq st serena 
léxeinise stom mori aurtes aoiveded Iniinovsttib eit 
wees ritiw fowew tsidinit ent to roibteoy oti pret 


. _ 
7 


—————— 
- 4 i 
















tye [sislai oft to alswov Isqredoi-SioW ete i 
yliexwtoutse s stasis Bivow apor ine a 

(RE Ter) Snapiincs0 ya’ Bebolt eat | 

hel \erseqyeaib aldeliya Usistnt tis 20 
 '~ itt eroti. "ge Yo eat a 3 
i tea af Si tl 3p wes a <a 





91 


differential behavior between word-initial vowels and initial-syl- 
lable vowels in medial position is clearly seen, even though cases 


other than the prefix ex- were considered for word-initial vowels. 


pre-pre 
tonic 





Table 5: Italian initial vowels; permissable positions 


3.3.3. Behavior of individual vowels. Also of interest to the 
study of phonological hierarchies is the differential behavior of 
the individual vowels themselves. As has been pointed out by many 
investigators, there is a great deal of variation in 'strength' among 
the various vowels of the Romance languages, and since the figures 
reported in Table 4 are based on the totality of examples collected, 
involving all vowels, certain sub-hierarchies have been obscured. 
Tables 6-8 give a breakdown by vowels of the various atonic positions 
under consideration, thus showing the pattern of positional hierar- 
chies more clearly. The figures in parenthesis represent computa- 
tions which, due to the small number of cases involved, are not sig- 
nificant. Since it appears that, before undergoing syncope, e was 
raised to i and o was raised to u, the pairs /e/:/i/ and /o/:/vW/ 
could not be separated for discussion. However, since e was fre- 
quently weakened to i in Italian, while o was frequently, but not as 


often as was the case with e, raised to u, the data in Tables 6-8 





a ; 


amit izog oideas ie (slow tbe ae re 
io 


. Of OF deexséai To cals .efowoy cinta i iy. kb 4 
20 saiveried Isitast2aiip onset =: akg a ae) 
yea yd gum bedniog nesd zed af x foemens efewae Ani ieiBars eck” f 
Gaems ‘inact’ at notisiusy io Inch tsexp 8 eh Smee coingigeonh “ 
cupid as scale Bae sogsuoce! <Onanest of 20 eleMeW aed | 
Badoallon aelqnsxe to ytilesot Sct no poosd ets. b) ahaa - a 
beamed neat ever zsitcrsraid-die mbsdaeS | 
agoisieog siaots avoisy ody to aloww yd @ P 
cet en 3h oc | 
“SIGHED SNBBFIGS* aleottinsisq stk aaxupit on : f sof 
wets don sut bovlouni exesc 20 wedi Lime Sek = 
ecw 9 penn ae 15 — 
Wwis\o\ bas \A\s\e\ exisg edt y 3 ab 


aia) 4 








tentatively point to the following vowel scale, based on resistance 


to syncope and raising: 










post- SECO 
inter- tonic meee me et oe 
tonic penult tonic tonic tonic 








Table 6: Italian /a/; permissable positions 





Table 7: Italian /o/-/u/; permmissable positions 


92 











eon ie 
7 > a ae 
oa 2 i + Ss a ) a 
























$e 


Se er ec. 





93 


post- 
tonic 





Table 8: Italian /e/-/i/; permissable positions 


3.4 The developments in Spanish 


3.4.1. Compatible consonant clusters. All the consonants in Spanish 
Iay occur at the beginning of a word; however, Spanish tolerates no 
word-final consonant clusters (except in a few Latinized or borrowed 
forms), and rather limited possibilities for word-final consonants 


and clusters, as displayed below: 


FINAL INITIAL 
(b) br bl 
(Sy erie cr cl 
d dr 
J xa fr fl 
1 gr gl 


n pr ofl 







Rewrcsod Yo bSsicttad wet 's ai tqsoxs) aratento ais 


'* 7 
* 


— he 
. 


; ae ae aes 


einanoento Lenit-brow 108) esitilidteeog Bosiatl satie wie 
:woled Boys fqeih es, 


JATTING 
id _ a : 
i> x 
xb 
. ae . 
c. the ~~ 
fi - 3 
- nite i 





9h 


Ss (v1) 
(t) 
Zu |. 6 


The consonant b occurs word-finally in modern Spanish in borrowed 
forms like club which, due to their high frequency of occurrence, may 
be considered as acceptable to the phonotactic structure of the lan- 
guage. However, word-final b did not occur in the earlier periods 

of the language which form the basis for the present discussion. 
Similarly, the initial cluster vl- occurs only in Slavic proper names, 
and may be excluded from further consideration. A further exclu- 
Sion to be made concerns final z, currently a voiceless interdental 
fricative [6] in Castilian, but which did not exist as such in the 
earliest periods of Spanish, being represented first by the proto- 
combinations ti and di, and later by the affricates [€] and [2]. 
Final [x] (e.g. reloj, boj) did not exist in early Spanish, and is 
not pronounced in many dialects of modem Spanish. The word-final 
consonants C (cofiac) and t (cenit, carmet) are found only in modem 
borrowings, and are often dropped in speech. Altogether, comparing 
the consonantal possibilties given above with the consonant distri- 
bution of Latin, and allowing for the homorganic assimilation of 
nasals before obstruents (e.g. np>mp), the following Latin clusters 
must be eliminated from consideration in early Spanish: nks, rks,lks, 
ks, ps, Mps, rps, and clusters whose first member is t. A large 
number of these clusters were found in learned Classical Latin words 
and probably were not pronounced in late Vulgar Latin and early 


Spanish, so that by and large the overall consonant cluster distri- 
















, ay wa 


= : 2 4 . ; : & a 
Beworred mt : oaege rerio ni vWisnati ‘row ezucoo d Jnsnoal 





~ - + =~ 3 > ins vy 
yan woneiaooo To yorsupe ripe rkedt at a® .toitw dol 
716i wt! to sinsorncte * sitoasonisig 7 a ct sidsJdgsm0s 35 
as a 
me 
P shysinvect 1s! > aft nt suo00 gon Bib d feni2?-biow .1s7e~oH 
3 EDOrSao liyss Sas AE Oo0 TOM Dip Ga iffiizwes . Swoon 
" : . : - ee : 
t ; an 
: , ‘ opce aa erence eft oc ace —_ tf att Aa 
steuamnalih treme: aft 268 erased afs mot mosey evevonsl. tc 
\. 
b- —— ~yeen-t oer fe, . foe etre = | 
(amen rsqoxa pivel? ar yino gumeco -ly tetvevf isipint a ,yoe 
~ - . > — = a il 7 vetine ae a r = we iy ~ 
“ASlLSAS WOETUWS 4 Jj? SISaearS 39 t moxt Sabu loxs sa Ye 
Petts ¢ spas fears Frye 7 ory aie > 
aM} oi fi i sBarsorov & SItSTss! ua SFitzi Bits 
x 
x = = ot " PS tim 
; eit om fine as sears zon! £5 spittw 3: art are 
_ 4, 4-5 “ed taust Pye Haesgyrc rr 7 es rt 
ra bans mn wed = ed a ' . . 
By ./3) bes, [5] aateord exit vo tai be 
Pte! os 
7 Pa - bes . wit mi st P a -t 
é Bet Bie .Geitisod visas JaieS Fol : ei 
= a Bes pena — r i 
isoit-Biow ef dasaaae rime to Baines 61 
miasbon ak Vino booot eve Gremts= .tioep) 3 Sas 
° fii a —— — 
ia = - —— ; £2. as - ae 
paiisgmeD .TeritepetiA .fosaqe oi bagged fie 
} vr or an 7 
-bueaif taaneanco. Sit ods savip 2 tof belteeoe Leal 
a 
; “ ; a . 
2m rouraekl favre rit pare eh octet. er ote Beas 
7 au Se! INLeSs. DiNBe Yann Sfisy: TOs prtiwol is Bas 
7 1 
tection ae? Grrenot eyri- =) zines Le ery 
assdeuio cicdel poiwoliot of ,(gn<go .p-.s) sinsuned 
oJ ete Daal ee. tin a ry , ak 
SAL. we wis -daimmge yee ci sonmsxzebieaco reat betenimble s 
eae Ga 
‘ : 
of 


; ay 
spist A ate i tates nh tartl seaiw srat aula ons Ee 


aa a 


7 iy nitel inoteesl2 bores at baacheel Fi Sts 
: rat ) 
ae , tc 


<< irs 
» 


vey bite tateais Snsnoanoe L. + satis 


my et 1 - 
ey oe : 4 
Pett ab 2 : : fp : i» oo ' 


: j A A aie ; y ' > " Th 
a ee. ie . -f{ 7 aia a Lat 


lisiorins nbtist asplnv aus bacon 





OD 


bution of Spanish is not radically different from that of Vulgar 
Latin, although the position of the syllable boundary in the interior 


of certain clusters has changed. 


3.4.2. Behavior of atonic positions. By selecting only those 


words in which loss of a vowel would result in a phonotactically 


permissable cluster, one arrives at the results given in Table 9. 


pos tS | second 
cae Pane reel 
tonic tonic 





Table 9: Spanish atonic vowels; permissable positions 


As in Italian, the hierarchical behavior of atonic positions is more 
clearly seen once only compatible environments are considered: 
strongest is the first pretonic syllable, closely followed by the 
initial pre-pretonic syllable; in an intermediate position is the 
word-internal pretonic syllable, while the posttonic penult and the 
intertonic syllables are diachronically quite weak. 

The relative positional strength attached to the word-initial 
atonic syllables may be further highlighted by considering the 
difference in behavior between word-initial atonic vowels and initial 
syllable atonic vowels preceded by one or more consonants. The 


results of this further subcategorization are given in Table 10. 




















: 7 / “7 ae vi iy 
eo ; xspidv to Jar morn? 3 rows Ib aba 


: - 6 ; : 
" - ‘es i ee ee 
roirednt ot mi yrabruod elds ifye ony 


“ 7 . 7 ma ae 
g, 

- 9 , ¥ , : és me 
Yisottostonties of tives y fewor B Ro OTS 
. 7 : ‘Ss 

| = 
2 olde? mt aeip esicvss: Smt js gavin “sno tected. « 


OM el enoitiagg v inode 30 toivaried (sotdexsisi ie f exit lei 


shetehienco sis etoamoccives sidtisgras vino sare freee ¥ 

: Tor a 

a ue 

a yo bewolfct yisnoin ,oi tat 72 S.brodexG 6 tar 5 <b at gos 


Yarfd ODT Aer bee te <P Poste? ove at 
Si) ei nottiesd ssetboerrsini os oft safdel tactong-< 


« 


et eg see =| he ee eee ee Se nee es pe 
ST IS 2.1Dey OOOO Shs Sittiw ,Slasiive off 
\ = 
“ fs kT le 4 try cl > a= 
 7ESW SI 5p VLC fnowibs ih srs 


fsisini-Gaoe ais o¢ Betosits nenenge Isnoitteod 
E ® - A * on * “ fo a i a4 e ca bail bd 
oat privshiance vi hedtipiipin ssddnd od Yet, 29: clive S I 
i { - : ; a — + 
 +Ssttiat Sas afew W oimosé Iaitint-fbaow iced xtolvariod ni aoe 
4. oe ; : " sits 
SS fete -etnsnIBttoe Som 1D sno > xe dairies: 
Ba | Ss 7 
ee «BE afeer nt tavip ox 


96 


pre- pre- | pre-pre- [ § pre-pre- 
tonic tonic tonic tonic 
#C-—— {= #C--- #--— 





Table 10: Spanish initial vowels; permissable positions 


The above figures clearly indicate the high degree of diachronic 
phonological strength attatched to initial atonic syllables in general. 
Cases where a word-initial vowel was lost may be generally attributed 
to wrong division or sinalefa effects caused by interaction with 
preceding words ending in a vowel, which frequently form multi-word 
pattems in spoken Spanish. Compared with the rate of loss of word- 
initial vowels exhibited by Italian, initial vowels fared comparatively 
better in Spanish.The main reason for this difference is clear: 

the great majority of cases in Italian in which a word-initial atonic 
vowel was lost involve reduction of the prefix ex- to s-. Spanish, 

on the other hand, does not tolerate word-initial clusters of s plus 

a following consonant, and consequently no initial vowel was lost if 


loss would produce such a cluster. 


3.4.3. Behavior of individual vowels. The data collected for 

Spanish also allow for the study of the intrinsic phonological strength 
of the individual vowels. As in the case of the Italian vowel sys- 
tem, it is impossible, in Spanish, to separate the pairs /e/:/i/ and 


/o/:/a/, since a considerable state of flux existed in the earliest 



















; | 
ge | 
; a 





witison sldseshmsg jeloavov. fering fakaEg 


cs - 
er ~ 


= . ie 


be tui tite vilersnsp ec Yar taof BSW. Lawov Inisianbiows § 


dhe notioaiediti yt hones etostte sisisnie to eens 


+ Cites ~roete Lerten tiotnets 
ere, mi eelcs: oinoais Jettint.at Dotmoisoss 





rs 














— 2 c. naked a ge ae. ar 

stmt YO saol to esser ord ctiw beisqmeo.  .Teriece rey 

= 7~ -~ . =e | * . -_> _ Cy 

fovitgeseancm bexst elawon lseistnt \nsifesi vat so sake al 


= : > aa = 7 . aT. oder 
:rsefS ef sonereitib 2irlt sob hoeser niet staan 


omnots fsi4 ind-pxow s roinw nd astisti ak esas 10: > ae 










1 7 
~_* . 
fininsg®? +e of -xe xitsxq eft 2c nobinsber’S 









con , ice 1 


-s¥ Rertpoiion sateb on’ 2 fay isubiviint 3 fe] zi 5 


“Ye Leroy mai istt exit to San one nk — FF 2. 


tae bas NAGS exisg af atsisgse 08 lima 


i 


10 sista sfdsteb: 
ayn es 





} 


_ ‘ te silt I62 ertt of Sate sufty 2 








nh 





a7 


periods of the language. The general patterns of evolution involved 
the following transpositions: Latin i> Spanish i; Latin I, e > 
Spanish e; Latin u> Spanish u; Latin ¥, 0>Spanish 0. However, the 
general process of unstressed vowel weakening took the form of 
raising, with the result that e was at times raised to i, and o be- 
came raised to u, as in Italian. Moreover, the uncertain grapholo- 
gical practices exhibited by the earliest Spanish texts renders any 
exact identification of vowel timbre practically impossible. The only 
vowel which may be accurately identified from textual data is a, 
which generally persisted intact, unless raised to e by a process 
such as palatalization or vocalization; e.g. f&ctum > fer techou. 
hecho. 

Breaking down the early Spanish vocalic system into the three 
categories /a/, /e/:/i/ and /o/:/u/, the individual behavior pat- 


terns characterizing these vowels are given in Tables 11-13. 





Table 11: Spanish atonic /a/; permissable positions 


Table 11 manifests the inherent phonological strength charac- 
terizing the vowel /a/, a strength which has been noted by many 


investigators. The four cases of loss of a in the posttonic penult 






’ 


















‘geet $B eau entt A “ .2 
J p ‘a , - : 
Y . : a : 
~OLotiesro nistisom or \1svooudk nek lest ak 8S at al 52ELBI am153 

>a a ial ‘ 7 a 
Wis orsivenr thet detneqt teerlass et ya Datars esokton iq Iai D 
: y 


» 


gine ecil idiesogmi \Lisoidessg suduts Lewov Zo pobtenttts mebk 3 DEKE 


let een fetes wae. Rida heb rexe # hes | 
6 at sesh fenrreed mort beitiepe eaenore Sc "yan 
eh ar eens - - — ms Ce r +> 
arc ; va o¢ Bsefsx aesinug \tostar bese. 


MiSs - (3 tiscs) -Ps3 “POmsstin: ~~ YD COE a2 ARiantane 


- 
8 i 
- 


= 


5 
c ryeat r } . = + mee - et <. r aS? 
oe he ae se 6S li SJeris yas TAS Hes. Tae: 4 vis 2 <¥J Te2S paLe = ers 
— = 
_ id 
a 
; wr TU ros evr set Ver «lr a r\ oo 
‘= EY Ditve;S DLV IE fit » Ke AO\ ris CXS Yea Aa\ & ‘or is, as) ae : 
yas 


I-II esfds? ci covip sxs-alewov Seeds onisizetossed | 


~ a i 


7 j = pn nore ee nee 
i 3 421 Hioose | \=Se0e 
' : es - 
ae : awake | = » ——— 
-ey j-org-sr¢ «=| o=stq)l | hee eee 
i coe i hl 7 
an @ - * - aa - ~ | a K 
at (| sino «=| Dine | jfiumea "Ses 
a = ater <4 PER ky cet Pm meso aemaneatl 







a coe 
j 


. 
so 
a 


anoitieng siiszanmnsd a sid ite os 11f aide’ 
a a Ee 


‘ ; and bs x %. , f rs 
Aine stead on iti 6 ni ae 2 


;e F 1 1a - t 7 


itp my ee if M ? 7 ' 
oa. | vos ie 
> ‘a 
Peray.: 
-_ 





eee 


98 


may be traced to the Vulgar Latin period, * where this vowel perhaps 
had taken on a more centralized articulation (e.g. coldphus » colpo) ; 
in any event, syncopation of a took place only during this time 
period. In the initial pretonic positions, all cases involving loss 
of a deal with a word-initial vowel, thus falling within the category 
of sinalefa phenomena. From these data, the vowel /a/ appears, 
except in unusual circumstances, to have been highly resistant to 
change or loss in Spanish, a situation persisting to the present day 
where, even in dialects in which the other atonic vowels are severely 


reduced, atonic a remains relatively secure. 





Table 12: Spanish atonic /e/-/i/; permissable positions 


Table 12 covers the evolution of the vowels /e/:/i/ in early 
Spanish. These vowels are much weaker than /a/, particularly in 
the intertonic, posttonic penult and word-intemal pretonic posi- 
tions. In the second-syllable pretonic position, when e was not 
lost, it was quite frequently raised to i, thus reinforcing earlier 
observations which classify i as a phonologically weakened version 
of e. Unlike /a/, atonic /e/:/i/ occasionally fell in the strong 


initial atonic positions, even when preceded by consonants. 












1 ee 
i ae seul 


nidhice, 0.5) sume slates ih 
itogioe < 8 Gel -P+ . uiehaae yee ats 


eal patviowni 26252 (eee 





yuoesyao srit mae: eiuitis2 auctd! .48 


ft 


¥eb siéesig ed? co 


Yietoves sexs afswov 


7 7 
ee et — ~~ ed 
; +, 
| se{ Jos 
Sel i ie ba 
| a be “sig | 
Shoo { Sic _) OLA sie 
Ep 
' 
i 3 
OS 3 
of | od 


ear . 7 


a bo A. 
“ 
Stiisiecxs oldseabmeg |: Eh ‘ik lagi ci: oe hi 


7 wil 





7 


a sia 


ri ytsniotixeg ABN net?! xedesw: rem, g 

“neg pinoteig Bs at | rie eriacan ipa i 

Son: asw > nordw (nok itteog- incaeng , eis “bitons ae airs — L 
viptobsrkes piioscintor aiit .t ag oles at ei . p ae 2 + 





yittice nb NACNS\, elewor Pe Benes at 


noianey bensulsew ddeoigoiormda 8 se: 
ie oie = me - ‘ 

WEEN OF 

aA Ns oe i 





99 







post- 


inter- taniee Soar eat ae 
tonic penul tonic tonic tonic 


Table 13: Spanish atonic /o/-/u/; permissable positions 









In Table 13 the data concerning the rate of loss of the vowels 
/0/:/a/ are presented. In the initial atonic positions, the strength 
of these vowels is clearly demonstrated; in the one case of loss, 
the vowel in question was word-initial. In the posttonic penult, 
the data require a bit of additional interpretation, for in this 
position, u appears to have fared worse than i, in contradiction to 
the observations effected in Italian. Nearly all cases of loss of 
u in the posttonic penult involve words of the form -Cilum, in which 
the process of syncoopation of the posttonic penult was already well 
established, and perhaps even completed,in Vilgar Latin. This pro- 
cess of syncope apparently led to the establishment of a series of 
canonical pattems, affecting any form ending in -ctlum, -btilum, 
-gilum, etc. In fact, eliminating all such cases from the data 
leaves only six cases of /o/:/u/ in the posttonic penult, of which 
only one example was lost, for a total retention rate of 83%. While 
it would perhaps have been desirable to have eliminated all the forms 
in -Ciilum right from the outset, there is no way of determining whe- 
ther or not syncopation continued into early Spanish, and a failure 


to consider such forms would lead to circularity and excessively 








a: pian. nen le 
ve G2 ‘ ro -oincs x a ‘ -_ ; 
















gnnitiaog nivatailes wee since £ 
| 
stew oti to aso to etet ay pnimiconco eis | 
oyesete oct , anak ns SRE féisini ait at aeey a 
f 40 sat san eit nb dbateriencned cieoke i eon ena } Io 
' Hote siandtecg ont oi raisin tc e50 igD Ok 
aids at xok ,aoltederqiatni fenoisibhs Io Side 
Gd nnidoibertosw oi ,i nedt emiow beret pected. 
10 a2ol to esas Lis ylase wmiinet nk badosithe €e 
Hits nt amie netad ont 26° abiow oviount Stuneg HY 
Liew vbsexis no sini ions = 33 a 
“wa ater .oldsd ueeiv ni pstsiqmo neve aqaitrag BAB. 
36 eabtee 6 Fo’ satenuiak (cies att ot Bel —_ - f 
mastitis ely Bh, wedene eee, We ae 
ea oe ; ae: 
Aokriw to hieeg steer a Sere a at a 


un st Sins ee ) he mee 
» te 


ata cies ahaa 


PAS. a. Haw S : 
oe i Es f 
; as Coe re 


- 


yi 
a 


a—<oo 


Skewed results. As matters stand, the relatively few cases of u 

in the posttonic penult position not conforming to the pattern 
-Cilum indicate that this vowel is rather resistant to change, un- 
less caught up in a more general pattern of syncopation such as that 


occurring in Vulgar Latin. 


3.5 The developments in Portuguese 


3.5.1. Compatible consonant clusters. Compared with Spanish and 
Italian, Portuguese exhibits a relatively simple and infecand set of 
co-occurrence possibilities for consonants. Prior to vowel nasali- 
zation, i.e. in the time period covered by the primary data which 
have been collected, Portuguese permitted only the following conso- 
nants in word-final position: n, r, 1, ands. All occurring con- 
sonants were permitted in word-initial position. Even today, Por- 
tuguese permits no word-final obstruents other than s (pronounced yy 
in most dialects), and borrowed words ending in an obstruent re- 
ceive a final paragogic e: English club, racket, team, sport become 
clube, raquete, time and esporte, respectively. 

In the time period before initial cl, pl, and fl were changed 
to ch (originally [&], currently [8]) Portuguese permitted the 


following word-initial consonant clusters: 


bl br 
otk cr 

foliag 
f1 fr 
gl gr 
pl pr 


100 


at Oe 


uo 20 deme wet he cde bai 4s 


















ristéaq sit of eal mance 4 
rr De ed | ice . 
my Sener oF InAs: aioe Secchi ea touov ent sent $60 tb 


2s rare aolte Vala rasaneg Lstoa8e “ep Sim 6 ri 


v 


f y 


: a erst Wi 


Sucsirsod mr atramaofoareb eat 


at Te gt Se 


bns detract chow begagucd .eitedents tnenoencp eidive 
= Se a ee - i — 


P 5 : 
a. aoe tab : _ F « 4AetCure. « +4 re 
to 302 breosios Bas oiarnte Vigiitsicor 6 Ss icin 


r - - : sr ar Pore my re - 
“il i Y OF Tors GIRS Moo IC! Geis li foie 
—— ; a = - ™ _ 5 ald, 
~ et = i os re is Ss > “nes ~ dan a “ fore «Ot? ist 
‘ paar £ 1 ~ TE ray or Via DSiI= Vir DOTS Shi iS hed 
; 3 rat f <a - 
MEMO LYitwollol SAS VLAD Det csne Saane OA paso 
— 
= —— Tr > evel ~ ro en 68 {s 
& l Lhd ati «@ £215 m 'G 1+ gia {Wis lfeog 
wed ~ = nd 
ar net pre a ee 4 tbh Yebenes wh. Mutat 
t Veit niswS = .coitibaog fertini-bicw ni Dbetiom 
as  - pert et — Ss > 7 ve 
NY . hae? be co £19 , x1 1 eM) ws a TS Ld ad a % al Az 
J al oy ¢ ss ~~ iL ¥ . — 
~-or Jamo I PiLions AF LOW ru IIE 
—— Pn oe ee 2 fwarsh F- es 
eee ih: ee 7 es 4 Pe hut artang? = ‘Iria ~= ~ SAS AE 
ania —— a a ta am ie . - nod = 
«p Ver _—o “7 ~ Lim f Fwre ones et 
= Wibte 1395 et fa tOReS RE Rit 
: inne ae |e 


Sepreth stew BD Bos .la .lo latiint sated b tisg emit ait 


Wt heacs hire sracentrot ~,([] vosemmas , 6) yitsaipsa) an iD 
x 


rareten(s. SrIBNOENOD fsttir nt-bxow p 


xf F ke e ; 


fi 





ile 


3.5.2. Behavior of atonic positions. The rather limited set of 
cluster possibilities, combined with the total lack, in Portuguese, 
of word-final consonant clusters, severely limits the potential en- 
vironments capable of sustaining vowel loss. Considering only these 
environments, and allowing for homorganic assimilation of nasals, 
the data were screened to sort out only those cases where loss of a 


vowel would produce a phonotactically compatible cluster. The 


results of this screening are given in Table 14. 


post- second ist ist 
tonic pre- pre-pre-| pre- 
penult tonic tonic tonic 


131 59 





Table 14: Portuguese atonic vowels; permissable positions 
In the data collected, no vowels were lost in positions where an 
incompatible cluster would have resulted. Of the word-initial atonic 
positions, one example of loss in each category represents a vowel 
in initial position preceded by a consonant; the remaining cases of 
loss involve word-initial atonic vowels, again highlighting the dif= 
ferences in behavior caused by the presence of an initial consonant. 

Table 14 demonstrates that initial atonic syllables are very 
strong, in terms of diachronic behavior, while the intertonic is 
relatively weak, with second-syllable pretonic vowels lying approxi- 
mately on the midpoint in terms of diachronic strength. ‘The rela- 


tively high figure for loss in the posttonic penult results from 
















“1S ToL Sanpgen, a3 ebimit yiersroe ets 


areit yl so wiscbeano +eacl pir cieat - olden : ® yr 


7 ae - ‘ = 
seleran Xo nok; Limite oineptoman GH ort. Eeette bas , aime 
- % 7 ae 


~ 


si eeol eratiw BsPRED seadd v abe s) duo #02 od : tt 
: : 


ef? .syateuin seidtisqmo yitsocostonodg 6 


[Af aide? ot navip om wiinssroece 











[ sal [| sel Vieee 7 ecg | 
“xy |-s1d-s7TG sig] Dittced. | eee 
ainct | Lyi ‘Sires _ | shurisg otf 
Va ee Joe eee oti —s © en hrleae me enc a 
Pie. | | | 
eor | fp |. G8 | a 
t ; 

+e PR Sa ee es eee | a ——~- 
: ' [ 

ay 20 ; x¢ . Ar : r | 
= ~ : ee - - 4 r 
| 


Ee 


—~ 
& 
t 


‘ 
: 
¢ 


he A) Reena aligte eatin 


anmpistiacgy ofdseanrisq :elsewov oinotds Seanpedaog ¢ 
= 
Sofas Lersingc~b ‘ow 20 et fyesr ogsrt bitow 


70 @8ee> painiass: ot jtnsnoecco 5 yd bebaoeta 
. ; ae x 

—> oF -t urs to cered 5 the 2 rt one a | oe a 

“ID ord prisnpiitipig mtepse ,efswow ali aS ipit iS 


snsn@anoo Isiccti as io & ag sa vd Boe BD 3 XODVE 
- , 


ow, ‘ > - £ y , oe = lH pub Pvt & 
\. “feev gs eeltiniive minots fettint’ ser 


bs SS 


‘ ef Sint i= tri act oft irtw Mates, vériod ota ciate ‘ae a anit 
se 


~hacracy fe pay elawoy sinodsigq oid Iye-baones site a 
nm 7 ‘ ; rae ‘ah 
Ll > ‘ 
a Riis ie ioc oat. JStenete Stach 
—~ P 
a moxt ead fess “Sleaaaiey | OLS: 
a) % o ‘ 


102 


the practice, noted above, of generally making no distinction between 
loss or retention of vowels in this position which may be assumed 

to have occurred in the Vulgar Latin period and syncope occurring 
properly in the time period designated as old Portuguese. For this 
reason, the rate of loss of the posttonic penult given in Table 14 

is undoubtedly too high. On the other hand, attempting to separate 
syncope in Vulgar Latin fomms ending in -Ciilum from cases in old 
Portuguese in which the intervocalic 1 was dropped leads to inesca- 
pable circularity, since it appears that, for a certain tim at 
least, the two processes operated concurrently. However, by eli- 
minating all Latin forms in -Ctilum, there remain 53 permissable cases 
involving the posttonic penult syllable, of which 39 were lost for 

a total retention rate of 27%. This augmented figure is probably 
closer to the actual rate of syncope of the posttonic penult in 
early Portuguese, but, given the complete indeterminacy surrounding 
the question of relative chronology, an accurate figure is impossible 
to obtain. Moreover, if one further excludes from the data all 

those cases signalled by Williams in which the vowel was lost in 
Vulgar Latin, virtually no acceptable cases remain, and the results, 
in addition to being totally circular, approach the level of tau- 
tology, since one is merely echoing the fact that vowels were only 


lost in phonotactically permissable positions. 


3.5.3. Behavior of individual vowels. Breaking down the results 
reported in Table 14 into the various vowels under consideration 
yields the figures given in Tables 15-17. Table 15, despite the 
relative scarcity of examples, illustrates the inherent diachronic 


strength of the vowel /a/ in terms of resistance to syncopation. 


Lae a le 
(ie 


‘ ra] ae wk | 
aes | poisons: on ori belay. ail 
















hemyees sd {pm ebsitciw nlite 


mirssinoe sce > bers ae té 


ei 









r ou 4 oh 


ie 
dai ei newb tips singde sop att ic ae 


* + 


airy fl -,sesopetroy bie 28 Besse 


=i 


NB aa 


PETE + oo ie =, eT TE Rae zaitto itt m0 * figs. nuhiees v oa 2 3 deo * qu ek 
ut sa8a0 mont mulso= ak priors. ara ae tel wplev at 2 om «ryA 
io” : ‘) 
Sees ct aDsol Seqertb esw I ntisoovxatns ene doidw at 254 


« z * * P oe 
: P yl a oe ——- \ +er ~~ : * a ( 
te mis Siseylas «£ MJ2 4 Bris ez ROGGE 5. ot sore ei = ~. bf 


-lfo od .vavewoH .\itdewitoneo Becsyeoo espesogmg Os © 


tebe 
j 
it 


1 “7 - abe Tree a ase 
rt Sami t SISOS oehs asyip sud yeum ~ 
on 5 r i = <u Torus " 
= : Ji LO sk « YRO A Krai 
~* otal oe a or ape _ £ a = ~~ 3 * 
16D Orch MOI Seni ioKe WS SiS 12 \aove 


ni tac! esaw ewou + 8 tiv ci omsil ciw: 


(e - 
J P . t+ , ? ore oye = +o owt ¥ 
gov Pe Gis, SOF TE 4 in A BEICHD Ss{ristqsoos a 4 ak 





7 7 
Hi 


-cnt te Isvsl,sd¢ desovgas ,telpotie yilesas paisa ae ma 


yine orsw elowov Jed tost ‘Sid piri vient iat 8 Sr 20 














4 
eroitieog af (oo tmersg ate toa One ig 1. 


etives: sid mwob on olesr@ > .elovenr Leni Bin 
baal ES TS CL — 
Pa 


moh pew oh a C 


texscblenco issu eloww scoORIeV Sri 
ens eticeob ,eI Sider .Vi-dl eek 
7 g 
q * 2 _ , al : 
siamies:b tegefci ait ectexctayiii 
7 : ; 


ri > 


rortegooaye of eostataet to & red a3 
= i 


aes ane ae 
~~ 





a 


103 


In this context, it would be interesting to know the relative 
chronology of the raising of atonic a with respect to the raising 

of atonic o and e; unfortunately, no information is available to 
shed light on this problem. Indeed, the entirety of the raising of 
atonic e and o seems to have been arrived at through a combination 
of purely phonetic factors and morphological confusion,© which 
renders any accurate time tracing impossible. Based purely on the 
inherent strength of the vowels as measured by their diachronic 
resistance to syncope and other forms of weakening, one would advance 


the hypothesis that atonic e was raised first, followed by o, and 
7 


finally by a. 





Table 15: Portuguese atonic /a/; permissable positions 


post- second 
inter-| tonic pre- sidineel 
tonic tonic tonic 





Table 16: Portuguese atonic /e/-/i/; permissable positions 





= 


LD 


| 2 . ep 8 + is . ae i, a ro x ,— 
of ofdsifsevs et aoktamoint on yuh Lei 
to pafaiex sith Io SEES 7 t= ie iit .meic 


mpitenidnco 6 saa 38 povines B 


Hae 













rs 







tbidw 2. coktAno 


lad rm viesuq Deesd 


: | 
eee) aa [sek | Bees nee 
 —ox@.| -sigrerg} srg 
ae | 
- 


Sinot orricd _. 





os 








| 


SSS Se 


104 


post- second first ieetrrst 
inter- tonic pre=  pre-pre=i" *pre= 
tonic penult i tonic 





Table 17: Portuguese atonic /o/-/u/; permissable positions 


Comparison between Tables 16 and 17 reveals, in most environ- 
ments, the relatively greater phonological strength of /o/:/u/ as 
opposed to /e/:/i/. As in the case of Spanish, the outstand- 
ing majority of cases of syncope of the vowel u involve Latin forms 
with the final configuration -Cilum. Removing such forms from con- 
Sideration leaves only six examples of /o/:/u/ in the posttonic 
penult position, of which two were retained, for a total retentiom 
rate of 33%. ‘This figure, while not statistically significant in 
its own right, due to the small number of cases involved, suggests 
that, in most cases, the vowels /o/:/v/ exhibit a much greater 


intrinsic phonological strength than is reflected in Table 17. 


3.6 The question of open syllabicity 


3.6.1. Introduction. The question of the role of phonotactic fac— 
tors in accounting for the data reported above is still not closed, 
for there is a further ramification of the notion of structural 

compatibility. Spurred predominantly by the work of Pulgram (1970), 


a great deal of investigation has gone into the matter of relating 





00L 










= 


anolsteog el dsaidinst 7\w\-\eo\ Cina’ sastipichtot ly & 


ag = 
51 GE 


sont 


—smitvis feo ct eisaver Ti fas of solder, nasted ooatrngeeO” _ ; 


es \w\:\o\ to ripasrte (soipe lonaiq *edssip yleviselex ent ee 
 abyrctetyo orth deinsa2’ to eas Sat ai eA « VE\s\a\ boon oa 
Ve 


merat alsml svlovnt + fowov ort 36 sqgsaye, Zo ens 2 he 2 
hon 
=o mors aect fore paivoms wii3= noite rupiinen fend 


- 


7 


obfmittecty ott ni \a\z\o\ Jo ealqmaxe xie yino eavesl aottetshte 
mitneson iedot s iot \bentstex srsw ot HotiW ao ee aan ? 
Rk Sos tte 2. ¥Liss iteLtete Jon sLiriw vote erm .@€6 J = | 
eteapwwa .beylovni 2scs5 + Sac Cie ae ee tiene. ‘ ok 
tedserp rium 6 + ‘efter WFNC\ BLO erty Saal a nk Se 

NE alee? ni batooMs: et asd atpiswe 


© 





yy 
; as 
en uy ue 
“re eoaY : a4 
7 Om + 
La | Js Le / 


105 


consonant clustering to open syllabicity. Pulgram has claimed that 
there exists a universal or aoe iReaal tendency which requires 
languages to strive for maximm degrees of open syllabicity; that 
is, syllables terminating ina sortie Open syllabicity is of nece- 
ssity determined in terms of the possible word-initial and word- 
final clusters of the language in question: if a cluster can begin 
a word it can begin a syllable. In the event that open syllabicity 
is not possible, Pulgram suggests the ‘principle of minimal coda 
and maximal onset' whereby the syllable boundary is placed in such 
a fashion as to minimize the number of consonants to the left of the 
boundary . 

Several of Pulgram's students studied the Romance languages 
in order to put his theory to the test. The study of Italian was 
undertaken by Tai Whan Kim (1965) who claimed (pp. 7-8): 'I main- 
tain that as many syllables as possible must be open, that is, end 
in a vowel, except when the rule of opening the syllable would create 
non-permissable prepausal or postpausal clusters’. Needless to say, 
the author of this statement finds himself in a difficult position 
when approaching the matter of geminate consonants in Italian, many 
of which developed from originally single consonants; the best that 
is offered is a suggestion that geminates somehow do not close a 
syllable as 'completely' as heterogeneous clusters. 

Klausenburger (1972), who attempted to apply Pulgram's theory 
to the diachronic phonotactics of French, also encountered an insur- 
mountable obstacle when considering the prothetic e which arose before 
initial clusters beginning with s, the so called s impura. ® In 
French, the s was eventually lost, thus creating an open syllable; 


e.g. sodlams escole, 6cole. However, as Klausenburger notes (p. 48): 






ih et. : 
ue agents sis a 
ee ue pais ep es 
aatit seatoidellya nego te : i ste ' 
~som fe ei ytioidetiye seq Sapna al resi 
stow Bas npn alee aaa ise 
isbilt'aas Sateto. 6 Ft srotiecep ai spéueast eB 408i 
Wioids Live asgo sett steye!ortt it Penal S a pate a 
eiee Leminin to éigionita' att etasptars onvelt seitieaog Jon at 
tex ot heoniq et yasiwod sidelive ont ydexeiw ‘Jean ne 
sto iat ot ot sitoaee a SiG i a — 
| eee (ae 
especiprial corgunc arid beibuse etrekute 2 ‘ci ih teeta a =a 
‘egw asilesl 2o youte afl .tasg ee TT ) ‘ig 
pane 7(8-) sqq) bemtsls oriw (2d@l) mix neh ko ye elects . i ; 
fs Bi tedt ymca of teum sfdieeog. as colds Liye ynat es Sect died be 
ete bivow olcisliva ais painsgo 20 slum ott eae | 
.yse cd esnlisst ‘ exrecen lS {seusgtaoq 10 Laer 
ee aoktieog tigoittib 6 ai tieenut ebalt re 
| : . yor edited? at eatrsicatco ecaninep to ~ niente ty 


dedt jesd ent jetasdotrco slpnte vi lantpixe 


















eotivpes doirtw yonsbaes L 

























ie 







. 












& epois too ob worlemoe aeatenimes cédt ; 


. : a 





; arateulo. aucsnisixnratsit es on 
raed em la ot atom et 


106 


If one considers the starting point and the final point 
of these changes, one can say that the role of the pro- 
thetic vowel was to reduce three-member postpausal clusters 
to two-member postpausal clusters, two-member postpausal 
clusters to single consonants, removing the initial /s/. 
The beginning and end of these developments show phono- 
tactic restructuring only in that the /s/ clusters were 
lost. But in OF ... these consonant groups were all 
sequences, with the syllable boundary after the /s/. 
Therefore, at the OF stage, there occurred a restruc- 
turing as the prothetic vowel converted postpausal 


clusters into sequences. 


Moreover, in Spanish and Portuguese, the prothetic vowel e has re- 
mained as well as the s, resulting in permanently closed syllables, 
although the moder tendency in Spanish for syllable-final s to 
weaken or disappear may indicate the beginning of a belated retum 
to open syllabicity. 

On other grounds, however, Pulgram's theory represents an empi- 
rical proposal with respect to the question of open syllabicity, and 
as such may be tested against the corpus of data used in this study. 
The prediction which would be made by this theory is that the rate 
of vowel loss should be higher in environments which would retain 
or promote open syllabicity than in environments which would merely 
yield dissolvable clusters or sequences. With an eye toward test- 
ing this hypothesis, and thereby eliminating another potential 
variable from the study of positional hierarchies, all the environ- 


ments in which a vowel loss would result in an open syllable were 


culled out of the data, and individual computations perfommed. 



















eS 
perme ee nao 
inenaese wleneet vevamet> Lneusgtoog soma oe 
Ae\ tetzint ot enivomen \edomqoenc oleae od ig 
oor site‘emamptorb sees 30 ine fe ysonged af 
cxow exatanis \a\ ott Jerk ak ylno privuctowstesr Bia 
{is ozaw aquexp gmsanenco sead . -- Sot ae ae 
\a\ arth tafe yusbauod ofdsiiye art riiw , (SENSIS ' 
-xuintites: 6 bartimoo sxakt ,opste WD art ts , gh Oe Ts 
Lsessqtecq betray Lewoy “obintoag ot 6 aaa 


ssoneupee otek ened <5 - 


sc tein. sacs scone aa sesiien-non ea daaye aa 
,eaidsliye beeolo yitcansarreq ci pritivess 2 ott as Dow ee Beka 
of 2 Ioni-ofdsliye 102 duineg? nt eoshot mzabem arty lgvedsis 
mario beefed 6 To paianiped ait sasokint ym sSeRBI® agi feAAeA ‘om 
pokey re a a 
ahha sanders amveady aebidinia tevewor some sate MY 
bos vestobdettye sop 2 motes art oo Yoaguen iw Keno ARBRE | 2 
bude aint mt hear steb to excroo edd Jarieps  Redted He § o 
dix atid Jantt at yroartt his yc cham od Buen Heiteie 96. ape 
isco bivow, doit coemotivas al sedpid sd iiocds eaol fay rm 
yioram biuow doidw etnemortens at npdtt winkdeltye . . 
anbaaitcdcn:s niall ecnmeges 20 satan 
seek sins snr tsi 3 


Pg nth 3 saga 


er | | e : e Bey erior at 
‘Pu ye ae — Pi wii 


i. 









“o 


ti 









LOFT 


3.6.2. Open syllabicity in Italian. The comparative results for 


Italian are reported in Tables 18-21, where the notation 'cc' means 
compatible clusters only, and 'os' refers to environments in which 


loss would produce an open syllable. 


jane | a | ee 
cat Bees 
See ie 
a ee ee ae 
Pacep ea) al Al || ef 


Table 18: Italian atonic vowels; open syllables 


intertonic posttonic 
Pat prrante 
weet [of 
sane ee 
poem fom [= [= [ele 


Table 19: Italian /a/; open syllables 



















The figures are not striking. In no case is there a significant 
preference for open syllabicity, and in some cases even the opposite 
tendency may be suggested. There must, consequently, exist other 
factors which counteract the drive toward open syllabicity in Italian, 


if in fact any such drive exists. 







| wom ts a tl 


Py pen 7 af : rs 


oe ene 4 
ii as: ’ ; ol. 











108 





intertonic posttonic second 
ene pretonic 
CS Os 
Reagent pe pm 


Table 20: Italian /o/-/u/; open syllables 


mapalgpeusd on 
Soe sae obens 
pemmenf = ft [ef fe 
occ SE 
poe fom | of ele | a> 


Table 21: Italian /e/-/i/; open syllables 





















3.6.3. Open syllabicity in Spanish. As regards initial consonant 
clusters, the syllabic structure of Spanish is substantially different 


from that of both Latin and Italian, since Spanish tolerates no 
word-initial clusters of the form sC-, but prefixes a prothetic e- 

to any such clusters which may arise. By means of this process of 
prothesis, the s of the cluster becomes syllable-final; thus stare 
#statre# becomes estar #esttar#. As noted by Klausenburger, the rise 
of this prothetic vowel stands in marked contradiction to the theory 
of maximum open syllabicity, since by transferring the s from syllable- 
initial to syllable-final position, a closed syllable is created. 


Many Spanish dialects weaken syllable-final s to an aspiration [h] ; 


i) 


* < 
ato oa 
eo gery arth .3eetiininesyslA yu Peon BA insite wustep sams 330 ° F 
‘ - 





or 


a 












se) eee 
Oe oes eee ee va ee : 


a . 
— ee OR ea . rt 
i : = cealdisgiive nsao s\6\~\s\ me i lest 2H ; 


> 4 


4 74 
tmsroenee Isitini ebrsepsr eA .felosas ras winidelaye 
ole 


a. 


greexettib vileitaesecve ai datnsge 2 xO i alas okdsiiva otk 92 


y 
: oe 
peteisfot fainsdae epnie. .met fete bre rtted do to a uit 


-3 nigertiorg.s eaxitsuq Jud \-De integt exi3. 20 wieeia isis ni 

= ‘a — 
Ro seepexc airtt to ehsam YH «Ss 2 hts vn dort: ua ws 
, xe is r{ 7? ; _ 
sista omit ;Lentit~sldsilye eemooad “seteuto eee 7 atesrh tox1g 


D eaiad ody of: misc oibsusseo berhrsar wt waste to Ne. or ot 


! As 
tekdsLt ye mom 2 ott pnbristensrt vst conte Re: pion et 
yes” ry gw 


-¢ od benacite at olds Live Bs 





is 
= 
ae 


thus estar>[ehtar], thereby possibly laying the groundwork for a 
future return to open syllabicity along the lines followed centuries 
ago by French. Without additional evidence, however, it is impos- 
sible to claim that this aspiration is teleologically caused by the 
desire for open syllabicity, rather than by a purely physiological 
weakening of consonants in an inherently weak position. In Puerto 
Rico, Cuba, Andalucia, and sporadically in parts of Central and 
South America, syllable-final i and r are often interchanged, due 

to their acoustic similarity, and are sometimes neutralized to an 
indeterminate liquid segment which is neither clearly lateral nor 
clearly vibrant.? ‘The weakening occurs only in syllable-final posi- 
tion, by virtue of the weak implosive articulation of segments in 
this position, and in the case of 1 and r involves no change in 
syllabicity. The same may be said for the vocalization of syllable- 
final 1, occurring in Latin, old French, and modem Portuguese, or of 
the vocalization of syllable-final r in Cuba and other Caribbean 
areas (e.g. porgue >[pojké]). The syllable-final position seems to 
be an inherently weak phonetic configuration for consonants, and 

to claim that syllable-final consonants are weakened or dropped in 
order to achieve maximum open syllabicity is to enter into a totally 
circular discussion, Since it is precisely the weak articulation of 
syllable-final consonants that appears to be at the root of changes 
to open syllabicity. In the particular case of Spanish prothetic 
e-, if the drive for maximum open syllabicity were a strong one, 

the process of prothesis would probably never have been initiated 
in ths first place, or conceivably, although not very likely, the 
prothetic vowel would not have attracted the initial s of the clus- 


ters to its own side of the syllable boundary. 


nal ® 


A 
ad 
«6 = robe ret 


~ 


Bo LISNSO Soynliot woanki « 























aogui ei 3i savewod , 


lags s A 
eds yd beauso yifsos jpotsatsn ai sedib aa 


Z gl 
r P é Po ae ’ “ 
. Espipoloteyrig yiewsg 6 a nerd eis hele 


; ae ba 
‘Ofrearl ot smart po a6 ae yl saforrarint nie = anise id “3 ae 5oK 
= it rr y. f 
cu.) Coir 


bre LartesD to attsa ar vf eee ‘Bie toate if 


: a | BY btice | 
eum \baonscbisttt asdto sis = bee L terse 0 tien mv we OS, 
, P. ¥, 


m «3 besciextuen 494 Triamige < s Drs’ % Wo ntet ime oben oe xrescdis 
e 





son Isie3 xsol> tstition ei roifw tasapee tinea tee Limes 
i Q on ' es ae oo : 
. ' on ee ee . aos = Me ‘a . : ys. r. . 7? A 
-j#og [anit-cldsLiye at yfino soo prieieasew Ss ef *.dneidiy yixsels. 
“ : Be ne eel ae) 


— a - 
- 2 ot * * - > 3 —— 2 A 
si ednempse to nolteluotthis svieolgnk asew ont Qo arriury xo Lok: it 


= 7 7 

. ; * 4 “yp -* 
. mt gonedo on eoviovnt x Bas 1 Jo sess a ak sind nemneis, jog 8k 
ve to poitesilsvoy ot! 102 Bise od yer some anf - inl : 
- 
20 YO .seeaupusrod michon bits ,ponert Blo .aeed ae pairmoco vit aka 





~ howe gt ae # ig a ee 

BOOK CE S906 Drip SLO mi 1 I Smti-sicger 
+ ore - see a - ° abn 7 in f peas 

OF Bee a ite — FL ! Live Sill o Seu 
f » ~ “¢ > - ® =er bry oa : oye 
WLS Cee) Geis Sw : JRL oITKO dt HAA 


ai beqgow> 10 barielsow sis etosroancy ai it -sidsiiye * 
; : aie . ‘ . 


wt Teartcen Agni: ; ot 2i v2 tie Pts oe 
yiletot & dnt uesGo OF 2f \JIoidbl ye asqo mon ead batten 


Say i 
to noiieloism Asow sit viseingrg Bf +i sonte ee i rnluo 


LS we iit ) 
——e ~ oh ne ~~ | if if 
appiisfip, to toot of 35. sd oo a169qg8 satis atnenoer Oz ati ~ef del: \ 


© ' i 


siserisorg dainsge tc 5269 4 sip bie9 oat oi 






ao 


Dc 


eno Brite 6 stew ys iordeliye. ne oH cea = ik IIE 


hada torre nsed oueit teaver yi ia, Biuow 








oite <jtams as F sy 
et ,yiskii yasw ton droits 





Leaving aside for the moment the question of prothetic vowels 
in Spanish, it is possible to consider those word-internal cases 
where loss of a vowel would have produced an open syllable. The 


comparative computations are reported in Tables 22-25. 


sal Siete tase 
pat 


Table 22: Spanish atonic vowels; open syllables 










oo 


In Table 22, the meaningless figure reported for the intertonic 
syllable must be disregarded. In the case of the posttonic penult 
syllable, the rate of retention appears to have been significantly 
lower in cases where an open syllable would be produced than in 
cases where only compatible clusters would result, without further 
considerations of syllabicity. Im the case of the second syllable 
pretonic position, the rate of loss was only slightly higher when 
open syllables would have resulted than when only compatible clus- 
ters were considered, but it is not obvious that this difference 
is meaningful. 

In Tables 23-25 the question of open syllabicity is broken 


down by the individual vowels. 


110 










aeidsifve asgo ;slowow Since. sane iss MB 


+i : 
oinoitsini ort so? bessoqe: ssupit easton incom’ ett se ade at 
thunaq otmtteog oft 20.5285 aint lRebibpstal> ed taum er ; a 


ie we cm 
| an 
« ghing 2 inpis nped saved at erpegeys nottnater Yo shea ort | : 


mt cpr, beoubhorg, sd a tae Sldsiive sgo 46 state eoaaD. mY wide s 


Led 
siti Spkierhy atbkee bucwtaragents Ravaschetover: |: ra 


sidelive noses sry’ to saB5 at mee -g3idtdeL ive 0 (eh ont bierico 


| vaso totic yforeite wine paw avol 26 oh i ac — “i 
~aio shies ylao nerie ro bed ives ave Dilut = aes 


aunisistiib aittt Jatt auoivdo jon af $i Sede a ab Fant& 






\ 














intertonic posttonic f second 
pene pretonic | 
cc os 


% retained AGS 100 (100) 


Table 23: Spanish /a/; open syllables 


soon eae ae 
sapery erie 

= =ap4 pel ere 

ae 

Sa aad ees cell a 


Table 24: Spanish /e/—/i/; open syllables 


poco sori [ontario Een 

penult Beet 
tess 
poe [tote t fo 
poo [foal 


Table 25: Spanish /o/-/u/; open syllables 















TEP 







(O0L) | 007 


“pes 






asidstive msqo NENA Aetna a 
: in o's ace 






WRROr ee 
ssldalive a ‘= ve ; 


. 


- a : mae y 
3 ia ve 
+ . vr i ia oe a ae 


Sleit “4 ? mt aie . 


age) ML were lL ate 
ge tee oe rer: Te 

















bE 
. 


ee 


Due to the dilution of the results produced by this breakdown, it 
is impossible to obtain any figures for the vowel /a/. In the case 
of the pair /e/:/i/, by considering only cases of potential open 
syllabicity, the rate of retention is actually higher in the post- 
tonic penult and second syllable pretonic positions than in cases 
where only compatible clusters were considered. The same conclu- 
Sion holds for the pair /o/:/u/ in the second-syllable pretonic po- 
Sition. In the posttonic penult, the rate of retention of /o/:/v/ 
is slightly less when only open syllables are considered, but this 
distribution again reflects the skewing of the data by the predomi- 
nant forms in -ctllum, in which loss of a vowel would result in an 
open syllable. This in tum accounts for the apparently low rate 
of retention for the posttonic penult in Table 23, since cases of 
loss of the posttonic penult vowel of -Ctlum forms account for the 
majority of the cases found in this table. 

Taken together, the results from Spanish indicate that the 
drive for open syllabicity evidently did not play a significant 
role in the diachronic evolution of Spanish through syncope, where 
general factors of structural compatability appear to have taken 
precedence. Even in modern Spanish, there is no evidence of a con- 
certed drive toward open syllabicity; clusters which are not struc~ 
turally compatible, and which result in closed syllables, are tole- 
rated throughout the language in borrowed forms: e.g. zinc, torax, 


actual, extra, abstracto, vals, instrumento, etc. 


3.6.4. Open syllabicity in Portuguese. Portuguese, like Spanish, 
pemnits no initial clusters beginning with s, which severely limits 


the possibilities for syllable onset pattems. ‘The figures in Table 


ismaatog te 20660 x Scam 


a ale i 
exit ptt 


. 


Bei <aiges10d exo = ee 


eis eoldsifye 1: 


i> 


xit to. pritwede arit 


asdf tokdw stk ‘ila mi 
| ‘ier 
- 


- * 
a 
= r 


Liye 


ota ryt tory 


Pe Prarie, 
9 eo 8 | exit coi a f D - rtrisdex 
, * - 
a 2 r 2 ad 
ye (esc + Fr 


as 
tes tO if FLIX] sti ret eOg, art , 
iJ — 
F ii “2 ea 
ttt nt Erwox sis aa 6 Oo Virotan 
r 


ee 


mosxa ailvesr oft otttauad « 


ib YLismebivs, ¥. ys iotdstive “= 
it. 


7 bal 
mie 20 moiiniovs tind 


. 
ad am or, 


fSsssqncao ‘Is nusouaite peda 


,f2ainsge tresbon ak ow on 
witoideth (a asp cae EX 


4 


beaeoio ak Jivgeor Siriw ‘em atm 


fan 5 


aid we sas ef 





26 indicate the comparative rate of loss of atonic vowels in con- 
ditions conducive to open syllabicity as opposed to the sole re- 


quirement of phonotactic compatability. 


intertonic posttonic second 
Set oe Sree 










27 20 deli 10 58 63 





Table 26: Portuguese atonic vowels; open syllables 


In Tables 27-29, these data are further broken down by vowels, but 


due to the small number of examples, no meaningful results may be 


obtained, except in the case of /o/://. 


intertonic posttonic second 
Se a Saas 





Table 27: Portuguese /a/; open syllables 


i bl 










at 


, ag) : 
- eae n oy 


nt ziewov_2inate to aba 3 oer 





= incase | sane 
sknossorr |  SEenSqy. 


echisilye meqo +elswov.ofmods sash ats iete 


. ae 
did. ~elswor i meob nevind aoftayh sis sdsb ead pe 


ae 2 
tines: iftoninsen on ,selquske Io —— ncaa 


al antes topes, be 


ad 2 rer on 
a ’ 

Q | ae 
nee) Mee eR 
1 i 

; 











11h 













intertonic — second 


pretonic 
oe Os mace Os aes OS _ 
<aiaied a ae a a 


Table 28: Portuguese /e/-/i/; open syllables 


aes poss 
Sige eae 
NS 
ee eee 


Table 29: Portuguese /o/—/u/; open syllables 


posttonic 
penult 


















In the case of the pair /o/:/u/, almost exactly the same rate of 
retention appears, whether only open syllabicity is considered, or 
whether the results are evaluated in the more general context of 
phonotactic compatability. From these figures it must be concluded 
that the theory of maximum open syllabicity has not been confined 
by the data from Portuguese, although in view of the relative paucity 
of acceptable examples, it must be conceded that the theory has not 


been disconfinned either. 
30 / @ Possible physical correlates 


The data which have been collected indicate that the early 










20 “it 


ee RE Ee 


: 
ex | 


| 
| 
| 


—ooO oor 
eo 


(\z 


| 
: 
Lt 


! 





2ofds Live asgo ;\i\~\s\ seeupitapd~ 38 








— a a 
a " EIN panes 
} : a Oe em | 





} i eo as of. 

to sfe1 omse | vigosse dsanis .WoN OO’. rhe at tos = ae 
ree 

1 ,Peteblanco ei yoioldelive nego Yino 790 ecw, eR Right 


30 4485 fe ote 
t 
yiionmo sad + sf aoa) ee te 
Sri To C aE BESO uno. 
wt boty Ss 3 if Y | 5 Siw 5 





“a 7 


eae by sustt adsoi Teta 
7 vee Pg J * ry i i wa 
oY Po ‘Dee 

of a + Cin oe os - 
LA hax Dee 

re oe “8 | oe ee eee 





115 


history of Italian, Spanish and Portuguese was categorized by a 
hierarchical behavior of atonic vowel positions, as well as of 
individual vowels. The question still remains, however, of the 
basis for such differential behavior. Since the notion of diachro- 
nic hierarchies may be introduced at the phonological level, it is 
not essential to the argument of this study that any attempt be 
made to determine the physical or empirical correlates of such pho- 
nological indices. Indeed, it is impossible to offer any empirical 
evidence whatsoever, except for the contemporary languages, which 
do not necessarily bear any resemblance to earlier stages of the 
same languages. On the other hand, in Chapter Five an attempt will 
be made to show the action of the phonological hierarchies described 
above during a later period of the Italian language, which pre- 
supposes a relatively constant phonotactic configuration throughout 
the time span in question. If it could somehow be demonstrated 
that one or more of the modern Romance languages also exhibits 
essentially the same hierarchical arrangement of atonic positions 
isolated for their earlier periods, it would be possible to infer the 
continued existence of this hierarchy over a larger segment of the 
diachronic axis, and would further support claims referring to in- 
temmediate periods. At the present time, given the almost complete 
lack of data on the physical correlates of phonological hierarchiza- 
tion, it is not immediately obvious exactly what parameters should 
be studied in order to approach the question. There is probably 
some sort of articulatory basis for the proposed hierarchies, since 
it is by means of spoken utterances that the various sound changes 
which point to them have been realized. On the other hand, the 


hierarchy probably also has a set of psycho-acoustic correlates, in 












ory #0 | seavewar veniemes 0. 


a ~outbeid 20 oniton sith Scale Os 
at ti , foveal inspec off de 





od tape sch a 


~aig bos 16 pEtibigiios Lait srt sna os 
| ? Pe 
/ - Sooduiivens ys tstto on ‘stelaiieoae ak a (pester. vessibad tec ey 
| ae 
| bisiw. yasnenpnsl EERO IED aid = ere: espana § on 26 ve 















a 
1 
" 
i 
+ 


: gi 20 eopéete oir ar consti Vi} 
a Tliw sandces te svil. xechget mi bass pitta oxtt a Serge en 
A bediimeeh soitssreis Leotpolonedg eft 26 nclbios ort cide 8 f 


fe Suodpierlt cotisums soo oitsstoaodd Jtaedenoo yieviseiot 8 seaog 
ws : i. ah eg 
‘se beceytanonéh ad worlemoe blyco SE ID »noitesip at nage amis 
“|. atichins oafe acpecpos! sonama ctabom set 2o elem forane: 


‘ fi 
oistaoq sinois jo daemeorsrts {soidoxs1eut ones 268 









7 / + - 
ca S$ ink. od aldraeog sd Shiw J2., ebofisq rotinse wedi = gS. Aa 
wr “se (om) 7 _ 
TH Lae 
—) - sit io tnempee ssp1el 6 xavo \oasxair etxt to oe SOLO 
. ° ; ; - 


+ q ti ot parrister entais-jzoowe, rst et 
7 stele temiis aii novip anid ene: oid 3A es 






 —§zirbousxoid Leakpatororig <0 pens eS Iso teyn vat 
Biuoie astomersd 2érhw voomts — ba iS 


view of the inherent auditory feedback present in all normal indi- 
viduals. 


In view of the near impossibility of obtaining any meaningful 
articulatory data, it was decided to attempt a rudimentary study 
of the acoustic properties of atonic vowels in modern Spanish and 
Italian. Unlike Spanish and Italian, the prosodics of Portuguese 


have undergone a substantial change since the early periods, since 


beginning with the 16th century, it has been subjected to an increa- 


singly strong stress accent, causing a high rate of syncope and 
apocope not yet recorded in the orthography, except in a few forms 
such as para> colloquial pra.1? In consequence, little would be 


gained by undertaking a phonetic survey of modern Portuguese, since 


the modern dialects, as well as being highly diversified among them- 


selves, are not an accurate reflection of the earlier periods of 
the language. Therefore, the diachronic data must stand as the 
sole testimony of phonological hierarchization in Portuguese. 

In keeping with the traditional assumptions about hierarchies 
of atonic position, length and intensity of the vowels in question 
were regarded as the experimental variables. The data were pro- 
vided by two educated speakers of standard Italian (Tuscan) and 
two educated speakers of Castilian. Each speaker was asked to 
read a list of words, carefully selected so as to provide all the 
required environments in the shortest period of time. The infor- 
Mants were recorded in a studio, utilizing high quality recording 
equipment. ++ Each vowel in a key environment was either word-ini- 
tial followed by a voiceless consonant, word-final preceded by a 


voiceless consonant, or word-medial surrounded by voiceless conso- 


nants. This was done in order to eliminate spurious lengthening 


116 
















Je 


stepdnod to solkdsoug, at ests 
sents ,eborreq yiase ort ah eae siloaelsneela 
el is ot betostdie naed esd Bia Vande AOE edt ‘eee 
bye aqcony? To tex dpi 6 pibadeo” esooe ate : r 
amxot wat snk seraoxe yogerpeniio mi cs bobacomn 30 Jon Sg a 
of Hitow alistt | S2ATSTPSRICO a OF su niwpotton <a as deve ie 
a coimaely s2etputsod nushon 3c yen, aktonorey: o eaesabeus el Boney 7 “ 
—marit ‘peas boitietwib yinped ‘Priied eS, fiew 26 vetoes i mssbom stg ia 
16 afloat: aq weiltss ott to notigsiies stemoop ts om, sesetintna, 7 | 

eft 2s baste jean ateh oineniieskb ond sxdiiensdt ee ae 
Ssasuputied ni sites icotersint isorpolonoig 30. yronttags adem’ * a 

sei tertenp ts Jods anoitgnsses isnot iheat atte iow pabgoent AE ary el 


maigasup at alewov eit to‘ rensdat ioiangeareii 4a 


sony exew steh ed? -eoldstaey [sd remnsgKs es 








¥ 
re 


. 
7 
: 
a 


-_ 














bas (msoewT) oorleti. ictnadeo aie . 
: o¢ Devas abw toslssce doad mena 
si ee cm Sa a 
~roagt sy +aRSt Ao oles 
ptibrone: lee sina tae orem ee 
sink-byow 2sttis me eon pene — a ena 
6 et babapesg Latit~Guans 9 eae" ovis od vant 
pane Soot cbem-biow 30, SriatORO 





babe 


effects caused by the proximity of voiced segments, and also to 
ensure that the vowels (as voiced segments) could be clearly ob- 
served on the tracings. While it would have been desirable to uti- 
lize a greater number of subjects, the results obtained from the 
informants converge in such a fashion as to suggest non-randommess 
of results. Each word was read twice by each subject, to ensure 
that no production errors had inadvertently been recorded into the 
data. The figures presented below represent the arithmetic mean of 
the results obtained on each of the two trials, except where a pro- 
duction error, such as a woiceless or deleted vowel, made such a 
computation impossible. In all cases except those involving actual 
errors, the values for each of the two trials were sufficiently close 
as to make the arithmetic mean a significant figure. When more than 
one figure appears in a given box, it is because the set of words 
used as test tokens offered more than one example of a particular 
environment. In general, an attempt was made to study the behavior 
OretieavOwelSeaysc, 0) ,sOrdiaueinethe initial pretonic, initial /pre- 
pretonic, second syllable pretonic, amd posttonic penult positions. 
As a cross-check on these figures, measurements were made of the 
stressed vowels and final vowels; however, these figures are not 
presented in the tables below, but merely serve as part of the veri- 
fication for the data of central relevance. 

Once recorded, the tapes were subjected to an analysis on the 
Mingograf.1? The Mingograf is a device which records, by means of 
a set of inked styli, various aspects of an incoming audio signal 
on a moving roll of paper. Four tracks are provided: the fundamental 
frequency or voicing track, unfiltered intensity, filtered intensity, 


and an oscillogram of the entire audio signal served as input. ‘he 



















eg cals bra -atasmpse beok 


~do yfusaio ed bluan (esas pp 


— a 

“ity of sfcdsziaab reed overt bia’ eae cfd ort a 

dey . sere: 
es. i ovis testne 
“sures of a > 
to ofa Bebrooer nocd yinec-evied Bed Bae 

“ony 6 sxarle jqeaxe — ee 


to mem oitemttins art. tnseenqex woled’ betneaang. 
& Howe abam , fewoy bavelsb to zaalsoiov 6 2 es hua tors noidodb at ; 


pin iertle ea 











Laurin erty Lerwrnt S20! it to80xe eoeB6) ifs a fainoomnt ieoate): : 

a 
deol yisnetoEtne ars alsttt owt aes 30 doke 10% soulev ew ort oe a 
nectd ‘exam neji = .smpil toanoltinpie 5 nent. oitanit ius et y eee 7 * 


gbaow 20 gee off sevens! ai tL cleci attest cause 
tsleobhieg 5 16 sigmaxs sno ssi srcar Berzsitto anoles Pet a 
sorveried ert yiurt2 of absem sow tegnstt5 NS texena ot “stramoxives 


eA eae oh 


~eng [sisini olicisig ieising ye nt. ran S 4 2 ' alow ait 20 


‘enoisteog tfgneq atnodtaog Eris pines 


"4 
Bh 





ee 


aft to sbem orsw etrameivessm . oor oat 
a: 






7 


jon sxe eewpit seadt ,xgvowor + abowov _ 
.? +s) 


-frev offf To Jug es ovise yistan jpd \woled aefctis _, i 


ru a 
—_ - 7 
= : ran “@ ry, . 


oe 
ord no afaylais ne ot botoopdve sr ooo 
30 Siete ropa clin eb af ‘rpc 


oun as 
a : 





118 


time base at which the material is recorded on the paper may be set, 
thus allowing a ready measurement of segmental length; in the present 
case, a time base of 50 mm/sec. was found to be adequate. It is 

also possible to run a calibration of intensity for each tape that 

is played through the Mingograf, permitting relative differences in 
intensity to be computed with a high degree of accuracy. 

Vowel length was measured in terms of the fundamental, or woi- 
cing band, and cross-checked with the intensity bands, thus necessi- 
tating that each vowel be placed in the vicinity of voiceless seg- 
ments. Intensity was computed through the filtered intensity track, 
and cross-checked against the other tracks. The oscillogram was not 
utilized at all, except as an additional check on the time base. 

The results for speaker A (an Italian woman) are shown in Tables 
30 and 31; those for speaker B (an Italian man) in Tables 32 and 33; 
the data for speaker C (a Spanish woman) are found in Tables 34 and 


35, while those for speaker D (a Spanish man) appear in Tables 37 


and 38. 


initial 
pre-pretonic 
95/100 





Table 30: Speaker A (Italian); vowel length in ms. 













sanaiaak su ite its mn dape mi Na 

jnseany add ai en on ssi x 7 : 
saa _steupebs 9d_at/f mot Fst ony ‘ 

tert sq63 fone 108 Rent cys + 


ni aaoteist? ib euiseiar che amen a | 
rate 3o Joerg) 2. rd oe 
niow 10 fecromabeu? ott 9 Bama ene atk. 
Stesaoen audit ,abriad ‘ana A, Deo ‘bas —— 
~pez lazaisoioy 2o Ytidioivishy i beostg sl fawew tse Sect enbtet 
ober wieasint horadle? St REO Paden) ecw \itenadal adnan 
joo usw mexpolticeo od? ..aiget sertio si Fenksps Bevbet>aaoro Bas a 
aed emit sd+ mo pero Ienoisthbs mm a& Iqeoxs vile 29 OL * 
-egide? oi mete oxs (aemaY rsifesl os) A sereege 10h ativest at 
:€€ bs SE zoldsT ot (new obi ietl ms) @ odaege wot peor 16 fas OF 
Bus S€ Usids? ni Bavot ots (stsmow stakes? 8) > xaritage w2 stab at 
TE eolds? ni ansqqs (nem deinsge 6) 0 sedssge tz enor) oLsiw 2 a 
he 


en ms 


a) hg ee rome 7 
Paar El 

ap edtsts thot s —4 cab a : 

 oeyoe | OL | ag ee .. 

Se rca " a: ca Re 90 “a ine 2 : f 


; - WS) ee haan 
Oe a "00: ae. eet 


ees ee 0 | ones fe 


is 























i. 
“i 











| Sinedseccy 

















Table 31: Speaker A (Italian); vowel intensity in db. down 













initial 
pre-pretonic 






initial second 
pretonic pretonic 
00 





Ae) 
mere 2 | eee 





Table 32: Speaker B (Italian); vowel length in ms. 
|| repretmic | pretonic | prounic | femte | 
pre-pretonic | pretonic pretonic | penult 
Se ee 
Geel le ee te 
Ss te 
Ween |. ee 


Table 33: Speaker B (Italian); vowel intensity in db. down 





119 















Es As 


mob .db ak watoesi Tawov ; (casting) <snets “i ater oe 


, $ ee) aa, viseen 
Miaitc esp loo] | Broose a 
. tfuneq joinetaig. Ot Ssro 


omc | OLE OY Re 


eee 





a arn ainoe-aillll MT = 
DLOGTIEOG Srcoss isichir 
phurisg | Sinotsxg pinay 


aa 





Oper 
eae 
a 





120 


pre=-pretonic pretonic i 
A 






Table 34: Speaker C (Spanish); vowel length in ms. 


initia aaa second posttonic 
adi Bees pretonic 








Table 35: Speaker C (Spanish); vowel intensity in db. down 


posttonic 
penult 





Table 36: Speaker D (Spanish); vowel length in ms. 


7a ee " A iT i ; 
ie last ‘ ot & ‘ 
rary can y 
i ¥, gt a "i -_ 

, aA 7 

| oie =" via OSE * 


+ 


omeseag 


siunsg 


_ 


a 


pan 


oa 
— 
Ue» 
a 


ee 2 ee yy ot . 
<< 
as is alia a) ree | 


‘; 
_- aes 


4 
~ ; . 
or on © 





lai 









Initial initial second posttonic 
pre-pretonic pretonic oretonic penult 
ils aA Gas inte 


Table 37: Speaker D (Spanish); vowel intensity in db. down 





From these tables it may be seen that no significant correlation 

in terms of vowel length may be observed with respect to the various 

atonic environments; indeed, in many cases the longest vowel was 

the final unstregsed, while the shortest vowel was the tonic. In 

Italian and Spanish, vowel length is largely determined by the kind 

and number of following Sonsonants,** and hence appears to have 

little to do with the hierarchical behavior of unstressed vowels. 
Turning now to the relative intensity of the unstressed vowels, 

it may be seen that there is indeed a correlation based on the various 

atonic positions. The figures reported are in decibels down from 

the intensity of the tonic vowel. The figure in decibels, a measure 

of the relative intensity of two signals, is ten times the common 

logarithm of the ratio of their powers. That is, given two signals 

S; and S 


, with powers P, and P,, respectively, the difference in 


2 2 
decibels between S) and Sy is expressed by the formula: 


‘2 
Thus, for example, a figure for an unstressed vowel of -10 db. means 
























aetna each 


noijsl orico jes tase on tert ase ad a eae 
Te. 0 2 bee 
BuO rss9 ent oF sosqest ae parisedo ext ean sthoeat Igwow 20 
eu 
egw Lewov Secpnol oft akeao spite and »beehat retasnnoatvns ¢ 


: is el 


aI .otocet ort asw iowov teedyode ont afiriw \boetontery I 
bait ott vd borcimrateb mics? . ritpnel Lowes dake? brs 


- tein 


aver ot e1ss0qs sored one © \atnemoenao pniwollat to 
.2lgwov beeacttart-to toiveried Iso tbusxais erlt (iw ob oe . 
* oe 


,aletw Borastsanu ods to es gt eviteles eit Od wore peuierasT 
“ikieRey arti no boasd notte totxos 5! besabat ai stedt | 2 me ao 


mort rin afedioeb Ss 
fmasem 6 ,alodioeb at siupsrt ort -fowow > 


ee en et alsqpie) ait 20 9 


efstpie owl corto vei 3enr a le 
ni somexeitib oft .yfevitseqger 3 fs 


: eee a z 2 ae a ; - hee > 
6liimoi ocli yd basesiqes, eb ce & tS {¢ 
dy | (orn a i Ve ae ei “a 

i * Bi are 


> & — aan 
ae ; : > oh 


7 
a 
es 


prea 

















» : 
wil a 


‘a 


38 


that the stressed vowel has ten times as much power as the unstressed 
vowel, a figure of -9 db. means the stressed vowel has approximately 
7.9 times as much power, a figure of -3 db. means that the stressed 
vowel is approximately 1.99 times as powerful, etc. From these 
figures it is seen that there exists a clear hierarchical tendency 
with respect to relative intensity, with the initial pretonic syl- 
lable being the strongest, followed closely by the initial pre-pre- 
tonic, and then by the remaining positions. ‘These differences in 
relative intensity are striking enough to be of some significance, 
Since they demonstrate, in a rudimentary fashion to be sure, the con- 
Siderable difference in acoustic power between the various atonic 
environments inItahian and Spanish, thus hinting at one possible 
physical correlate for the hierarchical behavior of these environ- 
ments. These results are non-conclusive, but they do lend support 
to the common suspicion that hierarchies of position are somehow 
related to relative intensity. 

As an aside at this point, it may be added that, in contrast 
to the rather consistent behavior displayed by vowels in other posi- 
tions, final atonic vowels displayed a tremendous range of varia- 
tion, with relative intensity figures ranging from -2 db. to -9 c., 
in an apparently random fashion. ‘These figures again highlight the 
fact that the behavior of final atonic vowels in Italian and Spanish 
is dependent upon more than purely physical correlates, perhaps in 
view of the morphological functions carried out by vowels in this 
position. 

It may also be noted that there is no significant correlation 
between vowel timbre and either length or intensity, thus implica- 


ting possible articulatory factors in the determination of the 


122 






panera an? 26 seg st 
Vietenixemnags eer! ao 
bewascte act dadd emmeit + £~20 
aaortt mort .cdS {Bitehog as 2 | bo 
youshtar feo rbekregas if aio al isang ; ) ted can ce 
Hiya stinciinag 16tAHAR ore attr aes oviieler oo 3 
awto-axy feisini sdt yd vieadio Beto Lt ee a ted - oS dad 
fic atoustSitib Seon " paoksieag preithentes ads “Noi nest brs, 98 | 
condo itinpic oz to. ed af Aivode Set abr Rps- 2 <) 
ype erit .ogue sd ot moites? Vase 6 rte sextencmab, yard sonie a ‘é 
nirets svobisy sit «sewted yowog oi Fas at cocenetib aldebie he 
aldiaaag sno ts ont daiba eairtt. datas, bas on Ube atk ‘ejosmonivas 
Labi s2ar} to azNive df Leokremembia ont tol anelersbo, Leokayg i “3 
asain ab yards tud javian iaeus oxs eclgset s2odt Snsm fl 
woriaibe sis noitiscd 40 eSitorssefit ets roioiqeue oro Pa | 
| | “ys Legace oulgad ay <a “2 
peervtacio tit tert Bbebbs od yer 3k ,anieg ebr Sey, shies 5a ’ 
4 “hog teio i afewou vd hovel jail tohystied: satin tes et 

























en “pimey IQ gprisx sackneneys-é bevels ee. — a 
7 (<0 8- & .db S= fost porteass eon ikanaiai By ‘ 


ne at sietiiptd piaps zecp Et seat? ae checemaneall — 
' 


neinege big nahedi ni! = Lowe a ual 
mc Bgetiod sede Loria. isnlevrig 
aint nt aflawov be said eer 


ot - 


















= 





7 


4 


——- pe 





micaninmiibch i 


Le 


RP selemen ak 
| de nregnen sit 


nea. ie) 


a * 
7 7 » - ar 4 - i 
ais my > : y die v ail is wv i 
~ one ; : 7 7 ‘uhe : 

J ae) on a ae 


123 


relative intrinsic strength of the individual vowels. Bull (1965: 


71) noted the following, in regard to the Spanish vowels: 


The so-called degree of audibility of vowels is entirely 
irrelevant to their phonemic fumction. The [i] of 

pino is heard just as clearly as the [o]. ‘the pho- 
nemic contrast between puso and piso is just as clear 
as the contrast between puso and puse. ‘The strength 

of a vowel is not a phonetic feature which can be 


related to its phonemic function. 


While these remarks are to a certain degree justified by the lack 
of correspondence between vowel timbre and such concomitants as 
length and stress, it is still true that, froma perceptual point 
of view, there do indeed exist differences between the various 


vowels, aS pointed out, for example, by Navarro Tom&s (1971: 27) ae 


A las vocales mAs perceptibles (a, oO, 2) se les suele 
llamar fuertes y llenas, y a las menos perceptibles (i, 
u) d&ébiles; la naturaleza de la perceptibilidad no tiene 
relaci6n ninguna, sin embargo, con la idea de fuerza 

o intensidad articulatoria que estas denominaciones 


sugieren. 


Moreover, perceptual differences clearly exist between stressed and 

unstressed syllables, and between the various atonic positions, but 

from a strictly phonemic point of view, the mere presence of a vowel 
in such a position is sufficient to place it in phonemic opposition 

with the other segments which could potentially occur in the same 


environment. 





lee * Abentbu LeaphBt 


= pay wee 
salawcy 





detmece” ett” 


2 H 
er C cari zoe) apals 2 4 
. 0) ehh es vianelo 28 38) 


boii tiiauvr sexsi nrariss 6 td Bare) aokxems 5 


rLidtt+asexsq 6. eh saeiswioan ef ¥ sob ici : 









a, a 


>it FO 
rhs it be net 


gh 


1 he ae 
Nhe y twa . caletal ict 4 
PLE fie 
‘ ‘ ve “es 


na ae pie Us 





- 





eeu 



















116 8L wfleiwov, To, oo ehaee nated reab batito- 


ah 
x 


ot? Os Yoru noe or hed ab 9 


f 


uj ei deig Rit ome ae nscwied deexinen * > ima 


i 


_~ a - 
foidw sticsst nitenoig & Jon =“ ined C 
Ae ioe! a. 


~ ie 
ont oitesrorky att at be 
ai 


fT? .sevc Bas. cage viepiatees dd Junk 42100 cI 
; f : _ 


f 
’ - 


a i ; = 
Soe eicmit’ fewer aéendext op 


. é | oo i 
se (9 ,0 ,5) eatdiiqegrsd eam 23 LEOOV BA 


coq ecoyen abl 5.y .esmeNl Y¥ Beowent ascaeet 3 ett 


, Cae 


oa 
b asbt at me voprachs) pte. paemrie Dio a 


BIS | vVitsit masog sae fs 








4iGo Sates 


sis off ye aiv qos 


sa | 
a ae ; 


Si sealq oe pot 
eel foi é 
: : wi) 


Si 


Notes to Chapter Three 


The complete study is Anderson (1963). Earlier definitions of 
phonotactic compatibility are found in Saporta and Olson (1958), 


Fischer-Jgrgensen (1954) and Vogt (1954). 
Cf. Tai Whan Kim (1965: 31-2). 


C£. Devoto (1930) and Brgndal (1940) for a more detailed treat- 
ment of the origin, evolution, and contemporary manifestations 


Ofetheyhatingprefixtex= iin iltalian, 
Cf. Grandgent (1934: 101) and Menéndez Pidal (1966: 76-7). 


For a complete survey of the evolution of initial pl, cl, and fl 


in Portuguese, see Malkiel (1963-64). 


Cf. Naro (1971b) for some considerations concerning the possible 


routes of evolution leading to the modern raised variants. 


A possible example may be provided by vulgar Puerto Rican (and 
some dialects of Cuba) in which atonic e is frequently raised to 
i, while atonic o is never affected. Thus, for example, the inha- 
bitants of the town of Lares, in Puerto Rico, refer to themselves 


as "di Lari". See also Navarro Tom&s (1948), Malaret (1955: 27-9). 


Other problems both general and specific, engendered by Klausen- 
burger's treatment of French phonotactics, are noted in Rochet 


(forthcoming) . 


See, for example, Alonso and Lida- (1945), Robe (1959) , Matluck 


(1960). and Zamora Vicente (1967). 






















i steett belisteh aon 6, 10% jal se 880 
enoitnjest inn yusxogned (to bas vastsatore ,aipbx0 ato da 





- STR ELS boeiex mrradcn orks « or giao 


oF boetsx  iebiialoite at Ss oinbite uth 


“pitt of ,olqmexe sot , oir ih 


‘ savicemad) of ar9tea \oois odtaud ai 
TARTS 228OL) soxnlem , (2deD edn i a 8 
ah at 
rigs lA yd fig abit ri 


¥, cdi +. ae - iA 
| oe ioe sen ere 
tarot a bason 6° “82 Biot aC ill “ . at te da wc ; L fan ; : 


= i)" : 
a ‘ 4 





10 


alban 


12 


3 


14 


Leo 


Clo Williams (1962: 58)x 


Thanks are due to the staff of the language laboratories at 
the University of Alberta for permission to use their recording 


studio, and for assistance with the recordings. 


The Mingograf was made available through the courtesy of the 


Department of Linguistics of the University of Alberta. 


For Spanish, see Navarro Tomds (1916, 1917). For Italian, cf. 
Josselyn (1900), Metz (1914), Gregorio (1932), Parmenter and 


Carmen (1932), and Rosenzweig (1965). 


For English, similar data are offered by Wickelgren (1965). 


, wombs ae) 


Cae 


si Io yassioo edt ioveratts efdel tev sien Re 





a) 


5 to vite er) ors o eniveiepait 2 


72 


cs 
=. 
. 


i 


A ; =< a i al i - 
sk Sd Ams cies T id @ > . (PLleL ol ref) aia iT cr I6V5 { oni . 
7 4 '> ; 
hes 1atnamed , (SECL) otopeta , (Per) stat , (O02L) o 

— a _ 
. (280L) piowsnesct bres. = roms 


er ze 
ere et 
detemng? 10% 
- °F - - 
cary t 
baie 
Sea0u 


J 
ay 





a i ae 
-(@0@[) nayoiexbsW. vd Bexeito exs ateB > 
; 
1 
= sy 
x 7 
o 4 
cae ( 
- oi Ay) 
Mi iy ’ 
7 
ny i - 3! 
‘ on - 
ot oad ; 4 8 
[Es eae ae) ae 7 
av, eee 
> oe hoe rr ie 


CHAPTER FOUR 
FINAL VOWELS AND THE EVIDENCE FROM CATALAN 


4.1 Introduction 


The preceding chapters have served to establish certain hierar- 
chies observable among the various atonic vocalic positions comprising 
words in early Spanish, Italian and Portuguese. From the relatively 
high degree of congruence of results which emerges from a considera- 
tion of these three languages, one is led to believe that investiga- 
tion of additional Romance languages would yield comparable data, 
given the common denominator of Latin. The potential eae ofthis 
diachronic strength hierarchy has been signalled in various word-inter- 
nal positions, where it has been seen that differences in intrinsic 
phonological strength may manifest themselves in a highly differential 
behavior of segments occupying the positions in question, thus account- 
ing for otherwise puzzling events. The rudimentary phonetic observa- 
tions which have been presented suggest that, in most cases, relative 
phonological strength of atonic vowels is rather closely related to 
their relative intensity with respect to the tonic vowel. Throughout 
the course of the preceding chapters, however, there has been one con- 
sistent exception made to all observations and measurements; namely, 
the final atonic vowels. Final vowels were not included in the compu- 
tations of rate of loss or retention, since it was asserted that addi- 
tional factors, probably of a grammatical nature, had prevented vowels 
in final position from participating in events generally characterizing 
the fate of atonic vowels in other positions. Moreover, phonetic data 


concerning final atonic vowels showed an extreme degree of variability 


with respect to relative intensity, and the relative intensity of 


126 


: OE or 5 a 7 " 
—— 7 — 
: : _ a Bae ic 
~~ ; yd. i. a 


- 


« wiioena a) woes. ¥ ore 7 a aw { 2 2 Mic. OV ld vrs 
f ; 





















s 
‘ ~! Ne a i aes 
~tatein chetre get [dst2° oF beret vat esters ert 4% 
; P'S SP eal 
- a 
viewitsiox ods mor . oes TG Sas mss fu vince z 


-ptebianco 5 orl eopisme mm inw dalle ia 3 ciated 0 
: ce 7 ea ae 





: z 
55 sidevegnoo bloty bivow espegp ony SO MSOF tar 


art 20 noitos Isitmesoq ad? .aldml Io tedetiinoneb « 


“rstiti-biew auoizsey af boelfsapia need eed yess sisis repaende 


olaniviat ait 2oonsrettib tant nsae asad ast ta exotiw mee 
« ; as da P ‘ Baia 
isisnexstich yirpid s ot aavisanstt fest Canm Vem dicate! SLX 
eG. 

“dmyooos aut .noisvacup ne anortteog art patos eiramgee 2 07 


avyeemw saitenor istnambsy otf jechevs phd sire 


ot hadeiox yiseoio sorter ei elewov oLiods - eens Lae vise potos 

duotipucnt wov cinco sit o¢ tosde St, b aw atic os 
a 

km snd need asd it tavern tae 5 0 pal tasosig, ad to 


a Ret 
Yiemsa <etosnenuesem £ erriseddo ile od shen 
aa 


Aes ait mi Soboions Sou stow ale ae Keni Ca 21 eWOV DE 
. ¢ air 7) 


nibs Jet farszes ecw 4i sone aoinetes ao. ot 36 
’ mar’ mle | 

diewov Bednevexc bel .emrtss Isotte Gg Xk ian 
7 : 4 ee mse - Pe ae 


pnisiuedsaiusio yilsreade atneve rt 


7 


— rosie tevesiemM =, afiots teoy een ws | 
_¥ ; Be — 


ep 


| yet Cider Lay Ro Qezy spb ONL 42) 6 beworle 2) 


pi Ro yotanetmré a ay acid “BOs 4) 
2 —- ” : oe 
Pia , 

ne | 


Bh ihr: Sic lee 





final vowels was generally mach lower than would be anticipated in 
view of their diachronic phonological behavior, which affords final 
vowels a very high position on the strength scale. It now remains 

to explore in somewhat greater detail the relationship between the dia- 
chronic phonological hierarchy already isolated and the inherent 
strength apparently enjoyed by final atonic vowels in the Romance lan- 


guages under consideration. 


For the sake of point-by-point comparison, it is useful to rei- 
terate some of the facts which have been stated in earlier chapters 
regarding the behavior of final atonic vowels. In addition to the 
three languages which have been discussed up until this point, a fourth 
language, Catalan, will be included in the comparisons, in order to 
place the data from the other three languages in a clearer perspective. 

In Italian, as has been noted, Vulgar Latin final vowels were 
generally preserved intact, although with various phonetic modifica- 
tions depending on the dialect and on the time period involved. Excep- 
tions to this general statement are quite frequently the result of 
haplology , 7 (e.g. piede > pie, mercedey merce) or a consequence of being 
found in the middle of a breath group (amor mio, ben bene). It is, 
of course, noteworthy that in many regional Italian dialects, final 
atonic vowels have been lost, or have persisted only as the merest hint 
of an audible segment. This matter will be briefly returned to below. 
In the majority of cases, it was final e which was lost in standard 
Italian, although in a few established cases, the vowel o was also 


lost, in the middle of a syntactically-united breath group as in buon, 


ane, 
















~ta; ot iifisen et 35 UES ! 
pare vabiuss nk badete nosd avert ius as eee tg 
edt of molsiboe sr atowoy ‘otners Lsatt aoe 
dew?’ 5 dniog etdt Lhiaree Béenueih Heed eva ose cco awit 7 : ) 
et yebro ai ,eno2tteqneo Sih ae pebutssti ad Ifiw , pee 


Bisoageiog iowsslo 6.1 espruprpd costs to wt war aia el i 
Yer me 
even-alewov Iscii nided tapi Bacto need esd as ae y, ; 


er 


soltiban sitenorg avotiey i iw epevests Ls fost boraooaa 
—qeoxt 3=.fovioumi bors anit tno bas JoaetB wh mp if - soe} 
wat nae 
to sivesi ab las i ediup ots jummeseve Snxonee « 


pitted to soitaupoenco 6 so (Soom <sbeervem. Sig! er) aver oat ¥' 
» oe “sy 
, ab +I . (gitect pec ON ans) wore disend 8 ears ui ben eas . ) 


hart . ghvstsih net iss T Lemmotpen. re ne sent aa ac o* 
; 7 >» » _ : 
2301 - = s Ss 05 | al on ee 4 7 


deck Jaane ois ap vse base aod own oboe. 


‘jeied os Reerpsson yitetud ae i se — : re oe oe us 15° 


— 











burbanle oi 320 sp cote 9 i) | 3k ei 2 eee an 


ouls iy a Tomar ads anes : se we 


eet ow tn al ante 
haaek 
ra id 


i 
eo 





Pp : a 





‘ , : 


128 


bel, quel, andiam via, etc. 

In Spanish, final atonic vowels remained intact as a general 
rule, with the exception of final e. Spanish final enregulanly are L|. 
after intervocalic c, d, 1, n, r, s, and t, remaining in other cases. 
In a few cases, final o was lost, possibly through the action of addi- 
tional factors; the most well-known case is ministeriu > menester. 
Final unstressed a, in keeping with its overall behavior as the dia- 
chronically 'strongest' vowel, remained in Spanish. 

In Portuguese, final vowels were, as a rule, retained more fre- 
quently than in Spanish; however, final e usually fell after intervoca- 
lic 1 (in the Vulgar Latin period), n (in the Vulgar Latin period), r, 
Ss, ae a Later developments, however, served to severely reduce 
final (and non-final) atonic vowels; a is generally reduced to [?] 
throughout the Portuguese-speaking world, unstressed o reduces to [u], 
while unstressed e becomes [®] in European Portuguese and [i] in most 
epess OL brazil isthe modern dialects; . the reduced fom of je and 
© are weakly articulated, and both often fall in casual oe The 
tendency toward vowel reduction appears to be more dominant in European 
Portuguese, while the phonetically more conservative dialects of Bra- 
zil tend to put greater articulatory energy into the pronunciation of 
atonic vowels. In the Carioca dialect of Rio de Janeiro and its envi- 
rons, however, the relative strength of the tonic syllable is rapidly 
increasing, with the result that initial and final atonic vowels are 


frequently truncated in rapid Asean ea 


4.3 Atonic vowels in Catalan 


Catalan” provides a striking contrast to the developments just 


Epteiep 5 36 set 
[ist yiasluper © faut A 
.eeck> otto al palsies ee “ae es 

=iihs Yo HOLDS aig euros alee 
| Agteattedn< cca ei aapihiphains 
~s iD sil 25 tolversd Steeda atis athe ehieoot mi 48 
Aetieae, mck bamisine Lewy | Miheiie Isoinexio’ 
“wat gucm baniesot Slot 6 Be \Stow = nomena a 

Soavxetni watts List yiisues = Ieart cetbhia wiaineg® nt acts | 

ra (bos3q ridel sspiov. sit ah) iv \ (one pens 

Sone exes, of panes revewor ae | 

a ot beoubsr yitsiensp siis selewor inchs (eetnabn are ; 


«hin od eepuher o besvextderny bitow pnibissqe-saaueutao8 er yuckt 
geom fi ij] bts sesuputicd naeqesiel ab [&}- esmoad 9 boeascter iw » 






























) ‘ Dis 9 90 anor boovbs: sri ,etoafis ib rested aid mid - <r e 
; - - v1 ‘tz 7 

a ont : eadt fau2so ot List nedte tod pris baa eaeaien bar ; 

bat } ei sear! 7 

a (Ssqoms cit tranimob. sion of od Core a caisson o 24 


“B18 to atjosisih sv Lisvreence sic wUiaaldocor 
to noLseisrunera ot odat yorerts: eoshlnias 8 
“ivac, ai Bus oxtensl ob olf 30 isle ageited) 


Dita ae 


wWoiget st sidsltye oincd oy 20 are laighed 


ons alewov aknbits Leni? bos Ssisime pi 







129 


mentioned. It was not included in the primary data computations for 
two interlocking reasons. First of all, Catalan has, by virtue of 
its political and geographical history, been strongly influenced by 
Spanish, as well as by Italian, Provencal and French, thus rendering 
accurate etymological tracing impossible in many instances. Second, 
and perhaps as a consequence of the first reason, a long controversy, 
still far from resolved, has raged over whether Catalan should be clas- 
sified as Ibero-Romance, together with Spanish, or as Gallo-Romance, 
along with Provencal. The result of this dispute has been a series of 
conflicting, restrictive, and highly opinionated works on the Catalan 
language, which makes it extremely difficult to trace its linguistic 
history free from extraneous considerations. These two factors com 
bine to yield a lack of sufficient etymological data upon which a sta- 
tistical investigation could be based. Nonetheless, since the consi- 
derations to be offered in the present chapter are placed on the mor- 
phological plane, and since they deal primarily with synchronic 
considerations, Catalan data may be profitably utilized to provide an 
instructive contrast. 

Final vowels generally fell in Catalan, with the exception of a, 
which survived, reduced to [9] in most dialects; e.g. d&ma> dona, 
5 naixer, etc. There are, however, a few well-defined exceptions in 
which the final vowel was preserved. Final e was preserved after the 
muta cum liguida combinations stemming from Vulgar Latin: acru >agre, 
fratre> frare, masclu>mascle, dibru> llabre;wetc. Finale alsocne- 
mained when the fall of the posttonic penult turmed proparoxytones 
into Paro tees e comite >comte, decimu > deume, hdspite > hoste, medi- 


cu>metge, paupéru> pobre. Final vowels were also preserved after rr, 






























to adttiv wi , Zar + nba a 
isre soit wae er 4 i = 


—, eat: — 
oes sgeonet ent -yAem all afk: vt x 
Vee 


yemmebenetnico prot 5 nosnee text : 
“ents ad ‘Bivode asieted satitoriw Ivo fates ast \bevioes: mort ist £. 


naekieol {any as x6 ,dakasce thw rerttogcs panera a 
Zo estube 6 19s asi ‘sangeb airtt 20 tineot ‘ear -Lpgnevoxt agkw ene 


detavtin’ iy no atin Batsaoiaiee yltpict be soviioiajens oe 4 
oigecpail ati coset oF $fup Etib vlonterctss dae soon Hoh «Soeur 7 < 
“18 | Bato + Gad Sebdr vanoisereblenca : cane Idxs ified - | 

‘sie 5 rtotriw sogu 6385 Lsotpolangs sinasi Pera to eae « Branca "4 = 

fern ext conte ,eeslarttennk iene od Biles colosplvewnt ieoben 

jon Sri no beostq ets. zed@edd dreassq oct mt : bextuite cf 02 BIOEHRES iv 


oinoxtbiye diiw viiwemixg feeb ve: sonia. bie veil Leptpotrtg 


(sé Sore Ce | ot oekt ear vides Bore od yan cab ease sero tam| 


(6 30 noligqeoxs, odd itiw \nibisteo, abuse Asan | 
BHD SM 3p.2 yatoalsib vacm “3 ae : 
sasevsn ity <uiso 4olob <solap 930 « ——s 


ai Srobiqeoxs beiteb-[lew wot 6 ore 
ott satis batteesrg e6w 9 a a0 ; vise 


wel © < 7 A : ae a ~ | 
i ds, | SN - eee: = 7 













- a ites 


4 


a 
F, — - . 


VEREPE < kt ris trided 0 ae BOB CLS een 3 , 
~si cals 9 ee nods Ry: oe ti. cies presrais 
Bis oe wits 


a <a eee f bias ee 
Re Cs hele oes ee Ve ee 


: blak rae 


130 


most probably in order to preserve the trilled sound: ferru , ferre, 
turre 5 torre, verre > verre, and so forth. Final u in hiatus with a 
preceding vowel remains to preserve the diphthong: Deu, meu. Ihe 
preceding examples show that Catalan is characterized by a signifi- 
cantly higher rate of loss of final atonic vowels than the other three 
languages previously considered, conserving the final vowels in general 
only as a phonotactic 'support’ for a preceding consonant cluster. 
This high rate of loss is not confined to the final position, but is 
characteristic of the entire language, in all diachronic positions, 

in a manner strongly supportive of the diachronic strength hierarchy 
isolated for Spanish, Italian and Portuguese. 

First-syllable atonic vowels in Catalan remained in nearly all 
cases, whether pretonic or pre-pretonic. There were, however, a numr- 
ber of phonetic modifications, depending upon the timbre of the vowel 
and the following segments. / Second-syllable pretonic vowels fared 
much worse, sharing, in fact, a fate comparable to that of final atonic 
vowels. The vowel a was usually conserved in second-syllable position 
(although reduced to [®] in the eastern mainland dialects and in the 
Baleares): calamellu> caramell, paradisu > paradi - The other vowels 
usually succumbed to syncopation: farinariu>famer, bonitate > bondat, 
cerebellu > cervell, pectorale > pitral, laborare > llaurar, etc. Once 
again, however, there are a number of well-defined cases of retention 
of internal pretonic vowels. This vowel was preserved when, according 
to Moll (1952: 97) 'de su desaparici6n resultarfia un grupo de consonan- 
tes diffcil de pronunciar': avellana> avellana, hortulanu> hortoldad, 
petricosu >pedreg6és, etc. It becomes apparent by considering the 


data in greater detail than is possible in this brief survey that 





















fexeome ab elev tate? ot pobatieg blatant a 
| redento domemene parheosig 5 aOR kine om 
at tud ,noitisoq Lanct odd ot Beatties Jon ef weet 40 eet Ao aba wid 

20nd beoq cingaedbsif Lis nl .sosupnst cnithe elit 20 ottabsteeends: | 5 P 
Yrorsxeind dprexts Stcrshis ot Yo nrictroggue weal : 
-sdeupectio? ne) MibLisg? daisge 30% teapot —- 

Hs, yiusen ni benismex neleteo si slow oteste aiAsEiyeaneR ist” oe 
map \ Faved S19" sxedT soknoteig- 91g "0 oinerency xed see By 


few eit Yo sucht 9 goat pABERSS (ASDA ibem oiteandg 209d ie 
beint alewov otaotexg tie Myesbaeose ay Leonenpee pawoLto® edt: Bas “ 








; ofecds fans? ed TED, og, Slusisgicy 5b6 jostot \paixede eT - 
MPitieeq sitteLlyebroose mi Sersans ytlavex SEW B, owe ect town 4 
edt ai bie atsatsib babldten trades ot az, {9} ot beoufen dpuvadats) ry 
=n amen aaa 
iJibaod <sdktinod yxocust cunts tiss SOL BQEDL 
gon ote aps! < sxbmodel ea <stsctoet b 
noitnedten to eeseo nae teed alana | 
| vee oh opens ay meee: ra oo 


wig 


: 
7 
7 





fs 
* 


13h 


Moll's statement is equivalent to the phonotactic compatibility of 
cansonant clusters, providing the same general patterns observed in 
the other languages under discussion. 

Internal pretonic vowels other than a were preserved before s. 
this particular situation are not clear, and will not be considered 
here. 

Morphological considerations appear to have prevailed in preser- 
ving internal pretonic vowels in forms followed by a derivational suf- 
fix, as stated by Moll (1952: 98): 'por conservarse el sentimiento 
de la vocal del primitivo'; e.g. dommitoriu> dormidor (cf. dormir) , 
molinarius moliner (old molner; cf. molina), sentiretmentu> sentiment, 
traditore > traidor, etc. In the verbal endings -icare, -indre, and 
-ulare, basically the same sort of considerations must have operated; 
Moll (1952: 98) believes the internal pretonic was conserved 'por la 
analogia de sus primitivos': carricare > carregar (presumably influenced 
by cdrrec and caérrega); terminare > termenar (influence of terme and 
témmens); termulate > tremolar (by analogy with tr&émol). 

In those cases where more than two atonic vowels preceded the 
tonic syllable, it was generally the atonic vowel closest to the accen- 


ted one which tell:° 


amicitate > amistat, recuperare > recobrar, comi- 
nitiare > comencar, etc. This pattern is in keeping with the relative 
placement of the secondary tonic which has been observed in the other 
languages which have been studied. 

The fate of the posttonic penult in Catalan is somewhat more conm- 
plex, and the details will be omitted from the present discussion.” 


Offering a reasonably accurate generalization, it may be stated that 








=a 
UR. eroaset-sit ey TSOInSL. 4° 
= - 
} | porabirenoo oct Jor “on ne mae 
| 


~ mm eS 


“wa isnoitevrseb 6° yd peat St ‘enzo? fata a 
odasimitines [6 sexévasesoo “x00! tae 7$eeL) ‘unaocaa 

| (ximob .39) ~robimmas sopietol alas ee D9 5 ‘ova ania, Leis Seem eSB 
Jassitnse si < usoontst tase \(entlom . 29 \zonton Sic} saath d = 

- . has.«sxkate Bak eel Ta a ote sob <moatbars ~ 
| ees event t2tm anabeieteas 70 tios ome oft visentesd .sxiio~ — a 


or 


| ) Bi xoat Bevibente. ecw oinetexd lenisini 4b ewetied (8@ SO Ae 
ik ‘Peonee Tit iene eS TSPSTISS < egos : ‘sovidiming eve Sena 
bots. sms to sont Sine) sepsis < eusaonpt i (spexxS9. Sas, 
(Lees Hit Yee sed inns mae aaa 
orit Sica 1qQ alee ines on fasy ani an 
| shop ort ot tesaolo Iowa aiipte, oe sammy Se saris ; 




















a 
a 





Pal 


vv 





. 
— 






ve 
ie 





" 


—in0b susudooar < SISTOGUOSL vies 5 atetinime “sist te bet 
ovis lox ont rin onitqsex ctf a ceniaial oom x 


vorto ot ot ieee ar aed 







Wi “moo. siGn, terlwamoe) et 18 
; © adtesine ip taseBa < 


‘ged. ediase | od “ey 


132 


a was never lost in this position, while the remaining vowels consis- 
tently fell, unless checked by considerations of phonotactic compati- 
bility. We have, for example, andte > anec or 4neda, asparagu> esparec, 
orgénu> orgue, scanddlu> escandel (modern escandol), opposed to cinéra 
and so on. The vowel e was preserved in most cases where a nasal clus- 
ter or a cluster of the form M would result: cophinu > cove, fraxinu> 
freixe, homine > home, etc. Occasionally, it is also possible to dis- 
cover the competition of two processes: apocope of the final vowel 
and syncope of the posttonic penult; realization of both processes 
would have produced a monosyllabic form replete with consonant clusters, 
and so in most cases only one of the processes operated on any given 
form, as noted above. Several examples of the triumph of syncope over 
apocope have already been given; some examples of the opposite develop- 
ment are: nubilu> nfivol, strigile> estrijol, apost®lu >apéstol, pisulu 
>pésol, *trifdlu > trévol. 

This historical panorama of the Catalan atonic vowels is clearly 
no substitute for the statistical analyses which have been undertaken 
for the other languages, but it does offer a number of general con- 
clusions of comparative interest. First of all, it may be seen, by 
considering a larger cross-section of data than that contained in the 
above examples, that the vowel a was by far the 'strongest' in the 
history of Catalan, followed by o (now reduced to [u] in most dialects) 
and finally by e (also generally reduced), thus conforming to the vo- 
calic strength hierarchy isolated for the other Romance languages. 
Moreover, the overall rate of loss of unstressed vowels was much 
higher than in Italian, Spanish, and Portuguese, creating a large 


number of clusters ; thus, Entwistle (1961: 84) remarked: 


a 
| io oe 
sie sto ents lt a ti mall saat 


Deniges < upbzeges soph sent ¢ pe gine 


aie eos i Peay a Ne 

at a+ Saaomm. «¢ Sims Age £2 <e on 
ate <uitops syle ah at ale 
itis on a 
eucxsxt BOD + Fras oll -tlvest Bvemy (mc id sete a a * 
~eih os aldizacy cals eb ti veLleaazeesed 028 <eteHt < «soled actos 

; fowov Laci orf to SIONS :eceesooty Oud: 0 opishtagmns Sets xBioD 
: eseasoorg dint to mobgesilesr luisa cinctaog orf 20 sgcoiiye’ Bas 
i 
| 





























, 
1 - 








(Brectaniio Jracoancn rite adalgex mrot ois [yeoraat 5 bescboua seed Blow S 

Ret en-vo Wie ee 
a | Ԥ$awo eqoorve to tigmixt adit To is a fsiayse ee a 
“—- —=aplieveb siieoqqo ert jo aslqnaxs embe attovin eed vont s par age: ot } 
a sale nae dosates - uistaoys vlocitias < al ipinte Lowi < cucu “ip de a 
= od < hs tae 
: Witesi5 2i efeyor ofodds neta axtt 20° aie too fsolrovaid abit” ae 
revadieiiw mex! oved rotdiw sohtehet {soiteiteie ota sot ee 1A 
“e800 Lsterep 2o xzSscdun 6 asfto ase $i aud ae a ' 


ya gee od yet ti \ Ils $b tent Jeetedat : 





a tt benkeinco Jen. ods etab to notinea-Becks, ¥s a8 
7 ett ci “teengcrte’ ot ast vd. eaw ss Lame ofid 


°# (atooleih saan at [tj <n wil nage: 
i ~ sev seit ot primigtacn suri iia : aon eae 
-eeqeepanl eonmmakt sorte itt i i 


if a 4 ag : 5 
reste: cow shear aaa niniad 6) ; Pe, said on q 


sessed ma 
ianagelecae panty: 
si 8 
oa: seine 


! a si 7 7 - 
BA. ; ; 
*T CY ee ee fee ee © a he A ei Py Ae Oe 





Le 


If we compare Spanish with Catalan we are at once aware of 
a difference in the form and accentuation of words. Catalan 
is more abrupt. Both languages share the strong tendency 
of Western Romance to eliminate unaccented vowels which 

fall between one accented syllable and another, but Catalan 
has carried through this change more thoroughly than 


Castilian. 


Thus, the loss of final vowels in Catalan is a manifestation of a 
much more general process; it remains to be seen, however, what con- 
sequences these developments might pose for a more complete theory 


of diachronic phonology. 


4.4 Preservation of morphological material 


It is fundamental to the tenets of functional-structural linguis- 
tics, and more recently, of generative se ainiae that morphological 
considerations may often intervene to check the flow of otherwise regu- 
lar sound changes . 11 It is not necessary to reopen the time-worn de- 
bate concerning the avoidance of merger to prevent 'homonym clash' to 
see that homonyms can happily coexist in large numbers without signi- 
ficantly impairing commumication./ Nevertheless, it is almost uwniver=__- 
sally acknowledged that, all other things being equal, a sound change 
will be more effective in cases where no morphological information 
will be endangered. 13 More specifically, it has frequently been 
Claimed that grammatical endings are more resistant to phonetic modi- 
fication or loss than corresponding combinations which carry no such 
morphological function; several such remarks have been quoted in the 


preceding chapters, and similar statements abound in the literature of 


wo 2 ee 


5 20 mortedest Lhe & at antes Gas dee dibits. aeol « ie ih 
-ncb aadw revewod 1982 sol af eaismer It se ten oan 
ieee 


“ 


Yioes 


Lee 


~petigpnict 
fsoipoladgrom sds 
“foot seiwxorttio to wolt-ard ee as MATLSSL ado a 
= cicw-onts. arit ‘maqosr of \isdeapen don ar; ae ii mt 
oF ‘feéio »mynoucd’ sasvsorg os tae to caaen patos $20 
an LPs an 7 
—tripie juctGiw asda spiel mt teeeeo Paine: dias. id G2 
esxevir secmin vt 34 esata azigt? ‘ 5 


epreto fapoe 5 » SetIpe poise cenit saa ° 


ga 


nod yannit ot ixogolorigen on nad 3 





Romance linguistics. It is, however, imperative that any such claims 
be substantiated, regardless of their intrinsic common-sense appeal, 
in order that the functional factors operative in sound change be 
rigorously accounted for in a comprehensive theory of diachronic lin- 
guistics. Kiparsky (1972: 224) succinctly summarizes the challange 


facing historical linguistics: 


The great difficulty with functional explanations in linguis- 
tics (and partly in other fields as well) has always been 
finding the general theories without which functional expla- 
nations of specific phenomena can have no empirical sub- 
stance. It is easy to point at a specific historical event 
or synchronic fact and suggest an ad hoc "reason" for it. 

But however plausible such explanations may seem, they 

carry no force until backed up by general claims, which go 
beyond the case at hand, and which for that reason can be 


put to a test. 


To date, no satisfactory method has been devised for providing 
a numerical characterization of "functional load', although a number 
of quite detailed models have been proposed. 14 In other areas; how- 
ever, particularly in the field of sociolinguistics, it has been pos- 
sible to arrive at quantitative data which support in a striking 
fashion the claims that functional factors may take precedence over 
purely phonetic processes. Perhaps the most noteworthy of the recent 
studies touching on this topic is the work of William Labov on various 
dialects of Black American English and non-standard white American 


Me 


dialects. The data collected by Labov indicate that, in long 


13h 


























2 3 on 


anisio roga ye ides. site samt) seenenOnt yee 
7 ; dy. iy a. ‘ r 
, : ' TU rte © oe Rte 

} = Pumas : ig! 7 
ac! aprstio Beuoe ot ‘sviterego se es eee 


iy “ » 7 af : ey ee . a 7 att A 
“jt tJ otecstps ib Io eaoeREs aiiaat 2 mK a 5 i os “ said) . vd a UC ; ss | 
‘ : : Sa per er > aha 
— =a oe eee os : iP 4 ' a we : 2 ee ae 
spre List ait on yisoninoye (BES :SPEh) yt ~ -o shea 


= 


: , Tre: a ie - ade, Sas 
’ ; = & id LETS & 


= : : its 
)  =BHprtl. at anni asrisiges {snc itonit SW Want bes iee Be v) nett & : 


read eyswhe aad’ (thew as abisit eit i bree Bis) sane i 
~f bets fénoittonst  dogiw front edtroott i nestaias aeap exited | 9 AD 


Pat. a 
—dis teeny x avec ABD shemosrarig nitiosae to SEI5iT ‘ he ; 


at 


‘dnavs Isoixoateid oittooge: & 38 Soiog as yess ak 32. eae ag 
* ti aot “noessx") ood ‘Bs ns +aappya bs tet oinouteaye: x0 


= 


 yortt «meee yan ecoltensliqns save sidievsiq xevewod Sud 
> coidw ,emiels Lsreres yaou ‘bavnsd Goer 33 sorte? an sy fl 
ed neD> noesex teddy 1032 roitiw bas \bosa ae aiesaie att pret re 


a i 
ane wee 
eo aa 3 


7 





is 


; ; a 

rma a eee ey 
pribiveig toi hseived need eat bortanr rosette 9 on \ateb OT ‘. 
. a r Seep a 
ysdmun 6 ouort is 7 ‘deol parole tus 36 nokta Dia, 


“Won ,2pe%s warto AT 


—aoq need asa 4% .201-tatuna tS bsoRIae Bai. rit at yh 


Eats rut “cts 5 “Wi trooque ota ont 9 rrr OF an 
; ae 


ar 









10 sonsiessiga sled yen axosdek warty i 


Shis0s1 ong 20. yitrxas sean ad 


muotisy no rer mae to Sal es “re : 


13D 


expanses of recorded conversation, the processes of reduction of word- 
final consonant groups (e.g. -ing to ‘in, or the dropping of final t 
or d in past participles) occurs less frequently in cases where a gram- 
matical category is signalled than when the identical combination 
serves no grammatical function. The number and variety of data adduced 
by Labov and his co-workers leave no doubt that functional factors do 
indeed provide a checking force against the otherwise powerful drive 
to weaken or simplify final consonantal clusters, thus providing the 
first vestiges of the empirical support called for by Kiparsky. 
Additional support for the hypothesis that grammatical suffixes 
may survive otherwise regular sound changes may be observed in Spanish. 
In many dialects of Spanish, particularly in the Antilles and the 
coastal regions of Central and South America, syllable-final s is re- 
duced to an aspiration [h], and sometimes dropped altogether. ‘This 
process occurs both word-internally before consonants and word- (breath 
group-) finally, and it is perhaps in the Cuban and Puerto Rican dia- 
lects that this phenomenon reaches the greatest number of speakers, 
appearing frequently even in the most cultured conversations. How- 
ever, by considering the interesting and highly illustrative data sup- 
plied by Ma and Herasimchuk (1968) and Fishman and Herasimchuk (1972) 
concerning the speech of Cuban and Puerto Rican speakers in the greater 
New York area, it can be quantitatively demonstrated that s is lost 
less frequently in word-final position when it signals the plural mor- 
pheme, or the second person singular of verbs, than in cases where no 
such morphological distinction is performed, typically word-internally 
but occasionally also in final position. Moreover, it was shown that 


final s of noun and adjective plurals was lost more often than the final 


s of second person singular verbs; a possible reason for this discrepancy 


; “now to sclsouhet to tovesscna 
; + Ineit to preggers sett te 
¢ “Metp 6 setery eoeaD. Ab 



















becebhs odsh le bas east 
‘* ob sratost Lars onan Rit bce: on a8, ersten 
evinh {utbrowor on twianso aie et et pain 9 
a ‘ot pet bivora ants erodes Liiianoenes isnt valignie 20 cele. 
| Liana vd 103 bal iso nee txpttioms a8 20 

eax itine Leo dtecmetp Int @ teorkiceryr oat x08 perro » Sate 

ares Rear ae 

fetasqe rit bavigedo Sth yam éugnaro Bauoe Seto -eatvmaitio 6 ry : " 
ef? bets sot ited arlt re: Usalim Ea tains SY apse wm ak a 
BT | “st ai 2 Innit-atdative ,eo-sramiytue? Bis iawn to snstper intaecs ae * 
sti? .1erdspodis Baguesd ey ay brs sa] cobseies mee Bat 
d¢esrd) ~fnow Bas atuanoetino sugied UL feentain bac, porta 
-~sih msoif ottesT bas nsdyD arkt nt agsitiod at af friws- lamang 
, axadsege Be ‘cect Feot6 opi stk eorosex ronenentlg aide 308 
; ~oh .chottsexovate howls lod team, 20g tk inate ween ni 
; R que steh ovitsiseuLit * cirfpsst ris eolieaneink ott ya . 

- (SVeL) icbmlasrstt bre utemiel® Bas) cae y a | 3 J 
j tsissip odd at etacissge neort ofascmh pee . : 


teol ate Jed ha ah vies cn 






a 






~fom fexuta ott elenpia at whoa OT 


cet eamnind Bere rer ners ae 





136 


will be discussed below. As an even more striking demonstration of 
the power of morphological function, the differential behavior of the 
definite articles may be noted. In Spanish, the masculine definite 
articles are el and los, while the corresponding feminine articles 
are la and las. Thus, only the final s serves to distinguish the 
Singular and plural feminine articles, while the singular and plural 
masculine articles show a much greater phonetic difference. Not sur- 
prisingly, final s of los is lost more frequently than the final s of 
las; for example, the singular-plural distinction in such pairs as 
fel mu%&co]/[loh mué4¢éoh] is easily preserved even if the two final 
s's are completely lost, while in a pair such as [1a kasa]/ [lah k&sah] 
aspiration nearly obliterates the singular-plural distinction. 

As a further corroboration of the Spanish data, Clegg (1967) con- 
ducted a spectrographic analysis of several speakers of Cuban Spanish, 
and found that, in cases where final s was completely deleted, the 
preceding vowel was slightly lowered, enough so that the morphophonemic 
distinctions could be preserved.1© 

Returning now to the main subject matter of this chapter, it has 
been repeatedly hypothesized that final vowels were preserved in such 
languages as Spanish, Portuguese and Italian in view of their morpho- 
logical function of signalling grammatical number and gender. More- 
over, the preservation of final vowels, particularly in Spanish and 
Portuguese, conserved the phonotactic generalization that no word- 
final consonant clusters may appear in 'native' words, since in words 
ending in a consonant, an epenthetic vowel is regularly inserted be- 
fore adding the plural marker s. On the other hand, if one accepts © 


this conclusion, it becomes difficult to explain such cases as French 


and Catalan, where the widespread fall of final vowels apparently 

















ety dabucn ish ot vices gle allie. wer. 
fexiiq bis selenite 543 ol idw neloisae sciiime® Leste ass 
wae toh -sonsisPib oijsnatky vataenn) deen’ 5 wee eninia antime” 
to 2 tanith efit coed vidnecapeet ‘Sate deol eh agh 2 a Dani velenieisg i 
| Z ae wxieg owe nt enkisnitenb erst ha~ueempente volemexe 308 gel a 
; fants om ott Tb newe bevaedsig yiiese ef SB dot Nob Le 1 
(ries fet) \ [sabe c. 5 rove isq s ai slitwuteel. lesetgm ons ote a 

“moitonizaab Lewsiq-ss fiat. ont anverrsd bdo, vee moksetiges 2 
“190 wave) posi? ,aveh defneae agd to! not sLodor1on sorts adie | “sy 
i. Heinece sadeD io snsissge Isieyae Jo siayisas Dusseoe Bato oo 
arti Bessie y iedaigite Baw 2 2 Laat sary ages mi ee Bre bm 7 
\ voieAcdooriqzan sctt sett oa 2009 bexawal Vitdoo te! aaw fonov pribeosag td - 
: se al af ~ ben eseStG oct Bilson siderite 
. aeri +i cesnteaEck afdt to sede soetdued stem, api od vm eee 
o tote ai bavrercyq gtow elewov font? dathts oe 
; soctorscm ~irns to weaicy rei meilaaT bas seegpus4ot | isi 2 5 as 
| 
: 


7 



















“sia .2abhee fins 1dr Jsoistuemp pRnlisnpie te. noke a 
A sent 
oe a Se") oh 


mee. 
wey 
bis reinsderd Wnclucbizsg .alowty Lei Ronee _ 
=ixiont ont sits notices: israta vismetonetg: oe b : 
efsow ci sonje ,yeixew 'svitten’ a siete ane sola eee -t 
“ad beitrgenk visatomar af Lewor ns 
atqarie: one Bi! far: septte soar ration te 
done es p e386 doa <2 


la a ae . 






an 


ene t¢ 


St 


obliterated much of this grammatical information. In Catalan, the 
almost universal loss of final atonic vowels led to a large number of 
final consonant clusters, in both singular and plural forms. In order 
to suggest some tentative reasons for this discrepancy, it will be 
necessary to briefly trace the processes of pluralization and gender 


marking in the languages under discussion. 


ee 


In Italian, Singular noun and adjective forms were taken from 
the Vulgar Latin accusative case, while the corresponding plural forms 
came from the nominative case. As a result, Italian signals plurality 
by an inflection of the final vowel, rather than by adding the desinence 
s. The most conmon pattern is the following: masculine singular: 0, 
masculine plural: -i (e.g. libro-libri); feminine singular: =a, femi-— 
nine plural: -e (€.g. casa-case). Nouns and adjectives ending in -e 
form the plural in -i whether masculine (professore-professori) or fe- 
minine (lezione-lezioni). Words ending in a stressed vowel (cittd), a 
consmqant (sport), the combination -ie (serie) 17 ende—t (Crist) remain 
unchanged in the plural. In addition, there are a number of apparent 
irregularities, generally stemming from Greek borrowings, or from a con- 
fusion involving the neuter gender in early Italian. For example, the 
Greek masculine nouns ending in -a (poeta, fonema, etc.) form their 
plurals in -i. The feminine noun (la) mano forms its plural as le mani. 

In the cases cited above, although apparent irregularities do 
exist, the gender and number of any given form is uniquely determined, 
18 


and may be further verified through the use of the unambiguous” arti~_ 


cles, concord of adjectives and verbal concord, thus providing an 





















aor 
; to nocmum sornl 6,08 Rel 4 
| “39680 of -eomiet emily Gog tal 


ee 


ed Li tw Bh; PuIBqEOBL 2 


xtizise Ss noisssifeulg Ro « 


oe 

a Rate 
a A > = aus : 
coulest i ya fae sa al 

moti nevtet oxow amet eviyosebe) int met ela seo LT a 
ny oe = 
foro? Lexus a a acid able 9aLo 2) Re « 
-* fo 
ateanlt siaate nek ied Huron 6 aA «2850 oy keine ‘att 


Soneciccs ott poihbs yo aa vortex iN {enett ahs io. sotsoatind a 
Pak 
.O ¥sloprite. orfepean :paiwolion ort ah revastn acamoo ‘joo ot a 





Pee ee © 
mrs 


ied ¢ > % 
se ~iiet .6~ ‘asivpiie ois tet  adlieanaes “Pp. 3) is Lei tg 
P + &- ai patbne ssvigos~hs fins eaoM » Gteoraies a) a ‘oe erin ; 


Me? 10 (toeastorg-srosHsic1g), Sr Imeet xertorie 2 2 fa sew mach : 
ae : 


ec 


& , (6i3ip) Iswov beaeowte 5 Al peribe abi a 
) afemex (eis) i- brs V4 Gets 192) Sis aoidecthteo : 

$nsaaqos. to. 1Sscimin 6 ox nots roi bs ia ws 
ates 5 mart to \apititworxod Aes mort primate , 


: rit yotomexs tec Biiss] yitss if, xab0e pe 


- stot mot (.9t5 arabes vB380q) 6 erat 
os | r= ni Sa 
7 ob ealsiteliperst SaAsISAgs pater 


_Miilearae cemeis ot we as 





138 


extraction of morphological material through external means if necessary. 
There are, however, a number of conmon nouns whose bizarre singular 
and plural forms are the result of confusion with the reflexes of the 
Latin neuter gender (cf. dvum-Sva). Such nouns form their singular 

in -o and their plural in -a. Grammatically, the singular forms are 
treated as masculine sails. while the plural forms are regarded as 
ferinine plural seterg-eilebraccio-le bracciayyil dito-—Je dita, l"uovo- 
le uova, etc. A determination of the pecularities of these nouns 
requires prior knowledge of their unique behavior, and constitutes 

the only true point of indeterminacy in the process of number and gen- 
der marking in Italian. Thus it may be seen that, even in cases where 
inflection of the final vowel does not yield unambiguous results, the 
language provides further recourses for determination of gender and 
number. This is particularly significant in those regional dialects 

in which final vowels are reduced to schwa or deleted, for in these 
cases only such additional morphological signals will yield the re- 


quired information. 


4.6 Gender and number in Spanish 


Compared with the situation in Italian, gender and number marking 
in Spanish is a relatively simple affair. In general, the masculine 
ending is -o and the feminine ending is -a; the respective plurals 
are formed by adding -s. Nouns or adjectives ending in a consonant 
or diphthong (always of the form Vj) form the plural by adding -es, 
except for words ending in an unstressed vowel plus s, which remain in- 
variant in the plural (el lunes-los lunes). Nouns and adjectives 


ending in -e (and those relatively few ending in -i and -u) form the 




















kina 


6 Sartor pics = rh eal ‘nal 
8s bebisper sts aa Laxylq se olin tien aiasccnia il be: 


: dh 
Shiai!’ .2gin ol cab ginseng etraloosrsd Lt awe) since a 2 he 


| 


amon s2odt Io dolstvelimeq edt 2 aottca inmate 4 88 \svoy at 


esdutitsnos bas Joivertsd sepia td pie) sehpRrOnt salag 
c) iar 


“tsp bas tecrua Yo sued" sit nt varie ct te tniog suit 
te “hs! Yd 


ster 29a6> ci neve \der neee ad wan #r soot tok tgt me oxen 
‘ot ete. eprogrotcimerntt adie gon @66b Sener fea act 20 ae 
= 4 1“ : aA ash cs 7 


7 





br tshnep to misscinaisb tat Saemioost sadoan aebivasg ¢ 71 
70 


secet kip isaotpax seod+ nr jnsoitinpie ‘eesibiniaties ai argh 


acxtt-it 12 \boteleb 16 swibe oF Beotiher exe alow Iend?- “ue 
"oma ae 

-si ai bloty Litw alemie Ieotpolesignan: Lanoitiibs rua Yin asasd © 
; 2 Tse * sas fort J a 


ex Li 





enitcam sedis bos septtep yasiledT mk coset oct 1.0 Bi 
siifsoaam odt ,fetonso nl iets Skombe elavitaton 6 ak 


aelewnlg ovitSeqest sit :h- 2b pains wna og 


7 Pras ae 


ae Junmmeng> sot pris eeviioatts 30 NO af a ou 
a fame. ye Lease a mee iets | yee et ib 


“fi absmer, doitw . 2. sul Lewy beeaescta — 

a ak ey hinab ha Bas antic peers fam i iia ~ 
za meget a 
) > eet mut (y- anit sathamene: Lent 





plural by adding -s; there is no inflection for gender. Nouns ending 
in a stressed vowel are required to form the plural by adding -es 
(domin6-dominoes) except for mam&, papa, sof4, and words ending in 
stressed e, which only adds. In most dialects, however, this is re- 
solved in ordinary speech by merely adding s to all such averSe 
Finally, there remain a few recalcitrant cases, mostly of Greek origin. 
Mere are a few feminine nouns ending in -o (mano, radio); these form 
their plurals by adding s and the plural forms remain grammatically 
feminine. In addition, there are several masculine words of Greek ori- 
gin ending in -a (poema, morfema, [narrema?]); these also form the 
plural by adding s and remain grammatically masculine in the plural. 
Providing the final s is preserved, the articles in Spanish ensure 
unambiguous identification of any nominal form, together with such 


recourses as verbal and adjectival concord. 


4.7 Gender and number in Portuguese 


Gender and number marking in Portuguese is relatively complicated, 
although sharing several essential features with Spanish. No attempt 
will be made here to analyze the pluralization roses the data 
will merely be presented. Portuguese shares with Spanish the basic 
patterns -o, -os for the masculine and -a, -as for the feminine; how- 
ever, the o is reduced to [u] and the a to [®] in most dialects, while 
the final s is normally (¥ J]. Nouns or adjectives not exhibiting the 
o-a alternation do not generally inflect for gender, except for such 
irregular pairs as bam-boa, and certain adjectives of nationality or 
profession. It is in the realm of pluralization that the most complex 


developments ensue. All words ending in a vowel (except for the nasal 









D Reins ent .xobrisp sch p 






























: nen ia 
, sar patbbs ye Eetnly ontdim a any 
a . ee 
‘ nk petthns ebscw bop \Btoe \RaBe y sero 
| Pe - 
ay “ot ek Pid) ciaveiont atgelsih aoe 
| - Mie 







a ~amot rowe Eis at 2 pnibbs am ay 


= niin int Bevin 
we .nipiio AesxD Jo wee , BOBBED ssssiotaen wk.» abawer ened: tel 
; mie snort ;(oiber .otat) OF niet Aree ectininet wet 6 936 et 
uilsottenmern aise amet emubg okt bee pee. etemtg sist 
~iio %enD to efrow sativeren fereves Sis siony noktibbs al -sninime 
oj mot oels, seeds + ( [Sass } Di vsmgog)  &- at pats ity 
init ort ok onbinosem “Yilsol tems: diame bos 2 pode eh temaley a 
sivueas reinsg? at esfoisis eds Persea at 2 feait hae) 
Hoge tthe xetteped jentot Tanking yas: to rorseo iS Einsbh auowpicecar , 
: pion Eevitseths bas echuoy as eoemooS 
“ear 
-2curcttom ak apd be ses Wt a - 
wbedsoiiomn ylovitsist ei seanpeiiot cab pnitene ae te ates 3 
tqietis of -.felisqe rtiw semrtest Isitneaca isvevee pebssde i. i 
; sdsb oct ni ieceneng forest Leteiq ery. eagisne-ad cia 7 
‘oiaed ont Heinsge dtiw aewAe S2sQUIIOT, oe : 
~weil jenictims? ort toi 25-\5~ bo rar or aliens | gm en | 
oliiw .eioeisib teou nt [#7] ot 5 ads, bas fa 'od Reouber; 2" 7 ae id a 
= Siok one oot deat 


ty j 


| opie x02 tasyx9 (rebisp tot S38 ee 0m Jo 


7) 10 ilsnoltsn do Be iHostbs fren bi ioe 
Ag ag") 
| aie iniieahe dt noises oh Rane ees 


Reh aS . 
‘pee f j 


of at as baliees 
+ a aa 
Lar 


on “4 
‘bth e 
vio 


> 








diphthong -&0) form the plural merely by adding =S:) _livro-livros, 
cama- camas, cidade-cidades, immaé-irmas, pau, paus. Forms ending in r, 
z and s (with accent on the final syllable) form the plural in -es: 
favor-favores, rapaz-rapazes, portugués-portugueses. Words ending in 

S with unstressed final syllable remain invariant in the plural: o 
lapis-os lapis. Words ending in -em or -en (phonetically [é], [&3] or 
fej], depending upon the dialect) also add -s, after orthographically 
changing the m ton: homem-homens. Words ending in -al, -ol, -el, or 
-ul drop the 1 and add -is in the plural: animal-animais, farol-farois, 
etc. Words ending in stressed -il drop the 1 and add s: fuzil-fuzis. 


Words ending in unstressed -il drop this ending and add -eis: auto- 


m6vil-autémoveis. Finally, words ending in the nasal diphthong -&0 [ e# 


form their plural in one of three ways, depending upon the word in ques- 
tion: (1) By adding s: cristéo-cristéos. (2) By dropping the diph- 
thong and adding -ges [#j%]: alem&o-alemées. (3) By dropping the 
diphthong and adding -des [538]: licg&o-ligdes. As in Spanish, there 
are a few cases where the apparent surface markings for gender are 

use of articles and gender concord disambiguates any construction in 


which such words are found. 


eee 


Marking for gender and number in Catalan appears, on first inspec- 
tion, to be highly unusual, but in several major aspects the situation 
is not radically different from that of Spanish. As noted above, Cata- 
lan preserved only final a with any regularity. However, the vestiges 
of Latin -us, coming from the accusative masculine singular ending, 


often remained as -e [%], especially after muta cum liquida groups. 


1h0 






















"| pat hanes 
eal. mar - 
ig be A Fs ~ mas 


o povek paying ger 


I ni, pnifigs ems? pay ae 
tap- at Lert edd mot: (ah 


‘ce 


ak poifns ebtow  .2sesontidtag 





o thsi ott Kins iigeuel | 
0 10] tS] etx eng) so a le i “aa a nee 
. amit rigsxpore+s ists .2- Bee /oats, (osteib ok oe a 
| #6 Use Wlo+ \Js- ab pniiias apo" ~gnpmadtrsmed oom th id 4 
2 \Bieuet-lowsi .aleningql eens : Levey ont ot aim (Sbe Bes L ett qa Lye i ° 
Sit Lime i bee'Ras | ott aome Si beuaeste ak entbne aba, oe a | 
“ods :gis- bbe Has ons ‘ote etrit qovb Li Seeeouan eb erties ac : 7 
ee} oB- entieih fact. srt cut pribae-ehtow ,yitenkt atgvonitusLivtm ; 
waeip ni Brow sri nogy prithraqel \eyswissals oe arm ce tomiq alot! soit | ' ‘aha 
Ge ib xt paiagotb ya {£) _sosteino-obtaies 38 patbbs, a. () smoit 
edt prragoxb ya te) .coansls-onmgis + {ats} ee ei Som pm) 
sia ,Jetimge al 2A .eos:if-ossek +1866) gee eruibhs 
ote YOdoNSse ol apnisttam somitwe +nsrbog6 ery oraciw conn wet 5 ‘as 7 
eis ,eseso. stowe fie al ,Gote | ampog® 20h Be-OmN s pete 


- el, 
if fnolispexiaroo yas astsupidmsaib B1cons6 tebnep. Binks 


i 


- 








ar 
26 





"esta ak 


~oagedi Jrafl no .exseqas as léte. uk secon ns | 
. 
; “ aolisotia iii edosqes softs isxevse, . a 


“56 wads baton 3a snes reeney pire biti i“ 


} pean: 





All forms not falling into one of the special categories to be enume- 
rated below form the feminine merely by adding a, although the phone- 
tic distinction is obliterated in most dialects by the reduction of 
atoniGzasto-[-2)|*- altre-al'tra, ample-ampla, mateix- mateixa, verinds- 
verinosa, content-contenta, blanc-blanca, etc. Quite frequently, the 
final consonant is voiced (agut-aguda, amic-amica) in the feminine; 


this occurs in words ending in t or c, coming from Latin -tu and -cu, 


respectively. Words ending in s, from Latin -ssu, -cyu, or -rsu double 


the s (orthographically) before adding the a, thus preserving the 
voiceless pronunciation: gros-grossa, escds-escassa. Words ending in 
a stressed vowel, originally containing a final n in Wulgar Latin or 
Ibero Romance, form the feminine in -na: huma-humana, ple-plena, 
divi-divina, etc. Words ending in the glide [w] change this glide to 
v before adding a: nou-nouva, meu-meva. There are also some more ir- 
regular feminine forms, such as roig-roja, mig-mitja, where, however, 
the irregularities are ‘often orthographic. Finally, as in the other 
languages under discussion, there are a few adjectives coming from the 
Latin third declension which show no inflection for gender: alegre, 
feel, gran, pairal, veinal, etc. Originally, Catalan provided a set 
of definite articles embodying distinctions of grammatical gender; 
however, in most major dialects, these have been replaced by the pair 
el-els, providing only for the singular-plural distinction. Therefore, 
in many cases, gender concord is destroyed in Catalan, both through 


the loss of final vowels and through the reduction of unstressed a 


141 


and e; however, as has been noted, many adjectives (ending in consonants) 


still allow this distinction to be realized in speech, much as in 


moder French. 



















waitin ofl}: i ‘(alias "li 

ia o- brs us- ee hxxxt einen 12" Pay 
| Sickie oat- x0 \uyp- .aaa-| abIed aban” 
ot piitviseerg edt 5 att avthbs suoted 


mi pnitas abr? -seaiies eties Sebetp seme initsioruneng Se 
10 ited FEL nt ies tant s prihistoce vilstipine | Fier 
\siisies-olg ncergsel ~ fires  - at “enniretoet ald mrxot 

ct abifp afte opnerts [w] abtip aig fr rcatt ven ebrroit 08 


| “Li som emoe opts. srs sear -avgr-iam XBOR-YOR Pr) a vont wiley 
Visvewod .sxorw (étt inspim sita~pioy as cove .ammod ae ie 


~sbeito Sri ah as \yiisait ob igempaddto, neste" ous aeittxelupersi a as 
mt Sic ‘ 
ony mrt prittico Sav isosfhs we 8 on Saath eta ste wren mig 
eh oe 
Sipels =vehreo ot patlsasan fered worl sie sass tel 
; as “tases 


¥ tes. 6 Babhveng ns fete \yifsnipiso Joge setioy eee 2 ae ; 
(wehisp Iso btenneap’ t6 RAEI OLE TBED eniyporie sale kras of f 1 Pr: + ; 
wisg sdy yd héosioes mesd overl eeot > toast soem aha): ss LE - ay 
<Sieisrst? .sorioritelb isuuic-skipetie wi ot no pnb ong yale 2. 
front (tod .hsinte at a ‘yam eck 
- "8 Beateutennt to noitonbst sft i he fi 


ni 


>: an leaped 7 
fi a i, — ac 
ye | eg ee 
(etnencaneo mi pects) eileinteice' aon mised aaa <stialieadaieal 
as . hos PEL: ud 
nk ee. ova eae bok: bat ? . 
psi a” » aa > < ie 


a“ 
Yoke 





lhe 


The plural of a noun or adjective in Catalan is formed by merely 


adding an s to the corresponding singular form. 


4.9 Information theory and redundancy 


In the preceding four sections, the processes of gender and num 
ber marking have been explored for the four languages forming the 
basis for the present discussion. Careful analysis indicates that, 
regardless of the grammatical potential of the final vowels, each of 
the languages permits a certain amount of redundancy and indeterminacy 
with respect to these two grammatical processes. Italian makes the 
fullest use of final vowels, utilizing them not only to mark gender, 
but to mark number as well. In Spanish, Portuguese, and Catalan, there 
is a relatively greater number of cases of ambiguous gender, consider- 
ing only the forms taken in isolation. Noun and adjective pluraliza- 
tion is generally unambiguous in all four languages, especially given 
the additional recourse of verbal and adjectival concord. In addition, 
the widespread loss of final atonic vowels in Catalan has resulted in 
a phonotactic pattern significantly different from those of the other 
three languages, yielding an expanded number of final consonants and 
consonant clusters. ‘Thus, it may be conjectured that, while morpholo- 
gical factors may perhaps have prevented the wholescale loss of final 
unstressed vowels,these factors are by no means of an absolute nature; 
moreover, even in cases where final vowels remain, considerable ambi- 
guity often results. It appears, therefore, that the relative role 
of morphological function in determining rate of loss of final vowels 
often gives way to purely phonetic processes of weakening, to a dif- 


fering extent depending upon the language in question. Given the 





7edt (félstsD Bris \o8bebutnod feta: fl .ffow es 1sdmin 


ae 


“ae 







M4 bi: co, 


/ - = al 
y ae i. ’ ar If “2 ; 


peat pan 


j . 
t 4 
wae Sheek ie 


_ eta om 


La, 
























huh ns raitrap 1¢-asersootg okt potas 

Sd} ‘thericl aapepprBl set td xt Pee 

dent aoctso ties giaytens tered -noiaanelb shoesig 3 

Yo rss low [atti od? to Las tesidedimenp ‘rtd of = ss 
species shark Dos voosGndber to Jauass nhageso 6 ‘etimereg eonaupint au 

ott seven nist isd -BSeRQ019 iscitemnsrp owt saert ot si a 
“tebya> whem of vino don met ecis LEM aiaww Yan to sau a 


ey Anco: ,tabnep euarvides 3p ases> To Tedmon tedsexg a 
-psiiszuig ovisoochs 5f is moh .totteipet rit aelet amc? oid vo | 
mavip ylisicsgss .zspenprsl quot Ils me aucepichasces yLisweasp 2 

noltisbs aI - .Beome isvidostis Bos Ledsev 2o sence tevotiteee 
mi betives: ast melessD mt plewov Dancds Sept? $6 aot 5er 
25fo stit Jo sexi mont taswsttrb visemitiopia misd36q = 

his adoendenco fsernit Io radian bebnsqxe ms, pribfaty bree 
—oicodaxroan sf irtw dart ett ett ad yam ot wt aueduis Sa 
Danks: to g2a0i sleoasloiy ont Ss i ea 
tsush sti lotds'ie Zo onsen ox OS eae att ei: pipe 
-fdns sideathesne ontémer efswow Teed? sear aoa ak a arooen a 
sicx aviselos att tad orotate \aisogas Fae | wom ° 1 @ 
alewev feni to Beat 40 patos . : tied. so ES 
_ ib s ots prittosisew he 7 | 
any nevio ee at 


aad oe 
aint Te 


os SiqfA 
’ 7 - a 


8 eit 7 







Pity, 


; . iy 
ele 






@ a ; 






: ; ; or i 
=. vo oP AY mas. ek 


current state of knowledge of diachronic processes in general, and 
also given the necessarily limited scope of the present discussion, 
it is impossible to offer a comprehensive theoretical statement con- 
cerning the relative weighting of final apocope and retention based 
on morphological function. It is possible, nevertheless, to suggest 
a highly plausible area in which thedirections for an eventual solu- 
tion may be found. 

Information theory is in essence a formalized study of the comm- 
nicative capabilities of natural languages. As such, the basic con- 
cepts of information theory may be profitably extended to both the 
synchronic and the diachronic study of language, a fact which has been 
noted at least since the time of Zipf (1935) sot Speaking in general 
tems, one may define the information content of a message, from the 
standpoint of the receiver (where 'information' is the primitive, un- 
defined term) as that quantity of data which is new to the receiver 
of the message, that is, that provides him with a hitherto unknown 
access to material. The information content of any given message is 
largely determined by the receiver himself, and also by the pragmatics 
of the communication act; for example, the message 'This building is 
on fire' would carry high information content if shouted outside one's 
window at 3:00 in the morning; the same message would carry virtually 
no information if uttered to a person standing outside observing the 
fire. Similarly, the same message, shouted in English, would carry 
no information content for a person with no knowledge of the language, 
regardless of additional circumstances. This definition of informa- 
tion content, while highly rudimentary and analogical, is nonetheless 
sufficient to visualize the action of such grammatical processes as con- 


cord. The opposite of information is redundancy; a message is redun- 


143 


is Ares F 








: can ‘, es: i 
bom. Genoa at sano Sly #3 | 
seen betieiah es, aS ae Abythd enon 


3 “sfieo thamsdss® ‘ceepeilealine “aia ‘saan 


| heeed noignsiex bas RO, Lani 20 
j 
; 
: 
; 


| 



































tesgwe ot .aeslondroven ae oe #1 estas pote 
; Pee » Aianats 
a “toe Intthnevs. ms 102 cl foistw rt —_s 


4 


| “immo ott 2 Yiite hontismet s sonsBEe mt ab yuoads + ‘ soini 
nics pjasd ott Hie Am -eopaapnel Leurten 30 uae 
srg -citocd ot bakes eldeyiterg sd yam roads “ookssmacial 2o 2o edgso 

nsed asl rbidw tost 5 .SoBtpnel Jo yfedde pinomtosi edit Bae cies vie 

isxenop Ar pared K (REL) Igt! I onus erlt conte Han Bo | 


ree bs vu Oy ci ai Ae oe 
Sits bio? 2oseeamn 6 to dashes robtemtotut addy eniteb yam sno 


wha 
—rm ev tt inte seit si ‘mm tsomoint’ exet) tevieoer edt to . 
7 Ofivy * 


tevrisoes sdt+ od ver Si Tete ee a nae St ae a i 
{ a3 
mein oftattind s tw isd we 2abivenq ted et hardy, 


Si Sosasany dievip yas Jo tasina> cottamrcint ott tats 6 
roltsimerd eit yd cate bas jteemit teubeoa arte oe] inilretse: nt 


a1. pribLind “sire! eS2am ait ,olqmexs yot 20 snksestmames at 30 
spans. Skene . i os r 
Bio sbhraijsa besuodla 11. taedineo notreersciiat i a 
pe Wear ray | 
Viteosttiv rico bivew Spesaan amie, sfis a a at 00: * obit. 
' a , si 
exit orivicedo eePP pr fpnssa Friatss top] «ot aan L 
. ©. Biyow Paton ni. Datuale F aeaiael: : ek ime 
.” : by fia y Raise. . 
j seston i 26 Selon oA ln he | pices i 
cs ae) 
i? “mrotnt ito nottiniieb re Bb tiscangr cidibds 260 
i: ) easloritsngn ei Jaotpotens: bas. bas 
i in bc 
) . ieo"es eagesbose TREE 9 
ee | Armibes Ba Leena 5 
) 


i : ’ a : 2 i] 
uy Were ih ate roe 5 


—— iP 







dant to the extent that it supplies data already known to the receiver. 
The concepts of information and redundancy are intimately connected 
with the general diachronic behavior of information-carrying sequences 
(i.e. morphemes) . 

Starting first with the concept of gender concord, it is obvious 
to any speaker of English that grammatical gender is irrelevant; many 
languages happily survive with absolutely no vestiges of grammatical 
gender marking. In those languages which do employ such markings, the 
gender morphemes serve the function of check morphemes, that is, mor- 
phemes periodically inserted in the message to ensure that the message 
is being correctly processed. Check morphemes play an essential role 
in formal messages; for example in computer programs or automatic data 
transmission systems, by providing a constant verification throughout 
the expanse of the message that errors of transmission or reception 
are not being propagated. As a simple but illustrative example, con- 
sider a code language in which messages are transmitted in bursts of 
five digits, each of which is either 0 or 1; i.e., a binary code. lIet 


the first four digits of each burst be the information-carrying mor- 


phemes, and define the last digit as 0 if the sum of the preceding four 


digits is even and as 1 if the sum of the preceding four digits is odd. 
The final digit of each burst, then, serves as a (parity) check that 
the information-carrying part of the sequence has been correctly re- 
ceived; a discrepancy will be noted if an odd number of errors has 
been propagated. Such check morphemes can obviously be incorporated 
into systems of any degree of sophistication and complexity, and the 
probability of error can be reduced to an arbitrarily small quantity; 
a much more complex but basically identical system is built into the 


book numbering code of the Library of Congress. 


14h 





























badoartids orérepeeas ou yotsbuvbet Bas soisemrotat -* 


esofisuped PiLe iiviao ab eee to > rola otek 
| <a ary 
gotvdo et ti .Bxooncs ceabggay to Sollee ort senda 
One ; 
Vwi < feocnt ef asbrep Isokvemmssp = deiinna to rerisege: ve mi 
. ae. 
ls Bete To esprgasy on zhedidoeds sb av ivwe yliqasrt eit 
ert vephizkem dove yotorms ab motdw aad ont at -prikstaem 2Sbnep a 
“rom .et 3ac3 \somaclqvan 2gstip 20. nei boat edt eutse sce tea an 
apseezen sit tact oivens ot spsasam ent @f betteank yLisokboleaqg eomorig ‘es 
aior iskhisess as velq Sree eet -iezasorg ‘eLtoerrap qiled at 
JENCTUS IO SMSIPCIS ° 1907 qmap it slamexe 162 7eepseesm fsa} nk ae 
Juorpyort nbitsoitiicy teense B yaibiverq ya vemeteya’ “ict benniiel: : 
peabveenee so notzelnateit to ster gard spssean ork: oe sansgxe rth 


a) 


“fo ,shywixe oviteariaul( i tod efgnte 6 eh <bedepegorg pated For oS 7 
jovatenid qi betsy inansi3 sis espseaat aoiiw nt a eh ard oboe sasbia ae 
tay _SB00 isn id & ,.e.f <i 16 0 vorhtte St — to dose \eabpib evi a 
—om pniiyries-noemiciat orlted gears % atipib wwot sex at : 
sO paibsborg a tome odd tr Osa5 tipib steel af} ented brs’ anol 
abo ef adipith wot puibeosig adit to me ah az f aS bas re oh etal 5 
Jéct. tosds (ytixsa)'s & Savitse : “asta etude feo jo tipib £ iow er 
“91 yidesiioo need agri sonsrpes 2 arid to Fase 
aatl sorte - tad bo 16 Zt actor sd Phi § 
Pecticeoetotett ed ywieyolvdo so nemeriqquem cot te 
add bas .Wixelgies ‘bus nottaoitaidade nae me ey 
sgbistciteyp Ifene ylinextidis 1 OE ‘hednbet se 2 east 
: 7 ice bal a ad * ; 
spies ist enien ional { tuck acategnen 


ey Peed | 
st% gS ; 
es a a ; Ts , 

i. : Leas ion ,S, 


yr 


145 


Returning to the subject of natural languages, it is easy to see 
that gender concord provides a system of check morphemes, indicating 
that the correct grammatical relations between the various constituents 
have been perceived. A language which makes use of such markings is, 
all other things being equal, more redundant than a language which does 
not. A certain amount of redundancy is essential to the efficient func 
tioning of any natural language, and the lack of gender marking in a 
language is generally compensated for by additional systems of check 
morphemes . 77 

Maintaining the singular-plural distinction is usually regarded 
as a relatively important semantic distinction, and consequently many, 
but not all, languages overtly mark number in some fashion. ‘The Romance 
languages are no exception. What is unnecessary, however, is the mark- 
ing of grammatical number in more than one place in any given sentence, 
as any student of English will recognize. Thus, in those languages 
where number concord operates both over adjectives and verbs, these 
additional inflections may be regarded as check morphemes, serving to 
ensure that the proper relations have been processed. As noted above 
with regard to Spanish, loss of word-final s results in no loss of in- 
formation when, for example, verbal concord remains operative. Similar- 
ly, in those Italian dialects in which most or all final vowels have 
been reduced or deleted, verbal concord, together with the articles, 
demonstratives, and similar check morphemes facilitate recovering 
semantic information regarding number and sometimes gender. In this 
sense, then, it may be affirmed that in any language in which verbal 
and adjectival concord is operative, and which in addition provides 
check morphemes such as differentiated articles, overt number markings 


on nouns and adjectives may be, and quite often are, redundant from 











Son OF -yEBo act FL 2apsseneL ma 4 
_ ‘paitaibnt casinos Sead el 8 2eb be iil 
on cnn sat 






an : 
,@f epriohiem move 10 sa neem Heidi agegenal 8 £ Fibg & ad i a4 
eedb rioisiw spsupastt mits snabeiben ae Saipe poked eg 7 aa 





















aot josisifte oft at fetoneaes’ ei yorebebst sa a 
S ni piitien ashesy to soe art ‘Bris <spaapnsl Isuctan ys’ 


Jhon 36 anateys Landisiibs yi sed seaman tet ania ——s 








pacispes yilsoas ef “ poktond sai: iewiq-~xsimrte ert sorta A 
« VerEM. esas pasiios bas ee Sitaenee toacaognt vioviteier 5 a rei 
Bone sift inotdesd snes : 2 techn stem yirsyo espeapast Lis fon suc ~ 


= 
B 
- 


— ELST att pi ceveword .ywsessosmnr af aada nolsqeoxs Of cs “lee if ia 
Jorainee awh vis at sosig seo medy exon af isdn Ieotsenmeap 30 Bak 
aapegoist sacdt ol , 2c? . .ossnpeoss iitw tat ipnt to treferte yas 26 a 
eort ,ecrey bos esvitosrhs aqve ried sedgyeqo Brocnoo ee ae 
od ‘pttivice , zansriquon toe cs beiieper sd yar enoijoolint Lsnoistibs — ‘a 
aves bator af .bseasoorg shel qven asotoaier naqeng oft teeth Starteme 
“ft th Bebk on si etivess 2 leni3-buow 36 aeot lieinisge od er : 
statin WES IEqO ertemest Prroonics isctay ,elqmbxs sei etn noi 
Sint’ Biowov tank Is 10 Seen fe Lite ae aioe ini’ aaltaat ect at St 
vaaloitts of dst tortiepat ,bicond> tector Vbetefes 
retoet Ginet! tos aemottescm: 2a bacraehinit 


“Bids ql .zabdisp setts Sas Bas paRracsor 
MORES - 


: ssoivers coltithe oi Asie Bhs. 3 
“serio stitar Hove 42 th bat 
mone Shabbona 


a 





yew v oe ae eee eee) we) See Se 


an information-theoretic point of view. 

These same general considerations also account for the behavior 
of the verb-final morphemes signalling such grammatical categories 
as person and number. In the four languages under discussion, the 
highly differentiated verbal forms have permitted the optional dele- 


tion of the (redundant) subject pronouns. As a consequence, this 


minimal redundancy virtually precludes the obliteration of the phono- 


logical distinctions embodied in the verbal endings, and sheds light 


on the observation that final s in Spanish resists aspiration 


more strongly when signalling the second person singular morpheme (-s 


in all tenses except the preterite) than when forming part of the 


first person plural morpheme (-mos), the nominal plural marker, or 


other similar cases where additional check morphemes generally preserve 


the necessary morphological information. It may also be noted that, 
in Spanish. dialects where aspiration of syllable-final s is wide- 
spread, the use of the subject pronouns tG and usted is much nore 
frequent, thus facilitating the unconditional aspiration of all in- 
stances of word-final s. In languages such as French and English, 
where differentiation of verbal forms, as well as (audible) verbal 
and adjectival concord, are minimal, retention of subject pronouns 
is generally obligatory. 

Sankoff and Cedergren (1971) present a set of data concerning 
the deletion of 1 in the French of Montreal. It was found,.for ex- 
ample, that 1 was deleted 97.9% of the time in the pronoun il when 
it served as impersonal subject (il faut), but only 92% of the time 
when il referred to a masculine singular individual (il m'‘a dit). 
Similarly, the pronoun les exhibited an l1-deletion rate of 46.8%, 


while the plural article les exhibited an 1-deletion rate of only 


1h6 


























a tohwitled oft) xo? intonos OPER Sm 
geiazopeiso Isoitaamstp rue ont 
ei} \apizeaseib yébru eapecpael wich, ot ot 

\ si [enoitg ied paldiiteoy overt ame Leckie he i 
| isis gets 2h .ameccog soettum mabe), ei Yo malt J 
: eter exit. 20, nolidexstiido ond ectulsia ilaenee sunita! - i, . - 
Sdpel abste bos. . 2 ribs Ledtev ec Gt hetbocme enaijontteis Leokpel, 
_noiteriges eteresr deinsee at e femtt same anisevreedo tt A 
2-) smrigzom tsfupnte noersyg bnogse sit webiste cece vipaowte sxam 

a3 Zo te palaot rarir pens (adizadorg ai Fqsoxe eoanat Le mk 

o Siete an isuule Tene ot , (Sor) siedgqron iewig aoeneq tex: 
| e@rrpeetc Vilorsasp eatin 2harb Tanoi2 1bbs srerlw woman seLimia xadso | 
inc Baton sd oats yen 3 -nodteimBat isotpoforgnan Yrommeonet, 908. |) : 

~sbiw-2i 2 Lenit-sldsllye 2o mitetiqas szedw eiseisip, ORR 

ovom bunt 22 Detey drs Dt amuongig Jostdye efit Yo om ot bose 

“tt Us 20 moiextaes Isnottibqeom st paitstiiinst aut tape, | 

eifpott Bos doneva as cote espagprel Ar 2 Lsnit-rov 30 esoneta 
- | fertey (stdibud) as Tow dee eee Lachey 30 apkiaisasnat tb exert Le, 
 Savotetg nofdys to nobinsder Lpenthdiin. ees irroonm > ame a 
poierisonco sdsbh to toe 5 scioetttg, rrr? | i 
“¥9 aot ee ee op ee (eertectt 26 ecient act ai L 
nediwLi muonorg edi ct ants ot a bleciesngetl ad 


sais sit id #fe in gud — Ef) deat tse eregi a6 | 
Bs Gib b'p U1) Sevbty tet setae eaten mr $A ox Li 
is: SE eae 
188.98 Fo ster gio ig ; max it 
ee see -o iy pale 


SEE AD » Dearie 


eae 


- 


a 





a 


oe — 


a 





Dz. 


18.7%. Speaking of the former case, the authors note (p. 67): 


This difference in deletion rates cannot be explained in 
terms of phonological environments. It must be due to 
the different syntactic function of the two forms or 
the quantity of information transmitted--so that the 
personal il, whose main function is to replace a noun, 
and which contains more information than the impersonal 


il, is less susceptible to /1/-deletion. 
The latter case is explained as follows (pp. 68-9): 


Welcan interpret the low deletion rates for the article jes 
in terms of the information about number it transmits. 

This information is carried by the noun only in certain 
irregular cases--for example: cheval-chevaux. Usually 


it is carried only by the article: 


le garcon-les garcons 
la table-les tables 


Given the importance of the article les in transmitting 
information about the number of the noun, the /1/ is less 
susceptible to deletion ... Of course, the pronoun les 

also carries information about number; however, in order 
that the referent of the pronoun not be ambiguous, it must 
appear ... relatively close to the pronoun in the utterance. 
The information 'plural' is thus already present in the 


context and the plural pronoun loses its importance. 


Bearing the above facts in mind, it now becomes clear why final 


vowels were often reduced or deleted among the Romance languages, 





‘eat Sfsstte * tis + act eatat noidsteb veel ai pemqrstnt Fac) a ity = 
ea viel 2 


“i 


aah oat - 


nae aad ak ee t 
is Pins pare 7 gen 4 
a poh pesto fe) 
4 wr! bik; ' ‘ 


= Per = a 
ihr om by 
_ - 7 oh 


Aid pod) 


atts ee Dpto a8 


,fuen B sosloes ob et mbit lim osc « vit Is 




















aed verge ms 
feckosrseptt edd nartt not SeonseRinit src ekatnen rity as ' 


ro HSish-\V\\ at alelisqevene seal et a ag u 
: (0=88 $q@)> ano Lick es obiniciebhe od wuss cule B/: 


a 
- e wed as ne 


. ivae Ke 
2 etimestt Jt sedan 20685 aoissanotat eds to bart hay 


wier50 af ylno mon of is borraso et noisemotnt eifff 


a 


ach Nal nds 7 . 
yiissel  .xuR pieces eve rolqupxs uei-~eee8d xe Lupertt Ss 
a A, y 5 7 


:oloitss off yt vino Bebriso af ina 


LS, Les a5f-n0 ’ 
ae ek 


paitsimensxt ai eal ofSitas ait 2 oonusrxorgi ade enna 


eset at \I\ oft \anon edt to SckTEn. ach Se voltemotni 

2eL mencid sd¥ .sextioo 30... ivenaanes: sldisqeomia ei ; 
. $SRro mi ateveadod soda suds ecient eaieso mM aaa , 
Jeum $i yeuapidns od ton none paeaeananibe's ie Pa 


sors aity sad ni quanto. ods of re We u 
ent nt snes Yaserls anit et “isslg® Moi 





148 


even when grammatical signals were thereby obliterated; the systems 

of check morphemes built into the various concord mechanisms in most 
cases more than compensate for the loss of morphological information 
occasioned by loss of final vowels. Considering only markings of 
gender and number, it is apparent that Italian exhibits the highest 
degree of redundancy, of the four languages under discussion, followed 
by Spanish, Portuguese, and Catalan, in that order. This fact now per- 
mits a more general statement concerning the behavior of final vowels 
in these languages. 

From a phonetic point of view, final atonic vowels occupy a very 
weak position; therefore, it may be concluded with a reasonable degree 
of certainty that it was the grammatical function of these vowels 
which prevented their more general loss in the Romance languages being 
considered. For example, Catalan is the only one of the four languages 
in which the reflex of the Latin termination -u, characterizing the 
masculine singular, was lost with great regularity. In Spanish and 
Portuguese, only final e, which does not serve to distinguish gender, 
was lost, while in (standard) Italian, no final vowels fell. the 
relative behavior of final atonic vowels in the four languages here- 
tofore considered, therefore, seems to be a direct function of the 
overall rate of loss of unstressed vowels in other positions: in 
Italian, where unstressed vowels exhibit the lowest rate of loss, 
final vowels fared the best; to the latter observation must also be 
added the dual morphological function performed by final vowels in 
Italian. Portuguese, exhibiting a slightly higher rate of atonic 
syncopation, deleted a correspondingly greater number of final vowels; 
in many modern dialects of Portuguese, an increasingly strong stress 


accent is dropping or severely reducing the remaining word-final 





anateye oct vematettldo, yaaa s 
deur nb anateetoan Baoan nsisay aioe 9 at 
noittsunroteil isnteotoriqyen to! aaal att ict 2 


io 


teorpin ocd i es net fegt sect Yassegqys ai $i zed bsp we | 
: ae 
Rawolfot , note ease S aahe sopsupiBl mot ect to ree 


iey 5 


Sarit 


sit * 


oti 


afdsnoassy 6 dtiw babylones sd yea Jf ,ouoiemads vnoitizog Asow — 
abnenr-as gant to ooktoeul Isotdamniszsp off dew of gat wainties %0 ce 
pais sopsut nsf onmémA eft ok aeot fetemep exam tis bedrievertq odshe 
‘dpaaeast Mot Sit 3o sao yIno ‘art gi assists ,alqnexe 107 beta ence ia 
pilixinsdosisid .u- noisesriumist nited ott to xeltex ais ri rk a 
Bas deinsge xl vii aa iat Jssip sie s30f 25w talupote onilinesn 
xsbirse deinonitech of ‘avr 1 2e0b toby ys Seni yao oor 
oft .fiei alsww Isat? bo past fsal (bnegmedta) mt elicw ,teol aaw 




















pete: onan “10 | 


spritem yinie oa, jt fentt 


ei 
doe? ent? .adba0 Jord sak yates brs ,seeerpus104 siding ox | 
feniit to toivaded ant prirmcams samara fsxedap axom 6 etim | o ae 
ta 

~ eepsupcsl slats 


yquocc slewov binods Isnt? .weitv te Jaieg pitsapdd 6 moe, oh 7 


Bepecoiel wot ott nt slew oie fare ea Ca 
7 Sap 
20 naitomu> Josh s af od ames. lh OR aa Ae 
5 
sero: tized isto mt aise heersuteoy to Beal 30 ec, ds im ; 


2201 30 Stet teswol orth chicks clewtar hepsarten cont mit 


od cals s2um apidinepco todas t ott ot jdesd ast fee ; v Sent a 


~ | Be Slee Sask Ns nanicigeg oe 
obtets to Sher Iscgid vistigdte is 
jakswov Leni? to sreageern <edso9 er . 














; 4% , 7 ~ 


1h9 


vowels. Spanish, in which rate of loss due to syncope is higher 

than in the preceding two cases, shows even greater numbers of cases 
of loss of final vowels. Finally, turning to Catalan, in which by:: 
far the greatest number of vowels fell through syncope, virtually all 
vowels, with the exception of the phonologically most resistant a, 
fell in final position. By viewing the process of gender and number 
concord in the light of commmication theory, the behavior of final 
atonic vowels among the Romance languages loses some of its mysterious 
appearance, and a clearer insight is provided into the interaction 

of forces of positional strength and the drive to preserve morpholo- 


gical material. 


4.10 Conclusions 


The remarks in 4.9 must, in view of their rudimentary and 
non-rigorous nature, be regarded as only suggestive. They do, how- 
ever, correlate rather well with the other observations which have 
been offered concerning the behavior of atonic vowels in the early 
Romance languages. Each of the four languages under discussion suf- 
fered a ceriain degree of loss of unstressed vowels, and the rate of 
loss is different in each case. The statistical data presented in 
the foregoing chapters are not sufficient to offer a numerical func- 
tion characterizing the exact rate of loss in the four languages; 
indeed, it may never be possible to precisely formulate such a function. 
Whatever the cause for differentiation, the loss of unstressed vowels 
operated, to a greater or lesser extent, in each of the four languages, 
being checked in non-final position primarily by considerations of 


phonotactic compatibility. In final position, morphological consi- 


































swtipid ek « ecqony? ao oat 
geasn Ao eisdmin raider ne 
ao bist tt iefadss ct oti alent 
ay Pin eetcitiulv SUCoITye ince ti it 30 ac 
| (& Sistalest secrn vt tiehiahaeabanaiad te. solace 
(yedira bis 2bncy to Begomia otkh pcdwaty a -rictibony Le 
fenci? to sofvedod. ett, vaca adéstaeninco to gripat at ae 
een eit tosempe jaqeot ReDELNVIEL sonar “it pooms: elowov tae 
nfoftinexcint ars ahtr Bebivorg af SHpiani rmisal> 8 Eos: de AY. 


+ — 2. r- ; 





— 
ak 
“oforimoan aiw2esg at vE edsrbas aiceieeial fsootsieoqg Io’ asoxct 20 on 

-istiedsm Inokp a 


f ye aM - s 
a 

anoiantongd Olt i a 

y ‘ 4 7) 
: bas yietnemtfurs tiait io wotv mi Jem 8.) me wiser ont 
‘ Wie 
j =—x ti . OD your Vise Spies vireo eb bebisyor od SRvsr coscet-nat 
; r = i> _ 


sf fotrtw enoissvrsedo wore omg Hs ty [low tertter ats Ioxidd ove - 
— 2a ; 
yiass oft ni elswoy sin a To aaa ey prisons berate = 


“tie notseersetb ywebow 3: seep! wot at to rised -eepaxpanl — _ 

ae ee | 

to-svsx sti? Sas ,sleww besaswtenr, 36 weal io ssnpat nasizen _~ . 

), SIV ee Cae we 

‘mt Sognszorg ste . fa5 elinie of? ,92a89 Tings HE snes tb ak eeol i) a 

. 7 a) ei _ é 
int isotisamn 6 xSito ot der ey SO Se via elas he 


near 
) mre rr Pee rar r 
; akon ao 


Pas ae y : 


,eopavonst ect. oft nt seal to sea oExS ap 
MOotsont & fous ats Lemaoz yleeinera ont er asc, oS raved 
~ efewov bearsxianuy to Baol sat wo nt fx hath SEED as sovetari 

7 : 


iis ah bee ohh : wie ak 
>. : sSopeeanL Biles orks te floss mi Saath, “to sectaer 
to ' aa iy 
7 ae enotiabitench s vrais citi x47 Tan Mes 


i cenit uaa ae are 





150 


derations appear to have triumphed in most instances, except in Cata- 
lan, where the greatly increased rate of loss was sufficient to 
result in the deletion of a large number of final vowels. ‘Thus, any 
theory incorporating the notion of diachronic strength hierarchies 
will have to incorporate the additional limiting conditions of phono- 
tactic compatibility and functional load. Only by precisely deter- 
mining all three of these factors will it be possible to offer nume- 
Yical characterizations of the diachronic behavior of the languages 


being studied. 


. ett Om TL 

a —_¢ 

% 
‘i - 

71 i 
i i 
ni tywoxe .ssonmdernt Jeet: 

is eat 









ea fewor lent? 20 > ert ap: 8 to apres ab ttt 
neo ee Ms , 


_ - 
Le o toate ; WY 
me 4 


™* nail, 


Srezhid ris in wits <a fos eb 20 moito 
Oo Brolyst eID eatsimel {aration oid eh pueey ad 


g : 
(oe ; Lal 
leeinorg vd YIao..Beot Lanatinaat Bes vai tiditeg BQ 


(izes od © +i-Lite's modos? seatt 2 


; 
we 





ait to wOtvedad oboe IB end 26 enpigest: 
ae 
4% 7 ’ ¥ 
. a 
7 De g J . Ly a 
_ - . -_ : 
ng ara |} _ / of . is aed 


15% 
Notes to Chapter Four 
Cf. Grandgent (1927: 47-8). 


In various Continental dialects, a fleeting paragogic -e may be 
heard in words ending in -1 and -r. Cf. Saco Arce (1868: 20-21), 
Soares de Azevedo (1903: 122), Williams (1962: 47, 112), and Leite 


de Vasconcellos (1970: passim). 


Cf., among others, the following sources: Goncalves Vianna (1887), 
Leite de Vasconcellos (1890a, 1890b, 1896a, 1896b, 1899), dos Santos 
(1898), Dalgado (1900, 1906), Nunes (1902), Fortes (1944), Rogers 
(1946, 1948, 1949), Strevens (1954), Batalha (1958), Thompson 


(1959), Herculano de Carvalho (1962b). 
Cf. the various observations offered by Houaiss (1958). 


The data concerning Catalan have primarily been gathered from 

the following sources, in addition to a number of minor journal 
articles: Mila y Fontanals (1876), Brekke (1888), Fabra (1906), 
Schddel (1908), Griera (1913, 1931), Pereira (1915), Rosketh (1921), 
Fouché (1925), Meyer-Litbke (1925), Badia Margarit (1951, 1962), 
Moll (1952), Alarcos Llorach (1960), Garcia de Diego (1961), Rus- 


sell-Gobbett (1965), Lle6 (1970). 


In the early Catalan period, that is. Syncope of the posttonic 
penult during the Vulgar Latin period did not prevent final apocope; 


€.9-) OGULUS.celup ill. 


For a more complete treatment, the reader is referred in particular 


tO¢Mol1. 4195239 3-6). 





















¥(ES=08 28931) PIA HR 33 — x bw a eth 
i ei tes (Gi! cP FS0et) ane LLn SBE ROE 


,(888L) snasi’ eovisoncd, :esauvos: orciwl teat orit athe al a 
ay be a 
Botess eh , (CeRL AEST Boel , dest BORBL) ne >i Ae x 
. i 

giopott , (bbe) aairoa .(SOPI) aon ,(900L ,OOBE cbepiell. | an a = 





7 a 
a 
apeemcti , (B2eL) srifeted:, (S2CL) anevada 4, (abel yeber arery es 


a 
. (SARL) onthe? of oimstsiotel. « (GHEE) B* 





a : ak 
if a 2 
- >) oe se i 


og : ‘ 
“ ; (822!) eeisvot yd bexstto eaabievreads epabusv sit 20° 8 y 


i 


4 7 
a a 


mori beredtep mood ylang avert neledeD pmimenass: etsh of 2 ‘a - 
tarertot tofin to rede 6 ot onitibhs ab ,esotmoe pniwollot odd 7 : ie > 
(8085) sxdsi ,(85ef) ables , (0180) aiamesaey eit en we a 

M480) rifexecH , (2leL) srroed Geer ff) SSD. . (ee) tebe 
k SBOE \f20) sinepust etbed (eek) sieti-teyet A220) Bows 
wast , (LaF) opsid ob slowed , (O3PD) dosreli acoA vise) ie a 

(FL) dealt ee . ." 


ahs 


a 






obnetsena. atid 30 egoonve sal sed einai 


uy ne ane 


a Reqpegs Innit inevesy son bib bolveq after sa Ns ce 
‘ - : ieee one re 


we 





a 


10 


aa 


12 


The unique exception to this accentual pattern is the posited 


development *pellicicare > pecigar. 
Ct. Moll (1952: 98-101) for details. 


An excellent example of this development within the framework of 
generative phonology is provided by the work of Kiparsky. In his 
dissertation (1965), Kiparsky went to great lengths to refute many 
of the working principles of structural linguistics; in his latest 
work (1971, 1972), he speaks of 'functional considerations', 
"paradigm conditions', ‘minimization of allomorphy', etc., ina 
manner highly reminiscent of the work of Martinet (1955) and other 


structural linguists. 


For more detailed discussion of particular cases, see Malkiel (19- 


60, 1962, 1963-64, 1966, 1968, 1969, 1970), Rochet (1970, MS). 


In particular, it is the use of the term 'tendency' by structural 
linguists which has come under severe attack by generative gram- 
marians. Thus King (1969b: 193) remarks: 'Any claim containing 
the statistical qualifier "tend" is usually so weak as to be value- 
less'. Kiparsky (1972: 195-6) is even more categorical: 


One of the reasons why functionalism has generally failed 
to get off the ground is that it has been content with 
making vague statements to the effect that there exist 
certain "tendencies". It is necessary to give a precise 
interpretation to this claim before anything can be done 
with it. I would like to propose that a tendency for 
some condition A to be implemented is for a language 
meeting condition A to be more highly valued, other 
things being equal, than a language not meeting con- 
dition A. 


Generative grammar, however, has also developed a number of theo- 


retical concepts, such as the evaluation metric, global deriva- 


152 
































e 


eo 
; ee, 


; 


7 1s va 
) oi mn 4 wit A Ls ee - 


whines ae ah erred $6 Yount sistnn ; 


1 : , es oo an od 
Bie ine. 
 WSpiesg .sren in slisg* step 


> 


af 


46 Avowsatsri oct ainttiw 9 
Sint .ydeisgh ae stow odd et icles ak yeolorcrig < 
yrusn — of adzpast Jdssxp ot Sits ersqgit v(eaek) 
Et : tetal aid ni += Seironil isunends => a cat easton ad 30 ; 
F | , Veroitsr Abilene isnok sont" bra) latins Sd srer {tet | bin a ? 


aA _) 1 aL ..od8 ,vrinoniNtls Fe noresiminio’ , "enol eeee wee 





’ ferttc bes (ceer) [tiem 26 =e att to Jneceiriine: lipid rl 


A 
ar / OT) fSixisM ose .eheno isloobhirsy Jo neigauparb idols som ag i. 
a Li 
x (Hh OOH) senses (Over Cet eek ANAL OMEREL EBON OORT ; - 
, i 


‘ | 
x : ree eu : 
“/ BespaSuitza vd ‘yoneicst' mrat oft fo opt) Beet ad ti Satie ic AL " 
| ; ; ae © 0) \ : 
: ” 


; Het sviderenen yo cbetts sisvae she, eine apd Aobriw atekupalt - : Ale : 
Taw Rit Je - 

z 3 
printed. misio yaa! oe (ees, sdeS@L) pat aude ies: 13) 


al 
vf 
=—misy si G3 ds tray oe Vileder Bt "bossy sal = b Ege a teats ont “ a - 
5 or 
- 
; fs Op 335 S ft 3-28 A 
7 | Dit 2 Sion neyo Bi. (O=60L BY Wereqia . ‘i 7 


7. 





.) ) 
ts belis? ylisreisd asd metlenoidont! tlw 2  Srit - 4 * 
ls : ftw taste tesa aszi +r tery cat waiaiepe el < sae 

Seine sues terit tosits off co ehrs or 

7% - SISsTG 6 SVS od yipaésmer ef 3T "anit 


, ; of sd nko patdevyis sited mie (, eit; 
a Yo? vyorshiss «6 Jack S2000TC , oF | 
Seatonst s 20t et basienst 
wero, ,betrisv virpist Son onl 
10 pet itsean 3on. Spempriei sn 





1S) 


14 


ED 


16 


i/ 


18 


tional constraints, and markedness conditions which, while perhaps 
possessing empirical content, remain to be definitively established. 
Thus, when the dust settles over the terminological debate con- 
cerning 'tendencies', it can be seen that the basic methodology 
of functional-structural linguistics and generative grammar are 
rapidly converging, especially when it is necessary to offer gene- 


ralizations based on a large corpus of data. 


Cf. Kiparsky's (1971: 602) proposed hypothesis that 'morphological 
material which is predictable on the surface tends to be more sus- 
ceptible to loss than morphological material which is not predic- 
table on the surface’. Note also the use of the word 'tends' in 


this paper, which postdates the 1972 paper, presented in 1969. 


One model, and a good survey of the literature, is provided by 


Meyerstein (1970). 


The results in question were first reported in Labov et. al. (1968), 
and have also appeared, in different versions, in Labov (1969, 19- 


Otel ids, 972 a0. 973) xe 


For Cuban Spanish, cf. Isbasescu (1968). This same tendency has 
also been noted, although not experimentally verified, for other 


Spanish dialects. 


With the (orthographic) exception la moglie, whose plural is le 
mogli. Here the i of moglie serves as part of the combination gli, 


representing the lateral liquid [)\]. 


Unambiguous, that is, except in the case of the elided article l' 


which appears before words beginning in a vowel. 







eaprtzog oo ata Ait inatssahuli acid 
inibttidesne ylovivinPteh ed \td arate ; 

+tno sdadeb Isoipoloninmet act me 

- ypekohettan oimed acs stad nisea sid igo gh 


Sis xem te evitereansp cee coseieakt ecuntonnct ia 
~aee afio a \isseamen ab 3b node bfeioeges '¢ 
-sisb 2 Brgnes xis! 6 19 oa 


a a ee 














cae 
| vis A. 
Pi ca * : WA 7 ( 
‘aa Esokpoforgian’ tact eles 1oaye besowoxg (se “FOE. sheineait 138 ne 
7 ; > > = 4 
“me enim od  abiet soeillave st no sitktomieng ai Acta isixeten ’ 
x i P ae ae i 7 _ 
+Sifeig gor <i (nbitiw Isinoten Ino ipolotnon aeae Biol ed eidisqeo +“ 
ys . » AAEM 
ni ‘abnet' biow ort to sar sft coals Sfo  .*ScRge Sef ao mie = 
: 4 a a ee | 
Sain 





an : ' ; 


- 





f Og@f nt Datjastexq \xeqeq SVEl off eeceiieogy cee 898g an 


‘ : 
sia = 













P a) 
Pare a ris | 
wi beBiverq ai ,sursistil ait to yovise Boog 6 Bere Isboa ps br 
a 


itey s 
- (OT AL) nieces 
i (gder) fs .39 vodisi ot bomoget dexit sew Ree nk astventiatt” ey fs 


“Of y@e@i) vodai oi ,enolersy srsisd2ib ar ae r Pa 
; | (ere over Iver 


aps OE ei vs & i 
ua asd’ yonebaat anne oir , (s3en) ieaeaciet ~13 a 
oe tao 10k ,boitixey yifstr (eats 3On alii belies mead © 


oe, 
‘ake 





a7. oA 
. - 


. 


ae gl et Leaulg szorw at ipod et a Se aioe (otek dl + iW \ 
ae q ae: e * "% Bie : 
“Up anitentence ota 30 st16q as avis 3b | oe dh xm 


oc are Le 
te sae pans 


sates ; 
4 athe eye acs °0 oa et ‘eicbes Lae ahs a 
a 2 a s a j 








19 


20 


Ze 


22 


154 


This confusion occasionally results in multiple variants of a 
Single form in the plural. For example, in Cuban and Puerto Rican 
Spanish, the noun mani 'peanut' variously becomes, in the plural, 
manis, manfes, and manises; the back-formation manfs in the singular 


has also been observed. 


In addition to traditional accounts, found in any Portuguese gram- 
mar, see Agard (1967), Hensey (1968), Saciuk (1970), Brasington 


Cay) ote ter 1oyl) and Lipska (1973). 


The following remarks have been largely stimulated by the work of 
Beckmann (1972). A number of other considerations are provided 


by Cherry (1966). 


Cf. the general discussion in Beckmann (1972). Information theory 
in general treats language only as an information-carrying code, 
and consequently makes no provisions for the impact of any message 
on the receiver (cf. Frumkina 1963, Paducheva 1963). However, in 
order to apply the theory of conmumication to natural languages 
regarded as error-detecting codes (as done, for example, by Beck- 
Iann), it is necessary to consider the information content of a 
message from the standpoint of pragmatics; i.e. the effects of a 
given message on the receiver. Beckmann's book provides an excel- 
lent discussion of these concepts, together with a number of use- 
ful illustrations. For remarks on a different model, see Afendras 
(1970, 1973), Afendras and Tzannes (1969, 1971), Afendras et. al. 
(1971, 1973). Some remarks concerning final grammatical endings 


in German are found in Chisholm (1973). 





























4 6 atinixey sald ace: 
nents odterS bas msciO nt /Sioreesrao 

,femslq oct ni ‘Selecta ‘aba 3 yng’ be } ne | : Ke . whee ~ . 
‘alupnis edt nb e2nem a sane pass | i 


<r 


a 2 





ri eee 
ae “Mere seeupriod yas (0h bees .siaeont baanistbeert oe. 


et 
on 


’ iveneaee 
7 . Sofpiaest , (OVEL) fare (8304) weaned. fTaRE) Buses ese 


+ - (EXE) psgigthas , (erty Saeed ade! Ge , vires 
_ ine on cy. 

cm rc] 
Fo strow a1 vd bedsionite yWoprst mead syed asgneae paiwoLlot ear - 
‘ie ca 

hobiverg <1 =noltezabisace tarkto 7 recht A (OT20) mots 

: A sik ; 

ee rmsd 

3) ; oar ay’ * 

: Wroorit: potteamoicl « . (SV@L)’ cries ni nolaawoet® lepenep ortt 20 


tage pe 7 7 
: .AbeS pis Semon oe 66 vino snaupretl atsoxt Ipisese | 
ee ah 
| Spseean wis ico toaqni st 103 enoiervorg ony aeien yi Jnexrpeenop oe 7 f 
tes - wy ich 
GE \w6vewoH. . (fas xvi us °¢ .£3@f srtice ob tevtecs ad 90 ie 
ras Slate) Set cee 54 fer | iszy ot fol ai ig LORE tH Yao ort “thor of tities: im ay, 
—$s6 vi .altstsxe 10% \onob es) sshcm pa iioetebataay ae. bebrsper | 7 
& 30 teasing, soissimctant ody sebienco- oy ‘asassoan ai si ae 
; | ; i's — a 4 
E to adaait Sot .9.1 peoiiempetg 10 ished a or : a 


: cae oie i al . ae , — 
leoxs 715 asbivosa ood 2" nneiiost tavieoss ont os Wr ; 


“set Jo usdhiuct se thw aextopo ,2agoneo” Saad) ao Aoi 
is asthe ese ,fsbon tasisiith © 0 aitigmsy 10% - 
ae da aisthesta , (iNet yeaee) eres! 


‘ “4 ya =) LN ube 


iv 

BE 
; 

hie - 


OD 


CHAPTER FIVE 
A MORPHOLOGICAL CHANGE IN ITALIAN 


5. introduction 


In Chapter Three it was seen that, from the standpoint of synco- 
pation and weakening of vowels, occurring in the early history of a 
set of the Romance languages in their evolution from Vulgar Latin, it 
is possible to establish a tentative strength hierarchy based upon 
position within the word, as well as a more general and samewhat less 
clearly defined hierarchy of intrinsic vocalic strength. Data from 
the other end of the diachronic axis, representing modern Italian and 
Spanish, were adduced to hint at the possible bases for such hierar- 
chies, and to suggest that essentially the same hierarchical structures 
may characterize the modern languages as well. In this chapter data 
will be considered from an intermediate point on the diachronic scale, 
involving developments occurring in the early periods of literary 
Tuscan. The data to be considered center around a complex set of dia- 
chronic developments which may be most easily characterized in terms of 
the hierarchical arrangements isolated in the preceding chapters. To 
the extent to which this demonstration may be considered to be at least 
partially successful, support will be lent to the hypothesis that the 
phonotactic development of Italian, and by inference, of other Romance 
languages, has been constantly affected and guided by the existence of 
certain well-defined hierarchies. This in tum will lend empirical 
Support to the methodological practice, widespread but seldom explicitly 
justified, of assuming an invariant structure for any given language 
across a chosen time-base, in the absence of evidence to the contrary. 


Moreover, by demonstrating the sustained action of strength hierarchies 





















stonye to Iniogbosta ‘ors a pawn uv #8 ae? 20 
WE | ; & 10 y1c eire s7h0 a 5 ee Lined elena 2 enim ' 
‘ $i ae | pl mors “ otis eed asd “Fepsuppnisl Soman “ spit oe + f ; 
imp) Dsesd yiiorstezdt fitonste euitesied ™ iektaegne b3 ll 
; 
aeei Jafwarm2 bas [sie 39 2 Sion, 6 28 aes 26 brow ‘att cabathiw tx 
_ "esi etal . pase nifesoy pieatitnt to: yiosstreid beniteb visa 
BAS teiiss misbay wiitnessxges 2b ninctrosih aft to eae abdahae 
essed douse sot aeeed aidiesou edt-ie sft a Beoubbs ois 


: : 5) 


escourte ieotdozeroid Smee edt vilsisnses tery teopowe ot bas eel 
BAB wedasro aetdt at Liew as esp patie mebon orit ostratoenao ar 
: ~abeoe oroorinsib oly no detiog sta ibarmsidi as mort harebierno ed 
: \isiedil to cbatteq vitss adi at pabrage einemqaloved. § rae tJ i, : 
- 2 : Se 9 - - 
“Bi to 352 xSlynee « boos tadneo betebli arm ‘ad 6d. ede ait sat a 
to setts? nt hewitatnsis viteke ¢eom sd vem Aerie soemplevsb 2 ual 
; B: 
Gl .axstaci prtheosig off of Bedsioer CA TEMOMTSTTS ination 9 ie, 


ry 


be 


aaty 
- 














. i 
Jessi ts si.ct Ssrabienos ed yet nobectancmah: air tists 0 | set eee e 
= ; ov si ae - J 
Sii3 dads aiaerttogyd ort od Joel et Lity eax pny A i 
y a) 1 = a 
ennieest tee Jo ,aotsisini yi an ‘nw thes io snaneptinsb § is: 
to sonsdeixs alt wa babiup, bab hadootts vfsnaianee set as a 
a 7 74 > 
ra ¢ : aa 


« - » fe reigne bret EL bw rem i eidt - Bak 






Vitis tiqus- mohles suet Aegrqasbiw es 
era 

loa 
evorei favip ys 362 ousurtie 4 i er 
: ; ; Cnt. Te i : 
Sg TS 






156 


in the diachronic evolution of Italian, it should ultimately be pos- 
sible to coherently extend the theoretical concept of phonological 
hierarchies to the characterization of developments in the other Ro- 


mance languages, and eventually to other language families as well. 


5.2 The problem 


Every known language exhibits phonological and morphological 
alternations. Some alternations stand as clear reflections of general 
phonotactic processes; for example, homorganic assimilation of nasals 
to a following obstruent. Other alterations, while synchronically 
rather anomalous, represent the results of previously occurring pho- 
nological processes; for instance, the loss of stem-final 1 and n 
during pluralization in Portuguese, which is the result of an earlier 
change deleting intervocalic 1 and n. Still other alterations appear 
to be purely morphological in nature, and while serving as part of the 
automatic language competence of speakers, reveal no apparent phonetic 
or phonological basis. One apparent example of the latter type of al- 
ternation is offered by the formation of the future and conditional 
tenses of Italian verbs of the first conjugation. In these tenses, 
the thematic vowel, normally a, becomes automatically shifted toe, 
while all other forms either preserve the stem-final a, or replace it 
with a morphological suffix; thus, for example, from parlare, the 


following forms are derived: 


future conditional 
parlerd parlerei 
parlerai parleresti 


parlera parlerebbe 


——_ 
— 
= 


ee ae 





e Py ae 
manqge anoissoist{6 torte ifie .o bas Lf oi fsoovsedat priteled ere ang 


. Viole here 

ert to disc a6 provise of Eth bos ,otuden mk isoipeforkgram vewig sd rel : 7 
> 
4 x Soy k Ch hea “i 
gitenorky jnsisqus on [sever ,etedsege To Ssonetaqam@e spsupnst obtemodus: } if 
wi Cre 
“is to ogyt ~eidsl ert to sa ad Sneed aro hanes {soipofonodg 10 7 
( pares fs - 


-eog ed visti far pete: 

Isorpofornora I6 tqSONO © . ee 
~ serio eft ni stnemyofevei 2: Nall trtiemrin be errs: 
-Lfew as acilimst SspaupAsl aatito od vilsctaas fins. . mm 

















@ 
, SS a A TT} uve aie 
& ; j ' Ps Wa 7 


feo tpeforcrem bas ‘ao gotonatgs athsiiite sescteost went vet Th 7 
> a Ab Lipa ee Pyy, > : 
fexense to encitoslisi 1685 25 Saeed enc jucastts ans’ paobsemisd Ls: 


ar. e i 
afseen to noitelimieas Sinspxoe ,Sigmex> 752 saonsenoesl obtostonoda ‘a 
‘5 =@ at ¥ 1 
Wleolnoutery: alidw ,enoitsmeadis ta0 -anerieds oniwatio? 5 a 19 
at aA pene a 


“org en TH000 “YE Serer Yo atinasr adi Sa at .awoLemorts Seciet s a 
bre I tenkiinees to eal a \soapdent 102 rasessoorg Isoigoton ~ 





a ay 
saitixss oe to tluaer oft 3f chicw' ,seouputiod AE noites itemta erties 
ce, 


fsdeitibte bes satu? add 20 omitemrot ott yd fenedittio et Lampe ni 

\Beenst seett nl .noivaourae tanct one Go otis? amb Let 30 pete 

2 of Bettida yilsoitemcdus asmooSd ,s vei Femxore secon aidamedt ot ae if 
i soplgsx wo. .s Isnit-pate aft Sara xsdtis amok serio Lis ae aie 


7” r “= . ai voth ~ As oo ey 
edd <suabieg meri ,Siquexs Io} yams axtve fi digqron 6 iw 


a 


: _ Essodsihao 
be | isieIxsq 
a itesneinng 


Loy 


parleremo parleremmo 
, parlerete parlereste 
parleranno parlerebbero 


In the remaining two Italian conjugations, ending in ere and ire, 

the thematic vowel does not become altered in the future and conditional, 
although it sometimes disappears due to earlier syncope. We have, for 
example, from ripetere the forms ripeterd, ripéeteresti, etc.; from 
Capire such forms as capiranno, capirebbero, and so forth, all with 
the thematic vowel unchanged. Alterations of this sort have tradi- 
tionally been noted by Italian grammarians and philologists, but no 
systematic attempt at accounting for them has ever been offered, al- 
though a number of partial suggestions are scattered throughout the 
literature. In this chapter an effort is made to trace the possible 
Origin and development of this series of vocalic alterations. The 
tentative results which may be established on the basis of such inves- 
tigation seem to bear directly on the problem of diachronic phonologi- 
cal strength hierarchies. In order to visualize the conclusions, how- 
ever, it is first necessary to remove several obstacles in order to 
obtain a clear view of the primary data from which theoretical specula- 


tions may be extracted. 


Dee Le IOnILGINyOL tie Change 


As a preliminary observation, it should be pointed out that the 
shift of a to e in the future and conditional apparently had its 
Origin in the Tuscan dialect spoken in and around Florence; + even today 
there are many local Italian dialects in which this change was not 


effected, and in fact some in which the opposite change of Cpuera. 
















| Wes i 

; | omnes ie 

. |  etestelthg (1 

fy usin 

ih en ai “< i ot 


,Si bas 919 ek 1. onions notte en cuiladt oat pat / a : 
| BenoltiGneo bas omy art i Ste sal em 


Fi 1 .qved SH - .Sqoonye reritss ot eub arsoqaseib nthenat af ie 
Ras ¢.O99 seseagrs acs emt att seipeaasa weet he mcs 3x9 

site Eis thiol o¢ bis ,oxeddemige>. yontisrigas as emo sage oxiges 

-thext eye toe 2ivtt Jo snoitemeglé .Sepnerom Lewov oitemedd act 

on. Suc ses bas ansttsmmexp meiiedt yd bedon weed yLisnoit -. 

“fg ,baretto ised tars asd mech xh aa laed 38 sqmetis oltemessye _ 

| alt tuortguomts herattabe x6 airs bblaigre isktrieq to vedaun s douods s - 


a he 
7 ah ib 


sidigeog eit sos1 of shect al 1088S me tetged> anh at sutsred et 

‘ 7 

ea? .anotisnisdia vi lssov to esiyse shat to saamgoleved bas aiptso 
~siycni ddue to eiesd a4 co Seieifdstes od wem poidw etiveet arenes _ 


x 


uk 


7 
= 
‘ 


Sipatonning oinostiosib to. asftigta otf no visvemsh wsed:os mace coiinpit 

Wier! anteas(one sits seid cove ty di seabhie .29ttoasreid dtpnsxte fo 
od shies, ni. aslosdado Istevsg yobs od yiseseher yeu ai ti v9 

: ~siposae feoftcicets doiiw md soteb yremiqg Sh wee weiv aso s een 


‘isiala eacaedeaie 
ti kee Sa aa 





occurred. The date of this shift within the Tuscan dialect is nearly 
impossible to determine with certainty, since it appears to have coin- 
cided with, or even preceded, the transition from Latin to Italian as 
the literary standard. Migliorini (1963: 140) gives an example of 
one of the earliest attestations of the vocalic alternation in a docu- 
ment dating from about 1250, in which the form manderd (from mandare) 
is found alongside remnants of the Latin synthetic future, such as 
dirabo (from dire). In the works of Jacopone da Todi (ca. 1236-1306), 
we encounter variant forms of the verb avere, including averai and 
avera, and also ard, indicating the early syncope of the pretonic 
vowel.* Other Tuscan writers of the same period show evidence of the 
shift of a to e in their works. For example in Il Libro dei Vizie 
delle Virtudi (1272) by the Florentine Bono Gamboni, the change is 
already in progress; in Chapter Six we find porterete (from portare) 
and lamenterete (from lamentare). In Chapter Eight there appear 
durera from durare, in Chapter Eleven consiglier6é, from consigliare, 
and so forth, alongside numerous forms in which the thematic’ vowel a 
has remained unaltered in the future and conditional. 

In the Istoria Fiorentina of Riccordano Malispini, dating from 


approximately the same period, the forms tornerebbono (from tormare; 


158 


note the archaic conditional form) and manderd (from mandare) are found. 


In Ristoro d'Arezzo's La Composizione del Mondo (1282), forms ina 
are found almost exclusively; this may be in part due to the fact that 
the author was not from Florence, but from Arezzo, a short distance 
to the south. 

A century later, the change of a to e had become firmly implanted 
in Dante's Divina Commedia, where, however, one example of the change 


in progress is provided by an unsyncopated future form of andare in 





ws To aan 

2. yizsen ai tpelsib nesavt <n i a , etsb va | 

7 atiee overt ot exssgte Jf sonke: \ebnitedre Aitiw 

.? es neiiss of aijsl inca? cotta edt .bebeosxg t 
Ro signex ms eovip ey 2£9RL) intvotip is bushes Y 

aiotb 5 ni nolthmadie obisody ot Dovennbasandts pea eee Seen 


edith r iz) 7 : 7 


(onsie co) Sintec 8 te ab te 
@¢ fovea .ouctd oLsartinye GE SD: Se nee: REE ay ye 


~ 


, (BOEI-2ES5 .so) ibol sh stogoost Zo ecko ent al «(ei a) md hae 













es @ ney 
i pene oa 



























ae 


-5 ae 


bre istevs 5 prthuiont :SISV6 csay sd¢ to wkedt sonreor seanaceee 


; item ; 
Simotery alt to eqpome yiobe Sd} enitiso iat oe cals — EIS a 


attt 30 sonabive wolfe Boreq amee Sit do axadiaw ana? stert0 eon ‘a 


wy i. Di geo aA 
3 és SEV fob oxdid fT al a fipnsse 102. -2ixow sie i 9 ot 8 30 — 
7? gy S rite 
et Sonera odt ,inodnsd ond entiisxolt act ya (stst) ) dhertatV Lieb - 
= 1d eee a 
a Most) ateteizoy balt sw xie sodas? ai eae ni ybsoris ‘a 
wii. Ly! a , 


tseqgs sxedt tipid iscsi al . (opagnemel © mont) Score bas 


; oe. at; 7 
tstiptenso net ,Gistipienco nevela rSsdqgem a ,gzemub oxi oy 
we SG 7 


& Lown “nitsrerd Sd¢ coidw nt amok sires ‘Sieleuenaiae »fix02 “i a | 
oY os eh gy yy 
Lenk 3 thy <0 ne ca: ot ok Benes Deniemen ef a 
cal}. | 7 
rt partied ini babies onsbacooia Yo Eaitnesomt sritneroly sivoder wit al 


ae | iy 12 Aisghgy 

yeierIot mor) onoddisrantod amor ads , bobs ‘anse ott ° ap 
eau 

sbrot S15 (Suaboun mot) Gysbcem bas (nro? aie ages sade ane bet: 


& tt eirot , (S881) obroM Ieb sols teogm a ed €' osss4h"b ouodeis a 


ey _ 


a5) be es ‘ 


an?) sonadeib jtode 6 \ossexA mor? jud eto ~ we we 






déris Jost ott od <ub 16q oi ed yenr eirit % 





159 


the Paradiso (Canto 30, line 144): Non andera con lui per um canmino. 
Here the a has been raised to e but not yet syncopated. That some var- 
iation still existed during this period is also suggested by the appea- 
rance of the older forms in a in the work of some of Dante's contem- 
poraries. For example, in Boccaccio's Decameron, we find (Giormata 
seconda, novella decima) the line: Tosto ella mi si gittara incontenente 
al collo; only a few lines later, however, we encounter such phrases 

aS voi mi perdonerete, qual donna cantera, etc. In the Rime of Petrarch 
we also encounter (no. 28, line 36) the line: Con Aragon lassara 

vota Espagna. 

In the first known Italian grammar book, the Grammatichetta Vati- 
cana of Alberti, written towards the end of the 15th century, the change 
of a to e in the first conjugation future and conditional forms is 
listed.? Bembo's Prose della Volgar Lingua of 1512 also states this 
alternation, although indicating that there are many people even in 
Florence who still pronounce the a.4 Later Italian grammars, such as 
those of William Thomas (1550), Henry Granthan (1575) and Giovanni 
Florio (1591) unanimously record the raising of a to e in verbal forms 
and in some noes By the 1600's, the Tuscan standard had become 
the norm for all of cultured Italy; while one still finds forms with 
the a unaltered in the writings of non-Tuscan authors, such a practice 
was stigmatized by the grammarians .° 

Once the change had been at least initially implemented in Tuscany, 
it spread gradually to same of the other Italian dialects. As early 
as 1374, we find traces of Tuscan influence in the Rime of the northern 
Italian poet Antonio Beccari; for example in the poem "Primo che '1 
ferro" one finds (line 18): e1 non ve mancherd finir questa opra, 


while a short time later (line 44), the 'normal' northem pattem 


jb) a ee ob 
“i “PEL os | sai ae i 
is vine eH a all 
A “sy ence Jed! .bsdsqosmer toys 
\ ~saque 1 Scr le et 
i aaonde 2 dnsd Zo anne 0 ten an mn Bynie x08 a | 
scteetxo iD) Batt sw pomempoea lo beeiaot iE: adc te 
_ eas an eas saat at (orice overs babes 
+ aoestrig fers tatntorta tad arent | aatet “nam wee 6 lao: sole Se 
fiasives to anist ond nt Jods Besa soi Lap semen en 7 
oe. Gassasl nOpeXA #00 oeikt ott (0G Omit (88 -ca) sednvonns oss ow a 


i 
-ecpeges to 




























q 
“hjeV satotbissamei of iced LEMTRETD neiindl swonkt text? ert at. ae a ; 
apres ert vpurices setl ott to bro ot ebastod neddiw bedi£ oo) eat a i 


9 


si fot ([anciaibroo bas swdut AO pepe oD textt ot mi gate to 
aa 4 
2} eotete oels SiCL to srpaid xepiel sified seoxd 8 odmadt "\betalt | a 
ni nse efgosq yam sts siedy gard petaeiest ‘Spuorls fs nottennsd Is a 
: ’ 


_ 
i , 


25 rove ,arammbip mstfatl iste] *.s edt Somvedeeq fiers atv i ane Gh ve 
i 

rarsvorw bas (eran) nertasxe yaoeH ., (0eek) samod? meiLTEW 30° eer aa 
amot Isdiov ni 9 ot 5 to piiigiay sid bioosr ene (£eery ‘otzoLt ar 


Snoosad bec TE DOF Be meoau? StF Pi! * OOo ony = © emion ame at be « 7 


i 
_ 


dtiw emnot 2abnkt Litte ano sitriw windt 5sio3 feo Ao Lie: Tod uncon at 7 
édisosig = touz enortigs SeDenT—non to apabt inw ay wake VOUS F saal i 


-nsiranmsxe ott ete 
\wisodv? ni bainensiqni yiistvint Jasef ae reed Beri’ ae a 


yittss 2k <etosieib astisdt tito eid to 003 Ot re be - ia ‘ 
- A : ve 


a Bel 20 Sms édt nt consuftat nsoeuT cm | 
aes 


Ps Ce a ol tes 0h : fs a gai 
Au 9 Bs = Ae al 
veto seit tage sesso aay isi 

| 7 a a ce 

if niall xsd Speeds ae 


ae Z bani a 


Me x ; : 4 - : 4 
ae | Aco a en | 


Uy ww * 





; : : vali ¢8 : a - ‘ : * 
7 eV toe eee eee? EP kee Sy Pen 4 oe 


160 


reappears in the conditional form bastarebbe. ’ 


By the 14th century, the Tuscan innovations were beginning to 
leave their trace on manuscripts written in the Roman dialect; however, 
Ernst (1970: 59) feels that the appearance of e for a in the future 
and conditional stem of first conjugation verbs is largely due to the 
analogical influence of the verbs in -ere, which, as noted above, 
retain the thematic vowel e throughout the paradigm. Since Emst's 
study was based solely on an examination of documents, one must of 
course add to his theory of analogy the most likely source of influence, 
namely the writings of the Tuscan literary figures such as Dante, Boc- 
caccio, and Petrarch. 

Returning now to the Tuscan dialect, where according to all avail- 
able evidence the vocalic alternations originated, one must address 
the question of precisely why the change occurred in the first place. 
The attempted answers to this question have been many, although all 
clustering around a few common opinions. Before examining the assort- 
ment of scholarly opinions, however, one salient point should be 
brought forward: the shift of the thematic vowel from a toe in the 
future and conditional affected all verbs of the first conjugation ex- 
cept dare, Stare, fare, and andare. In the former three verbs, the 
thematic vowel remains unaltered in the future and conditional; in the 
case of andare, the thematic vowel is syncopated, although originally 
the theme vowel a was changed to e as noted in the above citation from 
Dante. In addition, the future and conditional forms of essere exhibit 
the stem sar-, presumably by analogy with the forms of stare. These 
exceptions must be kept in mind when evaluating the opinions of the 


various investigators who have considered the problem. 




















saved ysoslsii names add nz eiaeeitiedal 
suqut eit ni 5. 10% 2 to _ oases ecb tars atest (Rk at 
act cd c0b ylowsst ei, adiay nolteaubtigs 22522 20 ae: pate: obs 
ae onl A eet ter nik atte. Sct 6 gas : 
2! sane aoe MLS IEg, ony soedguosds © a iowow obdiemetta:. ¢ 
oo > So tam sao ,einsnae® 20 anitanckatee NS ‘ales boasd ae ybute 
| eaten fo sxame ylotii taom odd ypotans Zo yrosdt ald oD bbs se1u00 | 





— 


-909 ,ainxl cs roum esumit yretadil meogsT edt 20 apaitiw eds yLemsc ms 


> a 7 < a og 
qLisve fis o¢ paiinesos suorw ,joalsib mepenT ot) oF won paiva hee brie: 


exexbhs 3 trun sno .bedenipiso snoitartsdie nites edt eonsbive Side - 

eeig s24i3 ay ai besna6 sprsil> ‘ert ule viseinexg 20 ooitzeup eds : > 

om Tis thwortia  ynem miesd oval ooLteS 1p eit of 2aaweas batomesss ect 7 7 ; 
i ~\agtoees edt prisimsxs s1cted .anoiniqo neemes wet 6 Daur eakzatauts yy 


ad Biccrie tatog tueiiee Soo ,tevawsr vanointee yinetora 0 aman 7 
ett nt & o¢ Smit low oitamerth oy Jo dtadg st shuewrot se 
~xe poitewhaco terki oft to adxev iis beineits taspttibac Bas ac 
old ,achtey soudt temo? scl fi ae has \ Saas poate 
od «i ;lsaoitibaoo ns suctut, eta wenitilactind arate sora sri 
-yitsniplzo mgworitis boateqoonye et Lowy Dis smeact ot asa oe 
i ienth coigetin ovods adt ai bsdon a6: 2 oa beens aw 6 <7 . ft a 7 
_ Shdtexe sipeas Jo amo} fenoctibnes ater ier ait Sb 


161 
5.4) ile reissng ineuence OF) Tr] 


By far the great majority of authors discussing the events in 
question have ascribed the shift of a to e in the future and condi- 
tional to the 'raising influence’ of the following r, which occurs in 
all future and conditional forms. Such an observation logically fol- 
lows from the fact that each of the verbs under discussion exhibits 
the alternation ar-er, with r being the common factor uniting the 
variants. Some scholars have been content to merely categorize the 
before and after effects of the change in a purely descriptive fashion. 
Thus d'Ovidio and Meyer-Lilbke (1904: 671) remark in passing: 'Vor r 
witd)a ... Zu e)bleibt'. Rohifs (194924230) states: “Was a betrifft, 
so wird dies vor r im Florentinischen in der Regel zue'. More pre- 
valent, however, has been the view that the presence of the following 
xr in some way reguired the shift of a to e. Meyer-Litbke (1890a: 290) 
Stated: ‘Von den Sonanten is r der wichtigste ... vor Sich bedingt es 
Cine ltabreniscnc: . CHa ei hae: 354) remarked that 'Nell'italiano 
Hehe vaole-daventt aseeunve = “Elsewhere (py 321) "he amplified this 
remark: ‘Cosi il r preferisce essere preceduto da e atona, il 1 dao, 
il mn e altre consonanti da a, purché la vocale atona non sussegua 
ad una palatale, e la finale sia a odo'. Bourciez (1967: 485) noted: 
'En effet 1'a lui aussi 6éprouve devant r une fermature en e, soit a 
la pénultiéme ... soit devant l'accent'. ‘The same position was also 
taken by Lausberg (1956: 155): 'Wo der Mittelvokal erhalten bleibt, 
kann er durch die umgebenden Laute verschieden beeinfltisst werden. So 
verlangt das It. vor r die Qualitat e'. A similar conclusion is 


echoed by Battaglia (1970: 169): 





















~In8 yifsoipel notésvrsado on age 


atidtvnsa anieameaib weiat aki ont 20 fipse stadt toe: edt m 

_ set pasion tOs36i aQRNGO ‘eats prod x dike at 

art as itopsdss, yloten oct +naariod need sored aislorbe: ame ae = 
-moitiast Lwitqirmesd vi Sau s ol spnerio it te ejost@e 28536 nes el 
“2 i6v’ = :pateasq nt anemer (V2 =BORE) svidiki~reyom bre ofbivO'b 





disied & 2ew' fae (OBS sQhOL) hinacanin .gdkeld: 8 us vee rer <3 : 

ssig gioM .'9 us Ispat teint nese iidaniecete mi 3 IOV eat brio : j 

paivoliot mit jo soneesnq sit ted wey oi meed asrl visveworl taalsy ue in 

(OCS =8008E) oxkifiI-toye 2 Gf 6 to stida ott Soiivpst yew a hs 

as spnibed rie IOV «+. step ito he 26D) ay ne naNoOe nab ony" pate ¥ 

omsiisu i Lon" sfertt Beotrsmor (808 8000) obtenisi .*adosins i iast eae. 

eine Hotiitons sn UISE .q) qisresels .'s co 8 6 itapveb efor aes 

0 65 | Lf .snoje 9 sb c tibateaee FISEEE cceinsteag + * tk fecd' 

2 siipsaace oom Bhots Sisgoov sl srinang 16 ab a ana nits 2 oan 
rian a (8b :$80L).‘sofsuiod | 'o bo 5 ef2 slanc2 st ad abi ms De 


cS tid \2 1s oases? Ban gs Sous RMHSIGS ideo sot Sky 


can 
a 


o#ls asw mort fea ame edt 501560511 Fasvsb he: te i 5: 
Adis, nad Lert Issey bossa eb OW plac: ® 
‘ . “02 .neires Jeatitdateed sebetroaray ie 


ar cies ties 16linte A 


~ 





162 


La formazione moderna del futuro 6 percid assai semplice 
e chiara ... Il cambiamento, per la prima coniugazione, 
di a del tema dell'infinito in e si deve alla presenza 


eh Rs 


Bec (1970: 152) is of the opinion that 'il s'agit d'un changement 
conditionnée en position de a devant r'. A somewhat more moderate 


position was suggested by Grandgent (1927: 55-6): 


The choice of e when r follows and i alae is probably 
a phenomenon of association, e being in the whole mss 

of words the commonest intertonic vowel before Very sale telat: 
commonest before other consonants. The Sienese prefer- 
ence for a may show the assimilative power of the first 
conjugation, or simply the natural tendency of r to 


open a preceding vowel. 


These remarks are, Of course, largely circular, since the reason for 
the predominance of e before r is precisely the fact that a was often 
changed to e in this position, and reflects no overriding Latin phono- 
tactic heritage. Scepticism has also been evinced by Pei (1941), who 
notes the following in connection with the vocalic alternations in 


the verb stems: 


In the first conjugation it is to be noted that this restored 
Or retained vowel is e where we should expect a... a 

general rulewthat pretonic a tends to become e before» r has 
been suggested. Before other consonants the change seems to 
(oot LIME LectlOneOleite e(Collocare> ws )i>tcorricare), 


oboedire > ubbidire). (p. 39) 













iy a | n=) ve 
5 "0 _ rae ; i ‘ me ae 
_ wiiqnee soe Gioxed 8 it J 


ee 
.ent ixepu Lew Ent rina ~ 
at ee 


' 
pis 


es lcci 


fIwanoe 






sjginiian Sxcn 
~ {Rape +f COE 
e (& < ~ + 
v , 
= t 
: 
ou * iy 
iweaals it Bos Bwolfo® x a ww t 
Diisa Ss ,AOLFHIooSas 320 fe 
ant 


Mi MMe : 
cxder t = 

way 
laced . 


Bil SLonw 
[ Towow obnodaedtet 
%¢ IO. anh died ean 
5 a 






Ad 
- 








eit b ,.t stoted Is 
- wicterg Sesto: onff- SetnenOEnGS 
" . 
“ tarkt ert to wewod sv ijsd imteas ory wexie hom 208 § 20 
tas ’ a ears e, 
lemeben ‘edt yionie xo. wiolspas 700 


. Lowe 








n : 9 1365 “j ost 
hee) al peiws 
7 





({B€f} fot ye Bsonive 
2 iissoy Sd4 dtiw ‘nofos 
5 
toteex eit tar) baton od oF SE St 
5 ... & JooaxS Dicore 
3 sino! 





Cd 2sse Sp ie 
LEEENOD 


| ip 





163 


The e for a in the first conjugation infinitive ... is some- 
What doubtfully accounted for by the fact that it is followed 
by r; an alternative explanation is that amerd is analogi- 
cally influenced by persuaderé, perderd, and so forth; this 
is equally doubtful since there are more verbs of the first 
conjugation than of the second and third conjugations 


combined. (p. 103) 


Mendeloff (1970: 14) mentions the alternation without offering any 
Opinion other than that there was a 'tendency' for the vowel e to 
appear before r. 

To the above observations must also be added the fact that in the 
third conjugation, the thematic vowel is i, which finds no analogue 
in verbs of the other conjugations. Nor have the verbs of the third 
conjugation ever shown any inclination to fall under the influence of 
the dominant future-conditional er pattem represented initially by 
the second conjugation verbs and later joined by the verbs of the 


first conjugation. 


5.5 ‘The lowering influence of [r] 


All the above studies, and others like them which may be sifted 
from the literature, have inclined toward ascribing to the consonant r 
the potential ability to raise preceding vowels. There exists, however, 
a far from negligible number of linguists who feel that, to the contrary, 
the dominant characteristic of r is the tendency to lower contiguous 
vowels. Grandgent's speculations with respect to the lowering of e in 
southern Italian dialects have already been noted. An even more pro- 


grammatic assertion was offered by Deferrari (1954: 29): 


a mn 
' 


7 sat } - , ; . ia ry is . ou 
it | bawallod ek ot desis fost sorts 30 SIONS YALE ee: 
- rh ill 
; | -ipetens ei Grom tert ek cena Ss ex 
uw aids wuitw? oe bre < cr obra, tse 5 
: sit ec 23 ia: aces AD 
tha ee Britis Bis Bese: eit 30 asckt 
en 
(COL .@) 


























Ute paixsiio tues iw eee oft eanoiinem (BE s0ver) Holebaat i 
od 3 aaah ovtt x02 tySnebtet! 5 aow osortt Jedd net. sertio nointg 

a oxched cael 
orf} mi. tort} Jost orft bebbe' od cals Jemm empiievasede aveds oct on gc 

a ext abai8 ipirky e: Ea fate > itements ods .sobaeqatae bxidt 
Brids ot) Jo achray at aval 24 veasoitserpaeo xsttto att 20 ediov at be 
yy Yo soneftet ont yebuy [Ist of poltsnifont Yas mvede atove sokteeutmo 4 
Wi ylisitint boinsesxqet mating 39 Ms opts ibres-serte? semectncs edt : a 
ets to actisy oct yc bactot segel bas ectey meigmpuiaon ‘baqoee seit 


reat eases ee ES * 


wredy 5 a 


re. 7 | 2 2o sxrewiied pulsewnt edt | 2.2 


= »* 


: 


er anges - 

bettie sat yem dois mods siki exsiito rts rial ists: wil Ls ie 

| ys ao an 

~ tnprSenco ort ot paicitoes brswat benoit ove vsarssred Lt | 

; teveeod (eteixe got .afowov pribshorg, seer o8 hid . 
Westin et} of .terh Lest ordw aseiyenit! 20 “ect 9 fe 


. excopietnoe eine o+ yanshasd ae (et 230 9 ferse 






“ge 9 het 
pee hSOI) 1 Ists a} NE i boas 


Gc Soul 
hy 
U 








> - 
$4 
ca iS 
if ee ic ul 


oo PA ad 


7 Ls ‘ re a 7 1% : : ¥ - e 
0 * oF Ya aos ae | — eri ol - —_ 


164 


A vowel is sometimes opened when it is adjacent to a variety 
ofr: ‘This opening effect of ris strongest (most frequent) 
when the r follows the vowel and, at the same tame, 1S 
followed by another consonant. ‘The Opening effect of ig 

is less strong (less frequent) when the xr follows the vowel 
and is, in turn, followed by another vowel. The opening 


effect of r is weakest when it precedes a vowel. 


Needless to say, several of Deferrari's conclusions stand in marked 
contradiction with the observable developments in Italian, a fact 

which Deferrari himself was often hard pressed to defend. For example, 
in considering the change of sarmento to sermento, which would be highly 
unlikely given his posited hierarchy, Deferrari (p. 152) alludes to 

the possibility of analogy with certain forms in which the opposite 
development ensued, such as starnuta <Sternuta. More baffling, how- 
ever, is Deferrari's attempt at accounting for the shift of agtOre in 
the future and conditional of first conjugation verbs, for, in referring 


to the development of cantarA to cantera he states (aueelG4)r 


The above-indicated development is consistent with our general- 
ities for the opening effect of xX... Since [i] is the most 
frequent result of V. L. [initial syllable pretonic] vowels 
which were kept as vowels in Italian ... the [e] of the 

above rule is very probably the result of the opening 


effect of r. 


The only way to make sense out of this rather confusing passage is to 
infer that the thematic vowels in question were first raised to i, by 
whatever means, and subsequently lowered to e by the effects of the 


following xX. Such a pattern of events of course remains purely 

























Cis Gem he 

a a a) 

¥ ve vgs 
J8ACB EBS, i voll 


i 


Th : a 
hides + actit) gan enol | at 


ywiekt ey 6 cw 


t lott ese ‘att 46 i 


ae $b recto jasiisow abt 
‘ i : sd = 


a ‘ ave 
Feed "ists to is seve 


Hessen oi pisss anol suigmee s af al 


IDE tsarisdl Ar GITSRE oles 
SJ & : » we ae 5 
c Kicte o+ basesiq SYBR Gayi0 smw pene 
vefemexs 101 = -bisied Of Doss 5x6 i 
a= ar or rad 
pe cl — wy . ic 
f) ! ~earion Gt coonamiee 4° — ‘a 
2 ie i An iae— ~ hs 
: * 
J 
2 c ah ack © mace 
i S E fosetein pasteeq 
= 5 ar. a." | eve 
: 
He i-th <Rond ere 
he OO spryol nists fol Vee o 
Se 74 c) <1 iy t “HiT Th). 
ed «tS ; , 





’ : c Res os 
i - 4 . — = yrtTs hwy 5 sasuitiesye £6 rye wSlei SD a 


oe x A , ts ve ins ; ; jalion = ~ —— nial 


“3+ onriqpoooes gS joes 


= >» i we 
aotiqay-notdneuides 220m G0) Eanod aa Pe 


7% a sad Brsdoso Oo 616i 159, to 4 ‘prams 
taf .q Sistisl | " p 
: ? L% 
; a ny 
; dtiw jnederanco ai Fn eanoiovab Dagso. 
{4 2! wiie «s+ 1 20 Soeiee prvirsqo 
sktnos pis fds! L\ 2 fersernri re av to 3! 
J ? = 
eit Yo fa] ad ..- naiLingt Ae ieee 26 Ff 
~aa 1 JA... -| atk ° i 
; vas 
oninao oft to tives att yidsdoig 


esq piiiaviacs rsrisx ekds 20, tuo” 
: 7 -, = 
bakes ¢22i7 sus Cs breast preity i 283 
ied Was 
rieROL. Ys 


atoms sci: od. eo 





165 


hypothetical, and is in fact contradicted by the abundant documenta- 
tion, which fails to reveal any such tendency. In addition, one is 
left puzzled by Deferrari's reference to the initial pretonic syllable, 
Since the thematic vowel of the future and conditional, while pre- 
tonic, is certainly not in the initial syllable, except in the future 
and conditional forms of dare, fare, and stare, in which case it 
remains as a! 

The lowering power of r has also been postulated, in another con- 
text, by Otero (1971: 60), in speaking of a cited form sarao from the 
Galician-Portuguese serao. He notes that the former results from the 
latter 'con la e abierta por la r contigua, como en el Fr. marché 
"mercado"'. Gili i Gaya (1932: 245), in considering some similar data 
from Catalan, also speaks of 'la tendéncia general del catala a obrir 
les vocals travades per r o a . This proposed power to lower vowels 
has recently been nominated for language-universal status by Vennemann 
(1972: 883). Vennemann considers various developments apparently 
influenced by the presence of the alveolar trilled r, including data 
from Spanish and various Germanic languages, and concludes that r 
exhibits a definite propensity to lower vowels, although not without 


exceptions. 


5.6 Further remarks on [r] 


Seing the field of inquiry thus divided, between those who feel 
that r tends to raise vowels and those who maintain that vowels are 
lowered in the vicinity of r, not to mention those who ascribe no spe- 
Clial Statusewiatsoever to r, one is left witha feeling of confusion 


as regards the effects of r in the Italian case under discussion. 






























. ; va : 2 “ve Aa) wy ; 4 ) me) bh 
f Lae | \ | y No mY me) 1%. 
: AP - 2 ne ae ae 


ep wetomonmob scbkertis ott yd Be KK coring 


| | lane ie}! 

le. ai ono vigitibbs at pediocin ins sie 

os nae ape 

eidellye ciactstg isttint acts apn ei , pies 
wirlds 

| meets ality Tsnis ibang Bre ous i to fower 'o: | 


ia satu a pi tqeoxs Jefdsuts Settings ote rn dor ‘i 
te ee cater Ete os a ‘pase 2 ene 


smb yeritom: ai betaluteoy need oets eel 2 ad ee pekiawol set 

\\  s#8 mov} couse mot batio 6 Jo pabisege nt , (08 :8er) oxedO 
| ‘ae most ativesx remot oe aah eqdonsH .cmt9e eccnan 7 
4 fs ta cmoo seotine t ar “19g sctrmicis 2 BL edt zatsel : 
stab: ssfimi> anne piiisbienbo af . (2BS; 2SECT) wee £ ib oo 4 v mn e 
"Tied ; Efetso [ob fersnop sionSbasd sf’ to ales@e cals « A 15969 mova as. 
slows wawol od “seo gq basegoig eit ar ox 19g ameanacd meee! eat - ig 


riibesine! yd cotste [serevinu-spsupns! tol Sade iio ceed visasoer ca 





Sistem - 


be 


‘ yidrersogs etrsmoloveb auoitev ershienco annmemisy . (88 sate Lite i 
’ " oe vil rs 
e#eh prises [ori i7 bellizs xsfosvis sit Io acai. exit wa penne Ek “4 

; = ‘Tes 


> 
’ 7 


sty estirionco bos ,eepenpisl oinearied BIOrIBY bas faLasge mot 
strexitiw ton dpuoritts .alewoy tawol od vitemseoaq eeeiaeD & eoidine 
liar 


td 


+ 


7. 


7 


ey ‘J 
a i 





; ar: ka 






[2] so 


iF 
a 


; tai : 
fost odw eserlt neemted \bebivib eurd yxitierk te pieit wm tee 
vA eae. 
! = atswoy Jant misanise ogdw seont Das alewor a ox tat 
sit ma a4 
Fit VFR 3 hen 


au al Pas ie! ay id ei 


= 


“2qe on sdiioes onw seods odtnem oF 46H ato ae 
ae f - fokauitsce to pnilisst s riiw S390 aL. 3ho 


| a ee reteneit sch se  a Sahn 9 Sad 


H Fs ray, Wie Ti 









7) uf is 7 : - tiie 
Save AY at] ere} el eh. ak ‘dia 





166 


Each claim as to the possible phonetic effects produced by r has 
been accompanied by at least some concrete examples adduced as evi- 
dence. On the basis of these conflicting results, several tentative 
conclusions emerge: first, that any particular effects to be attri- 
buted to r are language- and even situation-specific; second, that 
in evaluating the data, errors have been made regarding the possibility 
of an r either raising or lowering a vowel; third, that r exerts no 
influence at all on neighboring vowels, and that all reports to the 
contrary are erroneous. The third possibility seems to be in general 
refuted, owing to the large number of observations which have been 
reported in various languages; it is still possible, although quite 
unlikely, in the particular Italian case under consideration. In no 
other phonetic environment in Italian did a become e, or vice versa, 
with any regularity; hence, the conclusion implicating the r in the 
change. As regards the actual direction of any change motivated by 
r, one must delve into the phonetic and phonological features repre- 
sented by the segments under consideration. Perhaps the most useful 
way Of visualizing the relationships between a, e, and r is to display 
these segments as simultaneous bundles of features, thereby manifes- 
ting any Obvious traits the segments may have in common. Starting 
first with the vowels a and e, there is not a great deal of possible 
variation in the description of these segments, regardless of the 


descriptive framework chosen. In the Jakobsonian binary framework, 


the two vowels stand in the following relationship: ® 
-~dif Cult: 
+cmp Jaf —cmp /e/ 
+OIVv ~9IVv 


Here the shift of a to e may be interpreted as a simultaneous raising 




















anda yd edwieng eiiata’ si 
-tve a6 beowbbs salqne ssoiestillaapaubi 
evitasaat Lorevee ed vee pritpi Pine, sasxtt Io # 
“igs oi ot eivstie rsiooisieg yn Set erkty 
sett \bacoe2 sede p Scabies 
usiindisece art prt reste oben need aval aomrs .stsb ots priseukeve a 
oer sitxaxe 1 tect .bxidd :lewow 8 gitttewsl te pnieitex radtie 21830 — 
wert oF oot fe tect Bas, yelouaw ertiodelp ban we din ae sons Bat > 
leneneo ni sd os enipoe yt idiesoq/ bei ent .suoamerse Sms yisxtac 
neat ved doidw anobievresds to tetmm spapt ee 
Ssiop tpromtis , aide 2800 Lise ai aes 72eomupred avoriev ai bedrogor | 
em al. .soLtsxisiienco x= feb aes9 Meifetl ssigvotsiusg ent al wefan 
Bets sory 10 yS Smhosd & bif asilesl of: daseaomives: olbtenorig sadio: j 
; wate ai ied piutsotigzi com sy fone3 odd soni citinlupen yas: rw 

wi batayvitom spnsdo yns tomodisersb Taodos oat efispor 2A sents a 
—x1qet eoulssi [soipolonorig bus oigadelg at otne evieb, Seam wo a ‘a 
bitesu dor ad piartsett .colistebiancs, Wahqu-aienESe ott wi betas | 
yelqeib of 2i 1 bos .5 ,5 needed eqitianpideien ost} prbsbieunty 30° Vw a 
—#Stinan ydorsd:+ ,zouuctsst 2¢°celhnud evosme ties ae sinamse east 
pattxate .somoo- ni overl yeu atriempse2 Sct einer solves yn pais 
altieaog to iseh jeaxp 6 ton’ eat dradt 2 os else att ti Sead 
etl 20 acolfiisps: ainegmpee saad Bo ccleseeee slants > _- 


eroheanst yisnid nmeinoecoist sit ak . 12a 3 






<a 


® giranoiisior pakwakiot orth + aaa, 


Peae 


en 


on | Se Le 





167 


and fronting, which is more clearly depicted by the articulatory 


features of Chomsky and Halle (1968): 


+back | baer 
+low Vay -low /e/ 
-round -high 


As might be suspected, the problem lies in the specification of /r/. 
In the Jakobsonian distinctive feature framework /r/ has generally 
been represented as:” 

VOC 

-cons 

— Camp 

prc © 

~GXV 

Soni 
Upon seeing such a display, one might be tempted to regard the change 
of a to e before r as an assimilation of the values of gravity and 
compactness. It must be remembered, however, that the features grave, 
compact, and diffuse refer to different parameters depending upon 
whether one is discussing vowels or consonants; consequently, the 
apparently similar specifications of /e/ and /r/ lose much of their 
plausibility; note that the valve of diffuseness is not assimilated. 
Intervocalic [r] in Italian is a single voiced alveolar flap, much like 
that occurring in American English ladder. It has no overt similarity 
with either [a] or [fe], other than sharing such general features as 
voicing, and therefore a measure of doubt is cast on the practice of 
classifying this sound as [+vocalic]. 

Within a more modem distinctive feature framework, Harris (1969: 


46) has attempted to describe r as follows: 


The segment [r]is simply a voiced apicoalveolar single flap. 


We may confidently assign to it at least the features 


; Pe O=EB 


















yg 
NN Te fort nai at a ett eign a h ae 


Yileysies asd \\ HOWEMBx? “once erisorisels. a ae 
>, ; | Sy ed “0 J 
ae ) ch oer) a ie 


ny ( 


: : _ wi i 


; 1 Oe. v7 





4 Ss _ 
dma * 2 
ae: ‘ait Exspss o¢ bedqmat sd srpins eno \yed@eib tous priese nog 

a 


Bie viivetp ito esuley ait to inhd-tespeptalaee 6 St eacisd. s om owe op 


a 
} 7 s — 
Ps i. 2 


Wrap eexisset at ject  tevrowor! | bovadnarnisr sd tam ot <n 4 
- fig: pribsoqs siateitg tnswsitib of 2Ster eee bas soem 
om .%! *iaupsence Sie ‘ro to elewov pnisesoat et enn: sodttoriw : : 
tstit Go doen seol \r\ Bes Ns\. to ancitsoitiosge salimie Yisntecrs a 
Bede Llimiee: gon ei aaensaaiib to selev sit janis een eT 7 


‘ede L Hou .g5It xeloevin bsotov slpate's zi asifagd at [af 4 


a 
: Witsiimte ove on esi I -tabbsl Mak tong aia elas anc : 7 7 : 
ap aéfotse? fstensp rove prberé nerld nado ac Nadia 


#0 Slew only 0 +260 2k tdyob 26 susesan 5 =. z 
ae | ) J[obfeaoie) 25 fig ar Gieesio 


my a . 
i or a 
: aul © ao Meakt $< a, 
s : as p bee ig 


pQ9°L) ehrisH ,showemst? oucisst eats te prabor Ewe: nicsiw 
bikie » ye 


ma fe 
La ae 


: yaaa 
. sawol. 25 
| eed hae a Nie 


es on | 
bd Pe 


a 
“eet sions wplowisoigs tai oe 
ee ote Ay get 35 Ht ee 


. tae 
“a0 . we ¢ rw. 
. oe y ~ 7 








2 , 2» 
Ret Ae 
7 ae 





168 


[+vocalic, -consonantal, -obstruent, +voice, +anterior, 
tcoronal, -strident], and with only slightly less con- 


fidence, [+continuant, -tense]. 


Chomsky and Halle (1968: 177) adopt a similar specification for iG, 
except that, unlike Harris, they specify r as [-anterior]. Andersen 
(1968: 175) feels, however, that r is characterized by the specification 
[- continuant]. Vennemann (1972: 886) who, as noted earlier, considers 
xr to exert a lowering influence upon neighboring segments, suggests 
that r be represented as [+low], at least in certain environments. 

It is obvious that discussion of the specification of r could be 
prolongued indefinitely; equally obvious, it seems, is the fact that 
the presently available distinctive features tell us little about any 
possible assimilatory effects r might exert on contiguous vowels. In 
order to determine the latter, it is necessary to consider the articu- 
latory details of the Italian single intervocalic [r]. Although arti- 
culated in an area roughly between the points used to produce [d] and 
[z], Italian [r] is a flap; that is, its manner of articulation is 
characterized by a rapid movement of the tip of the tongue from a point 
somewhere in the middle of the mouth against the alveolar region, and 
back again into a more central location. The full effect of the flap 
is produced only if the tip of the tongue travels sufficiently both 
before and after the point of contact. Thus, for example, in the 
nearly homorganic clusters [dr] and[zr], the apex of the tongue dis- 
engages itself from the production of the first consonant, lowers 
somewhat, and then returns to produce the flap, after which the tongue 
is lowered again. The point at which the tongue stops before and 
after the articulation of an r varies somewhat, but it is invariably 


in the mid-front region of the mouth, i.e. in the area used in the 



























: | <x “tot toktsotiiosqa aah pure & Luan et 0 | 
; _— Repemeials ita ial Fs + Vives yeds ,eirrsH tint = a) 
‘h moitesitiosge oi yd cast axosyarn ee jes vesvounl acral Be er) a 


; —e ~zreiltes Besfon as ,Orlw (ae 2ST @r) reigned ewe 
; or ‘Beeps ,etnenpse otiitodel ten ADT ‘sorte LU pabxowol Pas a 

‘ , 77 4 a 
chemnoivns fiistiso nt tesei Ga ,|woit] es a ark sett Are F 
ied Bivoo 1 lo. sotseottinoge ont 36 aoieeoeib t$6d3 awoivde zt ry. i. 
gerd 022 ot ek ,atss2 Ji‘ puObvdS (ifeups +yiedanBtebak sweets 
Wns jocds ofs+il et Met sowdsot ov Lon igaib sided tevs yiinezsig ott 
A; fit - tai euoupitnes co tyexs tipin + eters vrowel imieas see, 
‘polis: Sit tsbienco o¢ visesteon ai. 3b \teteel st enieedeb a isto 
(=iiwe tpyortiA . fr) >ilisnavrsdat Sipat fa neiistt arkt 90 allstab ae na 
“/ Bes ee Souboaig ot Boa ayniog ait neswrsi | VEER SeTS hosed nd Bada i «vy 
i 


ei sorsslioitis to xsnnem ati \el Set sqeks 5 ai fx] astiegt is) 
SRiSg S mt soprnot art to i+ orf Jo Smamevon biqset B we best 


bine .notpsx isfoavis sd? tenispe ion sti to sfipim ont ak pererins 


te i a 
dali sit to dostie Lint sit) .notysool Lextias Sram & cont prrree a 
2%. ‘ ii ne 

fod yivnsioitine elovext sepaat ot to git at ot ‘wlio & eouboxs q el Ly 


4 


od} ai yelguexs sot” evi toginoo Yo: satog Pree ge ted De 
“eih suspnot oft to “sas srit (ss ] ris (x5] le 8 f 








ite extol ,dosricenc yaxdi ait 26 Te ott % 

F >, sdpiod sdf dobiw sedis elt erty sopborq.o = se re 

a ‘rte ancted wes a tt hoe ae f 

on e  eidoicsovat ak si gu athagnoe 2 bess ihe ars ot 

ay Tis a ‘oat moe ni .2.4-\dtuom sx 30 ae 
re side 


_ et : . 
; o> me Ps eh a ~ wert la Pek > 


articulation! ofle] or [e110 It seems, therefore, that any assimi- 


latory effects which Italian r could possibly exert upon a preceding 
vowel would be such as to achieve a degree of tongue height like that 
of [e] or [e] . In the case of a low vowel such as [a], therefore, 
one would look for a following r to raise the vowel somewhat; pre- 
cisely this occurred in Italian, and the full extent of the develop- 
ment will be outlined below. Conversely, given a high vowel such as 
[i] or [u], a following [r] would potentially lower the vowel; sig- 
nificantly, most well documented examples of purely phonetically moti- 


vated lowering enhanced by r involve high vowels 14 


Finally, in those 
cases where [e] or [«] is followed by r, one would not anticipate this 
consonant to exert a powerful influence on the timbre of the preceding 
vowel. Thus, for example, while first-syllable atonic e in Italian 


was generally raised to i, it remained unchanged before r: compare 


a 


prehensionem , *présione , prigione, phoenicem > finice, etc., with 
cerebellum , cervello, vericundia > vergogna, in addition to other forms 
like berbice, cervigia, mercante, mercé, sermone, pemice, serpentum 
and such infinitives as servire, cercare, Brora Meyer-Liibke (1890b: 
78) also cites numerous forms where i was lowered to e before r, in- 
smeriglio. Moreover, vowels were often inserted through hypercorrection 
or partial restoration, on the assumption that a vowel had been lost 
through syncope. Such an epenthetic vowel, being unstressed, was 
generally an i, except before r, where an e was inserted; e.g. supplere 
> sopperire, cithdra >cetra >cetera, cammdrus > *gambro > gambero, macrum > 
magro> maghero, mitra>mitera, separo > scevro> scevero, Germanic fodr > 


fodero, compared with blasphémat > biasma> biasima, spasmo > spasimo, etc.13 


























; fi ~imieaa-yite Sart, ou nat % 
: pmibsosny & fos ree" y tee 


D%. pmtts oattt Jdptert eagaod to penne ivnnaraa stous = oe f 
| eKits | [§]\a6 ab foe fon wol 6 20 e260 ott sll = 3] 


asng. {3erwande: Ighor ‘as oa x pribwot ict ‘ x8 Joo! & 


a ~qolovsb att to shedee ‘Lind iS Baws st nt wh 
‘ ae Hose four doin a nesrip se feeuIO «woled benkLtuo od Lihiw bess 
7 Pagel 4 Ze we 
. “pie yLlewoy ont zewol el fsutastiog Aluow ee piiwollot 5 sla] a: a a a 

oo 

‘  -—Som y fleoisan xia yfetuc, to- Bolquexs Betitetunab ee vetsemoitia | a 
f ay y ; 

eens ri “qu btenih Lt stowow ipid aviovai x yd pecnerne pivswl Bessy es a 


sich Stsoforins ctor bitow 2m" \X A Bead asst at. fa] wo [el suet S088, : 
age x NS | 
paifieosrg oft Fo srciis obs no ‘eoreustat iuitaeo. 6 etal o a ried 
milésl nt s sintts slosiivesdend athite ,slanexs 20% eonr ie anal 
ZA - oy a 
oIsgRx ‘x -SMteTsd bopnerioriy bertisnst +k to bseist rips " 
, a ty; a f 


aon int< MSOnLEM 4 SIOLIp in - mbit Sat SIOGiMN < medone woutia. etree 


a Wale oe. 


dyiw \.ods .spin!? « meclncodty . SOmpr ae sookesig® « a 


es ae f 


| : 


r 


Strict wzatio oF sortisbs of \Sipowi sy < sibrupitay « . fi Ia ’ 
muinecns2 \sobmisq \cicth3e, Sogem \Saseonan ein. wisest mt 


2G0CBL) oAciiidereyoM oh 29S, ;Su69289) iivisa 2S ony leantini fbue f 


9 ie 


“hi i smoted 9:as berawol asw i sr5ciw. RATIO? eucrcemme: éetto oeks 





Dns oats. foul , Corus I ,Cobkeirrao SQ0LE 7 ,8ifoiv 
_ a te te a - : 
+ ; te 7 A 


noitverioniscyil ieesexnets fedusent nedio'‘stew slowov 4 
aeol feed bsd iso. seth soz igmiaes edt #0 


; «BSW ,bepactiary prised ,fswoy oisetiogs es 


susigge .p.9 \bahreent Map 


"UT . 





ie 


The full scope of evolutionary developments engendered by the 
presence of r does not emerge from anecdotal presentations of the sort 
found in historical grammars of Italian, but may be strikingly demon- 
strated by considering in greater detail the statistical data collected 
earlier. It was noted previously that, in view of the general concen- 
Sus regarding a as a phonologically 'strong' vowel, the number of cases 
in which an unstressed a was modified, nearly always being raised to e, 
seemed disproportionately high. Similarly, e, considered to be a pho- 
nologically 'weaker' vowel, was not raised to i in as many atonic posi- 
tions as might have been expected. These discrepancies may be traced 
to the fact that, in compiling the charts for Chapter Three, no refe- 
rence was made to specific environments, except as regards general con- 
ditions of phonotactic compatibility. Consider, therefore, Table 38, 
which breaks down the evolution of a both before r and in other 


environments: 


initial DOSt= 
pre-pre- second cae 
pretonic 





Tables soss tallansatonic a before ry 


Here jLieselrectssor the tollowing ir are clearly seen, asijwell as the 
intrinsic strength of a in other environments, thereby strongly sugges- 


ting that the change of the theme vowel aetere before rin the future 


es | te So) eS nee ee 
_ eyt ne iN ea 






















Sid ‘ " : ‘* ae me 

— sonic: ing 

vor emo: Naa ks i 

were: 

ylprisitsss od yon tuck), cinch sos eeaaiel eins pea 

badvelicn ateb Lsolseitedte. of [insee retheome og ai 

~“fSDTIOND isvenap ert to weit Ri isc ylenobvany Asdon cow ty ; wi fy ; 

@3e6o ae ont oe ‘prose! vi Leotpatonarig be 6 por eda a i 
2 OF batics pial eyswis yWusard .beiibom enw s sisslballiede as inal we 

moily Ged o¢ boxwshienoo, .» vyfsal bine .Apid ylesanolsregouqaib bares 

-hang Sineds youn as ni i ‘o¢ beeitsy ton saw  leww “sepisew’ esaeeii 

feosat si yan eolorneqenmebh seed? jfedosqes osed eyed Sripim ‘2s efolt . 


a) a 


—_— oa ,seuril ssdqet) 2 adeno offs pit lignes mf" dem i aia oo — 7 
tite isxmep airisosr.2s tqsoxs .ctrenmeivas ofigaege GF shat anw cones ae : 

Bt olde rotors) jvobfeno) .\ LLiclBegqnes sigesieeande 26 anoktib 
60 nf Dis 3 sot sd tiiod & IO noisuLers ot mob esexd rire ie 


—jJeag ee if ss ae 





sina — Brosse | —suq-sig ishtint 
Signe; | obemderg 1] = Okice 04990 $0 
ib a a { 
RRL TD 
a Pi tee 
t 2 f 
Sane ee Weenie wena? Onn 
GOs | OOr (OWL) 


ET Sf 


: 
om A le le re 


and conditional forms is not the result of coincidence. Further evi- 


dence may be adduced by considering the raising of atonic e to i, in 


: yp 


terms of the presence or absence of a following r, as shown in Table 39: 


| # examples 
before r 


# raised 
cOmels 


toed 
examples 


% raised 
to i 58 63 





Table 292 Sitalianatonicse before r 


Hetesieimay be seen that the raising Of atonic € to i was erfectively 
prevented in the presence of a following r, thus documenting both the 
lowering tendency and the raising tendency which have been ascribed to 
r. As an interesting addition to the data regarding the raising of 
unstressed e, it may be noted that the presence of a following nasal 


served as almost as effective a deterrent to raising as did a following 


r, aS indicated by Table 40: 


initial initial | 
pre-pretonic pretonic | 
Le nde | 
before N 27 
% raised 
hee eee ee 
or r 
OL 


Table 40: Italian atonic e before r and nasals 

















¥ 


286 afde? cit awods sd eet it 1 





me ee 
















7 ne eaeaeiciee — 
| . 0 
A Pe 
¢ | OGL 
bes 
| . £3 
ere oe ae 
. E8 
- Es 
? 
3 : i ; Sa abd ian 5 
wievidosits esw t oF 5 oifots To pateres ont tert nese ee yen tt ora as 
MAS opuey - 


at recd pritaenrrs ob aur yt piEwoLlfot 5 10 STSesIq seit me Satas 7 


fh. 
 Bedinmes sosd Sayer rinieiw youebaad wiretss spit Bros Yeasbast sectional it 
ee 


fo paieies. edt ‘pibrspex ateb at of noi? bbs pectaonsdnt ms * i . 


ise6n pnimoiloi's to serisacndy ori tart Bistod ext yao 3 2 - 2 saa =a 
: is ae ee , 4 
= ‘pebwolict s bib as pricis1 ot Inexrs990' & svineis saad secmis as box ; a. | 


| 08 side yo Basso ae 
Le : a iY | eS aa 


initiste 
oF HIS 
ae 


ef 


pe a ae *) 


Pe 


ie 


This table indicates that, except when influenced by a following nasal 
or r, the vowel i may indeed be considered as the 'weakest' from the 
standpoint of diachronic developments. 

The above predictions conceming the assimilatory properties of 
r are therefore in large measure borne out by comparing the evolution 
and development of various languages, and in particular, characterize 
wi a high degree of accuracy the results in the phonological history 
of Italian, as demonstrated in the data presented above. In those 
cases where a reverse tendency is manifest, it is often possible to de- 
tect one or more additional circumstances which served to counteract 
the assimilatory properties of r, which to all appearances are not of 
a powerful nature. For example, the lowering of e to a before r in 
the future and conditional of some southern Italian dialects appears 
to be largely the result of morphological analogy with the predominant 
class of first conjugation verbs, as well as being influenced by more 
general phonotactic factors. Speaking of these developments, Meyer- 


Litbke (1967: 72) notea: 14 


I1 confronto tra amare-hd che muta l'ar in er (amerd) e 
dormire-hdéd che conserva ir (dormird) s'insegna la cronolo- 
gia relativa di questi fenomeni: LABYRINTU era gia diventato 
laberinto quando dormire-h6 valeva ancora per due voci; con- 
servando almeno un accento secondario sull'ir ( e solo piu 
tardi dormire-h6 diventava dormirehé, ma, poiché allora gli 
ir non passavano pili in er, s'arrestava a dormird); invece, 
solo dopo la completa fusione di DORMIRE H6 e AMARE HO ecc., 
MARGARITA diventava margherita (di conserva con AMAREHO 
amerd). Dunque IR ir 6 anteriore e AR er é posteriore alla 


fusione DORMIREHO, AMAREHO, ecc. 


Li2 












he ee 
hot r 7 ; 


io sebrisgotq wiodslunttaes atts: duabioain enincnlas c 
Pontulove of ornitisqna «i Jeo ond quai spas Ak b srot 
oabeginerato \xatubierreg mt sceijhgulicanalie 


Jnetaimoo ot Soyron coirw eesomtemonio Lseodsiiibs, exc 0 ' 
“be fon 213 asonszseqqs Lis of robiw «= ao megane. ox : 
nt. x-sxotsd 5 oO}. 5 to pairswol sf ,olathke 208 sours Ioewoq 6 
ernogs 2 stosisib osiletl qretivos ase 2o fsaaieibaa bos ste eth i 
 Srscbmpber ort sitiw ypotaas'’ [so tpalorigaom to Hives: srt vipast od cd 
Syon yd benasgint pated ee Iiew as  .edxev aoismutnee e183 20 seals ) | 
id. an meal, 7 ihe 


“sys ,zartengofoveb seed Jo pifisaye .ereeaet olinstonaig L ; apne 
: ‘ ‘ ef ; re ike eel 














odasnsvib Sip sro ULMINveEAT : inemonst saa Jb meisetar kp 
“foo (iptv sub “19g srcons svaiey Bat Di cence coe sgabtal _ 


; 
Sip oloe 9 ) at" Live oittsbriooee eemirinage) 


itp sxolic ae vay Seema 


osm * (Stimob 6 cvetesrs's on» 





29-2 tignn 0.85 ooaIOG 18 2g 


ook 
Bates 


‘Bae uno erstengo| Jb) - or 
rans a ck 

alle sniohtaieeg 6 ak : 
ne tt ee lp we v | 


all 7 rr ah x: a net 
# ra | ; ) rs aly ton 
H ' ' oo a 





. 
= 
af 


173 
Similarly, those cases of early English lowering of e to a before r 
may find an explanation in the more retroflexed articulation of the 
English and Germanic re thus justifying Vennemann's classification 
of this variety of r as phonologically [+low]. In any event, the 
detailed properties of the Italian r are not captured by the presently 
available distinctive features since they depend not so much on the 
actual point of contact during articulation, but rather on the accom 
panying movement of the tongue which both precedes and follows the 
actual moment of contact. To accurately depict this characteristic, 
one would need additional descriptive devices, for example the intro- 
duction of a new feature such as FLAP, together with a set of synchronic 
and diachronic interpretive conventions indicating the interaction of 


this feature with other vocalic features. 


5.7 The role of stress 


Returning once again to the data from Italian first conjugation 
verbs, it will be noted that the various assimilatory properties pro- 
posed for r are not of an absolute nature; that is, there are hundreds 
of Italian words exhibiting the phonetic sequences ar, ur, ir, etc. In 
particular, the very endings of the first conjugation infinitives, in 
-are, serve as proof that, if any tendency on the part of r to raise 
low vowels exists, there also exist factors which effectively counter- 
act this tendency. One obvious contender is analogy, but unfortunately 
the raising of a before r is not confined to verbal forms; in fact, 
Italian contains many non-verbal forms in which the same development 
ensued, aS indicated in the preceding section. To cite but a few spe- 
cific examples: cam&ra>camera, camfrus > gambero, matt&ris > mattero, 


Sarmento > sermento, comparare > (dial.) comperare, separare > (dial.) 





















aiet | f oscted 2 of @ to nivel th 
erin sitsasdiee ieee 


obiteofiimesio 2 'nemenhsi” peeon pi 


ry at Jrisve ys, nk =f wot) vile >. 
| ; yhineestg cuit yd berutteks +N SS eax otf to: volaoarag be 
eit co em GR $6a-ba: aqob yet sorte estatsst orisoitelb. 

smhoos aftt no usiitss stud ods ike piisub Jostheo No ano atom se 

egit awolfot Bras ait panetg ted ih.tte exned sit 20 iosnavor pakynng ry 

ss sandeay sgasiero airs toes yleismos of Josep 20 seme, Seaton i" 

-orsni «xi+ alanexe 10% (zeohvsbh avitgitmesb ipapiesihe Bear Biyow. an 
pheibados, to te2 5 ditiw rertiopod .4A78 2s tove saygset wate Io aoitoud 

Jo moigosistai scit. enigesibat enoisngynes ov Li srgistnk pidoutosib bas if: i. | 

@ eoutest oiisooy serio ativ swiss> ait a 

' 7 


encode 2p efor edt = 


; |= nag ’ pee ; 
Roitsneticy teri asilett most stsb a7 co nisps SOND priireutest ta 
: 7 ; a 7 7 y 
=oig esitracoty yrotsliniess asohtsey ent tackt bagen oe EEEw ti eter 
. ; : 


sberbnut sis stadt .et ted¢ issue ecefoades a6 lo Jon ote 3 202 beer “ 


Sl wos \ai (ct me eeOneepen oom ony arlt prisicbriee eb3yOW opilssl- to 7 
ci \eovisiniaai noitepotnos sextt odd Qo epathas qasy eee lois ie 
eeies oF 1 20 tisq ah no bala sata Ye Tf sSerth, jomag ae avise — 
sedan yiav itoslis dotity srodpsi pee ofis sxe, aamianast fs" 
spiiabecsidhapttons gud .ypolsne si: tabasd. agp acivds aro bagech 
| OB} at 4 anol Isdisy o¢ ben tines tox aba smith 
tnemgofers sms es rioctiw rk aii angles 
Bi wet s Set: esto of .uoidose » aaa 





==) 
a 
ar 

‘y - 
of 

— 

7 

=_— 


174 


sceverare, smaragdus smeraldo, margaritas margherita,*°, lazzarelto> 
lazzerelto,*/ etc. Bourciez (1967: 485) cites albero from arbSrem, 
by confusion with albXrus, thereby indicating the same tendency. What 
then is the common factor uniting all these words? Closer inspection 
of Tables 38 and 40 indicates that in each case, the a which was raised 
to e before r was unstressed; herein lies the key to a final solution. 
Italian, like the other Romance languages, has been shown to exhibit a 
highly differential treatment of vowels, depending upon the presence 
or absence of stress, Thus, Rohlfs (1950: §587) states that 'Das e 
(statt a) in der Infinitivendung ist bedingt durch die unbetonte Silbe 
in der Stellung vor r'.18 
In Italian, it has been seen that vowels were often subjected to 
syncope, although not to the extent to be found in Spanish or Portu- 
guese. Pei (1941: 38) notes that, even when the vowels are not synco- 
pated, they generally exhibit some sign of weakening, often through a 
Change in timbre. Of the vowels that were syncopated, the material 
presented in the preceding chapters indicates that the most easily 
lost were i and u. Significantly, in the future and conditional of 
the most common third conjugation verbs, the i is also syncopated: 
dirdé from dire, morro from morire, parro from parire, etc.; this ten- 
dency was much more marked in the earlier stages of the language, as 
suggested by the remarks of Meyer-Litbke cited above. The vowel e was 
syncopated also, although probably after first being raised to i, es- 
pecially in the future and conditional of many verbs of the second 
conjugation: avro (early ard) from avere, potro from potere, vedrd 
from vedere, etc. Most interesting, however, is the fact that o was 
hardly ever syncopated, except when previously raised to u, and a 


ie) 


never fell through syncope. In fact, certain words contaning 






















ee . i ree Ie 


ie <odigustss! . .etixedoaes < sits 
| amieiictiey mort ose rey a 308 
| tet yotrebrat cmoe ene ni Sees ingest at 
| noisnegeni *220!. ‘Sebrtw. saat ican adie mo it at ined o 
bakes wey rivirs sid .9860 hs8 ni aay obhabient 8 
wiebsbise fsnkz 6 Fes sds 2541, nisxed egeaencteres Sow 
5 Yiivies ot nworle noad aéd Steele sonic sofia ott 93 ae 
gohsesiq sf} noqu priboegsb ~alewoy 0 JSnaagest ae a 
& 2st" tadt esteta (5 ROP :020f) | sien vauitt saeete iunleree 
elie atrotsdns sib rovub dontbed ter ariatrcianitis nk — 
@ Bedsetdre najio.sxaw Rees jer oese oped Bed ¢P ii a ie 
“went to détnég? at Biyo? ed's Gastxs SHS o¢ Gen tievorsis' \Squoeie 
e@oryve jon ss elswov “ads fady fave aeeie esecton (BE sfB@f) te4 22900 4 ° 
: & meow aaito ,painorssw, 30. nple anos Dice yileuetion acd osha 7 
tsiteem ott ,beysocmaye row jo elowoy eet BO) | + scant nk gers 7 ‘a 
yliess tecn ost ists eaten ibit eyadgerio en ortt ni batncesry i 
to Jencitbeco bos sma? oth \yienoGinete bas £ eee a 


egoonye oals at i srt .edroy noisteput coy kaakele mcomeo Jace ort? ip ee 
wed add ¢.t40 aritteg mozd Grd \ oe brent mot Gerson pathic e “4 
| as .Speimast “Ht to aspate Yetfuse ots mt an sie | 
new o iswoy sf? .svods bain AR. Yo: atet 


wee .i ot bapiss ‘pated jeri? sedis yisedera i 


. sete ot seen es 
a: Cabey .snedeg mon}, 60K sees a : 


eg ae 





ie 
Ati 
ie 


hii 
7 ~~ 
‘ 





_ 





unstressed i or u changed this vowel to o, in the process avoiding 
an encounter with syncope; for instance thyminus >temolo, tvula >ugola, 
and so forth.” The tendency for a to resist syncope was demonstrated 
in Chapter Three, and has called forth many specific comments. For 


instance, Lausberg (1956: 156) noted: 


In den tbrigen Berichten macht sich eine starke Tendenz zur 
Unterdruckung zwischen toniger Vokale bemerkbar, was mit 
ein Ubertreibung der Druckabstufung zusammenhdngt. Am wider- 


standsfahigsten ist allgemein der schallstarkste Vokal ie 
Tekavéié (1965: 141) added: 


Per conseguenza, benché i verbi in -are facciano oggi il 
futuro e il condiz. affievolendo la vocale caratteristica 
a ine, la contrazione non avviene. Da cid possiamo con- 
cludere che al tempo in cui é avvenuta la sincopee la con- 
seguenta contrazione del lessema nei verbi in -ere, il 
futuro e il cond. dei verbi in -are conservavano ancora 

la vocale caratteristica a. La vocale a, essendo la pil 


resistente alla caduta, ha resistito alla sincope. 


5.8 Raising of unstressed vowels 


Intimately linked with the shift of a to e before r in Italian 
is the more general process of unstressed vowel raising. It is a well 
known fact of Romance philology that, ever since the days of Vulgar 
Latin, unstressed vowels have shown a tendency to raise, a tendency 
which has operated with varying degrees of regularity among the various 


Romance languages. Raising of unstressed vowels attains its maximum 





















bi batextenameb e5w agoonye rai ba Si Ne 
WOT .edremmo 3itinege yabne Hera bette esc & 


~ 


spa er Reet 


a: 
Ws siabast otisda ante mois ‘oem nt deren won ioe 
vim aw tacatremed of si0V opines eoee y 

| —tebiw m4, cis rire AMMBAUS peuttrtedlegbarst 13h pri 
dls isxoy stedisdefilerbe seb nismepife gar 


: ; AS r raul . : 
_ dt ippe ons ine eis nk ise & Saboed <sseopennmn af 

: piles Soa 
,S0iseiyedisuso olscov sf obdelovetits, .sibr0 ti 2 omut ae 
; : : on ona su 

z AGO Ome La e0g Of5 Ba .ssivve mon eee sf id be , 

z : vod le ahd 
—w2> al os sqoonie 61 sjunsves 5 io me ognat ry ‘aio Sone nee 

: ful 


Li . Sis- ni idvev iso sresest Tem aroisexiaco sinaupee ' ; 
sscons onsvevicerco oxi tdaoy eb Aaa A « ee - 
A Gig sl cbnszes: 5s ays SI 4.6 ‘otieleinainie tela 

re eqpoornte slis ottteragr ed S40 san in 
al 


) fsHadl nix sted a ot & to FLide port a 
> flee 5 ab Jl; .paleisx [aay foanegreay 
i epi on a aa 


4 eR Gs cere ) 
© Reerter ort erene 





176 


dimensions in Brazilian Portuguese, where [a], [e], [e]s [i], [e]> [fe], 
[Oo] >[u], and [> ]>[o]. Similarly, in Catalan, unstressed [o] and [°] 
are raised to [u], and unstressed [a],[e] , and [e] are raised to eye 
In Spanish, the tendency to raise unstressed vowels was much less pro- 
nounced, although examples may be cited. The developments in Spanish 
were sufficiently striking, however, to prompt the following comment 
from Makoto (1971: 38), whose full impact may perhaps be ascribed to 
a faulty translation: 'In the age of transition from Vulgar Latin to 
Castilian there appeared a tendency to hate low vowels ercioreren the 
most idealistic vowel system ...'. In Italian, as previously mentioned, 
e€ was quite regularly raised to i in pretonic positions, except before 

r and nasals, and sporadically raised, although more often syncopated, 

in other atonic positions. In literary Tuscan, there are few examples 

of the raising of unstressed o, but in the non-Florentine Tuscan dia- 
lects, as well as in the wulgar speech of Florence, this change is common- 
sie Unstressed a was generally not significantly modified in 
standard Tuscan, except in the cases before r. Somewhat raised variants, 
ranging from [e] to [=] occur in the spoken dialect, however, and are 
also found in other regional dialects. It is interesting to observe 

that unstressed vowel raising, which in itself is a form of phonetic 
weakening, followed roughly the same vocalic hierarchy characteristic 

of syncope in Italian; that is, e was most frequently raised, followed 

by o, with a seemingly being the most resistant. Of course it is dif- 
ficult to discover the phonetic status of a during the earlier periods 

of Italian, for unlike with e and o, no letter existed with which to 
represent a raised a, much as in Portuguese. Equivalently, there was 

no other phoneme whose allophonic domain would be infringed upon by a 


raised variant Of a. Nonetheless, the existence of the raising process 


un 


arr A ab AL aR 





















| fe} elst, 4) <i} (8) <6) oii of neil ise 
| Pie cglnaconmanh stay 8 ds ae 
* fa 5 heater oss bo).aa el “Seecentem) bas 5 

“wig eaei heen baw pfewo poccertith deter od no) 

Mepreng: 9) aynsmpplevah ott .pedeohed van elon stgeods is: seem 

Siti priwai rot Cts t¢poOxq oe even” .paiirase yWneloitive orew 
¢ Bedivode of eqaviveg van tosh Ofek Beart , (BE etter) Sule 4 
oP mess) epi mos Aoi tbenert 2% IPS ert et’ oobtsfenssd lust 6 : "i 


oan 7 
as 


vol 


ce 


eit 19tsu0 bos 2lowor wol sted ad toashasd = bemsegye ‘ore? ‘meifites® is 

; F 1 
wbenbisiten vlevoiverg as.,c6iisjI nt .'... meeye Tovey nivetfsebt secon 
4 ” 


Sidted jqsoxe .enoitiang ae hia i o¢ Beviss, yloeiupes ativp ssw 9 
beisgoonve casio sxom sooorttls .beeisr yileoshesege bas valsean bas 2 
esignexs wet ets syed lasoavT vieedtl ot wanoitieoq inode xortto ai- 
+sib event ox Litnenol’ tenon sit! nad arsci Oo howwontens to onizisx ed to 
epomno> af opnero aid sonstold to rossqe tepluv ods at a5 Liow es wack 
e! ai Saitibon yidreoliinpie tom yliszansp =sW # beesontant “Scoala a ; 
eonsixsey boatsx Een I sxOtsd 2oaso ery mi Sqsone vnsoen? bisbasia 
ets Ans .1aswort ,joolsih aovedge sot mi. mac te] ot [s] mest peti i 
evisedo oj prissersuni ai st -etoslsib tenolpes seto ai Braiot cals 
5ifenotig to mot 6 at tleett ‘ai roiidw .paietee foww beassrtany jastt 
oitelretoeteio yivisieid oi fssov ama add yiteoon bewol fot vninedlsow 


Gawolfod  beaisy yidneqpert team aswis wah sels erases ot ae Sl 
-2ib et ti vatues 30 .aestetast J20m oft ented eamanibeeieek 


eh 









fs aborted t4ii1ss sit paiiwb s to stigse no enone) at: si ae 
= an ; 7 
| oo daily stim bavains asm Df om 49: Bas gw ating a aay, 
; a | ee.’ ae ay : 
a 260 sash .yltoslevinpa oceans 26 ri | &. he ‘ 
~ 


i 


Ya S clisodieaaianeiaa hana aecaabince: | pit 
; (iam iat tt 


genong patties edt to -einstaine arth 





= 

is a documented fact in the history of Italian, from which it may be 
inferred that unstressed a was subject to a constant tendency toward 
raising, which was often resisted in reflection of the high position 

which this vowel occupies on the vocalic strength hierarchy. The com 
bined tendency for an unstressed a to raise plus the potential raising 

and fronting properties exhibited by r, could then be expected to 
interact, and the e which often resulted before r, while not totally 
accounted for by these two factors, follows the established tendencies 


of the language. 


Even when the assimilatory properties of r and the general raising 
of unstressed vowels are taken into consideration, there still remain 
certain data to be accounted for; namely, such forms as the future and 
conditional of dare, stare, fare, etc., which still preserve the atonic 
a before r. In order to bring these remaining forms under the rubric 
of an explanation, the phonotactic structure of the Italian word will 
be examined in greater detail. The key facts in this regard concem 
the distribution of stressed and unstressed syllables within the word. 
Except for a handful of unstressed monosyllabic forms, every Italian 
word receives a primary stress accent on one of its syllables. In ad- 
dition, words of more than three syllables often receive some form of 
secondary accent. Using the symbol (') to indicate (primary) stressed 
syllables, and (-) to indicate unstressed or weakly stressed syllables, 
the possible structures characterizing the conditional and future verbal 


forms may be grouped as follows: 


in, 7 ee 
Yhe ' hi Tak os tay Me 
si yen $f riibeiw mort mnt fad 2 

brawas yonaboet dnaverrco 5 8 sooteive ee 































a“ vn 7 
ictind tain 


nostinog spin oft Yo nolsosiiet tk satolons path td e 
sae ef! -yrrwezaisl dipasite vilsoay efit np eaiepmoe all ntcta ‘ Ete 7 
piisist etnies’ acts peti seis. oF S boeeettams ris IR | 

ot en éd Aiedt Blnoo ‘t yd bsoiditike aay a Be 


Yllatet ton olirtw ,x sxcted Setigess aatio wba 9 ae bas ostetn 


- ® 
ae a | 


eetonebost bsrisifdstes arts, ewolfok , etqdan2 ows saertt ye x8 b i 
nearest t edd.2 

eee 2 a 

: 

noisiveg inow 30 elo. eo 


tid Ai 
a = 
ef <a a 
bes sucnd aif 2s etrxot rove ,vienen. rot bednyooos ed oF ‘eae nists i 
met 
oineds edt evrepsty [lite nnitiw’. ots \otet sists ema to isonisibecn o 


ne 


¢ eae ve 
pateie: iste oft bas 1 to asidreqorg yuodslimiaes oekt nstiw nova 


xy 
r 


theta: {Lite exert no tisxebienc ont nSelet ox8 alaier bsesortams 3 


2 icin a awsebmy emo? ULES [ssc paiid ot +8710 at = excited & ; 
ti ; . Yi ay in we 
tify brow celbiscl sit To smutopite 2 stonedig ort wekseaelgee 1s 20 he rf 


itissno> Prater sirch me atost ySi st? \Lastiaa seated ak benite® a id 
. 
Betow ott cicttiw esldafive b: mike A bre pezesste +6 coisudiseib ‘se ” 


J a 
atl naa et: mem a 
ms kiesl yvtess \aetci side i tyeonon Arent to Ietiheed 5 20% FORE 
Pele 


—— . « _ 7 
~bs ml. .ésldsifye ati 3o sao no 3nso06 sess \amexg s Bovis TOW 


to mci samde ete natto enide Liye seid ash ae to st ag 


besesrte beeen stadipar o¢ .(') Lodimye odt pried 


a 
(sue 


a, 


aan i 
,aoldeiiye Boeesite yitsow 10 beeasttens eteo thutk at os a Ye 
ledtey sxutut brs Lents tino ould eciis emiaanasts SSIUIOUISS SiALeRC 
2 ~ fom pm 
: rawolicd 26 f ses aia 





Lee er als rigey f, 


FUTURE 

isa) dara 

(=i) daranno 

(as) parlera 

a) parleranno 

a) perdonera 

(soma) perdoneranno 

isizsie) dimentichera 

(—--'-) . dimenticheranno 
etc. 


(-") 
(-'-) 
(-"--) 
(—-") 


etc. 


CONDITIONAL 


darei 
daresti 
darebbero 
parlerei 


parleresti 


parlerebbero 


perdonerei 
perdoneresti 


perdonerebbero 


dimenticherei 
dimenticheresti 


dimenticherebbero 


In examining the above chart, as well as the data presented in Table 


38, a pattem is seen to be emerging: 


a was Kai sedetore.belorerr, wien 


followed by the main stressed syllable and preceded by at least one 


unstressed syllable; in those cases where the a was in the first unac— 


cented syllable, it remained unchanged; hence dard, stard, etc. 


Consider now the accentual patterns exhibited by some of the other 


words involving the development of the configuration of atonic a plus 


xr, exemplified by the following: 


(-'-) parare 


a.) margherita 


Li78 





ijteotsb 


Or os) 

fosstiey = ("—) 
Cae 

isestoliey rn 


4 i 
isisrobisg ('=-~) 
isgetsnobseg = {-'-==} 
greddessnabisg = (-—'—-*) 
et 


iFiss iin br EE us \ 


ee te 











itesiemsisnemi — (-'--++) 

aoe Ef | ie 
ouxidssiSnamib (--'--—) ts, ey 
— Behn, 

met pak “a 

AJ - , = y 

she 
Pali ie 


; ah 
ade ; SITES e 
er et 
ai ps Liew es. eartenrhess y 
ae a 
ipipieme sd ot riese al one 
erie. 
aldsi lye bees ae wt a ot fas 
ig ” e 
“oon deri ly al any 5 Std erin dees Sant ak :aIMnBya E item 
} iat oes 7 Ady: 
te Gutta .Gxeb sonst ‘bepiariony benismos ti voldaLiye bs patna 
all ah! ft a en 
terito etft 20 athe yd Betididks amediaq tsodmecos nn 


pa a 
wit 8 2inats to miseupitnad ads to — con 


% +4 
Pa 


SitisT ni hatrsssic stab 
iatw 1 stoied 9 ot hoatsx 26ws 


ono teest 36. Yd babsosxq Brits 






= Sree a = 


nue = | 


fg 


etc. 


In these examples, we find confirmation of the hypothesis derived 

from examination of the future and conditional forms; in addition, it 
is possible to see that raising of atonic a before r occurred in the 
mirror image environment: when immediately following the stressed syl- 
lable and followed by at least one unstressed syllable. Here then is 

a more useful sort of generalization, which specifies the precise envir- 
onments in which raising of a was the general rule. Given these data, 
it is possible to formally display the results of this process in terms 
of a diachronic eater? Consider, therefore, the form of a rule 


Gepieting the results of the interaction of a and r in Italian: 24 


VE rv 
1 - stress 
if 
—- stress 


Such a rule, while formally describing the observable data, is highly 





(1) a > e 


unsatisfactory for a number of reasons. First of all, it is necessary 
to state the rule in tems of two disjoint environments. Superficially, 
it looks as though there is another case of a 'mirror-image' rule as 
defined by Bach (1968); closer inspection reveals, however, that the 
notational devices proposed by Bach will not permit the two neighbor- 


hoods in (1) to be neatly and economically collapsed in an illuminating 


0S 


hovireh ateadogyd ety to ooltenmrfinoo hart Sw yaelquexe seeds nz 


4 


‘haces Bos earcstye Ste BO polbtseriimexs mor 


wee 
@ 
sal 
ret 
o 
Se 
bj 
= 
} 
ua 
. 


edt af Seriumoo 1 sroted 5 atnots to priater cad ese od eldieacg af 


-Iyve Baeastte ort ontwollct yfsisLbonmi aow sins opems LOTT oM 





at pert! erst .aldni ri" hidces Sit, SfC yasal +s yd bewollot Bas sidsl 
wie sainengq otf eeitrouge foirw ,nottssiisioiee 76 Pies ivisey sign Ss 

ab 05 i! nmevib .sisx Ietonsop =ci es 5 To prireism Mpitiw mi einem 
ayeied crt - ood et Jo etiuesr : 4 ve Ih yvilermol af sidtesog ef tz 
. eftry 6 to omot off  sxotetatt brand ES ota oinordos ib 6 to 


wWiLissl wt x bre to noljosretni og to ativess ort paigsigsd 


5 . \s + § ( f) 


j ~ "— 

: — - : wy ; 

‘ t - 

; ¥v 7 ; 7 : 
= 5 a) _ ” j 

\ jeearta . y 


Jrbpin @2  .edeb aldevisedS edt prrdtvo viiamroit alitiw ,elot s move 
G@isessosn 2i ti ,[is to saxti .enoessr io 16cm 6 aOR eisSIeweeeD 
Vitsioitusayes vateramocsovas trtopeth ow to emisc at sig aie soaga ao 

es alu ‘spani—torrim' 5 jo eaco xeon al sxe reodh ae eer Se 
“i+ tort .teweuor .aisever wexsuaaina reaato 1 (8eL) fos Vd ban ttsb.: 
-jodipten ows sed dintsa jon Litw dose yd beaousnit aso ive meen 
paiveninn fli’ ae i beexeLloo vil co imoncos bas einen ad ot (f) nk beer 


$4 
_— 
7 





180 


fashion. A new notational convention is needed which would permit 

the neighborhoods to be conflated. Such a proposal has in fact been 
offered by Naro (197la: 58), utilizing a notational device first intro- 
duced by Langacker(1969: 858-9) and Harris (1970). Naro has suggested 
using a double slash // to indicate that what follows to the right is 
actually two mirror-image sequences. Using Naro's notation, (1) may 


be rewritten as: 


(2) ary > er // VC cy 


In reality, however, even this revised statement is not totally adequate, 
Since the true consonantal environments are not correctly stated: 

the sequence vCercv is inconsistent with the Latin stress pattern 
carried over into Italian, while the pattern Vercv is equally rare. A 
few examples of the latter case might be adduced, however, which would 
allow the rule to stand on formal considerations alone. As an interes- 
ting aside at this point, it might be noted that, taken as it stands, 
rule (2) appears to provide a clear counterexample to the claims of 
Naro (197la) that all mirror-image rules are cases of neighborhood 
assimilation. Although one might wish to claim that the shift of a 

to e before r is due to a form of assimilation, there is nothing in 
the surrounding environment which may be interpreted as having exerted 
an assimilatory or dissimilatory influence on the sequence ar. 

The restatement offered in (2) does not add any value to the ana- 
lysis presented in (1), since the point is not to evaluate the simpli- 
city or generality of a proposed description. The events in question 
actually took place, and the motivation for them has already been 
tentatively suggested. What is needed instead is a theoretical approach 


which will clarify the relationship between diachronic processes and 


ae | ae 


+immeg bivow Roility babeen ai nokinavace Ieapltayor ber A notes 
cat gost nt asl kabpadio: 5 dov2 .Batefiae od ed ehoodhodrdiptan edt 


-catnt $2112 apiveb Lsnotéston s paisitatn . (8¢ sel) ous yd berstto 
fhadespows esd ons . (0TCL) ser ist bas (2-829 :ea0L)sebapned yd boon 
ei dtpebs sit of awollot sariw Jan ainoifar: o¢ \\ dese eidvob s pnitey 
van (0) .opigsatod < Yoel onia) ..2eonsupee opent—soriim wt yi tests 


7268 aedtixwex sd 


v»> W \\ 2 « SS (S) 


‘°C 


egepebs viletos ton <f Siaiscste hoeiver eit neve .wevewod ,yiilsex al 
vitoorico ton s1n atoasmmoievas isonaioemeS corrt ott conte 


: : ~ 
mts: aorte oidsal ons ctw taetetencont af VOreOV someupes ocit 
SL OE 


4 


& eve: vifsups ci Vous medgeq adi slidw «aeifetl oint tevo berriso 


= 
rts 
, 
a 
4 
he 
2, 
i 
“ 
“ 
; 
, 
- 
A 
Li 


fiptit sain raiael sit Io pery-¥ 9? wat 
seeedat fe 2A .snole enokterebienco temo? ao Bande of etm att WOEns 
phaste +i 25 nevet .tent becon sd sipim si ,dikog eh 3s obies ont3 
snisl> off ot Sf —_ anes. tssip 5 sbiveiq o¢ stseqqs (S$) siox 
hoortrocdcie.t senso c1s/eolir speni—rorninm Lis tems (1Vel) ox 
5 to atide edgy tard misio’ ot deiw tdpim ano fpyorttA nots Limiees 
eniction at srorct .coltslimrees Yo mot s od sub et = sicted s at 
bakiaye prtived fs betexqtatai od yam .crw creme ove pribaverive eit 
1 sonstpee ort, no’ sonadTing viobat iba to yiedelinkees 16, 

ers oid of sulev yas Bba ton e90bh (§) mb Dorsiie Sreiiegesees ss 
-Lignta sf stsecfevs o3 Jon et rq ott sonte ,(f) ok betmeesag eteyl 
noltesrp inf eines aff .noLdiqitos ab ‘boeogosd 5 to \eakngenaG ae ete 
reed. ybsoxrts aes mait +o notsévison « Bes ,sosite, soos vilendos 
dseciqds TsohiSrvoats 6 et beetant. baboon at osnW Reka: 


Li Pas 
ei rs 
* “ M 
a : rr 
© i _ a g hee 
y + a7 4 





LOL 


formal representations. More serious than the descriptive shortcomings 
of (1) and (2) is the fact that these rules fail to reveal anything 
about the true nature of the process involved. Even allowing for the 
full expansion of all segments in terms of distinctive features and 
making provisions for the assimilatory effects of r along the lines 
suggested earlier, there seem to be two almost arbitrary conditioning 
environments. There is no indication of precisely what effect the 
configuration of stressed and unstressed syllables exercised upon the 
development of the sequence ar, nor why another configuration would 
not have produced the same results. These are questions which must be 
answered by any diachronic theory laying claim to completeness, and 
in the case under discussion, the proposed answers lie in a more de- 


tailed examination of the unstressed syllables of the Italian word. 


5.10 Positional strength 


Although each polysyllabic non-compound Italian word receives 
only one primary stress, the amount of articulatory energy allotted to 
each of the unstressed syllables, and hence their diachronic strength, 
is by no means uniform. The preceding chapters illustrated the fact 
that there is a clearly defined tendency for certain unstressed syl- 
lables to exhibit greater diachronic strength than others; most note- 
worthy is the initial pretonic syllable, which generally resists 
syncopation or other forms of effacement. Grandgent (1934: 91) feels 
that this preferential treatment may be due to a ‘lingering influence 
of the Old Latin accent'. Final atonic vowels also fared better than 
many of their word-internal counterparts, despite their general weakness 
in tems of articulatory energy. Word internally, pretonic and post- 


tonic vowels fared worst, being subject to syncope and other forms of 


thd | : : oy me 
epiiimnotiote orctqghmesb ot osdd esokuee oxo .anoitetngesrge:. Learo® 
PALAS yrs LBEVSI ot fred @5iurt Ssaorks tert dost cit et (S$) Bas (1) to 

a? aot paiwolls neva .bewlownt seanoaq sit, 20 sasse surxt -ocit tuods 
bas eswmiinet oviconidaib to untied 1. eirismipse ffs 20 motansgxs, [Io? 
wetl ott ofols 1 to etoeits viodsl bieas att 10? enoieivorg peblen 
pLioitibres yet itive tears ow sd .ot mose stent .pelliss bedeeppve 
€ ‘4 TTS tHflw Vise 7x 7 sObtso tint on es sted? , etnennoLivis 

4 hoa fasioweso saldeilva heaeortaw bes Beaesate 2 mo Osivpringe 
buts Siterupltaco isi niw TON th sonsupes a Jo toompoleveb 
sf Yam coirw anoigaeup sie sead? .aiivesr ome set Beguboug sven Jon 
ms ,esonarsiqnce oF mts 1LyBl vei oboowfins LD yas yd. Setowerns 
ay sy 56 at otf src bse. ret oleeuok ib yesh ees off AL 


3 sot oie 26 : (ive heeescten apt Qo noidenimexs Delis 


Spneate Ispoitiact OLse 


si fet cica-rod aidsliveyviog doses mpworblA 
7 Bagso tory 4é tops Yo Jmuaes ond .eeate vasmiuq sno yino 
Hiensste oitnourfor Est axt bes ,eoldelive Besemeteaay sid 2o nose 
tas + Betettant te 5 piibeesrg off .mebing eanmsen on yo é2 
-ive betesctarny nistxso x: wohingd ber rte yiagefio s el. siedt Jak 
-ofon .t@an seuccdio osdé ritpneste abcowhbsib wissep Signe oF asidsl 
afoot (62 = KEL) srepbriw ramsontts Io aati sartto xq nokssqoonys 
eonaitini poiveetil' s ot seb ed yan shamtestt letireteiasd so tect 
(edt widted Seyret coals alowor oineds.Ieari  .'dneoos amael blo ait Zo 
| ‘ re 
spenKeow Ls — tied+ atiqeeb .esusqiaineos Lafrrooni-facew tied to rece 
-_ 2 ae - 


~Jaog bis olaederta ,viisomedal Aro . yerotie vaccieiioionne to ened al 
" « P ,~ Sip 5 
to emriot asro Drs Sgoonye at tnetdve pated tetow Boum alowoy since Pp 





weakening, On the basis of such observations, a hierarchy of positional 


strength was established, accounting for the varying behavior of the 
unstressed vowels in a diachronic perspective. At least three levels 
of diachronic strength may be discerned from the outset; starting with 
the primary stressed vowels and secondary stressed vowels, this level 
may be referred to as [strength 3]. Next come final atonic vowels and 
initial pretonic vowels in medial position, which may be designated as 
[ strength 2]. Finally, we come to word-initial pretonic vowels, and 
word-internal pretonic and posttonic vowels, the weakest of all, which 
for the time being will be called [Strength JJ oe Using this set of 
values, a partial reformulation of the above charts will place matters 


in a somewhat different perspective: 





(2=3) dara 

(2-5-2) daranno 
(ot) darebbero 
(2-1-3) parlera 

(Za laoee) parleranno 
(2-13-12) parlerebbero 
(Sole) camera 
(2-1-1-3) perdonera 
(2-1-1-3-2) perdoneranno 
(2-1-1-3-1-2) perdonerebbero 
(226-1) 13) dimentichera 
(2-3-1-1-3-2) dimenticheranno 
ao) dimenticherebbero 


GLC. 


MP 



























ai 


lenoliieog to ywhrell 6 \anoiseriend dene 26 abeed et AO pation ae 
rit 320 toiveded paivisy orl} it erctonte  beslet [dey | egw Hit ent Ree | 
eiovel sexist tesel JA . sv idoaqereq pinorrerb 6 ‘At alow Soaeodands a. e 
rttiw pnististe ¢teeive ort mort hentoskh sd -yset pects Sree ae 
level aid ,2tewow fozercte. Arairoos=: Bhs claw Seaesrte yusetktq adZ : 
bas elowov Sinate fer er mop txeii .[€ diterstte] 28 ot borrm@er od yen | 
es badsepiesd ed yeu roldw sobtreny Nekbem at efoww Sieotiag feteink 
bee  .aiawov oi0tsig fet} ini-fsow ct amon sty vifseata 6 $ ritproxte ] 
tb bew Lis Yo Jeclsow od} Valawor ohootedd Bas Sineteta Ieoreini-fiow a 
fo doz cit oniey ~~. [fk cipseste) belies of Biew pated ent? ai yor o 


eis sasiq tliw emraro oveds eft do noiteiemibter Isitvseq 5 \equisy 
; . . ” - 5 
neqeisg tniais tarwemoe 6 at : A 


Eat (+8) 
a (S-€-S) i 
msekdorsb {S-1-£-S) 
Bralaag - (e<f-sp! 
COUNTS Eg (€+£-I-9) 
creadere fit’ | (gemce=fcy it 7 
SISTBD Step * 
Exonnb1sq (et-p-gy" 
otha oasiPg (queegt ay <21P 
oxeddlan. annie (S-Ese<g—r=¢) 
Ornette tscremri 


183 


From this chart, and from Table 38, it is seen that a was converted 
to e before r in precisely those positions in which it was assigned a 
positional strength of 1. In these positions, the relative strength 
of a was weakest, and therefore the assimilatory effects of r, Which 
did not operate on this vowel in stronger positions, could act to raise 
the extoses 

The notion of relative positional strength permits an obvious 


method of rewriting (1) and (2), namely as: 


(3) a > e/ 
strength 1 


6 
The statement embodied in (3) is a simple and concise representation 
of the fact that certain Italian future and conditional forms exhibit 
a vocalic alternation resulting from an evolutionary development which 
also encompassed other, non-verbal, forms. To a very large extent 
(3) is also a productive synchronic rule of modern Italian, accounting 
for the alterations in the first conjugation verbs, and stating a 
phonotactic generalization which holds true for most of the native 
Italian vocabulary. In its present form, however, it is nothing but a 
bit of shorthand notation, standing by convention for a complex series 
of historical developments, but devoid of any theoretical content. One 
must resist the temptation to interpret this formal notation as having 
inherent significance or causal Seyi 

In order to propose that (3) is the correct representation for the 
events in question, it remains to be shown the manner in which such a 
notation may be made compatible with current phonological theory. More 
specifically, one must face the question of how the notion of positional 


strength is to be incorporated into the theory of diachronic descrip- 


£6 























barrrevnds ne Ee oP at $2 ,88 sida? moxt bas 

& Bampiees saw ti Bitter rik anoit ined pear ssbhaebeoiecciacle 

depnerte evitieicx orb (enoljiseg sean? ei .f to ogansecteootbeadts? | 

tinitiw .1 Yo atosiis yrotsl intees. oat exotered? bas .teedsgew enw. 5 20° 7 
exist of jos Bien ,2noltiacg Yronomenct fewer eli no sdexsqo jon Bib — 


2 4 


3 f« 
exotutio ce etinrog dtoderte Isnobikeag evrseies to motson off 
-25 viewer (©) Bas (2) eebeiawer fo bordsan 
-_ co 
¥ . \ 3 + 5 (€) 
: ; L. rieiprissse 
h.. ~<a , : 
Ao Tah ¢ WICO DOG Sate 5 (©) rT heathexdme tnemactete ett : 


* soit a feyab yvrsnol pat vo 76 mort paisives: aottentedis Si fsnov 5 
ineces. oorst vrev 6 ol aro? , feckrev-ren , yao deeasqnoons oals : 
prgtrceen ,felisti o whom to ofux ofnerdoays evitouboug 8 cals ei (€) oe 

5 mittgete Bus. .edr ‘ notisoubme tesit st a anoisemstia att 20t 
witan ait to teon yot «ant abled roidw soitss Lisrersp oitostonordq , 
6 tue auirtion ei +1 ,yavowod..miot jnsesmq ec al yisiodscovy nsifest 
eciise xgignoo 5 10 noitnevnos yw naibnete noktetonr basdtxode to tid 
a) <tnetaes Sestistoend vos to hioved-sud .ainemgoloveb Ientroda tel to 
rived 26 noitston femot eins teiqrsdn® at eolieiquad att teleost Farm 
ceo isasco w sonsoHtinpia anesednt 

4 

edt 2 mobistasesrqe: dostyo> odd acote) ee geogory of tshto a : 
5 soue dotdw oct -cstinem exit crwotle ad od erriameyx $i saoisesup nk snes 

F 
om syed Esoipet ocorig Sngrauss aie Sulake aban od “A 
Fenbistiaoa to Roijon’ ort vod to nobtesup ett eis? Jem SMO. ¢ 


sont sans vost Senge 


fi | vo Pri ‘al aa 


184 


tion. The values of strength cannot be introduced directly into the 
phonological representations of the vowels. This follows from the 
fact that in their underlying representations, the vowels are unspe- 
cified as to stress, which is supplied later by a system of rules or 
conventions. Nor can a stress-asSigning function be included as part 
of the general stress rules, since stress may shift within a deriva- 
tional or inflectional paradigm, thereby causing a reassignment of the 
relative strength values. What appears to be needed is a set of inter- 
pretive conventions that will assign the correct values of positional 
strength relative to the main stress; i.e., which will be put into 
effect after the primary stress has been located within the word. 
Such conventions would take the form of phonotactic interpretive pro- 
jections, which would serve to include a value of relative positional 
strength in the final specification of each vocalic segment occurring 
in a word or morpheme. Such projections must be largely language spe- 
cific, in a case such as the one under discussion, forming in this case 
part of the nmetatheory of Italian grammar. As visualized in the 
present investigation, these interpretive conventions form an integral 
part of the grammar in question, and augment the effect of specific 
rules. Put into effect, they operate over the entire discourse, and 
as such may be properly considered as a type of 'meta-condition' which 
characterizes any utterance which a speaker might produce. In the 
particular case of hierarchies based on phonotactic position, the 
values assigned will not always find an exact match in the directly 
measurable phonetic properties of a word; for example, although the 
final atonic syllable may be assigned a strength value of 2 or perhaps 
even 3 relative to the primary stress, based on its historical ten- 


dency to resist apocope, it is often articulated as weakly as any 


; 4 Py ei € 9 Lai. Fab tn Pe as i 
Bs ) ; 1a ae 


ot oni ybioertt benibosr ee rors tee 
ost oct Bwollo? aid’ .alewor a8 Ae: 

-sqanu sis alow ois. <enolisunecsiqe pilyiazebae ies hk 96 

Nd.eatcer: to texteve 5 vd zusE: boliqave at roldw ,esarte ode 
























gapq a6 Eebulont ad noi Py prLimitag-aactte 6 fig. 1c vaishpeeetale : 
siaeiuheh, 6 ruceitow sete We | easite gate ,polut aeoxte Iszan9p edt Fo 
att to dnatipiaescx s voiaves ydsmestt) modbewsq Lsmolsoslitnt 0s Rents f 
—regni to jse 5 25 bebsan sof ot eipeage SeGN .noilay dGpnsitea ov its for 
: Ismoisieay to eaclev tosixo> st opietes Lliw ted) anoitasydco. svieig 
cdni tuq od (liw rotilw ..o.f yaaeste aaa eft of evisslor ritpriexte . 
Ow ont 1 ne iw batsool ceed asd eteite vasmieg edt sedis sostis 
og ay ists Treint oLjostonong 40 not edt sied Sisow anokinevneo Tage 
Epnsiste xj ovitslen Jo sulsy 5 ahulonbh ed syuse Bilvow moiiv \enotsoot - 
 Pitieimoe Snamee 3: ispov foce to noitsoRaesqe Isak edd at paste 4 
-aqe spappnst vioprs! od deur anokinsfeig aed asa 10 Brow s ni 
sano Bic ai pmirrot .no(ecimeth zsber eno ent ap ee Sess 6 fe elt! oe 
sit fi BSetteuery 2A _temexp oasilst? to yomitesen ab to tq 
larostni as sarict 2notinsvico sricentaraiat aan wolssepiteount $rsag1q- 
oitinege to ssetic- of tasmais Bis .nOmesip at Tae adit To t35q 
prs ..semoosibh sx ind of tevyo ere Jo yotd? \jos8is ott tud- .eelix’ . 
foiry 'nortianco~sitan\ to smc 5s ep saat yireaqomng’ ed yar foue 25 | a 
off al .aceboug shim tsiesge's riskefie Osis 2m) 1) ys eas radostsi ty 
oft nok teog Disrmatonorig no boead sa tdore seit io saa ta 
Yisoer th edd at desem toexs ns) Sat RYBWhE: tor Diw a se = 
ott ripuorltis «slonéxs yor :brow 5 = astinedorg: saan ee ese 


oy v=" - 
: a 
7 7 eS a ws 7 
ae 


ne agsitreg 10'S. to evisu hinnsss2 £ eo od ‘en's Lele: aatiip ir a ds 
| igg4  IndluGteli ext no beasc alata aaa +e 


wr 25. ral a6 beta fro itis ato jek 32°88 


' 7 + 4 Mee ar ee: : 
7 7 or + i oy. aed is =f 
a “es d , “9 - + ra 


ats 1 ~5 & Gs : : s=— * 26 


185 


word-internal atonic vowel. The values of strength reflect therefore 
not the actual physiological intensity, but rather the phonological 
behavior characterizing the vowels in the various positions, and es- 
tablished on the basis of historical observation. 

Once primary stress has been situated in a word, either by means 
of lexical specification, or by rules of stress assignment, the rela- 
tive phonological strength of the remaining vowels may be naturally 
accounted for by phonotactic interpretive conventions of the following 
canonical form: 


(4) Vee ae | oe strength ] / x, 
1 areca Ty 


where each x; is an integer value, and the xX, are the specific phono- 
tactic environments characterizing the different value of phonological 
strength. 

The characterizations of word positions proposed for Italian are 
of course highly rudimentary in nature, and require many additions and 
corrections before a true phonotactic description of all Italian words 
can be achieved. For example, it will eventually be necessary to take 
into fuller consideration the particular consonants surrounding the 
vocalic environments, to account for possible assimilatory and dissi- 
milatory effects which might be discovered. Moreover, provision will 
eventually have to be made for compound words, as well as for certain 
derivational forms in which the accentual pattern of the root word more 
strongly asserts itself. Naturally, if the theory is to be further 
refined to include such additional cases, it will be necessary to speak 
in terms of more than three levels of positional strength, in order to 
accurately characterize words with two or more word-intemal atonic 


vocalic positions in succession, Since in such cases a sub-hierarchy 


issipoionorey arty sass, joc we sind 
+ee bis .enpitieog suoitat ort: ahiee wen 
.sobjevreatio Safe total to aianhiadet 

' 7 ot | 
Sane sil xecttits .Bbuowls mt ‘Dateusthe eee aad eeorte cineca 7 . 
aio =i} ‘unlgiaboe ese cta io colors yi to ,noitesfiioage: er 
¢ilevieten ed yam alowow er i Ro dposita ste 7 

ertwolicit ait to erieisneyac ov t4 oneopradin oigoasanorg yd yor bedrooOs 

srs Isoinonso 





























=a 


haat, 


f 
i > 





/ x \ p@onsrte x} + V (by) 


Pa ‘ , 7 an Ae a = 
—odoriy-oitiosge sit S16 x sit Bris: a laiail repent a6 BE) 4% 89 stot 3 _ 
se - i 
feslpolGneds to aufev Insts et?rb ot ei thresh CARDEN oftoad ee 


. ens metiscd sol boeoqorq anotjcepeigaaw to sooitss ixatasxaco : ott ae: | | 
aibes anoidihbs viem o1tupor Bre sagen ak yesdasnties yidpiet SE IOD: RO oe 
etsow nesissr Fle, to noise! ash obiosdenog scx 6 srtcted ano Ltogrx00 , 

o4e3 od visaesoan sd yiisodaays Phi ti .siqnsxe wot =. bevarrine, od 16 


» 


add piticiuorace einsmosago tsinoitnee sm seitexehcencp roll ott m: 
-f£a6L5 On& yrove linttees SLCLe RO tO? tnuoous of .ednemnosivas of > 
[liw noteiverq .x9v0StoM ‘havent sd tdtpim doiviw edoeite ein 


ftéetreo rot as Iisw . ohrow es “tot shan ed of ever 


. 


exom Brow joo avi lo matisa [eutas908 oct foniw ab epot 
godt? od ot 2: Vrosrt ont SE etteese -Aleeds mrcese vl 
—— od yaseesosn od) iiw Sr 29885) bse ce 


Cat et ont 


= oF xbbxo. ai \stpnese Lenolttog: meas a 
P23 


oiLneds. beet -biow tent x0 ona 
vibrate 65 BOUBO rower, at's 





186 


27 Even in 


is usually evidenced by subtle diachronic developments. 
words like parlerebbero, the stress is not evenly distributed between 
the pretonic and posttonic vowel, although these differences are 

usually magia apene Only a first step in this direction is offered 


by the remarks presented in this investigation, but the general direc- 


tions have been indicated. 
5.11 The vowel scale: a speculative interlude 


Before the matter of incorporating strength hierarchies directly 
into diachronic rules is pursued further, a matter referred to earlier 
may be returned to briefly. In particular, the groundwork has been 
laid for a speculation regarding the problem of relative vocalic strength. 
In Chapter Three it was noted that, as regards resistance to syncopa- 
tion in Italian, Spanish, and Portuguese, it is possible to group the 
vowels along a scale of strength, with a being the strongest, and i, 
closely followed by u and e, being the weakest. The question follows, 
therefore, of how this hierarchy might be represented in a grammar of 
these languages. As a preliminary note, it has been observed that 
this scale of vocalic strength enjoys a wider range of applicability 
than merely characterizing syncope in the Romance languages being stu- 
died. For example, in the process of unstressed vowel raising observed 
in Italian, Portuguese, and Catalan, a similar hierarchy is noticed, 
with i and u emerging as the results of weakening of e and o, respect- 
ively. In many dialects of Portuguese, the e which has been Yraised 
to i (or to [9]) often falls in rapid speech, while the u resulting 
from o maintains itself. In Spanish, only the vowels a, e, and o 
normally occur word-finally; in the earlier history of the language, 


final u was converted to o and final i changed to e.* In’ Catalian, 


aba 


4 1 
tesa batuebreed vinevs tomet semtta act ciasceratsmipadiscaiaa ears 


boietc ei nomoetib eift ni gate desis vino 8S gidipiipan yitsvau . 


-[exih [sverep sit sud ohtspitesvaremg i bacosee ug eaotismenr erit yd 
ee -bstenticect osead ever enokt ] 


ylivenif 2cirorsisid dtpnextes onidetoqreont to sesdem ed? sxotsi™ 


7 
~ 
S 
oa 
\e 
@ 
4 
+ 
a 


—SeeeaTy: ao socectaieor abner a6 tert Beton asw 2k sexdl sedge Ae hs 


oyit auoxp of sfdieaog ei ji ,sesupudtod fas ekwege ,fetlietl Ab nos 4 
awoliol nolieeup ef! .tesisow att pnied .s basa yd bewollos yieaol> 
30 semisrp 5 1r batnseorasr sd adpim veo tededl ait wod Qo \Sandarers 
“38 pited eopsuymie! sonenod sd} ni Ssoconva pris etosganoyiexen me 


havasedo priietex iswov hfeseseitany to eesoorgq afd nf \elge. aot barb " 


“desqegt .o fas © to piinsisew To adiveer od os pepe eines bee 


Kfrse of formsier sodten 5 .terihint beweme al aeliex ofnouibet® ciak . 


wii lidenifog: to cons: ~sbiw 6 avotne tipneide oRissoy 26 elsoe eine 





























ak newa °°. etre tavab oineradakii abseneaals sive Yiievell'e a iy ie 


eis cacnerotlib ssort cpuaiiie ,leww ares oknostesy ads Ma 


ulisdat sviteleosqe 5 igtewa Town sft Ife : 


ay 


need aed chowhaowe ait ,weinoiisag of ” .viteiud of bemuse: sd yen 
noldera ort paifisps: noidsivosge s 202 Biel 


; bas .tRepnowe ert onied s cbiw \dierete Qo eles 6 pois alewy’ 


Sect Derxsedo need aso +: sion yxsamiiiesq 6 BA vaepenpnel Ses 


3 


heatton ef yiowszoid rslimfe s .céleteD is \soepore (celiete ae 


hetia= ostd end doit o ott \searqng1od to eiosiek> Yi ' howd: 
pit ives: wert ofttw ,tinesce Digsy az alist nati ¢fe) edie 5! - 


g\Bts (9 46 alse oite yew vintreeiscde vet eT 


_f* 


vapseprial off Yo wrote tebiise odd nu Gy hiea tabs 
fa leISS HT 2 ot Heptario.t  fenuit ia ad 





187 


unstressed [oe] often falls, while unstressed u invariably remains 
intact. The three vowels a, O, and e are the three strongest vowels 
on the suggested vocalic hierarchy. Further examples could perhaps 

be adduced regarding vowel nasalization, vocalization of consonants, 
and other similar processes, but the main point has already been made: 
that there exists a recurring tendency for vowels to behave in a manner 
suggesting an intrinsic hierarchy. 

Following the proposals offered by Foley, values of vocalic 
strength could be incorporated directly into the phonological speci- 
fication of each vowel; for example /a/ might be specified as [5 strong], 
and so on down to the lowest vowel on the scale, /i/, which might be 
denoted by the value [1 strong]. Processes involving vocalic strength 
hierarchies, such as syncope or unstressed vowel raising could then 
be formulated directly in terms of the feature of inherent vocalic 
strength. Such a system would explicitly portray the facts which have 
traditionally been assumed in manuals of Romance philology. It would, 
however, contain no explanatory value, being nothing more than a forma- 
lization of a number of observed facts. There would be no logical 
reason within such a system why, for instance, o does not normally 
weaken to e, which immediately precedes it on the posited strength 
hierarchy, or why u does not normally evolve to i. Any explanation 
seeking to establish answers to such questions will have to be highly 
tentative at best, since there are no guidelines to direct the investi- 
gation. At this point, however, it seems that there is potential gain 
to be realized from advancing a few observations along the lines of 
an eventual solution. 

Central to the foundations of modern phonological theory is the 


concept that phonemes are not indivisible entities, but are simultaneous 


Ca 


ed ey 
cabins vide teva o fexeonteno ehidw 2itet medio fe) Ssemexiaad 
aicaw Tesoro: sant at sts o DMB iO 4% elewoy sextkt eft .Soadni 
aqari, Slum aaignes son modo vrbieees.tet Llaoov becesopue add mo 
einerotns to nokissifsoy ofjestiazen Leweov peutiexepen Bear bbs od 
ehem toed yPeaile 2af srtog atem oit-tud ,escesoosy a6Limie serigo baw 
tarsean @ ni ovecded Gog oy yo? vousietd privwwel 6 Braces grads tec 
_viorevebi ofenkxtnar as paitesppue 
sifsoory 2 uisv..veic? yd betette alssogesy ent ‘griwoiict 
-fhede Isctpofonod adit ont yitoaw betsiogiooni sd blyoo. dipase 
giotie ?)] es boli fosye sd tipi \e\ signee: S0e TAs tinse to noltpoit 
xi SCO > EN ,olsoe at ne Lewov taewol. art ag owob mo oe bos 


on - 2 ad be ’ mtr cf pane: TI | Bowe fy st LEV ort yd betore 


neds © ontete: Iowov Ba Io sgnorve 86 mous , aeirorsrs.td - 


wy moaiedat to sndsst ect to eased ab visoeekD Secs Lumro sd 
syect fevtsiw dos i+ sting vittolleons bilgow nedeye 6 chuse -fitonesise 
(bivow 71 pofoLrag soci Cunt tf hemes need yilenotsibess 
-ryrtcd u 5 1) . onted .cerlev veotere love on tieince qgevernd 
tem inet ox sd Binow o14adr) .etost bevasece Ro ctatmins fo mobiase 
fsornxt Fort axl >. is vitw madbve 6 tava alioilw ooesst 
“nnoute fatireoq ot no i eobsoseg yietsibernr dagiw 4S of nekeow 

NOt tars Lope 4 avievs yi lamar ton aecb oye so. \yiozeegt 
we od syowas deticewtes od putdese 
Peso: aft tootrb oF Sonelooine cl Sis orevit sorte ,geed te ovitasast 
‘iep [atenetod ai ore? teri) erese Ji (revewnl died ei 2A -noiL3se 
to eenti atk pnols enotisvreedo wel s pitiodavis mom bas tisex ed. ot 
Sie at yvrostt Iesipolfonoria mrshom Jo oresisabinaiiah evit “~ leucine. . > 


ayoamadiumie ons judd ,esktisre oiciekvibak ton es newreartorig, sack. 


gi 


é . meets 


7 

Ae 

- i. x 
Wa, 





188 
bundles of distinctive features. Therefore it is not unlikely that 
additional insight into the problem of vocalic hierarchies could be 
achieved by studying the interaction of the distinctive features com 
prising the Romance vocalic systems. Although the full '!vocalic system 
of Italian (and Portuguese) consists of seven vowels, only the five 


vowel a, e, i, O, and u normally occur in the unstressed positions 


under discussion. These vowels are ordinarilly specified as: 
a fe) e u i 
high = = = te t 
low + = = = = 
back 1 2 - a = 
round, |" = a = + = 


Within this system, behavior patterns characteristic of each of the 
distinctive features may be detected. Consider, first of all, the 
vowel a, which is unique in being specified as [+low], or some equi- 
valent specification. As noted above, a exhibits the greatest phono- 
logical strength of all the vowels under consideration. In the Italian 
dialect of Pescasseroli described by Saltarelli (1968), all atonic 
vowels are reduced to[®] except for a, which remains intact. In 
mainland Catalan, unstressed e, a, and [€] are reduced to[9] , while 
o and [9] go to u in unstressed position. On the other hand, Mallor- 
quin Catalan reduces unstressed e, [ce] anda to[e] , ando tou, 
but atonic [9] is only raised to o. The feature of lowness thus 
seems to carry with it a certain measure of phonological strength which 
separates the low vowels from the remainder of the vocalic system. 

In contrast to the status enjoyed by the low vowels, the high 


vowels i and u find themselves in an exceptionally vulnerable position 












- ais a) o . ’ ie ' 
a7 ain 3 nw a 
io oe 
uty voto Son ak 3 och lia o 
oe Cae 
“mo eoncieet evisonisedh arite4o 0 yee mci) 
Wedeye, iLecov! Lint ad} rpucdt£A, prescient ro ait a 
evit odd ylno , aimee sate toda ahaa at a 
enpisivcg beesderes att nt musce yiletaoe 6 Bre xo xBt 98 ; 
“ges beltinege yitisediaao bas alan earth > heme 















ats to thee 26. Seresersib eirrajsq soovidad samiaye aie mbsiot 1 a 
eds ,[fs 30 taxit (tebrenco Seat: 
. - | okie amos vo »lwol+| 2s.bsi3 inege pied ab cupinw ab rake ye ewe 
| onorig taetsexp oii etidirks 5. ovods baton “HA wiseoiioegt ameter 
deified? att nl .motisxebienns: tebe So ay He te ee 
obeots Lid , (G08l), ALferstisa qu hedizossb dene 
MI .josdat enisme: doitiw »s 103 tgsoxs) [ejed! bees os af 
Slicw , (81 a Beouibor ats. 9} bes .6 \9 Boevewtenny \ 
“soLlem ssi xarlio otit 90 - oh mee 
ar Wy of © bas, ‘[e}ot 5 Ss fa) 2 beazoxtens aa 
_ Buds aseawol to  sucinet si? ala 
Mise opamsse feoipotonns 10 sees nto a eee 
wmadaye oLiscow alt 2 xebrismen aitt most | 
ae Me ok ave eicie, eae 


- 


ia ih : = we . ae ue fs ; iv 










eee: 





189 


in terms of diachronic development. ‘These vowels are characteris- 
tically the weakest both in terms of syncope and as regards such pro- 
cesses as atonic raising and nasalization. In such cases, the feature 
of highness appears to contribute in a definite fashion to the rela- 
tive weakness of the high vowels. On the other hand, there also 
exists a noticeable dichotomy front/back which separates vowels of 

the same degree of aperture. In general, it seems that the back vowels 
exhibit a greater degree of phonological strength; e.g., in terms of 
resistance to modification, than do the corresponding front vowels of 
the same degree of aperture. This discrepancy, while clearly emerging 
from the data, is not as striking as that existing between high vowels 
and mid vowels, and between mid vowels and low vowels. 

The above remarks concerning relative vocalic strength, while not 
formed on the basis of universally valid data, characterize the vo- 
calic evolution of Italian, and may be augmented by data from other 
Romance languages. That the tendencies established above are not purely 
fortuitous seems assured by the amount of confirmatory evidence which 
may be brought to bear. An attempt may therefore be made to fit these 
observations into the general theory of Romance phonology. Since 
clear patterns pointing to the individual and combined action of the 
distinctive features may be discovered, any theoretical proposals con- 
cerming vocalic strength hierarchies would have to account for these 
pattems. As a preliminary and highly tentative proposal in this direc~ 
tion, one might suggest that the grammar of Italian, or perhaps all the 
Romance languages, be augmented by an additional set of interpretive 
conventions that specify a phonological strength value for each of 
the relevant features in such a fashion that combining the strength 


values of each of the features in a given segment will yield a measure 




















elraioetse sue elon Seat 
wi ot ehaigend Gan Bens 6 


omntset ott ,2ebso rove aT a 


mcaeiainapestinentips «6s: 
© eais oxeddt sina sertt0 2 a) Se 
[Ae ate en at RR 





NE te wiser ah: ce «cele Sanne to carp vatsoxy 6 sti 
“20 elaew snort gnibiniggecrie ort ob merit .noiteo Mibom ot son 
Piipuem yixsolo ofimww .\onedsmeib elf? .sauixegs so sab sage) 
Blower dei costed orideixe Jado es pnblite @& Jor ek \adeb mot 
.afewov wol bos elewor Bim neawied Bas vaiowoy Bim Bas -_ a 
toni elite vitoneate olisoov svitsiex pruiberinie aitenst aveds oat ies ui | 
she site ‘estvesosrans ,steb biisv enero to ahasd at no Samo a : 
sortto mort stebh yd beadnompus od ye bes nei tion balsa cialis — 
ging gon aus evotis bainildstes esionshast odd 3H -Sapsyprsl sonking 
‘tbirh conshive Wbdascit bac to Jjnvems aft yd Bemees amese piven 
ses) $83 of shan od ovctorsd} yam dqnest6 MA soot of sitguendl eam ; 
soti® .yeeledcrly soasncr 20 vaca bexenep Se! aim anekgereaee i" 
ett to moidos bertches fas nnn ee i i 
wieo Blavociag Ieobteosd? ye bexevessth 3d yam ae 7 od | 
eneiit xo trucos of ave Siow esttnisratl pone ae oe ® 
oti elvis nk Denocay enteties vinioist Bas Levey —— _ | 
sett Los sepstaiey ® .ostissT 26 ~ me 


| = wee 
me y i 


en 










ae 





190 


of the relative phonological strength of the entire segment. Some sort 
of interpretation rule will have to exist in any case, in order to deal 
with the various hierarchies characterizing consonantal lenition. What 
is being suggested here is that, instead of assigning a single strength 
value to an entire segment, the value of the segment would be a compo- 
Site of the strength values of each of the individual features compri- 
Sing the segment. In this fashion, the resulting strength hierarchies 
for classes of segments would lose some of their apparent arbitrari- 
ness, being shown to result from a principled interaction of the dis- 
tinctive features. Such interpretive conventions would take the 


following canonical shape: 


(5) a. i [x, strong] / ___ (X;) 


where a is the coefficient of the feature, whether +, -, or an integer, 
and the x; are optional environmental descriptions which might be 
needed to characterize the behavior of certain features. For instance, 
the feature round nommally occurs with back vowels, except for back 
vowels specified as low, i.e. a, in which case unrounded vowels are 
more common. In the case of the Italian vocalic system, one could 
assign a strength value of 1 to the feature specification[ +low], a 
value of *% to the specification [thack], and a valve of -1 to the spe- 
cification [thigh]. In this fashion, the vowel /a/ would have a strength 
Ciel ey O/ els) c/a Or 08 OL se randi/1/ Of —)..) Tne actual nume— 
rical values of feature strength are completely arbitrary; it is their 
proportional relationship which is important in establishing a vocalic 
strength hierarchy. When these strength values are added up, a total 
strength value for each vowel is arrived at. 


It can be noted that the specification of [thigh] as[-1strong] 


7 _ 4. . \ in a 
- Cae 
an Ns 

























See aie: eile tsin eel 
isab ot 38610 ai saaaaliaiie 
jan aoljlosl detisasamo yr 
ict. dnbdidadshtaitead cs AokdadaainA 
4 Seals. 0d. Sonahian: it SS 
re | “is ees evi 2. Ao 36 ee a Be 
a _ Spktderisa9is itengite paliluass: Slt, noire? aide of 

aietiichs trabatrs, 1ioelt Re mpc seof bincw osnamaee 30 2 
~eib eit 0 sotto ‘acetone Oe See Si 


oe ee 
ee ge me 


i atl 5 
; 5 cou - 
} my OY J Ve 
- 


5) 


- wepeded MIS \~ + zerterw cue at to solaom at Sh a 
ad tdpin dointw enoligiozeb (stremnmrivas Lanskigo ors. Eat bos 
-\aometenl 30% .sentss? aisdrao. to solverisd od ssiadoeusc 63 Ibeoh 4 
aoad 02 250% .2Lowy Joss lw eno iiamten Eegrt exatiat eas 5 ; 7 


os a oMcr! iiteiein sas, Abigw ab \B . 28 “sk as beRioeg ear 
Blue sno’ medeye oijsooy nebiedt orld 20) met eit a amelie | 


8. 4{ woOL+ jaoktsoit teas sustest. bat chit ee 


“oye silt ot I~ to auhev & bas. .) Hed 


a 
> 


ae > 
[> an ae 
= 





> ; cA 


Mme s qari bivow \s\ lewn sds notte? ett eff 


“amin [eputos ef? .t- to \i\ bas ee 20, 10.800 Beal 
pmtedt et tk 5 creas otsiceee soe o 
Ss caadscheeiaeiaiiahius | 
a er anliienannes sake 

. ee nae 


and of [+low]as [1 strong] corresponds to the three degrees of aper- 
ture of a five vowel system. If, instead of two binary features, a 
non-binary feature of vowel height were used, it would be possible 


to specify the strength value of the height feature as: 
(6) [n high ] [4-n strong ] 


Combining a scalar feature of height with an interpretive convention 
such as (6) involves an implicit prediction that vocalic systems 
with four degrees of aperture will also follow the above pattern, i.e. 
with increasing height being equated with decreasing phonological 
strength, and with the back member of each pair being stronger than 
the front member. Data from Portuguese partially confirm this predic- 


tion. Brazilian Portuguese raises atonic /¢/ to e and atonic /d/ to o. 


191 


Between these two vowels, /0/ shows a greater ability to resist raising, 


second only to the resistance exhibited by /a/. The maximum theoretical 


prediction which could be extracted from the above proposal is that in 
a four degree system with eight vowels, the phonological hierarchy 


would be of the form: 


ih, wes a 
Re as 
ge pa 
4 
Sy (ee 


At present there is not enough available information to verify or dis- 
prove such a prediction. In the event that such generalized hierarchy 


is found to be valid for one or more of the Romance languages, the 


theory of distinctive feature strength hierarchies, as well as the pro- 


posal of a scalar height feature, will receive additional support. In 


i 
we 


_ 
Sa 


pt 


> 












reqs: 20 exenpab serait, ott 
2 wekanicn ea ad 
aiidiescg of bidow.tb, .Saa0 st 


eee 


x ~ 













. fenotve stl - ; {doit m1) ve @ | ‘ bs 
One an a a m3 
spitngnce ov ijeroxetnt a tlw tinker to guxteot vedeoe 8 eriaki 1 
amadaye 2iiscov tery. “nottotherd dint ignt ts sembeornd (2) 28, ag 
~5.i evadt64 aods of wolfok offs Lib sactiega to eesxpsb wot die a 
Isipoloreds priescsb btw adeupe taked aiipled eabesemat die ta 
msd+ 19pnoxte pated rieq dose, 20 1ediem yosd ort ditiw bas wigeneads ; a 
—jibexrg. wisis mttnon yilsitrs seseputiot moxt sted rte OE BEE A a 
oo YA cinats Bis 9 oF \g\ cinods Beeker acsuoudiod asiiises€ okt am: 
spelieisy gekecx ot ytilids iedsemp © ewods \e\ alow owt eaeds moowlsd 
feoitgoxomlt mraixan of! .\e\ yd bahidisxe sonmieiegn ame, <2 yo baoose | 
Unb tedt 2b ipsa: evods. sumox?, Dedosxtne Sd hives dbitw aoitoibena ca a 
| YWiotarcid-Isvipolonoriq ati \elewov tripks aie ons ero 








™ ' 
- 


re 
) 


” (phil terettemeciee done, Sect i ei Od 

arty nas ot a ai an 
weit oth na. eae an, ae act | 
sol strom Kensie evibsoen EL 


aoe 


order to extend such a theory to include all the Romance languages, 
however, it will be necessary to make provisions for such vocalic types 
as front rounded vowels, which appear to be stronger than front un- 
rounded vowels of the same degree of aperture, back (or centralized) 
unrounded vowels, naSal vowels, and so forth. Such an expanded theory 
would of necessity have to distinguish between the features of front- 
ing and rounding, and might additionally require consideration of more 
than two points on the front-back axis. While nasalization of vowels 
will probably be determined by means of a vocalic hierarchy similar 

to the one outlined above, the feature of nasality itself might tum 
out to have a more feasible representation as part of a distinct 

nasal segment. 

Interpretive conventions pertaining to individual distinctive fea- 
tures would depict the fact that different segments exhibit different 
values of characteristic phonological strength, and would relate these 
segmental strength values to the presence or absence of certain distinc- 
tive features. The strength values assigned to the individual features 
are not meant to replace the features themselves, but rather provide a 
parallel composite specification for each segment. Within such a 
system it is therefore possible to maintain a clear distinction between 
the interaction of features and the combination of feature strength 
values. Combinations and modifications of features, in addition to 
reflecting configurations of phonological strength values, are also 
constrained by both language-specific and universal conditions invol- 
ving co-occurrence of features, implicational relationships, and so 
forth. Thus, for example, constraints regarding the reversal of the 
values of rounding and of front-back position will account for the 


fact that, all other things being equal, o and e more commonly reduce 






















val Z 


.29Rsupes S onions iis 2 | bits 
seer oifecov rue sot sso a tn pied tos” 
nw tno’ owiltt xapnertte of a -usleaRS: Heart | soar s 
(basiertae0 20) xed nana OT 
yioadt babasqxe as roud got ce Bae \alewow Iseen \alowoy Be 
i =< 
| exon to moiserabience eniepes’ YLlanOhibbs ddpon brs venlibrios Beis Bak os 


nee 
alinan Yo notes ifsasn StLiW: .aiks Aec-su08ks set? 0 eintog oe matt i 


7 
selimie yioxsxeid oflesov''s to unser vd Serimeteb’ ed yidedeng tit ; if 
feed dripim Meath Wileeen to sguss ad? ,svods hemifjuo Sno sdf af A 
fokteib s to tag eB sshd tics Resales aldiesst saan s aver oF SO 
itemesiigess 
ast evijoaitaih [subivibyi of painter SHO LITIVNOD owitenqreadal 
Sefe023 ib tidtrixse atnempes sneretich sett tos? sfif joke Biuow esiatt = 
etait Stelios Siuow bos \diproite Leolpofoaodg nite bredteters 6 eauisy 
ontiei® ihatico to <oreatis qo sonsesiq si ot eae erpmencte Uectersmgse! : 
sewed Kenibivibat O89 ar beapines aeulay cease Sit acattbey oe 
& siivoxg radisx tud , sevloenpelt. esiictse? ott atsigenier nse Son 8 
8 rowa aintiWw .jnempee dose rot mobisotiinsge siieagmoo tottewg 
asewied agisoniteib xsefo 5 cisinien oft eldigecg sxcterset at 3k moseye " a 
Mpnes2 gutss2 to noktentdnes ork bets asutast io aoitostednh ett a 
o¢ mOESIHGs mi ,eocuntsst to snot sso. kbem Sets noi spec ienaO fend a 
oale os ,ewlev ritpnette Lsotpolonong to enoiteus nee pa, 
owns. eeo8) tino Fevaavinw. bas 2H ~9 . 





ro 


ig 


- 








a3 


to u and i, respectively, and that the shift of o to e or of et 3 

is relatively uncommon, although involving contiguous points on the 
vocalic Beale: Viewed in this fashion, the vowel hierarchy does 

not represent a model of linear development for any one vowel, but 
rather a statement of the relative phonological strength embodied in 
each segment. The existence of a hierarchy does not by definition 
imply that an evolving segment will assume every valve on the scale; 
for instance, intervocalic /d/ generally succombs to lenition before 
intervocalic /b/ among the Romance languages, yet /b/ does not ordinar- 
illy evolve to /d/ during such a process, despite the fact that /d/ may 


be considered phonologically 'weaker' than /b/. 


5.12 Conclusions 


A theory of diachronic phonology which makes essential use of pho- 
nological hierarchies is clearly more desirable than one in which rules 
are stated in terms of seemingly arbitrary conditioning environments and 
developments. Based on the data from the history of Italian, an example 
has been offered of the characterization of a diachronic development in 
terms of strength scales. Further investigation is called for before 
such proposals can be admitted into phonological theory. Clearly the 
burden of proof falls on anyone wishing to introduce new proposals into 
an established theory. Despite the remarks contained in this chapter, 
| the burden still remains. In view of the complexity of phonological 
change, it appears certain that the current view of diachronic phonology 
will have to be expanded to include at least some new proposals; phono- 


logical hierarchies seem to rank in the list of possible additions. 


-Nenibio gon esob \d\ dey yeepsupant eonsmnd eck pagms \\, obLisoovisdnt 
] 


+ 

























op 2. By anes 


iL ohh Io 1 giodo to Hits: 
oti cing Seiiisince nati ae 
paob yihrwexsis Lewov oct hoi he | 
duh, Lewor eno ie sO Insmephowdh aon 20 deta noesiget 0 
ak Bechedss anata isnigcloondg audonios orks ts tt 
aoitinieb ya Yeneend viegeeoiel p To sontvebee oft .ausmgon fone 
nafeoe elt ac sulsv vreve smiées Dliw tosmpse wreviove gs sect youd ‘a 
sxotat noitine! at admeove yLisxecep \b\ oilcveumednk ,sonsteat Gat 


a 


a 
> 


Yen \b\ tec gost Si at iqeeS yeaoor & eee prin, \by.od. wiove yitt ‘% 
.\d)\ ster ‘xeolsew" vifsoipetonariq hesabienco sd 
ow “Gane bet 
ae to get; [sicinsass eoxem dolrw ypolonola oinortane %o yroart A te 
eels doiriw ni an nectt aldertess orar ylasalo ef ae ivntetretd indeaee fe 
Bios adreruron.tvne mitts 10/4 Ebro peerings vipnimess to amet at pajete ou 
Sfamexes ms ,ostisti to yrodsid on moxt sieb arit noe pee 
yaa “dhemqoLl: aveab obsoxibs.£h 5 46 notes medoetek aa to poate sed ast ; 
rs, 
: 





aupted YO belian at noifeplseevni tedhort calsce dttinie To amet cot 
aft yiaseiD .yrosrt Lanlpolorgdgd odat pasties edi ARS alsedqoaq Hue : | 
Stat efeenoocng wai soubotint ot Beer enoyns (0 etis seorq 20 nabud 
visdger> eitit at betisineo atsre ait atigest! ilies am 
fsolpofonerdy te yiixaiqnoo ait Zo weiy ni - aenheaert eer ibe 
ypolonniiy siseutpath Yo weir soci eds Sart, abacus is¢ 4 7 
worth) iaiseouing war ame Jasal 46 etrlont of Baa ie 

-enoks thie aa FabE alt a base of dae o6t 





1 


10 


ae 


194 


Notes to Chapter Five 





See alsomigzo: (1972: 164)* 

CE. BBech(1970 nel5i—2):. 

Taken from the facsimile edition Alberti (1964). 
Cited from Bembo (1955: 148). 


Cf. the facsimile editions W. Thomas (1968), Granthan (1968), and 


Florio (1969), respectively. 
See the discussion in Migliorini (1963: 471). 


The version edited by Bellucci (1967: 104) lists other variant forms 
occurring in the available manuscripts, including marchera and 


mancherra. 
Cf. Mulja¢ié (1966, 1969: 396), Saltarelli (1970a). 
Ibid. 


Actually the central point reached in the articulation of r may be 
somewhat more centralized than the front vowels. For example, in 
Rumanian, the vowels e and i generally exhibit the centralized allo- 
phones [°] and [+], respectively, before and after r, variants which 
correspond to the vowel phonemes orthographically represented as 4 


and ap respectively (cf. Vasiliu 1968: 130, Ruhlen 1973: 39, 41). 


See Vennemann (1972: passim.); for Spanish see Navarro TomAs (1966). 
Newton (1972a: 569) notes the lowering of i to e before rina 


modern Greek dialect. 



















— Pe 


(hort) i gl cs ga oe aI 
ORY s22eD) sane 


- eps Me 
» ies Peak 
(Leto 7 Ww itt 7 i; 
bas , (8805) casdiaeso - (Beet) aanoet anott ibs alimbsoel af ; 


7 irsqess »(2et) OFF 
ml ; va i ol 


(ITB 26985) inixoblei ah nokpmmeth att 998 8 oe 


a 


airyor? Brey sotto afekl (MOL <Tael) foto lies yd Bed ibe éctshed ane 1 a 
7l<& 
‘hos Sxecdoxent enifwilont veiqiipevnen oidslievs ett nt ‘eotrueon 


» 


({g0V@L) iffexstise , (BCE seal ,dReH) Abistiam 20. 8 
ma sa sat 


| ‘ - at oe 
od vam x 30 doitalnoire ott «i bethesx $rtkoq Iex3nse a 
qi... sigmexs 10% .afewov aeoxt edt coset bos iiioeo eta ahs 
<ifs besiitatns srt +See aia F Pree > alowov <9 emmat a 
dott eindiney ,x testis bas enced lev iguoqess ae : 
Bae Seononenqe yLiuvinkgervert eamanetig Loew, sf dies pqeertoo. ay 
a ef 120 OL mein OCF 888 wtLteey .B9) y avi “s 


hn sat ine se wine ama atien 
: | | sit sn weil g.00 to pa 8 


© i 
- eee ara : tant 
' J - . in 


oe : os ro inf 
OP eee ae oe are 


19) 


12 Cf. the discussion in Meyer-Libke (1890b: 78). 


13 C£. Grandgent (1927: 55, 60). 


14 


1BS: 


16 


17 


18 


19 


20 


ak 


It should be noted at this point that there is no evidence indicating 
that the shift of ar to er took place any earlier in the verbal forms 
than in other words; indeed, traces of this change involving nouns 
are recorded from the earliest periods of Italian, and if Meyer- 
Libke's observations are nor erroneous, probably antedate the conm- 
plete fusion of the Vulgar Latin analytic future and conditional 
forms into the synthetic forms in which the stem variation ar-er 


is exhibited. 


Cf. Vennemann (1972: 872) and Kurath (1964: 78) for a fuller dis- 
cussion of the lowering influence of English r. A similar situation 
seems to have occurred in French Canadian, where the influence of a 
following [R] sometimes caused e to be lowered to a; e.g. merdam > 
merde >[ maRd } standard French merci >[maRsi]. For further dis- 
cussion, see Orkin (1971: 66) and the Glossaire du Parler Frangais 


au Canada (1968). 

See d'Ovidio and Meyer-Liibke (1904: 671) and Pei (1941: 39). 
mies bec n( 19/0 seloZ2)r 

Cf. also Lausberg (1956: 8293, 845). 

See Menéndez Pidal (1966: 67) and Grandgent (1934: 98). 

GE. Granagent. (19275358): 


See Griera (1913), Kuen (1932: 122), Badia Margarit (1951: 19), 


Moll (1952), Lle6 (1971), Vogt (1971), Phelps (1972). 


Ree 


VE. Joe 
S 


~ 4 
Avi 7 
on a if ‘ "s 
; a or Sai e ee } : io wee 





















baidsoibat somabive on ei sssosi3 seity acai bel 
wih ns ie sts a to Me 
ention paiviewt suiado etrt 30 ssasTd \Besbat petro waxtto ab 4 
<1yel ti bas .nsiletI to abolvag sedlinso tt mai Bebanbe ok. 


“mio sit eisbedns yldadaig ,eosnorSs Ton oth anoLievaseda aed P 


fsroigibnen bns sweet oltyisns ottel septtht ec 20 note? etal oP a) 
= 4 ot 7 
Ydeus noissixsy ada ot poliw ni emot ohedinye artt edad imi te a 


aBib tallied « rot (BY sbaeL) désxu Bas (SVS). Sven) itil al 3 a 
doltsetiz 1slimie A x detlpnt to sonswlinr enivowol exit 20 apkeawo | 
| 6 36 sonaudanr at orornw .cetbatsD donor of bere saved ot anees. | = 
<¢ Neiygn .p.5 15 ot Botewol od ot & DSBuED aemitemoe (1) eakwaliiels : 
“2i6 weritivi rot .{isisn) < iosen nonery Busbasve { fate ‘e abrsn Pe 


aisorartt rslised Wo sutse2cid oft Dns (98 :IV@L) nib see .nolsam = 


i lees 


(RE sLbOL) int San (1Ta :80eL) odes —nsyoit rts ObivO*B 808 aL oe 
banal ee Oo 


- Ly oa 
| - (Set ren oe ie 


4 mie 


- 


.(28B: \E4S8 +820) pvadet oats 39 me 
- i ; hyd 
i. % (82 :t€@F) Jnepbasw' hrs (Ve s2aef) Le H 


82 


; es aon. 


(er 1200) threpusl ekbed as 


196 


22 See Grandgent (1927: 44). 


23 The term 'rule' here is not to be interpreted in the strict genera- 


24 


25 


tive sense as advocated, for example, by King (1969a) and Halle 
(1962), who view sound change as the addition of integral 'rules' 

to the formal grammars of the speakers of the language. A change 
Such as the one under discussion is highly complex and evidently 
results from the interlocking action of a number of diverse pheno- 
mena. It is quite unlikely that any such 'rule' could be incorpo- 
rated as a whole into the grammar of Italian speakers; rather, the 
change portrayed in (1) is the end result of a process which even- 
tually required several centuries for consummation. As a consequence, 
the statement presented by (1) @nd the following refinements) is 
merely a schematic representation of a 'metachronic equation', giving 
the before and after states and saying little or nothing about the 
intervening events. In particular, there is no claim to the effect 
that (1) represents a single change which could be portrayed as the 


formal addition of a single 'rule' to a previously existing grammar. 


The development of smaragdu to smeraldo may be accounted for by the 
fact that, particularly in the early stages of the language, the 
initial s of such clusters had syllabic value. See the discussion 


in Bourciez (1967: 48, 156) and Andersen (1972: 34). 


Saltarelli (1970b: 94) calls for the secondary accent to fall on 
pretonic vowels and the tertiary accent to fall on unaccented pre- 
tonic syllables. These conclusions do not find support in the his- 
torical developments affecting the syllabic structure of Italian 
words. In addition, it must be recalled that positional strength is 


not to be strictly equated with accentuation, though the two are related. 




























“pransp tobcie et nt. baveusratal sd of 3 
eifsh bos (s0deL) pabt yo. amt 3h woot shetsoovbs a 
tages fexpstink® to nO LtERbe ort 8 span Setes = 





meet cy | 

a i's 
7: spaaio A soverrat ar to ethane 8 $0 eicmmeenp Ie sot at 
‘ eg oro: 
ive viinebive brs aie aoa 2k otamiaait xeben mgr 


-onsfic ceiwib td radu 5 Ro no Lfs gebsioot sand orld oi © 
> 
metssons i. bluco ‘ela! hue" vids sect yhoo fea edie at 5 ae 


abe 3 
edt emis: ,e1sieoqe ms init 45 halle ant etal date es beder 
tt sae 


isys Hbitiy aescoty 6, tb divest bie ay ei (L) me beyextied ¢ a A _ 
\songipeenco 6 2A ..noltemuencs 10% shies lexevee eetnees leat 
ai @iceseaitexr priwollel att bom (1) yd batnoneng dasmadad]: ott ad 
privip:, ‘oiteupe ntrouhstan' 6 to MOS BINT ST pisvemerbe 5 yforsm 
 ystit tuods paistion to sfttif priyse bus actade sae hee STE | i. 
Sostie sit-ci misin on ef staid .»:iotigea me Lesnewe poinvastak : 
os ? a . - 
a} as boysrtzog ed biveo doidw spas alpiie s aimemengen, (2), ped hk 


| “aemexp priteixs Yfevoiverq s of ‘els ofonte 6 to robsibbs Leino an a 


a Loma 

ais yd sot Detnuooss sd yen bisa vbposuime ete IO snsmploveb ott a - 

: aks qpsvensl ar to seuste yinse of AE inaivoiraed. se goat : 
pOipeimerh ert sez .colsy nicdelive bed eraauto cae to staat id 

. | (RE a8T PL) maerebriA has’ (aet oh isteth eleanee ai 7 : 





to [lst o¢ Sasoos yrsbacooe at tot Bifano (se 


sar 
S ae oon 


26 


Zi 


28 


29 


ii 


Cf. Nietzsche (The Will to Power, p. 294), talking of 'our bad 

habit of taking a mnemonic, an abbreviative formula to be an entity, 
finally as a cause', and Wittgenstein (Philosophical Investigations, 
pp. 37-8): 'The fluctuation of scientific definitions: what to-day 
counts as an observed concomitant of a phenomenon will to-morrow be 


used to define it'. 
Cf. Menéndez Pidal (1966: 74). 
This is discussed at greater length in Pope (1934: 112). 


Except by passing through the intermediate stage of a front-rounded 
vowel such as [y] or [¢], as in the evolution of French. This pro- 
cess of fronting, however, does not appear to be a form of phonolo- 
gical weakening; in French, also stressed vowels were fronted. More 
knowledge of the interaction of the features of fronting and rounding 
is required before the problem can be discussed in detail. The un- 
rounding of front rounded vowels does appear to constitute a form 


of weakening. 





Spies ha 
ee 


Poneman 












ysb-od Jaw -aiat sade simaioe eaacone is — 
pe veneionos Like rensmecarty 5 20 shim 5 hotone nea 


oa pat ary; aoe 
(ON sa0@L) Pepe a 


j 





; —. eo 


SEL :d€@f) sqot al pret setoote 3s tepmecelb ab abt so 


A i : ' ' ; , Aa 
feitwiox~tnocl s io spsie ataifemraiat of fquonit peiess yl tqsoxa « 
“Gq air § .cboerT to noitulova art at 2s .[R) 2 [y) eB fode low 7 : 
~olonorig te miot 5 od ad teaqgs Jon eso xovewod “werent 20 wees) 


5 


omeéM .Badroy? s19w alswov boesotte oafe toast met ‘prineoleow Lek 7 a 






petirwer San pritnoxt jo senidsst srt Io noltsrahtt ai fo epbafwort 
“ay eff  .fistei at beeavosif od aso meitiortg arts grotsd Berivper Bb a 
mt s séutictanto of 1es9qs = 0b eLsww bebruon Sree to palbawor a 
al ae 
——— : 7 
Dy o 
= ree ; 
’, wh 
i i : Ls - 4 
' e ; Att 
, ie | es i ] ; : : a ; 7 ‘“ a. 
oe eee ean ies 
ieee te a 
ral : »< a dito + rT 
roe PV a ee ea oF se ae ; ul 


CHAPTER SIX 
IN CONCLUSION 


‘apyll Summary of results 


The separate studies which have been realized as part of the 
present investigation have had as their common goal the establishment 
of a rigorous formulation of two diachronic phonological hierarchies 
which played a role in the formation of the Romance languages. The 
first hierarchy concerns the relative position of atonic vowels with- 
in the word; the second, the intrinsic diachronic strength of the 
various vowels comprising the Romance vocalic systems. Based on sta- 
tistical data drawn from Italian, Spanish and Portuguese, further 
bolstered by observations from Catalan, and considered in terms of 
resistance to syncopation and modification, the non-final atonic 
positions may be ranked along the following relative scale of strength, 
beginning with the strongest: initial syllable, internal pretonic syl- 
lable, posttonic penult syllable, and intertonic syllable. Based on 
intrinsic diachronic strength, the five vowels under consideration 
may be ranked as follows, in order of decreasing strength: /a/, /o/, 
(2/0 [Wan fife 

The final atonic vowels underwent a different fate in the Romance 
languages under study, perhaps in view of their function as morpholo- 
gical markers. While in most cases remaining intact in the languages 
surveyed, except for Catalan, the overall rate of loss of final atonic 
vowels appears to be directly proportional to the rate of loss of 
atonic vowels in non-final positions in each of the four languages. 
From these comparisons comes a relative weighting of unstressed vowel 


deletion and preservation of morphological material, two competing 


198 





oF to disq és bextlsox BSS suet tte seifuta edexsqse & 


- ' : a = 
Saf .esusupan! soneactt aitt Yo mestteeseh add ot afot s beysiq sbidw a ; 
stitiw elqwov 2:nots to noltiezog eyitetes as amrsons yhrisreid texkt> 
edt to ritpawite cinowbsib olaatiint otis shacose ads tee ort ai ; 
oe iL 


“ete oo bosses faye: oi Sexo Sofemon ocks prielargco eloww sate 


a ae 

,Seeuguared bas deinsge isifsiT mo ompb ate Ieotteit 
neleteo moa’ anoitsvaseds yl Bexstalod ~ 
itelox pniwolfot sd} paols Bestiex ed a opal 
olde.Llye fsitint staaprosta edit dotw ara 
sldeliye olnossedni Bas ,sleskhe 3 fust9g sinottaoq sitet 
afawov ovt® ort .pnerta’'o keoerkie ib pienixtnt 
.wolica as bednss od yen 
“VEX WA a 


Soe oft ri otst prised tip 5 tnewrabras aleww obaots dacs ar 


“exbetiepion es nobitonu? xiett 360 weiv Ak eqstheg yada tebe 2 
aopauensl art nd toettd pabaisoer aseso tec nk aliaw rho I 
Qinmds lsoit 


“ets “ht 
io eis fli betebiectcn Bas 

oimats Lanett -non scit soitse EY tbo bas nol seqognye ~ 
higesrte 2 


“De simeesiq Lartogni 


Q sisoe 


mo beess 
a a oe tex ablenc> seb: 
s WO\ w\S\ 


(htnrrete piiaésrsab to aabro nt 


at 






2e weol to oder Llsxevo! oid «as Lecied 102 tate: 


a ‘patel ete 


ws: 
10 Agex ont ost [sno isrogonq wise xd ” mails sola 


oa OREN. 
; ‘or benef Yo cate elton sai 


Io e@aol 


si & 
* tamiletidcs2e ot: (SSS nome. thert 26 Bal ayad aie ee 
: eabivens ici feoirolonadg sinctdon kb aad to pobre iso} eucopix 6 30. 


a ray 


-_ 

, 4 
mo 

a 


a an 


eae 
a 


os — ‘ 
a 








n. 


: 
Ae 
> 
























a! 


Pe r 


n) * 


, y 


_ s 
J 


4 
70 
oy ae 
aS. 
- 


at 
7 i 


a 
a 


7 
7 
UJ 


a 


factors attecting final vowels. 

Also taken into consideration were factors of phonotactic com- 
patibility, which act to ensure that consonant clusters brought into 
being by the loss of a vowel will be compatible with the phonotactic 
structure of the language. Atonic vowel loss occurs, in the over- 
whelming majority of cases, only in those instances where loss would 
produce a structurally compatible cluster. By further breaking down 
the possible clusters into those which would result in open or closed 
syllables,.the relative efficacy of the process of vowel deletion with 
respect to open syllabicity was assessed. 

In an attempt to demonstrate the feasibility of incorporating a 
phonological hierarchy into the study of phonological change, the 
hierarchies previously isolated were applied to the problem of vocalic 
modifications in certain future and conditional forms of Italian first 
conjugation verbs. Incorporation of these hierarchies into the dis- 
cussion yielded a clearer and more complete characterization of the 
change, illuminating the interaction between the influence of conti- 


guous segments and relative position within the word. 


6.2.1. Introduction. In the introductory sections of Chapter 

One, it was noted that, as presently conceived, phonological 
hierarchies could be incorporated into phonological theory in 

one of two ways. In the first possibility, the hierarchies them- 
selves were used to define phonological segments, replacing such 
theoretical constructs as distinctive features, morpheme boundaries, 


etc. Viewed in this fashion, hierarchies would in themselves 


19 



















ata dripuow! exsieulo tnproencs 5 ll od do8 
Slsosionorg sti Adiw sfetiysqnos ‘sd diay inoy 8 ‘She 
~yevo etd ai Rudd seal Lowy Sinada ree 
Bier nook Sic oisant sect mk Yam 129280 bo ae 
owob piidisesd rorliwi yi .teteifo oldidegms ieee Weel i 
_ Bedale to nsqo ni jLuesx bivow-tbdby seedt pint eredeplo éidiegsg ee 
iw mtteisd [swov to Beso aid to yori svitelon ott soldat | 
‘beeasens agw yioidsilve nego ot Soeqeer 
& ape eermoproon to wWilidiasst sit siswtanaeb oF septate (as a 
eit .gemeds. fs5itoolonoriq: 40 vbitte Si) ocak yisisxeid leoipotenarg’ 
sitsooy fo midoxs rtd oo Goilqas saw badelos: yladokyanl seirosexedt : 
Jeti? meifesI to amc? ismitibtico fis cumwt nisivep te enotysott bon 
Sib ont otnt sotrinsxsid seoit to noliexoquoonl Ladzey soiisgetacs 


4 





-) ait 2o nolisoisetosirsi> stalqmo svom fins teisets. 6 Rablely notes’! ra 
: é . a , oe 

sitties to sonsnfin! ort noowtsd noitosisdni oft paieaeeili jepieis ~~ 

. er L i 
.biow oct didviw sobtreog oviss.a brs ednenpsd"aderp= | - 


MB FG ee 


yaoet® Isvipolonorg ni eotdousreti 30 spéiig oct 5.3 5 


stetqet to snotiose yrotouhersnl ott al a ‘ha 


Lsoipetonadg fav isonao (isimeerss Bs <a r 

nt yroardy fsorpolonad, cint hadexocncont ot Bluey a 

ae nls “eHidieeog sexi?! sr onl ayy OF 
fewe” ptidelyst reagen Incinotanony aciteb ) tai 

7 _seeibaebrived emaritem: , esuidtset svitoniteb as sions oes i 
a je i a a ii 


eeeyy ie ri 's ets i He 


a fe 


aT 7 a ‘ ete ai : = 
a al a : Py, Lt 7 
t i 2 @ a ¢ bear) ¥ : ‘ 
ee ia oe Eur ot we it. pia | a Tay oe iy 





200 


define phonological change, since all change would be regarded 

as a shift of the numerical values of the hierarchy. The second 
possibility regarded the hierarchy as a metatheoretical index, 

an interpretive convention acting upon, but not replacing, established 
phonological data. The numerical values of the hierarchies would 
interpret series of phonological developments, acting camplementary 

to the actual phonological definitions. 

In Chapter One, it was stated that, for the reasons presented 
at that time, this study would employ the methodology whereby 
phonological hierarchies were regarded as metatheoretical inter- 
pretive devices rather than as phonological. features, a practice 
which was illustrated in Chapter Five. At the completion of this 
investigation, it is hoped that the reasons for this choice are 


more apparent. 


6.2.2. Hierarchies as theory. Fundamental to any empirical 
investigation is the testability of any hypothesis or claim that 
might arise from it. In the case of diachronic linguistics, the 
proposed concept of a phonological hierarchy is a theoretical 
hypothesis, and as such must be formulated in a fashion conducive 

to further testing, in order to remain within a strictly scientific 
methodology. While it is not essential that every observation 
realized in a historical linguistic study be immediately translated 
into a testable hypothesis, such a practice must be followed if a 
new proposal is to be added to the theory. Thus, the goals of this 
investigation dictate that the proposal of diachronic hierarchies be 
put to the test. In order to test the validity of a proposed hierar- 


chy, this hierarchy must itself be established on the basis of one 






















ceaaseaiel ad Binow epmssio: ots x : 
a 6a .victexonl oft Bo. donner: samuel 


debat Isottoredisten 5 ms te 
Berlatidetes .patoslqex tor Fhe . nOoRY prtidos nokimwticn © eh : 


7 eh; 


Biuow esidotsiaid 5 at 40 acu Dev oe am? .edeb eer 


ee = 


yisscereiomeo priitze . 2trsigqoievei Led ipod onorig to aaiyee : i it 
snot inte Isoipelonedg Tsdes we 


wae 


> 


bednezetg enoebeT eas x0? vtert Satete ea Ji , sO sesh oan a 
yiisrrariw yeolohodem ot votes bivew Viote ait omit jet te 
~yeini [soijgroatiesoy a5, habaspar oyow aoiiiyeisid fsotpofonodg 5 
isiq 5 ,aeustsel fs9 molonedd es. pend yeriey esoiveb eviteaq | 
uae: to soctel ax exit 44 <SVEF sadaseD i badsxtau Ltt esw toirw ; a. 
‘%5 solaic eth rol anossex eit fartt  Beagnd ak 9% ottepiteovnt _ 7 


aT a 
z fsolxigns yns of [etnomsbovt .\aoed) 25 eect S860 | Sa 
: Se = 1. Ae s) 
‘ E m * * * 2 hp 7 
tert citeak 10 ebeocttoqyd yrs to yoilidetesd sd? af soitepiteovnr an 
ad ,eoitetupril olisovinsib jo sas ont al >,35 ma Seige Jopme 
Isoisetoswl s af yioxsxsidt [sorpolenoig 6 to sqgnroS —- - 
evioubooo notiest 5 at Sajalomro? od Jeum dove 26 Gas yeteationt 

ofiisnsioe vivoitie 5s cirdiw mianmex of xebto ni ena seria 
YT a ia 


a 


noitevresio yuave Sui + isijaeees rer at tL oii “ipelcbertsant.. 


& 24 Hewoklo? ed temp soliosig s dove Ra 
Bir 30 alsop ont ana coat ai oP babe ach ot a i 
od soldeerott! ginosios tb 20 Laboaens sets ia wosoib Ne 


“ave 
“orld foncusg 6 30 rIbiaw a GP sabe a Ee | 


ae at. 


eto to. eiasd ate no bere Stinses a Moar? am 


: id oe ae 


‘ 7 Ae 
) ~ ‘2 - 
ate are 








eur 


set of data, and then tested by comparison with a second, distinct, 

set of data. In the event, however, that one assumes the hierarchy 
itself as defining the phonological elements, such verification is 
impossible, since one is assuming, purely axiomatically, that the hier- 
archy has already been empirically established, a counterfactural 
assumption. 

On the phonological level, statements regarding hierarchies take 
their place among more traditionally recognized entities such as dis- 
tinctive features. In order for numerically-organized phonological 
hierarchies to supplant phonological distinctive features, it would 
have to be conclusively demonstrated that the hierarchies could not 
only account for the same phenomena presently used as evidence in 
favor of the use of distinctive features, but also account for addition- 
al phenomena, or account for the same phenomena in a more enlightening 
fashion. To date, no such verification has been undertaken. Certain 
examples have been offered in which numerical hierarchies may con- 
veniently account for changes in a simpler fashion than in a descrip- 
tion making exclusive use of distinctive features. To extrapolate 
from such cases, however, to the general claim that hierarchies can, 
and indeed should, replace distinctive feature matrices in the phonolo- 
gical representation of individual segments is unjustified. The only 
method currently available by means of which numerical hierarchies 
can be made compatible with more traditional distinctive features is 
to define the former in terms of the latter; this practice, while 


yielding the desired results, is circular. 


6.2.3. Hierarchies as metatheory. The present investigation has 


sought to demonstrate the ways in which a theory of phonological hier- 


wipit edt todd \yileoittanoing ylomg wonimmecs ef ea 


aieiatene Gt Jruooos ceils tud \enudset svitonivel® to: se ait 30 yoved 
pehiettipilns stom s ot snomororq ame oft aot tavecas IO) ensmonedg 5 F 


oo a - 
~ofotorig aft ni acoiszsm sxutset sides ienacnan cOsiEiey. btu Boab 


























te ao ra gt ey eat 4 


Bet al ° ah "i Ap 
a ‘ieee 
yiusreid ot aemuaas ve tn 


si epitsoitives dave toenail isstpeFarody ; 


, 





lL 


wont ay - 7 


be we 
eave 5 su a 


uf —_ 





oles esiroisrsin prifiepes otremsdtade beds iso ipcienieab ential oe 
weiiows Houe esititie Besinpenes Yilenott ibs cro pone sos Hers 
frotphionaig bes tnspro-yi leo trem tot 19s6x0 vt agai eeu Bn 
Bitee +c .porctaot oyitoniserb Lseipofonorde Sis iquwe ot esitoagraidt | 
tod blues esiroxsisid ond tet hatetenonsb viev Feafonpe: od of eved : ) | 
a 0 : 
war 


% 
7 
yeu 
7 


ne ewishive 2, bea vbinsesia sremonong amse et) 2% tayooss yin6 © ~ 


: 4s 


(- 


ae 


Pirtpee wtesiesisbas. need ash noltsoiiney rove on eae oP 5, a, 
“(100 yam asidbarss: 1 isoivemin dotdw nk betstio aged ive ealqme wedi; 
-qiineeb 5 nf matt notes? ti 5 tt sous te dmmenos visoetney i 
Sis loas1ixe or .eoutiset oy idonitelb 10 ann " gvzeaooe ou aie : 
“Heo eebtoxexsis tert misio Lstedsp aft of eveuart 1 28EBD broly 

yino sf .feitisevtad ai minaipse Laer ter 
astioneteit Sentra doicw to ersgetn vd oitabiave 

gi wenadeed oriianifets epold ibecet Jamin sie SE 
of iste soisoe ebad citiel ody to amet ai as 


Bei. aces te 


abil pmitaplsevel saeesey ee 
“aati Enotpofonody 30 ost 84 


archization can be made compatible with a theory making essential use 
of distinctive features. By regarding the hierarchical structure of 
phonological elements as representing a metatheoretical level separate 
from the level of phonology, the concept of hierarchy can play an in- 
terpretive role during investigations, while allowing the distinctive 
features to do the work of defining the phonological system under con- 
sideration. By maintaining such a distinction between theoretical 
levels of discussion, the concept of phonological hierarchies may be 
fitted in among other possible factors which can globally affect a 
language undergoing change. For example, the behavior of vowels rela- 
tive to conditions of phonotactic compatibility can be neatly visual- 
ized when the vowels are defined in terms of their position within 

the word, as well as by their inherent phonological characteristics. 
On the other hand, certain changes operate over groups of segments in 
such a manner as to suggest the interaction of phonological distinc- 
tive features; there is no non-arbitrary method of capturing such 
behavior patterns in a system of phonology in which distinctive fea- 
tures are completely replaced by hierarchical values. 

The behavior of final atonic vowels in the Romance languages 
which have been studied can be viewed on two separate planes: the 
plane of phonology, indicating the behavior of the individual vowels, 
and the plane of phonological hierarchization, in which it can be de- 
monstrated that morphological considerations entered into the picture 
to effect a diachronic strength value different from that which would 


be predicted on purely phonological grounds. 


6.2.4. Psychological implications. More serious, however, than the 


descriptive consequences of adopting a system of phonology composed 


202 


i 



























x" 
ae aa Lsitnseas erubn Liisa 
x te omutorte fenieerusibid 


stemsaee Lovel Isoisexoniieoent 5 primase | 51 
wah op ysiq nap yhuexoithjo sergongo, Sekt <wpodienorig, 
witonizeibd ext priNetis oh iedw snob epiaeva palaub elox av : 

“10d ash mesa tes toot oft pniniteb to sew et ob os erat 

Isnidsiostt moniod nolnntieies nooe cricriniaine a gnome a 
od Yen eaicdoiws isis Lsoivolonoda io SE a aoteemeib to efeval — 
6 dogs yilsdolp .ns> toc enadoey: sldiarog orl30 prams ak bedsit 
—6fex vlewoy 20 10iveded eff ,slemexs 107 -sonssio paloyrebay epeupasl — 
~fsuetv- ¥itsen.od nso a al ciceeiinaicles to enoltibar ot vit 
recede moisieoq 1rodt to emrsd sof betitab ens alow ont corlw best A 
smal: aiosisi issipoicnoig soetandeb tieds yt as Lisw a, bsow act 
fate etnamese to equorp 149vo statego eopasrio aiediyeo /Ghel xsdto of 0 7 ; 
smuitetb isobpolfonoig io moltosiadal of tegppue of 68 ~oeniem 6. owe ; 
thie patusigss to bodtan visi icre-ron on ef oper vesustest evi : - : 
=591 evitoitsasb doisw oi yeofonaig to mateve sk ancsdzeq woivedded ‘ 
.aediev Espitedsrent yd boos last vistelemmo 936 sem 
eoceumisl ‘sonsmoH ot ni elewoy tacts Loti 10 solwetled of as : 
eft :eenela stsxsqse ow no. beweiv ad neo boihade need eved diy 
alown Lothivibn! sid Io yotvetisd «tt patteolhak tor 26 als 
“ob sd nao Ji birlw ak aoitasirouyets {sotpotsogta | A : 
smtoig ert otal honedas enoitmeebtenco f 
 -abetvorb tomtom en 


5 imide enrhaah 




















EUS 


solely of hierarchical elements, are the psychological implications 
inherent in such a choice. Most phonologists, with a few noteworthy 
exceptions, 2 have maintained that distinctive features must represent 
some psychological behavior pattem observable in native speakers of 

a particular language. The features that have been proposed are often 
based on articulatory correlates, and sometimes on acoustical correlates 
as well; their purpose is the elucidation of the bridge between the 
level of individual sounds and the level of significant sounds. Such 
features have generally been justified by means of groupings of segments 
observed during sound change, and by the intuitions of trained lin- 
guists or others versed in the rigorous description of language. Only 
more recently have experimental data been used to test the proposed 
distinctive features, occasionally with rather surprising results.* 
However, even given the current disparity of results in the realm of 
distinctive feature theory, the very search for a set of universal cor- 
relates betrays the need to see such features as possessing psycholo- 
gical, as well as physical, reality. 

In order to implement a system of phonology in which hierarchical 
values based on historical observations or indirect observations of 
synchronic alternations replace distinctive features, one must demon- 
strate that native speakers respond to, and produce, segments in accor- 
dance with such considerations. At present, it seems highly unlikely 
that any such demonstration could be made, given the extreme difficulty 
experienced in trying to assess native speakers' intuitions about even 
the simplest phonological matters. ‘To expect a speaker to phonologi- 
cally categorize a segment in terms of an index representing the cumu- 
lative result of historical change is to presuppose, for the human 


race, a sort of universal consciousness, a proposal which while perhaps 


- one 
» 


; © | fae 
fo misses ovissn at siciqrstds sig olivated spot eae aor 
noite wns hoeoaoue apolar Se) Game sel sama matostnen & " 


asdsisrios [e5ideyoos o> canijenca, bie vaeielorico \rotelesbyae mo Beem e a 

srt nacejod spbhind sie Yo nokishioris say ai ceoieg thed# (ifaw as a ig 7 

| re . ahyioe tnsoliiagha to level aci¥ Bre abrioe imbivibnl te tava 

einempes 4 apniquexp Yo znsan yd bél@itet need yilewtiep avad eomeet a ; 

Stil Bshiéey to Anmtstodab eet ya “iess cone bavee pric Bewreedo i? 

YAO. -spewpas! io nolagimedh guotephs ot Az bozaw exsde 10 bali a 

pions acit shat Go Beas ster Bdeki iechancisqe: omer yLneoee Som 

“atitesr orteinqmwe tetei iw Wilenokeeses outa! avideaivelh — 

26 misex at ac etineey To ws sr nad tosrwo oh nevip eve {2O7EWH 
7ieo Isevevirw fo 498 5 tot rousss yr a ope sunset evitonitai 

; “—olodbyeg poieeseeny as ectiutsd? fone ese a beak ert: eysrted pices a 

notion feokel a Cow 2m iinbip 87: 

Ieotriows 19 ici to iriw mek yoalonaey to meveys 6 secre Op: abe’ aT! 0) 2 

10 Sitbhisevisedo tosxifat x0 enpitevssedo ieclzodeht no beesd sauksy a iu 

“teh Jaim-ero \aduitnet witeaiteih aupteet cootsncced is Skene as s | 

~0008 ai streams: .whow bis \od ee ai 

Vistiiow yitiptd anose Ji \insasig 24 ceneitenebe iw Som a 

WHOA Rb emesixe ed? sovin sham Sef Bllob “an wii 

eve ty0ds enoititnt ‘sxsisoye evided saan 

~kpelcnotiy ot ‘enlgege 5. toeqes or ial 

. ae erence ec: soca abe ou 

b Calamus | onto feoioreid to. 


rrr ‘siptigparaplaaedh named. one nme Reo: 
LP ae 7a. 


ss a r 
oi! ¥ 4 a 
bhi. {or _ oh cael re : i mere 
























— 


, ” 
ay 
Zz 















20h 


true, is as empirically unverifiable as the controversy conceming 
the 'innateness' of 'linguistic universals'. Replacing distinctive 
features by numerical hierarchies in effect seals off the phonological 
system from empirical verification, and therefore is unacceptable as 
a methodological alternative to a system in which hierarchical values 


are viewed as compatible with a set of distinctive features. 


6.3.1. Segmenting the data. Once a phonological hierarchy has been 
isolated, and the level mm which it is to be represented has been 
agreed upon, there remains the problem of the manner in which the hier- 
archy itself is to be represented. Since the action of a strength 
hierarchy can only be observed through numerical measurements concern- 
ing the behavior of segments across time, or numerical counts of syn- 
chronic alternations, the hierarchy itself is numerical in nature, 
representing a function whose assumed values are the statistical re- 
sults of the observations. In representing the hierarchy, however, it 
is necessary to make the choice between two methodological alternatives: 
first, to define the hierarchy in terms of discrete steps of 'strength'; 
second, eens incorporate an exact numerical value into the repre- 


sentation of the hierarchy. 


679). 238 Hierarcilessas*step functions) “in all sthe literature *on pho- 
nological hierarchization which has appeared to date, it is the first 
of the above alternatives which has been adopted: generalizations of- 
fered on the basis of large segments of diachronic behavior are used 
to yield a discrete numerical scale of strength, resonance, closure, 


or some similar scalar category. Such discrete steps represent the 


Apter etramsuesen Laotian Topo Bsanteedo odd we £150 




























i? a 7 hows 7 

_s is a. ‘ : j ; 
re ovo a8 at we 
7) re a? a 
ie ne ae | 7 De ie 
; eee . 7 


oe 


awitonisei® colloids 2 
fabipolowony oft tio alee roa 
2 Skitqaonm er SASS Oe ae i: 
ecuisy ahr iiemd is aah aNy lite 
 pomutast tren ere 


' _ 


oly naan 
eolux slisisiy oistowtioeit spitninid ot menial 


(ised asi yibtersid Ispkpolonotki s son Sieh ott | i. 4% a 
: rim - : 
need enif badnsasiger. ed oF ak ik. Wiebe de Bape alt eae ic 

ro die sae 


“isin silt | Luly ai sgnnom or 20 nioKone alt aelammens exec soa 
: * ae ise a5 eters, .3 
niprsite 5 to Aoitss odd site _bednesencien ad at at vl do ‘ 7 


7 
(war igs os 4 










ws : 
- 


y 
_ 


Jf sae 
“tye to etnuoo. [soittammnr xo, smi} sacs aioamgee 30 ee 


awen ot sobre ef Meetk yiorersin ott venmisemed ie 


« 


“sx Issitetyste oft os eauley bemyeas esorlw ao sonst 5 | 


Ji ,\evowo! \ydounteit ot pritnsasros: al ademas edt | fo 


‘Ae oe ol Hane od 
rasvistsrrstis Isoipofabodsem ow nsewted sotoro srt cola od yisee 


i MHpnéets' to aqpede sternal te amet Ai siieomelil a0 oniteh of 


- 2 irae * 
renga: oft ciai aylev Baan eins) 


y 


sac sicacsbaiy gh Tis, at. ——s 
werkt att al ti ens 


arbitrary partitioning of a continuum, but if such partitioning is 
carried through in a principled fashion, it is possible to achieve 
useful results.* Halle (1954: 197) offers the following justifica- 
tion for one such partitioning, the breaking up of a continuous 


stream of speech into phonetic (and phonological) segments: 


Although the view of language as a continuous phenomenon is 
simple and straightforward from a strictly physical stand- 
point, it has certain inherent difficulties which make it 
undesirable as a basis for description, and investigators of 
language ... have usually preferred to describe language as 
a sequence of discrete events. Furthemmre, it is not neces- 
sary that a physical phenomenon be actually discontinuous 

in order to break it up into a sequence of discrete events. 
It is possible to divide it into segments if we can show 


exactly how it is to be done. 


It is this view toward segmenting a continuum of linguistic data which 
underlies much of current phonological theory, whether one chooses to 
define features as singulary, binary, or multinary. In a similar 
fashion, one may wish to segment the continuously variable data re- 
garding phonological hierarchization. For example, given a system 
containing n phonological elements, and given numerical data regard- 
ing the hierarchical behavior of these elements across time, one may 
partition the numerical results so as to arriveat m n-valued step 
function, defining a single point on a hierarchy for each of the pho- 
nological segments. More specifically, given data regarding the beha- 


vior of five vocalic segments, one may rank these data along a scale 


205 


+ 



























os atime ct aidieeca ei Jt motdes® Belqioning 6 
seo btitent(, etivol [or oft eretto (vel sb@eLl) siLen ° 
epeunitano 6 lo qigrucserd amy verinoititieq 

+ ecneMpee ea cibenee oizorig ofa fontaine | 





‘at noiatonediq sucuridnte « 8b apeiyist Zo woky axis feist 

“haste Ipois wei, yGelite 5 moxt brewtoitipiente bas ‘ofgnie 
: re - ‘ade 

$i asian. (oitiw ssidineit® ib jnstarin’ rikstieo meal +t aiog 
to serene: bas ynoiidinesh 20 ahead s: a oldextsobrar’ ae we 
25 spsupiiss edioanb ot barreltten ywhilaven overt... epsupnsl | Z 
“geoen ton at 3. someds + BI CEPIO atoielh te soneupes Bo pee thy. 
>auountjrmooeib yi levios ed comemonerlg feobaweg se dealt wise 
‘sigens o¢erpeib to sonsupee 6 oft qu 3i ceed od zebra Gy 


wade so ow ti etrempse atai oF sbivih of aiciaaacg ak or ae 
io ad od 2k 2 ved 088 Wis 


; pas 
jae - 
fiolriw sdeb 2: Hatupazt + cams 3 wollen a G 
5 Sr) soa: ee 
Oo aaBoord> ono “werledw .yrosdt feotpatonora joer Yo rbum sak irebeu bs 
hk WEE ee eee 
telimitea s al .yuarrsivm 20 .ersitkel y vis leone és emmitcot enite® => 
1s |" Sg ; 
=6r sdeb aldsixey viavownistnoo. sid jeiaieees os dain aon \nokiest 
Zz 7 oe 
metaya 6 novip .siquaxs 107 unites itbrusrsie cm eee * | 


Ww 2 ee 
-brepet adsb Lsoimsmmss near B16 xtasme fo tectvotonedg # paintetnas! 


qoie beulsy—-o np 36 aviTis’ cd es oe etfygen 
ork, att 20 tase st bioxisl sno gniopetpnies 
railed ot vathaspsr stab movin , yLieokitsege, sit 
| en eee 


206 


containing the discrete values i through 5; alternatively, one could 
choose to group certain numerical values, so as to arrive ata 
smaller number of discrete steps on the resulting hierarchy. In the 
latter case, some sort of principled criterion mist be upheld for 
placing two or more distinct statistical values under the rubric of 
a Single point of a hierarchy. 

Despite the literature apparently attesting to the contrary, such 
segmentation of a continuum of data is incompatible with a phonologi- 
cal system replacing distinctive features by the discrete values of 
hierarchies. This is the case since the primes defining the poihts on 
the hierarchy and serving to supply the points of segmentation are the 
phonological elements themselves, defined in terms of distinctive fea- 
tures. If such distinctive features are claimed to be of no value, 
however, it is impossible to non-arbitrarily segment the data to yield 
a new set of hierarchical values to be used in replacing the distinctive 
features. Thus, the replacement of distinctive features by numerical 
hierarchy values makes implicit use of the prior segmentation afforded 
by classification in terms of distinctive features, and is consequently 


circular and methodologically untenable. 


6.33. Hierarchies as continuous variables. In contemporary linguis- 
tics, increasing interest is mounting toward accepting linguistic 
behavior as continuously variable rather than attempting to discretely 
quantize all linguistic data, thus effectively eliminating much of 

the methodological distinction between competence and performance. 

The application of statistical methods to linguistics is not a new 
development: numerical computations have long been used to provide 


quantitative characterizations of style. More recently, however, it 





fhivbo soo vytovijersatts 12 duct f 
| -B ge vitesse of Be of ,esutav Es 

ae ; 

oath cil eet pasate 08 ee eel 3 en stam 

sh bistey od jeu ndized ts Baliqionkg 20 eae Gees te 

ab stair arid sae aavley. Lenkoelinys 



























i 


; -yibisusid 6 20 5 2 
tye (aeyinoo ot ot ontteotts yineteqgs entered tf nts eget 
steetshera s rig.iw alciidagnoont Bi ab 30 maison 9 20 onkdednamoee oy 


20 esifsy storoelb ont yd aout ceidoatande wroesgas mseye feo x 


fio Barttog oct pniniftsh eamizq at canis sqeo of at SMR “SORRY 


ety syn ndiiegaamper to asniocn arts yiqare ct paivise brs yibusieid edt 


ae 


“691 Sy itoriszetb Ro emis ac bemiteh J esiloanattt atnomale rete ered oe 
= a 
an on 


“PBisv Oa-30 of Oo Banssio ots aeiost sritoniheib rove 22 .eemit rad = 
_ 


os; Pa. 
“Billeky of sib ott dnenpea-yhixats Lous of aldtenayad at 3 give 
ayldsattatb ont oriosiqat at bees x! of segiev Sastdeemngaet gee ere ; 
~ Ser ivsnnict vd esuiteot rtionigeib toe tiansosiqet si vauelt cera 
belrotte fottsinaima: xroing edt Yo see tic tiga nar osuiey \pexelsl ' 


yisneepeens ei bis .esnctsst syiioniveib 16 anesed it coteottinasto ; oa 
77 










 sideratm vilsoivetabostten | . bo - 
| 710. nada 7 
~elipnil yisiogusiac af -saldecssy Guounitone 25 5 : fot Ei : { 
sivarupnit pnitiqecos. Creevaxt prrigrmom ar suonsent 2 rok = as 
Yietorelb of onitqnsdits cst sedis sldsivey yf pie i , * mee ed i. 
a Ane 


Maes, T - 






to roam pritterimils pateepie tit - me kt if 
wu ; ; Saris os is 
Sonam sg ‘bas Sonstaguns Gaeta: sparen . 


3 J 


mn cia» mantel Q 
" ' a a a ee ane 
- Sbiverg of baey mesd! pol svar: a 


Mn A Pes - on eae | 
: 


_ 7 _ 
= 7 ae 
qj , 


207 


has been more widely noted that linguistic behavior is not always 
characterized by discrete categories, that rules do not generally 
apply in every possible case, that sound changes do not generally 
affect every potentially affected segment. Rather than trying to 
compartmentalize these continuously variable data in order to regular- 
ize them, many investigators have sought to incorporate the statis- 
tical figures into their linguistic descriptions, to provide a fuller 
view of linguistic possibilities. This methodological trend is most 
noticeable in the field of sociolinguistics, where, among other uses, 
the incorporation of statistical results has led to one proposal of 


central importance: the concept of variable rule. 


6.3.4. Variable rules. The original proposal of variable rules, 
together with most of the concrete suggestions which have been put 
forward concerning their implementation, are the work of William Labov. 
Labov first introduced the idea of variable rules to account for 

the apparently random behavior of certain Black English speakers with 
respect to rules of copula deletion, final cluster simplification, 

and various rules involving negation. What Labov sought was an empi- 
rically more satisfying alternative to the concept of ‘optional rules' 
or ‘free ameiecne: by showing that the frequency with which a given 
‘optional' rule was allowed to act could be correlated with additional 
factors, such as age, social group, style, register, etc. In essence, 
what Labov proposed is that each such rule be represented together 
with a numerical quantity indicating the percentage of cases in which 
this rule may be expected to apply over large expanses of verbal pro- 
duction. Labov defines variable rules as follows (1971: 465): 'The 


first step is to assign to each rule a quantity ¢ ranging from 0 tol, 


, yy 
Vodet meilliw to Aiow off sis \noOLtainemalepee ise gaubscrenaco Page ey *? 


| fishy oi: acess to spedrmouse stkt pristeodbak 


<6 0m ss 9 © fa 





—telopet os 1ah70 pik ateb sidsitev iia: sear? 
Laitwe Sid wixiSooni ot tgyoe aves qoinpbimowmh | 
teLin? 5 sbhivony x eooljoiaie Belopnél sted? cant eimmge ie0is a 
Seon ai baex3 Leotpotebddtten- adit oi ti1ediveon aigeiueatl 30 9 a 7 
,eeen yerlio pac orate acidainpatiotoss to bieit ert ni aldssoizen - i" 
#0 issoucrg eno ov bal sed epivess Isobteisete to sobssxoquoond ett a z - 


.elsx ofdermy to tqsoneo sx seonstyognt Isxtne OT 
pa. 
ua 


= 



























“ eg 
aah ofdsitsy 3o Lsacgoty Innit tye oft wgelup Bat ME 
Suey need sved robiw eaosjespuoe stemono eat to Seon ds.kw eee 


a 
Aes 

at 
a 7 


‘tO jayooos oF eelsm alderesv Io 54bi et heaubortint sat: vodad: mA 
iz 
yr 


ste 


Tak ra 
ittiw exossoge deifpnt soela aisduso 36 yoivered erbrusy Utosxeqgs 


\molisoittigqnie xsdavlo Lauda \notielab sivgao to aelirs Pr a 
i 
fone m6 esw tifpriee vodat tari¥ © noirspec paivlovnt Belin siabaat Seal > Je 


i 
‘golut Lanoitgo’ to Jasmonic ort ot sviiversatis tetviebian or ee : ‘ 


7 ie 


gavin s roisw dthiw yereupert ant ted priwode yd ree: “sh 
isnoittobs. itiw batsferieS od Bived dos’ ot bewlié eaw. tars ise oe 


, = 

ansees cl .ode:, tetveipst ,elyte «quomp feimes op a6 dou Hie ao a 
~ ¥ ae: 

sortigeer Hetmoesneen od elute doua does turk oqena vodsT tert eh. 


- 
Pa ae es Lae > 


i 
- 
, 
- 
- 

on _ 


Piicscinaakiaedtzveriek 
pete (2d, E801). amasTod an wala shay 


vn 
i 


- 


‘ ' oi 
en 7 7 
a) : , 


Tia} =. be) rye 6, 


208 


representing the proportion of cases in which the rule applies out 

of all those cases in which it might have applied. FoF obligatory 
rules, ¢=1, and for optional (or variable) rules, 0<¢<1'. He 
further notes (1971: 469) that 'these regularities are observed in 
the production of language, but from the hearer's point of view, they 
may seem insignificant’. In subsequent work, the concept of variable 
rule has been further refined by admitting into the numerical quantity 
a factor indicating the exact nature of interference with the normal 
action of the rule. Labov (1972: 112) notes: 'To each [variable] rule 
-+.- we give an automatic interpretation that ¢=l-k,, where k, is some 
factor which interferes with the rule going to completion’. These 
factors of interference are further sharpened by the inclusion of 


"variable constraints' (1972: 116): 


For a variable rule without any variable constraints in the 
environment, we have noted that ¢ Sosa tbake! variable con- 
straints are written so that the presence of the feature in- 
dicated favors the rule; that is, they diminish the limiting 
factor ko: $=l-(k,-ok)-Pk9 ... -Vk,) where the Greek letters 
have their usual + and - values and the constants k) ---k 


n 


are ordered so that at >k, Syayete >k,- 

Labov has opened the possibility of adding a hitherto unexplored 
dimension to linguistic methodology. At the present time, however, 
Labov's proposals are backed by little empirical evidence, other than 
his own data concerning a restricted range of sociolinguistic pheno- 

6 


mena, and the proposed format of variable rules has led to serious 


contentions about their lack of feasibility. It has been claimed, 








' giiys (efdsiasy) rpes of eaten. (Sif S802) voden. valine ot 3o-noktos | 























Bh ot ab 0 eof (etdeieeyean) Teh to rae s 
j ‘ Eo uM 


re eal “\ 
bat ee on a ae , — Kei) 
a = a: : : 
Wet wo fo sing etvencod SHdHRORN da ao anak ‘- 
eldetiav 2 dqgoncs Sd. ohow tomapaedie al seeing * = 
Si 


Witesup Iscktemm ott cont pristtthe yt Detter nectae® va 
=) 
fartron. sit diiw sonsusivedal-26 ouuatan JpeKe odd pattsoibak zatostie . a 
a 4. 
ave , a ? 

Sipe at pl sratw , Qiao tadt aaietemqadat oi temntus 6 VED OW sey 2 

easa? . 'soidelqaco, od vetop aberr sft siniw aoastesin, doit, x00} 
to coissiont sft vi baaserede sedi? sp basapeeg sy 30 speed = > 


7 


4 








ait ai edatetienco sidsixey yas juodsiw ie tee 
onom sidslisy off sie y eis bean avec ae mma rr 

“at sunset oily to somszmig ant Jeet og rtedtiaw ous. etnbette: - 

PUBURE! ortttieinimib vert vei ase vedere 98 exoved beech 
ettel Aeos ati axsite (at éa' a ial dent oh sahil penne _ . a 
A... pl atossaneo oft ‘brie esd isv - Bos +i r ee: 
a “ 


ree 


pie: a 


i htins silks eine: Seeasinhaaaa aia _—— 


ceevemol Sanit tneserg. efit tA _ NRatobaiktsm ise: 7 
cou hens a a 
ET AEE ne 

eeauaicistrsient = ists 


209 


for example, that naive speakers do not respond to numerical frequen- 
cies to the degree presupposed by variable rules; moreover, accepting 
the validity of variable rules creates a host of problems in the 
area of language acquisition, all centering around the crucial ques- 
tion: how does a speaker learn the 'variability' of a variable rule? 
Does he learn this variability through listening to the speech of 
those around him and then generalize on the basis of what he hears? 
If this is the case, it seems to refute Labov's observation (1971: 469) 
that variable rules are consistently noticeable only in the produc- 
tion of speech, and are often not perceived as such by the listener. 
A much more categorical objection to the concept of variable 


rules was voiced by Bickerton (1971: 460): 


If we accept the variable rule principle, we must also accept 
that the mind possesses not only the apparatus necessary for 
forming two quite different types of rule, but also some kind 
of recognition device to tell the speaker whether to inter- 
pret a particular set of data as rule-plus-exception or as 
area-of-variability. When we recall that the data on 

which NON-variable rules are based is often incomplete and 
een aie the mode of operation of such a device must 


seem somewhat mysterious. 


Some of Bickerton's objections may be dismissed by refusing to coun- 
tenance the distinction between exceptions to obligatory rules and 
variable rules. In other areas, however, it must be conceded that 
the points that Bickerton has raised have not been satisfactorily ' 
clarified by those who would incorporate variable rules into their 


descriptions. The exact operation of a variable rule has never been 


hespeatt' Teotranmit Gf & 
pruitqaoss ‘inlined 
aftt at teidorg 40 Incl 
Seles. olds tev 5 20, ‘iia a acm 
| oe Hoeeqe suis of 2 rine Mowers y - 
eee ae) seiTerandp oe ie sad 
#0 aut ety notternsads nado “a! rods.J ‘adden Go emetie 4t yoesm edt et ahd ar 7 
“tuber eit ni yino ctcuao ison Yisnvitnads is le ata i ae 
Weredeil odd yd rine as bev ier tect 26 ee et a eeD 
eldsiisy to Jemoren sf} ot noliisetds tap Hepends! ote Hoon Ay 


Mays 
| Bb :ivet) coro alo eae | 


ae ed " > oe 
toePo6 “eis jeum aw .aiqioning sii eidetney ef? dasocs om 31 i ears 
Mok YIREESOSN Butateags oft Ying Fon eusesarog haim eit set. 
ferkd enoe osfe tod vole 36 Say? trestle atigp ow pabmsot toe 



































ie. 
~lediti GF woritertw Taiigece ackh LA ay soive® noitinpoosr 96: - 
no-sint) Se edd Linco owii ORR ee tomens | oe ite 
Sas scelqmeonl neite Si pees as Sales Seeley mE : 
a) > 
y r > - ars et 
. oot Shen mae, ial pt 
es = 7 = 
bre pepiers’ Tate oe hs on ee 
sett betondo 0a iam ai eet —— v 
: car 
‘ - Medd siti eters eter aie 
i ug pemnnicn i 


a 10 soliqsone-culq-alpt as. <teb 2o 3a yalerotniag = 3m” a ae 
ie 
Jagm setyveb s coge jo roihswsa 20 sham ett <euomnegoisted Pes 
: a a ee 
PROMMNCTRNI 53° 80 ame 
P eae ali a 
* ; Vimmodoetelee ceed ton “a 


vo": i. j : 
La 7 a ' ie 6 r 
ia i. ae rm . oe ty a 7 


clarified; merely proposing a canonical form in which such rules 

might be formulated, if they in fact exist, does not answer the more 
crucial ontological question posed by the proposal itself. In the 
work presented to date, it has never been made clear whether variable 
rules were merely a formalization of statistical results, or whether 
they were supposed to respond to some psychological correlates brought 
into play during speech. In the event the latter interpretation is 
offered, and only in this case would the concept of variable rules be 
empirically meaningful, a number of psychological questions are raised, 
questions which have not yet been answered. Perhaps the greatest con- 
ceptual difficulty lies in the quantitative aspect, that a speaker 
could learn rules which are in some sense 'optional', but still learn 
that some rules are more 'optional' than others. By way of an apology, 


Labov (1971: 494) notes, in answer to a query by Lightner: 


On the question of numbers ... the numbers game is quite 
irrelevant, ... what we are looking for is the same thing 
that all linguists are looking for, viz., relations. ‘There 
are times when people go to excess ... in looking at quan- 
titave relationships which are finer than the system would 
Beets support. The fundamental postulate of linguistics 
is that some utterances are the same and that implies that 
there exists free variation among these utterances that 
are the same... What we observe is that relationships of 


more or less than are all that count. 


Bickerton, on the other hand, has another hypothesis concerning 


variable linguistic behavior. He believes, and supports his beliefs 


210 























ee nite 
axom ot} yowens Jon aa0b tebe gas ait | 
att xi. 2foadt fesoqere, peneiceminiedeiied: 
eldsixsy ~vartsoriw is915 eit, rea sane eae a \Sdsb ot 
seiljetw to \eIiuega aSebtase to aabienLemter 's ca lire 
Siioid escalerro5 (eotpatodeves sate ross Saogeer ot | oo 328 + shh Sa 
Br mottetorgrsar ssfiel olf 3nave Sede it minis Gea es 
pices iis aide cxsy Lo Sieone Seth BOW suno abet ck gene brs! (BEES 4 
,bevisx cis snois2oup ieoipefotiver to aac Ss \Tearyninsan yiteobstee a 
ters tecdeoxn ont agerhiet Borawens ext Jay tor eved dolrw enbitaesp an - 
rpisege 5 tent \joages svitsd iineep ‘eth af weit yrlwolRtb Lesage ~ 
prisot tt ite jud ,"Isnoivao’ eanee: arate mk eas fokitw aolire erat Bako oe 5 \ 
sts te to yew yi -ayertto nerd ‘Esnotego” sien ded eelox ame J5itt a 4 


af sie) Se a4 > 


rectriphl yd vier s at xsSwens mE Bedont (ate 28@0) voded 





> 


etiup et emso eredmin eff ... axochmm 2o soktewp arty Ad _ fy 
patrdt- sme erks ai ot poidool sxs Sw darw . Sabansiarsinc th ie. a 
ered! anoiteloy ,.siv 10 pablook 36 TR RET ST 
“éup 26 poivool ni ... agsoxs at ob afqceq certw emit eas 7 ay 
: bluow medeye edt nefit sant? gus sipbeiy caitienoitelen eves 
Poitaiimontl to etefiteod leioemabro? eff | rogue GF mse an 
tect aed fopri Jett bas come so ee ea is 
tard aedtiersdsu seed proms notdeines ost 3 pelne c ta 
7 Io egirensitsion srt ai aren 0 MG per 
ins a Pr 


Rta x4 
a 












$4 


? he - 
an 
ain 


Ls 7 m3 se : a 


en 


with data drawn from the English creole dialect of Guyana, that 'inher- 
ent variability may be regarded as a developmental phase coming 
between two categorical phases' (p. 487). In other words, Bickerton 
feels that rules tend to change from categorical status to categorical 
status, but since most rule changes are not effected hah esse eheg:= 
resulting intermediate stage between two categorical rules accounts 
for the principled variation observed by Labov and others. ‘This view 
finds support in the data regarding the social stratification of New 
York English reported in Labov (1966), where the age-grading technique 
indicates a number of rules, each asymptotically approaching a catego- 
rical rule, in differing stages of variability. It still leaves un- 
answered, however, the pressing question concerning the psychological 
status of variation. Bickerton, while differing with Labov on several 
issues, is aware that variation can be of a principled nature; regard- 
less of the ultimate reasons for such principled variation, one must 
Still face the problem of precisely how such variation is manipulated 
by speakers and listeners. It is at this juncture that the study of 
variable rules has halted, for current linguistic research has not 
gone beyond the formalization of a number of hypotheses to account 

for quantitative data. 

The possibility of formulating some sort of variable rule appears 
to be a feasible consequence of contemplating the incorporation of 
statistical data into the representation of a diachronic hierarchy. 

In all of the linguistic studies which have made use of variable 
rules, these rules were posited as existing on the synchronic plane, 
being, in fact, highly individual representations. When dealing with 
phonological strength hierarchies, where the data come from diachronic 


observations, one faces the possibility of formulating a diachronic 


Ss 


7 
it 


: | ~stsi? ss nas, Yo aio 
| ann HRN 
redid wigow miltcnt .(Teb pijeatrs 
Igakinpedss edvausets isofrtustse mine Speiatib, eb 

exit tee ob baspathe ton aus “begins oli taon 
miimieode sole? Te girlz. oye Rested opal 
wetv aia? .avertto Bas voded Ss one lena 


i 


fet Yo notsisoit Herts’ ietoo2 add pitbrepon steb ody ak droge abet 
qupinied priberp-ses oct esatiy , (aves) vodsl ai besoqes dleiend OY 
“opens s prinosorqgs yiiesitotqnyes oss este 30 xodeen s eodeotbak 
~nh sevsal (lite 3r .\thhidertey To sogote painetih at olen tepk 
femipatoroyer ott peLinsonico nmoiseaip  ontaraig ort sovewod \bemewens. & 
Esuavoe no veded riiw priwetitb alLidw . noted ooksstuiv 20 eutede 
bases retoten bef{qiodixza-5 to sd neo noiterrev sori ousws 2b yasuoel - 7 
Jaum ony .noitedisv belgimaing rove 16k enoabet nerd po 2 anol 
| tarentisn ¢ at ppdaixey thre ‘worl viseioate Yo maidens ‘eri ‘ocd Lie» 3 
to vitite oS tens sossonup sids ts Bi oI “sexinagal ban as 


ee a) 7 


jon asd rouscesx nitatyoait fagrio tot bedfed asd ect 


> s) ee Pi... 4 


datess of eoeerduqyd to x9cmun 5 70 noiiss{enmot = jeeacee " ae 
Te gd _ 




























: 
a7 : 







wens 
a@isodgs olirx eldeitsv 10 +102 amoe Belts phonic te pepten : 
ae: are | | - 





¥o noitszoqroodt Sct: ont jedgiictincs io sorapesno | 


iad sands + Ss eoeuatane ae aa ab oe 


a ae ae 
sidsiisy To sau shen eved spit bisitaibnd 


vs ta oad, ae Lo : 


ae pe iy 


: : re aa ~~ Mae th as he 
be ol G ol ; - a YH aa 
Vari, ¢ Ried Bah > - 
7 v7 iy “ ey ' . 
ss ; i < i. : ’ i bof pine ev : a a 
a wie Sec Panes : ine - @ as 





variable rule, whose empirical consequences may be quite different 
from variable rules conceived in the synchronic dimension. ’ Unlike 
a synchronic rule, which may be individually correlated to the lin- 
guistic output of a number of different speakers, diachronic rules 
nearly always represent the overall effect of linguistic change on a 
speech community, although slight differences may exist between indi- 
vidual speakers. Thus, right from the outset, one of the fundamental 
problems facing synchronic variable rules, namely the problem of how 
to represent the various quantitative data in the competence of indi- 
vidual speakers, is irrelevant when dealing with diachronic data. 


Bickerton (1971: 488) noted, in this connection: 


It will soon become apparent that the only place to look 
for system is in system--in an ongoing process of linked 
changes moving through space and time which form an ab- 
stract backgrand to which the linguistic behaviour of 
individuals, groups and communities may be referred--ra- 


ther than in any concrete individual, community, or group. 


From this, one may conclude that data regarding diachronic phonolo- 


gical hierarchization, data which represent a large cross-section of 


linguistic behavior both across time and across space, may conceivably 


be representable directly in the form of a diachronic rule, without 
the difficulties inherent in trying to represent such data in the 
grammars of individual speakers. 

When attempting to apply the statistical data of diachronic hier- 
archization to the overall diachronic description of a _ language, one 
is faced with a choice of where to incorporate these data: in the 


rules themselves, or in some sort of supplementary description. 


Ze 


























alan masa yar 3 
ott  abapanah ois SAA 
=nt at of iets levxco (Meu Kind: St yam aot | 
asker oinorinett \sxorteege dnowsitib to wedmmn 6) 
5 no, spnero aldaivpntt 201026 eave ety. inte 
stint sowie setss' Yan enna Agile deserts <sinommes: deems) 


fetrsmsbrast ort! to aqo \tsetud arf monk ofpis) eu -tedsege Laubiy eat " 


Y ae 
Wod. to meidox at yEonin ,padrt aldeittey: cinovioaye paios? anaidorg ag ' : 
=tbai 3q sometognes ert af suab wittetirtamsp evoke edt dngeongeret | 

6i6h cinowbsib Giw prbiseb cedw onevelermh ef sexless Leibiy | 


-qoitosineo ated mi \badod (98d efV@r) ncriexota 





— 


yg 


; "a a , Sete <a 
> Jool at sosiq vine acy trade sreuedgs emeed noose Likw a : 


ein 7 i om - 


’ Bani! Yo eesooig palopmp ms ni-smeteye At at meseya ‘xot Saal 
car | any 
~tis m6 grot fhiriw enitt Bris sosqe ciperorsci3 priven espaste ,. 4 
to wobveisd oijatupail nds Aatriw oo bapxpbsed jonite | ‘a ) 
’ aA iiaete J _ 
npithecieres Sd yeat sotsinummeo Bus aqnome. veleubiv ibnt b ae 
fu » lS - r 7 Lr _ 
.quonp to yydiermmso \leuhohet sdoxoneayyns al nerkt sent a 


“gion -ioxuinsib prifyaps: atsh tere ebelonce: yam ee) vehsd, mat on ; 
to fotsvee~aecro spisl 6 dnseougex rbidw ada voitesiirsetd Leip (is 
viderisonce wan .sesge atone ine anid sate Ae sabia abaateaeE oe 

twortiiw alex vinoxtinsib s omic oat ia ee 7 
ec it 2nb cm niseae co ptlia aateal 2p © 

enb. \ptareie 6 20 0 nobsispeet 28 
pee ae mdab emery < 


iy mee) ° eh 






—- 


cea 


et. Fe 





213 


It is useful to briefly consider the difference between these two 


modes of representation. 


6.35. Hierarchies as variable rules. If the numerical data of 

the hierarchy are applied directly to the individual diachronic rules, 
the result will be the diachronic equivalent of the variable rules 
proposed by Labov to account for synchronic variation. In the parti- 
cular cases under discussion, loss of atonic vowels among certain 


Romance languages, such rules would be of the following canonical form: 


(1) Vitter. 18) AIK: eo <d3 


where X ... Y Signifies the particular environment specified by the 
rule, and <> indicates the numerical percentage of application of 
this particular diachronic rule. Methodologically speaking, such a 
rule would be easy to formulate: one would simply add the results of 
statistical computation to the basic diachronic rule to obtain the 
"diachronic variable rule', expressing formally the result of the 
variable application of the rule during the history of the language. 
Empirically, however, the relevance and usefulness of such a rule may 
be questioned. 

Since one is not dealing with synchronic data, nor with the 
behavior of individual speakers, such a diachronic variable rule is 
inherently tautological, and tells us nothing about the rest of the 
language. The intrinsic interest of a synchronic variable rule is 
that it embodies predictions aout the way a given speaker might 
react to a particular speech situation, as well as predictions about 
the overall behavior of an entire linguistic community under a ae 
fied set of conditions. Its diachronic equivalent, however, due to 


its relatively greater distance from individual, synchronic linguistic 


Ness 


> “<4 
. oe ae y 
tis —. oe y ue 7 


wie 


—— eer 

















; 7. oe ee oe 






6loz atiobels 3 zie ai | “ee . Died re nes 
poli side hisy air svoteviange ae eee na 
<fHish aii-nI .notistisv >invisbange a0 Amacoe ed veda yd ogouq, Lad 
, fhictrso pace alewov oineds Io eapt veviseosei® abr isi a 





sg a are J are _ mt 
" gpro? Leo.tronen oniwoLlot sit 2 ad ilvow madiox dome seepeuorisk a 
| eee |). 


ra 2 a De | ethos 


ort “yd Soitinage sotilnessiinss cies deat nee ioe ee : ed 


os roe 2 
to nciss>ifqas to snciorey lesinenn ot sate <b> em alk ; rs 
ww, 


6 fode welxtneqe Wiilsotpefoberkiet .ofor oinondosib rsivoite tau 


ee ws 


26, edineex sit She ylomte binow sno satetemegt ak gelnewnd “Hikesine webs 
ee 
Sd nistdo ot lta oinostbsth ubeic spl Co cksedpaes cealsabende a , 


ee - 
eid Io tivass aft yitemot weisiwoxats < ‘ofers aldpixsy oinoxwios Ib’ = os 


<speeois.£ salt Ro noteit ai achwb oft et 29 nobteotiqgs eldeiuay r 


rn ) 
yar olor 5 doe Io nenenge hrs oorevelerr ont aseage et ae 





OC: i : 
eatrpeaddt 
as dstiw ton owt siccsionge diiw guise Son wh 999 ee te. 
at ofirt ofdsizsy oinonbelb 5 seve erodesce Lf b ribet ‘pivaded . 7 





- 
= 


oft} to saat afi tuots geirtion ev obey Bast 







7h wy 
a UF _ Zea in af a a q 
é a 7 ia Ax" 
i . oe aN ‘ 
aa iv 














ai ols Sidaher staal © 30 aaaaat. em epeomant 
3 e- Be 7 Pgh Why si ey a} 
: erect Seen Dae: 5 ee a a ; ison abo 3 | 
’ Ba Parting - 


Tig ’ 


214 


data, embodies no such predictions: the rule merely states that, in 
a certain environment, x% of all potentially affected forms were acted 
upon by the rule. This statement, while more precise than a purely 
qualitative remark contained in a historical grammar, is of no inherent 
interest to linguistic theory, since it provides no new possibilties 
for investigation. 

A further difficulty to be noted in a diachronic variable rule 
of the general form (1) concerns the incorporation of strength values 
reflecting hierarchization of segments based on their intrinsic pho- 
nological content. In the particular case under discussion, loss of 
atonic vowels in the Romance languages depended not only upon relative 
position within the word, but also on the timbre of the vowels then 
selves. Thus, for each rule representing a particular atonic environ- 
ment, the function ¢ would have to assume as many discrete values 
as there are different vowels under consideration. Such a practice 
would completely obscure the essential similarity of elements linked 
by a strength hierarchy, whether based on relative position or on 


intrinsic phonological content. 


6.3.6. Hierarchies as variable metarules. The other possible way 

of Pen retenting a phonological hierarchy in the diachronic grammar of 
a language is to include a set of interpretive statements conceming 
the overall strength of the segments or positions under consideration. 
Since any hierarchy under consideration is a ranking of segments, or 
positions, rather than individual rules, the most direct way to 
express the hierarchy is by means of an explicit statement of the 
nature of the individual events. More specifically, consider the 


incorporation, into the historical grammar of a language, of a set 


; a 
a ab seit ostese viorem Sar a. | oiso 

Bains sow utttod bainstis debt 

when s asi seinen sun aktW 5 eeoxacthesd 

Seenstint on to et <rianete Senco orp 


asislidinnod wat:od mei! WO tL eoata! «yaoartt pete peng 
(ig ba o 7 


eatery sldsixsv olacerbprb & praetor od cia eS aren a 


ae OSS &% 

ven iev dtpnsrte to eee act b tateics ane if) atrot Isrene9p eit * 
vi. ee 

~ofiq digcizini xhett no bsesd edtieniob 20 cottmebsiosens int 
So eco! .noteemmelib vebrar sésp zélooidaee aft al Arteta fsoipoton : 4 
missles cous yino ton babreqab aspsupnel sorsmokl ads rk bemasil incu'x | SM 


stint efswov ait to entints ait mo oefs Jud \Sxow oxi atdsiw salam a 




























hee 


2 oe 
-foiivis otnois. telopitieq 6 pustiseerdsr Sie itis at seu. apni eh 
f hos Pp ba cy 


Seufsey sdexoaih yen 25. ames oF svsd bliow @ emer id edg .jngm 

§ mer.) iat LA ty 

Sotiosig 5 fove .noitershianco ashen atone snsten Ub os ovo eB 
See -Oc C8 hers le 1 

bent! stoansio to-irelimie (eitnseeo eld seeioedo fen Se 


ade Tee 

go Yo noitieog avissliex ac > Beded orld Wormald dapnent dspnsite 6 yd - 
insta ane 
snoso Teo ipotonndig 9 ie re.) ae 


yew oldieeog isto off .aaiotsten oidsinsy 2s ae re 7 

to +teniierp sinemdics.ob ait nk ydocetotad tackuetanoeis 6 pabmeasiqe: ca 
paiinboron ajiveniectiste ovbtswassdih 49 fee 6 a 
oktetsbienos web) eapiticeoq 1 aidpempae sind to, cee 

MO , 2trenpse Yo paittet « zr ventana soe was sone 
oS Yew Jost tebe etd. eats Leubivitert sactiee exbiite 
ec? 20) Yaseen kg sn Xo: ease at st np @ 

63 wsbigecd \phieoR.tesqe orice eden i aa | 


hi eis am ite 
cues) ‘dadabeiie santibeds romp Lf . ca 


Be i. ae i hi ¥ Pee Ree 
| % 






= oe 





215 
of meta-statements or metarules or the following kind: 


(2) [X, ] + (x) 


[X,] ae (x,) 


[Xx] > (x) 


where the symbols [X, ] represent the elements of the hierarchy, e.g., 
segments, positions, etc., and each symbol (x; ) is a numerical quan- 
tity less than or equal to 1, where 1 is considered as the reference 
value representing the strongest element of the hierarchy. ‘The quan- 
tities (x, ) are thus not percentages of application of particular 
rules, but rather quantitative statements regarding the overall 
strength of the individual elements. Such quantitative figures are 
not absolute, but rather relative, showing the behavior of each ele- 
ment on the hierarchy relative to the other members. As such, these 
figures are difficult to arrive at, since they necessitate the 
calculation of proportional values for a number of different phenomena; 
ideally, every diachronic process affecting the language and segments 
under discussion. Once arrived at, these figures would embody a nur- 
ber of predictions concerning the behavior of the language during 
phonological change. 

Since the numerical values associated with each member of the 
hierarchy are theoretically independent of any particular change, this 
mode of representation suggests that, all other things being equal, 
elements of the hierarchy will exhibit the same relative behavior, 


regardless of the sound change involved. This is suggested by the 


















| Saya Ae i 
‘ wie Cs: we 
mS .Yrowsein oft to etgamels ot jnozogex [ x} alodmye ert sxedw : 
. sfisyp Isotxamn 5 al’ (x) fodmye tess bas . of ,ecoitleog ,ainempes a 


Sonesta: ait es betsbtenoo af | oxadw .t ot Jeupe 2o apis eeal ytit 
ern yioxessin ect to dnenals saumiowia astt patsnoesrqen. eu lev Srey 
Aekbsisieq to Hoitsoliqgs to eapstasoieq jon agnt ays tp) eoitit 
ieetibie eft piibisscr ainonsdate evidstivnsup tarttex sud pelos 

es aswprt ovidsctisnep dope. .ateeeeto iavbivibat ed 2o aid - 
-5i5 ths9 to toOiversd ot sohexie (Svidelex aedter tod vestsfoads Joa eit 
eset) .dqwe eA .eredean woot od.ovideies ywersroicl oxtt 0 Sram 
Sh atstizesdon' yots -sonie ts suites oF Siero itiib ons eouspit 
jscammasig troxstiib to xsdnun 6 xc saley Isnottsogerg Bo noitsfroke 
atrsipse bres’ Soeuprsl oct. prfjestis zeanord ointetoslb eve wuilsek 
“mun s Ybodite bivow eaxupit sgodt \ts Beyiexs sont duleamnie naan a 
prehb Seer alt 0 toivetled’ oh peimeanc a 


wnat fe i 





« 


| seit 30 ised dose te Beaaisbaa omer asm a | ae . ¥ 
ekit \ onsets. taleois1sq yates to cele setanmries< i: ; veel 
chee aaah oes ea si 


r ms sain beviovmi epasc an 


7° ve 
, 


at é ji j , fi We 


4 ht yer = Sale a iy 


i 







* 


216 


data which have been reported in the preceding chapters; in fact, if 
the notion of phonological hierarchy has any real empirical substance, 
it is required that the relative proportions remain nearly constant, 
Since it is this constancy which defines a hierarchy in its most 


general temns: 


DEFINITION: A representation of the form (2) is a diachronic 
phonological hierarchy if and only if, for every 
diachronic rule containing one or more of the elements 
[X, ] , the relative proportions of the corresponding 


figures (x; ) remain constant, within a specified tolerance. 


The area of tolerance will have to be determined individually for 

each case under consideration and will of necessity be arbitrary, 
except in the case of an error factor of zero; as in statistical theory, 
it should be possible to arrive at a common concensus concerning the 
largest tolerable error possible to remain within the bounds of the 
definition. 

In terms of the intrinsic possibilities for further research and 
prediction about the diachronic behavior of a language, a representation 
such as (2) seems to afford the greatest promise for the incorporation 
of phonological strength hierarchies into linguistic theory. Hierar- 
chization of phonological elements is an area in which quantitative 
data must be introduced directly into a linguistic description; these 
data must be presented in a manner which depicts their relations with 
the entire language, not just with specific events. If in facta 
true hierarchical arrangement exists, then the generalized hierarchy 


should stand up to comparison with specific data, and as a consequence, 








7 . ‘ ie =e wer: = a 7 See) i 
_ bas } wee ta am yaw 
ae | at i 


- fl : ? : sg ; ow ; 7 
=e 5 a iy ee 


eoity Ww t 


‘- 
et en hahaa 























Ce 

- »tposronib « at (S) mot a Jo moitatnasege: # 

: wreve rot ti vino bos 74 youszatd Leotpolonoay 

atiremefe ait Jo s1cem xO eno piiniednoo eles otnomidoss® 

7 7 ws 

mnificcestice sit to enokimegeng svitsios ois , Lx) 
; eka 
Sonevelo Baitiosge 5s nidtiw .tostenco nce | ~ asm 3 . 
: Sten 


' ; ; 7 F = Ps ie 
sof vilaubivibni berimstab ed at svyad Liew aonsreicd to Bexs edt ty 
> 2 
anna to [ftw bas noits1sbi aneo aie SEB all rate 


aA 
. 
4 
"1 
Jd 
s 
‘7 
Z 
* 
“e 
iW 


yaoact (eoljeitet+2 mi es 1ozss Jo woos? torte as 26 eee exit ob ‘notion 
at} palarecim pbensonco commco 5 ts svitews of eidieeag sdf pinode 7 | 
« op To ebarcd elt ntthhiw aiensr og ofediieaog sorte sitiexalod saovant ‘e a 
bis dorseesx asritw 1ot eo itil idizacg otentxmnt of Io emzes ot it ‘ ° 


saktatieesss% & ,Spsupinls to 1oiverled oimoitosib ens tggde robsoibens 


nolsmincicon add 108 seiner, teatsexp a3 Entetts of emese 8) es 
“texeiH .\yroed? obtelenti otat esitfoterstA ritensase £ 
nvitayisneiup state fi sere ne et axnemsioc simminiietie cafe? 
anor nitetas eee oltaiupatl 5 atni yljosrib beovbown£ od te 
dtiw enoiieisx tit aatqed vbinw tema s mc bedinsaesg: 
6 tse at  edneve oft inage sitiw tant ter." oars 9 
yeistereil hesilexasp ort ranks weirs Jnsnepaasts: 
seaapsebs & ee ive \stéb pthionge ihiw maak 


eT i lee . 
ic? . rier! 
soe a ian oo Oe oO 


allel 


lend itself to additional cases where it may be used to characterize 


some phenomenon whose initial description is based on other data. 


Ls 
iS 


Mt CS 
rt 
Sen 

Fak 3 be 

: ; co 

va Das S289 Lanost ibe 

%~ me 
o 3 : 
ore | : 
vie ah Ji 
; . ee 
iy. 
Et c 
qitoasb 
fs 
SPU. r 
7 
. seartw 
ov 
gq ae 


7 
=) 
all 








218 
Notes to Chapter Six 
Ghee Halt. (1950,..1963), Posner (1967). 


2 For example, Hjelmslev (1961), Fudge (1967), Twaddell (1935), and 
Bloomfield (1933). For some ideas in favor of using physiologically 
motivated distinctive features, see, for example, Delattre (1967). 
Cf. also Delattre (1969), Halle and Stevens (1964), and Lade foged 
(1969), For more general remarks concerning the empirical corre- 
lates of distinctive features, see Brozovié (1967), Fant (1967), 
Halle (1964), Jakobson (1939), Jakobson et. al. (1963), Ladefoged 


(1966), McCawley (1967), Malmberg (1955) and Wheeler (1972). 


3 For example, the experiment reported in Derwing (1973: 316-21). In 
this experiment, subjects were asked to group together phonetic 
phrases (CVC in the cases of vowels, VCV in the case of consonants) 
by means of the 'concept formation’ technique. The set of consonant 
stimuli were divided so that one group would highlight the feature 
of voicing, another, the feature of continuity, while a third 
group would contain a random arrangement of consonants. Measure- 
ment of the degree of difficulty encountered in arriving at the 
"concepts' indicated that the feature of continuance was easily 
recognized, while the feature of voicing was seldom isolated as the 
key concept. Among the vowels, the front-back distinction was 


recognized more frequently than the high-low distinction. 


4 In another defense of discretely quantizing statistical data into 
the form of phonological rules, Householder (1967: 942-3) notes, 
with regard to the possibilty of a rule's expressing variation over 


a continuum: 





















bos \(2E@L) fiebbewr ,(Foe@L) aphwt (4000) velanieth 
ylisctpoloieyily priey to. 1owst of assbh. enwe t07 chs. batten we 
«(Tael) sutdsist cae tol ,9se ,eicteet sviltomigelb bejev.ivon 
Baahches Bins eae) anavate Bos sifakt ; (eaer) eridsied coals .7D 
ooo Isoiriane alt petarsosco avhtsmter fsveagp esom so% =, (REL) a 
.(WaeL) 3nst , (Teel) Alwosett ase ,eseudeet evisontietd Qo eedel ) 

Hopolebhsel \ (905) 16 -ta qoedodet , (C6Q0) moedodst \ (heel) efter | 
(Svel) xSsfleorw a. eee . (VeRL) yalwacah a i. 


mq .(iS-clt :fVeL) paiwied af Dbedxoger nena ant valquaxe 10T €. 
jisartorta 18rispoy auoxe at betes srew etoefdve ,Jaemiveqrs efrit 
(esnsrioenoo tc Sasn orki aL Vov moe 20 sozeso off mi DVO) esesiig 
seenoencn to +52 sl" .supimibad ‘aomemrot Sgeoroo ' ert to ensen yd i) 
siitsst sds idpitdprd bivew quoxp eno Jed of bsbiv ib gzow iSsentse 


Baidt s slidw ,ytiooitnos 3o sutsst edt \astioas s<pntotov to = 


-ayvesne .atceonence io txtemepne rts mobnés s misdaes Blyow quoxp > an 
_ 
eft 36 potvivte at Beradmoone ytleoistib Qo osreb ant 20 jasm a 

: 


vitesse esw onsuntinos to’ sure? sit teddy Bapsolbak ‘ etqsoneo" a. 
ent a6 batsioet mblee. eawW ynintoy to singe; ent ethtn .Sesingooes. 
pew ‘AOijonitach soed-trort od ,elewov add patims cece beget 


-Aottonieib wol-ipht elt, nett ~yorteypa. —_ iA 


te ts “oie 7 
a on " 










adnt éd&h {ssisabtate pis ieneup vietewoe.ib to. oaneiish 
piston. (€=S80) Vader) rab loresuol \eelus —— 
2900 ‘noltsiwY pritaaomps a'oler 6 to ies 


e 


‘ 


~—a = 


219 


There seems to be at least the possibility ... of rules like: 
[+ nasal] > [k nasal] (.94>k> .32, distribution normal) 
where k specifies the degree to which the velum is lowered 
or the proportion of the segment during which it is lowered, 
or something of the sort. If this were so, then the output 
of the grammar could be considered to correspond to indi- 
vidual utterance-tokens; it is more commonly supposed that 
the output should rather correspond to utterance-types; i.e., 


classes of utterance-tokens. Tokens would be relevant per- 
haps to a performance model, but not to a competence model. 


Cf. Labov (1969), Labov et. al. (1968). 


For further discussion, from a more recent perspective, see 


Cedergren (1970), Cedergren and Sankoff (1972), Sankoff (1972). 


For some rudimentary considerations on the possibility of diachronic 


variable rules, see Fought (1973). 

















HEE patie ta. -* 4 ve. ticle aq ; 
(lergon nobtudisgaib SE. <h be.) EE 


Bexswol sk tmlay att tot a at 4s 
besawol at 21 roliw pr liwb & xi} to soktrogg ect 20 
‘seeptug: Se! nied oa “stew eked ¥T tyoe seit 20 peur +4 "i 

= Sheet og Broweatteo x scale ad bitte oS as to a 7 

_ +63 Besoqaa Yigan som 22S. enedod-sone tu feubty 

<aSpt iasqys— ats oy poe ork et Bs Ls ote Jug. 


18 Sisvelen ad Bivow arrexat os aornredtey TO. 
haben sonadeqnan 6 oF 76a td | [abent SO NBITOS EH a 


(2801) fs ds voted , (@aet)- vod 2 ne : 

ee a ¢ a 

| y 

aoe .ovitosgeieg tasdsr sions moxt ,noteamei® besdamti ae a 4 
_(Sfef) Hrodae2 , (STEL) Molake Bas nerpwsied corer): pemgrebe 


“AD _ 


sinoxberb to Wilidieacg att no enokterabiehos yemineamiien enon ft k 
| epr>7 
(ETCE) Stripuct soe elon” oldsiso 


0 : © ' : ’ 
j 3 1 t 9? 
at 7 <_ re 5 oes 


220 
BIBLIOGRAPHY 


Afendras, Evangelos A. 1970. 'Can one measure a Sprachbund: a calcu- 
lus of phonemic distribution for language contact’. Folia 


Linguistica 4.93-103. 


- 1973. 'Some formal models for the sociology of language: dif- 
fusion, prediction and planning of change'. To appear in Advances 
in the Study of Societal Multilingualism, ed. Joshua Fishman. The 
Hague: Mouton. 


Afendras, Evangelos A. and Nicolaus S. Tzannes. 1969. 'Stochastic 


processes for diachronic linguistics’ - Monograph series of the 
Centre International de Recherches sur le Bilinguisme de 1'Universi- 
te Laval. 


1971. ‘More on informational entropy, entropy and sound change’. 
Studia Anglica Posnaniensia 3.13-24. 





Afendras, Evangelos A., Nicholas S. Tzannes and Jean-Guy Trepanier. 1971. 
"The dynamics of pattern change: ordered rules and feature entro- 
pies'. Proceedings of the Fourth Hawaii International Conference 


i ee 


on System Sciences, pp. 158-60. 


- 1973. 'Distance, variation and change in phonology: stochastic 
eSspectsmarolia lainguistica6.1-27- 


Agard, Frederick B. 1967. 'Stress in four Romance languages'. Glossa 
150-200. 


Alajouanine, T. 1956. 'Verbal realization in aphasia’. Brain 79.1-29. 


Alarcos Llorach, Emilio. 1960. ‘La constitucion del vocalismo catalan’. 
In Studia Philologica: Homenaje Ofrecido a Damaso Alonso, v. 1. 


—_ os 


po. 35-50. Madrid: Gredos. 


Alberti, Leon Battista. 1964. La Prima Grammatica della Lingua Volgare, 


——_ 


ed. by Cecil Grayson. Bologna: Commissione per i Testi di Lingua. 


Alonso, Amado and Raimmdo Lida. 1945. 'Geografia fonetica: -Ly -R 
implosivas en espafiol'. Revista de Filologia Hispanica 7.313-45. 


Andersen, Henning. 1968. 'IE *s after i, u, r, k in Baltic and Slavic' 
Acta Linguistica Hafniensia ale 171290. 


1972. '‘Diphthongization'. Language 48.11-50. 
Anderson, James M. 1963. 'A structural account of the evolution of 
intervocalic consonant clusters in Spanish". Doctoral dissertation, 


The University of Washington. 


- 1965. ‘A study of syncope in Vulgar Latin'. Word 21.70-85. 




























; f ; 
f Ber eg e's o5 

ig =p iso 5 saris sige a pprbret petreye ne pv oe a 
sion. Jeeineo: sparpasl 5 30 | 20 } 


5 +, 
‘ » # 


“Lib :spsdpmbl Ete) ees 03 
SSORRYDA ft IAMD | oF pais 2 
LY . peanite ra BOS 0b {bs Te ice 


ivjee 

siduersoic'  .eatl dotheigl Salsa Bets A ees assbnsth 

eit So 20rtse dgerponmt |. lap bie tprel: Sinoribalh 20% asezs01qg 
storey ic Sennett raat se “sorkyrarinsa of farnidemednl. stim ) | ran 
Level sd 


TT 


.eprish bovoe bus yoxrtds vvgonigns Iscoitemroing no gut! «lV@L: ych_s . ty 
-BS=Ef.& siansinsaagd soi tona sibuse : 


a al 
maya LEAST 1 Vise “eSt. fos sedtissT 12 arfertis or! antepasva: vestbastA “20 
“INS SULIe 2 Bis soley Bertebue  epnedp misagé6q 26 ao tlasnyh oa 4 : ”. 


Bossi on, fa teorisdny Ehewelt A sft 39 3s aleged fh Fil tera 
| ; ~~ Oe-Sat ahve 


Sijesfnose :ypolLoncdq mi spnsrio bas, nobsitey ,sanmidets” = ° 
“S-[.9 acitetipnht sifod .‘eidequs 


heeolD = .'sepsuprisl sonenc wool ot essste’ .VeRl . ebrebas’ (paeek eee 
-b0S-02L I eT 
OS=5.0 siexd ..'sieaigs ai doisesifess fedvev' .d0@k <P .enckeOEeiA ee ee 


."néietas canetisnov Leb noborrti janco si! .O0@L  .ohting \foszohl acousf[A 
Lt .V ,OanOIA casi wd 5 COL e830 i snemoH ‘eoip j mx Ws 


GG 






gisrlovy mpnil slleb solteagexD onixd 6] . Bees sitiiee eon vivredia. 

Teo: ib fsact Fisq Secteeiimo) ‘enpoied .nceves Lise) wi bo * “ 

& y a> shits} wine chef » Eos Obra Cast bes obeck . ome 
db-efe Aoisbaeih sfpole oli t 65 stapes . Lotpqes alec icseies ch 


. ‘Divsie bas bide nt A x wy i 1St8s BF an’ goer 
| ORES LE pteneinied 

oP alt -O@-11 82 ) meee Meat taaatin 
aed 
to moitelove wit o smicoos a eee ae €aeL- MM 
,Foitesaseaib Iexotoot .'datneee at LORMOO prem ror 


- ie? ) Ow eles ge ee ere y 
* 6 toate > Vv 


. One 





gh ae Sn es omnes 


ey 


‘i ce 
ao a ae 


eeu 


Anttila, Raimo. 1972. An Introduction to Historical and Comparative 


Linguistics. New York: Macmillan. 


Bach, Emmon. 1968. 'Two proposals concerning the simplicity metric 
in phonology’. Glossa 2.128-49. 


Badia Margarit, Antonio. 1951. Gramatica Historica Catalana. Barce- 
lona: Editorial Noguer. 


. 1962. Gramatica Catalana, v. 1. Madrid: Gredos. 
Balily, Charlesh@ W5iertiraite ide istylistique; Prancaises sParis7e3rd ed. 


Batalha, Graciete Nogueira. 1958. 'Estado actual do dialecto Macaense'. 
Revista Portuguesa de Filologia 9.177-212. 


Battaglia, Salvatore. 1970. La Formazione dell'Italiano. Napoli: 
Liguori. 


Bec, Pierre. 1970. Manuel Pratique de Philologie Romane, tome I. 
Paris: Editions A. & J. Picard. 


Beckmann, Petr. 1972. ‘The Structure of Language: a New Approach. 


Boulder, Colorado: Golem Press. 


Bellucci, Laura, ed. 1967. Rime di Antonio Beccari. Bologna: Commis- 
Sione per i Testi di Lingua. 


Bembo, Pietro. | 1955. Prose della Volgar Lingua, ed. Mario Marti. 
Padova: Liviana. 


Bickerton, Derek. 1971. 'Inherent variability and variable rules’. 
Foundations of Language 7.457-492. 


Bloomfield, Leonard. 1933. Language. New York: Holt. 


Blumstein, Sheila. 1968. 'Phonological aspects of aphasic speech'. 
In Studies Presented to Professor Roman Jakobson by his Students, 


ee 


ed. by Charles E. Gribble, pp. 39-43. Cambridge: Slavica. 


Bourciez, Edouard. 1967. Eléments de Linguistique Romane. Paris: 
Klingsieck. 7th ed. 


Bourciez, Jean. 1949. 'Notes sur la résonance nasal des diphtongues 
accentuées du Portugais'. In Mlanges d'Etude Portugaise Offerts 


a M. Georges Ie Gentil, pp. 61-7. Lisbm. 


Brasington, R. W. P. 1971. 'Noun pluralization in Brazilian Portuguese’. 
Journal of Linguistics 7.151-77. 


Braun, Theodore E. D. 1974. 'Phonetic and visual spirals in "Voyelles"'. 
Letter to the editor, PMA 89.353-4. 












2 
Come 
.20bea).  BithsM: of 


(Be Bi ated . setecnsrd supitellive ‘gb &b tite ERE elas vein 


> ; 57 
eandesrh of elsth ob faitos adel" - 20L vetieupan wielloeo™% (ertietsa 2) 


SIS-VES.0 sipololia sb seampeol esetvet = 


rifoqe .orweifsgT" {feb onoisamOt st .OTCL .enetavie® yeilesiied 
' . : 





i amot ,scamod sipotolint 4) ceplsint #2 : : erat yee a 
esr anole reabyst © _ 


oscars wes 7shs5upns! 2o. stusoueie Sat vel je niramnboed- hs, 


Beene ‘teleo :Obs10.00 tab [u08 ) - 


















eae 
| <2 oe 
—zintme) <sopofod .ixsoood olmctni ih-smit -.VeGl .he) amet stoouL Led : -~ : 
. “aipoid & test F 3eq snake j > ia 
i Sd 5 F _ 
; 7a 
.jrreM oftgM .bs siprht aeplov sileS a50xS «cee searere odmod em 
-5neivid umniete TP ako 


. Seley Sofas ivev bos yo riide inev dgeeverind’ .18@L .densd scams u 
5 elQBAVeh y Speupciel 3s al 2o Sepboabarre9% 7 % 
_ aa 


JfOR pHOY wat” ". opaopdad EGE -basnoel bieitmols 
. ‘pesce oben vat Be) ea Oe |e soins Loacdt), .B3eL es! Ste J vnietomeLe Bs 


\Saiebet2 Bit a Adsciqver pahoy xopzolard ot Be 
“Estysle .tsebizans> - \t-@F gq yefdda 






rs ia ered x. 


iy 
— 


pisaivpeky : 55. ettematal NBeE | 
| .b9 30 
eesti derid nf {sesa sornstost: ef we epic ia ~ ,f 
aoAStIO seteputzel shius'h z Sie a ee 


ee ee 


> j ~stigk Vala" ‘ey . 


a : r 


Boe 


Brekke, K. 1888. "Lie, (=2, D latin en ancien framgais et en mayor- 
quin'. Romania 17.89-95. 


Broéndal, R. 1940. 'La signification du préfixe italien s-'. Acta 
Linguistica (Hafniensia) 2.151-64. 


Brown, Roger. 1958. Words and Things. Glencoe, Illinois: Free Press. 


Brown, Roger, A. H. Black and A. E. Horowitz. 1955. 'Phonetic symbol- 
ism in natural language'. Journal of Abnommal and Social 
Psychology 50.388-93. 


Brown, Roger and R. Nuttall. 1959. ‘Method in phonetic symbolism 
experiments'. Journal of Abnormal and Social Psychology 54.441-5. 


— ee ee 


Brozovié, Dalibor. 1967. 'Some remarks on distinctive features, espe- 
Cially in standard Serbo-Croatian'. In To Honor Roman Jakobson, 
v. 1, pp. 412-26. The Hague: Mouton. 


Bull, William E. 1965. Spanish for Teachers: Applied Linguistics. 
New York: Ronald Press. 


Cairms, Charles E. 1969. '‘Markedness, neutralization, and universal 
redundancy rules'. Language 45.863-85. 


Camilli, Amerindo. 1954. 'Dal ritmo latino al ritmo italiano’. 
Lingua Nostra 15.90-2. 


- 1956. ‘Il ritmo intensivo italiano'. Lingua Nostra 17.26. 


Causey, Robert L. 1972a. 'Attribute-identities in microreductions'. 
Journal of Philosophy 69.407-22. 


1972b.- "Unitomn microreductions*. “Synthese 25-176-218. 


Caws, Peter. 1965. ‘The Philosophy of Science: a Systematic Account. 


Princeton: Van Nostrand. 


Cedergren, Henrietta. 1970. ‘Patterns of free variation: the language 
variable'. Unpublished MS: Université de Montréal. 


Cedergren, Henrietta and David Sankoff. 1972. 'Variable rules: per- 
formance aS a Statistical reflection of competence’. In press in 


Language v. 50. 


Chafe, Wallace L. 1968. 'The ordering of phonological rules'. Inter- 
national Journal of American Linguistics 34. 115-136. 





Chatman, Seymour. 1957. ‘Linguistics, poetics and interpretation:, 
the phonemic dimension'. Quarterly Journal of Speech 43.248-56. 


- 1967. ‘Comparing metrical style’. in Style in Language, ed. 
by Thomas A. Sebeok, pp. 149-72. Cambridge: M. I. T. Press. 


Binh .'~2 collet axtitug bt 





















SeexG atl seron. [ls ald J 


Afodnve sitouortt’ . 221 sine = 3A ba belt Ht 
Isince fer Gagrondh 30 Eapayeb By eet . 


= 
a 


metiodme oitseodg itt horse’ f By ee geomet mel os 
! Pt ; wed s£D0c , 25 . : 
@rid 2 ypolocved Isisoe hos: eee aie a 


“e929 ,aeuus6s? sviconitelib m Bate ATS eee" -ToeLt TOC , 
inedodk. come xonol of nt . "ne porD-rxree +4 
v | "neta -eemeH off OSSD ey, bigest rae 
bs one ere © ) 2 " 
eobteinpnit bolfagh :eterosst wok eg Agel mst Line 
ae ee, we ‘bisa vhot welt ane + 


teencvirw bos ,notsssilextisa pesabectias' eRe 4 valued) yemtisd ae 
> 28-£89.0b opsoper “eel youseem= St 


“enilfesi omizx is onits! cnttt IeG” BREE a 


cheatremA erede 
-£-08, or guest see i 


JaS. Vi atteaot suprh!l <'oumilath ovleini Opie sh es q 
ie. poe 
Wn =“ 
. ateiioubriorminm ni eskiionehs-san ita’ .sSVOD. at yt yous ws 


.Sc-F00.20 wiguaosadt 20 Leroy = 9 
Pac verses 7 _ : 
BESABTELES seschinye {NOISE OTA ae waasret) a 


wimosad pitsmetaye © seoneiod 20 yfk sce ai v88UE ssf ae) rag 
aS er baeseat ow sfatsoritd - a 


Sparel aif smoiteiusy sett Po ateesigzed' Oyes 
leervineM 26 Sd itervaviad’ <M . 














7 
s 


“teq ides sitikissy’ Seer AaeaaweHOnaG bee sited 
ni pest a= .sonisssqneo Jo noksoatlet isotteiseta i. | 
.V 9psyprsd 


~todret . apis fenrpotonolg 30 ipithsbre a ca i. 
ls LVI! PE eykzetepabt a ase de TOL, 


rio kdss cecptatens (bre anitoud ¥ 
LBe-MC 6b dooeee 2e ismmpet 


= aah tieamar: 






7 Ly 





* = 
-_ 7 
a¢ 
a 7 1s) 
‘ 





faye 


Chen, Matthew. 1972. 'The time dimension: contribution toward a 
theory of sound change'. Foundations of Language 8.457-98. 


- 1973. 'On the formal expression of natural rules in phonology’. 
Journal of Linguistics 9.223-49. 


Chen, Matthew and H.-I. Hsieh. 1971. 'The time variable in phonolo- 
gical change'. Journal of Linguistics 7.1-13. 


Cherry, Colin. 1966. On Human Commmication. Cambridge: M. I. T. 
Press, 2nd ed. 


Chisholm, David. 1973. 'Lexicality and German derivational suffixes'. 
Language and Style 6.27-38. 


Chomsky, Noam and Morris Halle. 1965. 'Some controversial questions 
in phonological theory’. Journal of Linguistics 1.97-138. 


- 1968. ‘The Sound Pattern of English. New York: Harper & Row. 


Se eerermneneetiilieeee eee pee et en 


Clark, John Taggart. 1905. 'L'influence de 1'accent sur les consonnes 
médiales en italien'. Romania 34.66-86. 


Clegg, Halvor J. 1967. 'An&lisis espectrografica de los fonemas /a e 0/ 
en un idolecto de la Habana. Masters' Thesis, University of Texas. 


Coates, William A. and M. W. S. de Silva. 1960. 'The segmental phonemes 
of Sinhalese’. University of Ceylon Review 18.163-75. 


Coenen, H. G. 1965. 'Zum Verstandnis von Arthur Rimbaud, Voyelles'. 
Zeitschrift fiir franzésische Sprache und Literatur 75.353-64. 


Cohen, Bernard P. 1972. 'On the construction of sociological explana- 
tions'. Synthese 24.401-9. 


Cohen, Victor B- 1971. '"Foleyology'. Papers from the 7th Regional 
Meeting, Chicago Linguistic Society, pp. 316-22. 


Cornu, Jules. 1888. 'Die portugiesische Sprache'. In Grundriss der 
romanischen Philologie, ed. Gustav Graber, avs ep 1o—o03% 


Corominas, Juan. 1954-7. Diccionario Etimolégico Critico de la Lengua 
Castellana. Madrid: Gredos, 4 vol. 


Creore, Jo Ann. 1971. 'Synchronic grammars, reconstruction and the 
selection problem'. Canadian Journal of Linguistics 17.16-26. 


Cressot; Marcel. 1959. le Style et ses Techniques. | Paris. 


Critchley, Macdonald. 1970. Aphasiology and Other Aspects of Language. 
London. 


Dalgado, Sebastiéo Rodolpho. 1900. 'O dialecto Indo-Portugués de Goa'. 
Revista Lusitana 6.63-84. 


@emexort® Setosopse oe?’ .0805 .svif2 a -2 .W UM Bre’ EL (eetecd oy) 


6 Faswot soludigns 


BO-VERF epaippoel to 


. ypotonodg it Beis Lette to 





















~olormody mi efdsixsy ania sei’ 


ELADN cua 


af iT AG teebprehsd ibis ai 


& Ja 


‘eatitive tnndiicy ire pntiso ne gilisoimer’ -€°GL .bived oe 
| B8-86,8 aig ims sake | 


ampitean (siexovquticg cee’ ae .alled alanis oe 
BEL+T2.5 po batmbT RS femtgot |". “yom iso tpaloroeig 
WA Sapa > OY wa wdetipal fo omaed baw Set Baer yt ’ 


———— 


asminenio>. ci wwe dasvos'! ob eooauitab'a" .cO@l .Seeppel mob: wa. 
.02-32,30 sige .‘pelladt ho ssiskban ~ 


\O. 3 #\ asmenct sof sb soliispartoegae eleiitnd’ ~ veel “Fiaewisl yoeets = 
pees? Io ywilecavint) \aleed? ‘ersdesM Jsoedel el 2b aio apne : 


h 


,PT-£al BL wabvel no ly. 30 wile in + “eaeilednte to - 
. “Sealfloy? | boseink? coritté ooy arnbattetev mg” .2gGE “(eee ee 
ABOMECE CN xoSeted 1. bay afhexge Soetecers Te Bee ok 


“soaloxe feoipoloisoe to moiinudens at mw! JOYGL .2 boa mae 
jO-FOR DS Seetage . she 


AS 


Pam spon fat =A fd i369 i° .'ypofoyaiot" -fVer 


mete Le — - oy em 


oe Toa WB Loe8 pide lupantl ste a 


Seb aeisbrerD ni. “siesige sroelesiousiogd ald’ -886L party a 
ft Onecle GQ sf .v \2a0u) Vern she < 


. ¥ 





ei 


Supoal i SO: coctix) copious 1 BeeeeeuO -T-DeOr an mgm 
>biubeM x 
yy 


ant bre notiowtenoos: , summa aoe ter. oe a 


_ 





(PE-OL. CL Bottetennay Jeiraek, ‘mak doa 
erm? .2supiniba? psa jo“eigs, er eek =, 
TeaupHE Te atheanA werdO bye ypolodistigs OV8E - tite 


"S00 sb efapstx0I-ochql ctoeisib O° re. é ey 4 obitesds2 ,ol 


uy 
: 
: 


us 


- 1906. 'O dialecto Indo-Portugués do Norte'. Revista Lusi- 
tana 9.142-66, 193-228. 


Darnell, Regna, ed. 1971. Linguistic Diversity in Canadian Society. 
Edmonton: Linguistic Research Ltd. 


Davison, Alice. 1971. 'A problem concerning the relative natural- 
ness Of phonological rules'. Papers from the 7th-Regional Meet- 
ing, Chicago Linguistic Society. 


D'Azevedo, Pedro A. 1900. 'A respeito da antiga orthographia portu- 
guesa'. Revista Lusitana 6.261-68. 


Deferrari, Harry A. 1954. ‘The Phonology of Italian, Spanish and 
French. Washington, D. C. 


Delattre, Pierre. 1967. 'Acoustic or articulatory invariants?', 
Glossa 1.3-25. 


1969. '‘'Coarticulation and the locus theory’. Studia Linguis- 
£icae2o.-26% 





Demattio, Fortunato. 1875. Fonologia Italiana. Innsbruck: Libre- 
Yia Accademica. 


Derwing, Bruce L. 1973. Transformational Grammar as a Theory of 
Language Acquisition. Cambridge: Cambridge University Press. 


Devoto, Giacomo. 1939. ‘Il prefisso s- in italiano'. In Mlanges 
de Linguistique Offerts aC. Bally, pp. 263-67. Geneva. 


Dingwall, William Orr, ed. 1971. A Survey of Linguistic Science. 
College Park, Md.: University of Maryland Press. 


Dos Santos, Felicio. 1898. 'Lingoagem popular de Trancoso'. Revista 


Lusitana 5.161-74. 

D'Ovidio, Francesco and Wilhelm Meyer-Liibke. 1904. 'Die italienische 
Sprache'. In Grundruss der romanischen Philologie, ed. by Gustav 
Grdéber, v. 1, pp. 673-711. Strassburg: Karl J. Trubner. 


Elcock, W. D. 1960. The Romance Languages. London: Faber & Faber. 


22h 


Entwistle, W. J. 1962. ‘The Spanish Language, together with Portuguese, 


Catalan and Basque. London: Faber & Faber. 


Ernst, Gerhard. 1970. Die Toskanisierung des rémischen Dialekts im 


—_— «eS, COCs 


15. und 16. Jahrhundert. Tiibingen: Max Niemeyer. 


Fabra, Pompeu. 1906. 'Ies e toniques du Catalan'. R&vue Hispanique 
15. 9-23. hie > 


Fant, Gunnar. 1967. “The nature of distinctive features’. In To 
Honor Roman Jakobson, v. 1, pp. 634-42. The Hague: Mouton. 




























“ieet preived «Sat Ob 


tm 


Ywoites maibans of yo beaver , 


Se RE enter re 





: 


~Lsnisen sv itsier ont ee catia 6 | Pati 
“Joa Leno toa fii ork nor 


—— a a) 


> $ 


= 

Gx : 
swtiog siigetporixo spins: sh ose A* «0002 A cubed eee wal aA 
S358 0 Snes heen! Shehyet «BRS eT 


a 16,0, i +f PS YEOL ONES See oar ,bceL A) rn | ixexisted cm 
re a. 4 -fonext =26¢, 


cs 
Satnseriovnt yvrotaivoises 160 obtauaos’ -TeOl .eeerd ,extceled 
aS-€.f sepzofd P 


srg 'wromt} euool adt bas no btéetooneson’ .03er + rea 
| AS-LES gold 


eexcd?t «teurtdeont BIS OSL & ipototerl evel -OfEmssi10t hae 7 


2 rad? 8 96 vemeD (eupeocam SURE scare ae ana 
ery’ yhfoms) :spbrideeD Noitecupod 


esorisictt nt . oom tistc nb -2 oes ET. v #@BOr nicakriae ee gi 


_ Bvene)  .V0-FaS: .gd .yited De sn tpaenhiae — 


Sonsioed. siiziconil to yerme A VOL -.bSo.a eee Ee, , 
eed ban lytem to ywiisusvind . $B ae ooo 


. ; J ig p ‘ 
Bdelvel ."oecones? s6 4aiygqog Fepsdannt’ . -SR6L or. 


of yeodns2 eo 
AT LO0,€ russ beak - 
sthseinaziast Sit! " leith ty mfoiibite Nejon ie abe ] 


pie f /B5 1. otsile ati poibekbans Seb «ecb 
“eadic? 2G Sea sprue = ve 


 %ads% 2 Yeds fhobrot “emus! cletnia: O3eE a amet 


Seemed TOT ney secQopet 






it etaSts td foie oer =, Rew 


| As oA tee | ce eed ei 
supiesre it suvet .'nsistso ub sand 9 oa ale PHA. ee 
; as Pel is i 
Of ai. "sautnst suisoattais 10. S36: ie 
TOSI 7aupSR ofl .SS-heS S04 


= ‘ ‘ “tel 2) 


i) 
ie | or 


ihe 
n ieee ae ge ws: aah 


Ferguson, Charles A. and Carol B. Farwell. 1973. ‘Words and sounds 
in early language acquisition: English initial consonants in 
the first fifty words'. Stanford Working Papers and Research on 
Child Language Development 6.1-60. 


Fischer-Jgrgensen, Eli. 1952. 'On the definition of phoneme catego- 


ries on a distributional basis'. Acta Linguistica (Hafniensia) 
7.8-39. 


Fishman, Joshua A. 1972. Language in Sociocultural Change. Stanford: 
Stanford University Press. 


Fishman, Joshua A. and Eleanor Herasimchuk. 1972. 'The multiple pre- 


diction of phonological variables in a bilingual speech community’. 


In Fishman (1972), pp. 162-178. 


Fishman, Joshua, et. al. 1968. Bilingualism in the Barrio. Blooming- 
ton: Indiana University Press. 


Florio, Giovanni. 1969. Florios Second Frvtes (1591). New York and 
Amsterdam: Da Capo Press. Facsimile edition. 





Foley, James. MSa. ‘Phonological investigations'. Unpublished MS, 
Simon Fraser University. 


. MSb. ‘'Nasalization as a universal phonological process'. 
Unpublished MS, Simon Fraser University. 


- 1970a. 'Phonological distinctive features'. Folia Linguistica 


4.87-92. 


- 1970b. ‘A systematic phonological interpretation of the Ger- 
Manic consonant shifts'. Language Sciences no. 9.11-12. 


- 1971. ‘Phonological change by rule repetition’. Papers from 
the 7th Regional Meeting, Chicago Linguistic Society, pp. 376-84. 


. 1972a. ‘Rule precursors and phonological change by meta-rule'. 
In Linguistic Change and Generative Theory, ed: by R. P. Stockwell 
and R. K. S. Macaulay, pp. 96-100. Bloomington: Indiana Univer- 
Sity Press. 











feo (20 Replys tOnCONeh mer abers a Lbom thesoul Reglonety Mee t— 


eS | 


ing, Chicago Linguistic Society, pp. 376-84. 


Fénagy, Ivan. 1963. Die Metapherm in der Phonetik. The Hague: Mouton. 


Ford, J. D. M. 1934. Old Spanish Readings. New York: Gordian Press. 


Fortes, Agostinho. 1944. 'Ndtulas acérca dum falar da margem esquer- 
da do Guadiana, acompanhadas de algumas noticias folcl6ricas'. 
A Lingua Portuguesa~3:128-37, 186-99, 217-25, 313-20, 353-60. 


Fouché, Pierre. 1925. La diphtongaison en catalan'. Buttleti de 
Dialectologfia Catalana 13.1-46. ie. 


225 


be yey 


Po a 
r 


















i. 
By 


‘ ‘> y) 
ee nM, pheow! abe ten = a on 
ih ecdrasry sattCO fazsént red Spel ‘2 





male 


? Mm. Tt TERE BH fens. tog pre cot oe t 
_ <a . j 


“wwatco, gremorky to seitiatiab erit mor ae 
. keris ps) ‘eoiteiowirnd SoA eine innokicited 


i os la ach | : 


: ® n : - 
?hagtast® =. apesd fatictt Imorsoe mt spayoner ST 
= OW : F; eeenrd 


ao hol 


i” ; | : tas cog, 


: 7a a t ro — i ul a = 58s —— —~ ra cal ‘ 7 aia cle: a 
| BQ sfgit tT yes . ViVILi2eten JORSSALY Sas . A sud aot Re Ls 


> 4. ’yskioummcc co: ferrantl et s) mu ah t= r Lsotpolondtig 20" noigoth <> 
: : ri-fet sq (STOR) ecole it 


7 - - P 
a eee ee ee ee eect cs 
= AAR | Ce fake Oh CRA BETES 4 18 «Gel 3 22 4" 5 ee 


- 4 . -~ 4 P 
— * un ne Pe 
-Seatd yYiteitev La “Bein bie tree 
a ( 




















@ 
brs . [) <eedvx . imott  .e6RL tansy 
r 4 yt sy ePerca 
QM botehickiant) ..'sncbtsotveantt Ledipotonoda" oaaMy ve 
, , — 
3 i (>yov in TORT co 
9 
F es a ! j ere 
V rd covaig 
ee ee 
Fert. ta — enn ge a ; 
> 
om ¢ 
cig 
“sited art to cmitese | ionorka ottamedsye a 
( et Be a ‘ad?id trstipenco 
seein - — - ¢ 
ee ae =% In 
tit so . notre ys Wa Serer [oo top locate” fT! sg ; 
Pere.’ . @FOos ostabrorid coaoid) Le fs mab ON a 7 
Pe iieemee anes “ ok — ~ pe mm ee z mat 7 
. Siiriredem yo sprert Sipolotide base seeps : abost" » aster Mas 
Trew yxy ah st ow } owes ~ A* Ma ¢ _ (3 f fetes oe} d vert). aide. ope ET ar if 
re ‘di oe —ernemp tate As 
at md: & ti .. The rx) ° -UUNL~be ay ma it rer i — ab 2% aes 
° , ae ae 7 ie 
S. ; 
R 5 4  h ize ' 
ee : z it2 I {+ i° 1s ib B 4 ET Scie z 'rreticd oc vian ' i [ a er 
&B aT ~ ~Weino® itparot 
/ sMGduCM ceubal a? 2Lvenoct 306 ak cee ett | .608E 
4 5 fi ‘ 
amet natin OY yet . eps ibaa Sige BO tee a A a Lo 


can - “teupas: nepyen- eb talet mb soxgosn esiiiiem are Oth 


2 ea REST Loie anioizoan asmpds, ab Paes MIO <6E Suif 
7 Py - (1 O062E OS-€! ey seS-TGS 00-38 oc a sees Pua 


‘ ce - P saesret sc 
> 


7a eae weit sue «rth BI65 49g leas zit al Ye er ot 


L@ as — ; : a 
7 a “$ ai a) a nes set 


Fought, John G. 1973. ‘Rule ordering, interference and free altema- 
tion in phonology'. Language 49.67-86. 


Fromkin, Victoria. 1971. 'The no-anomalous nature of anomalous 
utterances'. Language 47.27-52. 


Frumkina, R. M. 1963. 'The application of statistical methods in 
linguistic research'. In Hays and Mohr (1963), pp. 80-118. 


Fry, David B. 1959. 'Phonemic substitutions in an aphasic patient’. 
Language and Speech 2.52-61. 


Fudge, E. C. 1967. 'The nature of phonological prites'. Journal 
OL GINGULS ti Cs 23.1-30- 

Garcia de Diego, Vicente. 1961. GramAtica Hist6rica Catalana. 
Madrid: Gredos. 


Gebels, Gustav. 1969. 'An investigation of phonetic symbolism in 
different cultures'. Journal of Verbal Learning and Verbal © 
Behavior. 8.310-12. 


Giese, Wilhelm. 1931. 'Como os Mouros de Asfi grafavam o portugues’. 
Biblos (Universidade de Coimbra) 7.482-511. 


1964. ‘Zum span -ld- anstelle von arab. dad . Zeitschrift 
ables yomanische Philologie 807356-0; 





Gili Gaya, S. 1932. Estudi fonetic del parlar de Lleida'. In Misce- 
lanea Filolégica Dedicada a D. Antonio M. Alcover, pp. 241-55. 
Palma de Mallorca. 








Glossaire du Parler Francais au Canada. 1968. Prepared by the Société 
du Parler Francais au ara: la. Québec: Presses de l'Université 
Laval. 


Goncalves Vianna, A. R. 1887. 'Falar de Rio Frio’. Revista Lusitana 
aloe oor 


- 1906. '‘Quantidade prosédica das vogais em portugués'. 
Révue Hispanique 15.24-7. 


Grandgent, C. H. 1927. From Latin to Italian. Cambridge: Harvard 
University Press. 


1934. An Introduction to Vulgar Latin. New York: Hafner. 


— 


Granthan, Henry. 1968. An Italian Grammar 1575. Menston: Scholar 
Press. Facsimile edition. 


Grégoire, Antoine. 1937. L'Apprentissage du Langage. v. 1: les deux 
premiéres années. Liége. 


Gregorio, Giacomo de. 1932. 'Scoperta della genesi fisiologica delle 


226 


cosidette consonanti doppie'. Studi Glottologici Italiani 9.3-15. 


Act eny 


ay, ye 
2 af ; 1 

= 2S 

: 


sorpolorert 
sre LinsT f 


(JF enh 


ra J rh) 


Horst sb 
; : 


LBONEp Be. 
rolpolosgoS) 


t ri tel ebrne afte’ 
Sah geaeaeia 


clemchonny eft ~fVel 
Te sgauyanit 


vet. .o ade. here 
. ypalorordg at so 


-sixotoLy Aes ae 
. 'eaormsisisy 


oir’ .<00L. WM od tele in u | 
‘"oxesest otatopon tt 


Heo * | Brved wrt 
lé-é.$ fosege bow spssposd 


SSIS ~=Sett veer sd 2h Spb 
00-16 epigptpnid Bo 


= 
cree 


ei 3 


aver ri 


S to oOLTR 
rie tris 


. C2PT 


* J 
i J a. ahd 


foe4 stmpolV¥ .opeid ab sicrso 
: enber> 1binbeM 
¢ eantir cs .fe2L w.vedeuwd ,aledep 
< pnees sy , voy ivm toereziib 
SI-OL12 2 2obverted 
| ae reel wolelli EN .seelD 
(sek % oabsbissavints) eoldss 
Jy i I { woe ims * bael 


é 2 1 6 ne fe 2S Sf SOUerT wT 
©) /@ \syeo Lip 
pofos oLiS sans t 
sart ite & sh amied 


tpt ay as 


alisd ub otisaeold 
14 346ivsa gb” 
isvel 


nae 8 foustis fe < Ett ne2 


Cp ( A , sre iV aeisoncd 
.oa-Bel I 


1q ebebirtaseQ’ .d0@f . 
" maeta suv 


.. ,dnepihosx 


Resa yoiLexev ind 
.beeL * 


ee ae 


mi .8a8! «iene . meting 
Oia tps alimtpastl -2aeT 


ATEOL ,aerbodni. \entooeao” 
25ers seu Stmexg 


-ob cmapos.iD, .okxopsiD 
edtebLeco 


erp 


A\S@iL 


Tsoi ot moctowhenigs 7A 


Wes) Metisn, 


¥; 2632} bay SISTER, aed 


ar | “ah LI 


SO & 4 age, . . SEL 


thine .'sieqob iabnoenmgo 





eek 


Griera, A. 1913. 'Notes sobre'l parlar d'Evica i Formentera’. 
Buttletf de Dialectologfa Catalana 1.26-36. 


- 1931. Gramatica HistOrica del Catala Antic. Barcelona. 


ee 


Guarnerio, P. E. 1918. Fonologia Romanza. Milano: Ulrico Hoepli. 


Hadlich, Roger L. 1965. The Phonological History of Vegliote. 
Chapel Hill: University of North Carolina Press. 


Haiman, John. 1972. 'Phonological targets and unmarked structures’. 
Language 48.365-77. 


Hall, Robert A., Jr. 1950. 'The reconstruction of Proto-Romance’. 
Language 26.6-27. 


1963. Idealism in Romance Linguistics. Ithaca, N. Y.: 
Cornell University Press. 


Halle, Morris. 1954. 'The strategy of phonemics'. Word 10.197-209. 

1962. ‘Phonology in generative grammar'. Word 18.54-72. 

1964. 'On the bases of phonology'. In The Structure of 
Language, ed. by J. A. Fodor and J. J. Katz, pp. 324-33. Engle- 
wood Cliffs, N. J.: Prentice-Hall. 

Halle, Morris and Kenneth N. Stevens. 1964. 'Speech recognition: 
a model and a program for research'. In The Structure of Language, 
ed. by J. A. Fodor and J. J. Katz, pp. 604-12. Englewood Cliffs, 
N. J.: Prentice-Hall. 


Hammarstrom, Goran. 1971. 'The problem of nonsense linguistics’. 
Acta Societatis Linguisticae Upsaliensis 2:4, pp. 99-109. 


Harms, Robert T. 1967. 'Split, shift and merger in the Permic vowels’. 
Ural-Altaische Jahrbticher 39.161-98. 


Harris, James W. 1969. Spanish Phonology. Cambridge: M. I. T. Press. 


- 1970. 'Sequences of vowels in Spanish’. Linguistic Inguiry 
1.129-34. 


Hays, David G. and Dolores V. Mohr, trans. 1963. Exact Methods in 
Linguistic Research. Berkeley: University of California Press. 





Hensey, Fritz. 1968. 'Questdes de fonologia gerativa: as regras de 
pluralizac4o'. Estudos Lingiiisticos 3.1-10. 


Herculano de Carvalho, José G. 19624. 'Nota sobre o vocalismo antigo 
portugués'. Revista Portuguesa de Filologia 12.17-39. 


- 1962b. 'Sincronia e diacronia nos sistemas voc4licos do criolo 
caboverdiano'. In Miscelanea Homenaje a André Martinet, ed. Diego 


Catalén, v. 1, pp. 43-67. La Laguna: Universidad de las Canarias. 
















v.¥ .4 ,BOefST . Bok Fee: 


COEF I .00 fxoN sap or to yosemite, oft! bee enn 

















8-82, 8. bioW ‘semmexp sy.bisiensp me ypalanants' See =e mee > ae he 

me | 7 = a 

te, swaps adi pl ..  wecaenit to coand odd 7 ae 
wolted ” “E2-E Voy fsce® .U 2b Gomi aonod .A LU ye abe ae 


_ diated ws <0. M4 @22 

mit! rmpoost ieeqe' 680!  ,aicaste- 4 ea Be sat ta : 7) | 

133 ees Tore a aie aT > a ett 3 su 6! 
tteaivonts eeneanan to maidosq. ap ' 


-BOESEE Ga \h:8 a sso kigiugs 


. ‘afewov oii! oft ni seers Bos site Bee -VaeE AP todo 
-BE-laf PF . - 


Sapexs oP LI LM soph rrdns> pot onoid oie, 
Ytivpal sce tipredt 






> 


at . » .£000 canst? (een Vv Rese 
; ElMOtELsS tO. vhiax=tst! ‘whee 


Sh asipst as cavisergp Sivofow? « 
: -Of-L.£ sonia 





Hjelmslev, Louis. 1961. Prolegomena to a Theory of Language. Trans. 


by F. J. Whitfield. Madison: University of Wisconsin Press. 


Hockett, Charles. 1958. A Course in Modem Linguistics. New York: 
Macmillan. 


Houaiss, Antonio. 1956. 'Tentativa de descrigao do sistema voc&lico 
do portugués culto na Grea dita carioca'. Anais do Primeiro 
Congresso Brasileiro de Lingua Falada no Teatro, pp. 217-317. 
Bahia. 


Householder, Fred W. 1965. 'On some recent claims in phonological 
theory’. Journal of Linguistics 1.13-34. 


- 1966. ‘Phonological theory: a brief comment'. Journal of 
Linguistics 2.99-100. 


1967. ‘Distinctive features and phonetic features’. In To 
Honor Roman Jakobson, v. 1, pp. 941-44. The Hague: Mouton. 





Huber, Joseph. 1933. Altportugiesisches Elementarbuch. Heidelberg: 
Winter. 


Hunting, Claudine. 1973. 'La voix de Rimbaud: nouveau point de we 
sur les "naissances latentes" des "Voyelles"'. PMLA 88.472-83. 


Tordan, Iorgu and John Orr. 1970. An Introduction to Romance Linguis- 


228 


tics, its Schools and Scholars. Revised, with a supplement "Thirty 





years on" by Rebecca Posner. Berkeley: University of California 
Press. 


Isbagescu, Cristina. 1968. El Espafiol en Cuba. Bucharest. 


Izzo, Herbert J. 1972. Tuscan and Etruscan: the Problem of Linguis- 


eee ee ee ——_———— 


tic Substratum in Central Italy. Toronto: University of Toronto 


eee 


Jakobson, Roman. 1939. 'Observations sur le classement phonologique 
des consonnes'. Paper read before the Cercle Linguistique de 
Prague, 1938; later presented at the 3rd International Congress 
of Phonetic Sciences (Ghent, 1939). 


1941. Kindersprache, Aphasie und allgemeine Lautgesetze. 
Uppsala: Almgvist. 


Jakobson, Roman, C. Gunnar M. Fant and Morris Halle. 1963. Prelimina- 
mIeSttOropeccht Analysis. Cambriage: M. I. Tf. Press. First 
published 1951. 





Jespersen, Otto. 1922. Language: its Nature, Development and Origin. 
London. Sn) = 


Johnson, Ronald C. 1967. 'Magnitude symbolism in English words'. 
Journal of Verbal Learning and Verbal Behavior 6.508-11. 


med 


Bp somaecs ct pOLiouborsol 


vias” sdsmeiqove 6 iw hechven -exsiorie bas af pL okt" 
we 


Jjeorsioed edu ne Lofeons [i .8eRt -sriteixd ynaegsdel 
—— a s a <6 @ 


ess. i: ak ie 7) ake 












. , is 
a is sag : _ yee a ; 


sanexT Ee teal 36.1 ot tS Sos __ -ainad walk 
 Beeat rife Pv ie a ae , oe «SESE hac’ 


:7OY WH wep iveinpnts ietett tk send & HERE 





ooiffocyr aradale ob eee 7 svisesngr' .deer 
» shite BexBsa offer 


orkid Gb eienh _ . ‘soe. 
WEE-TES qa ortbet on via tae eared 2 css a — idee 

























ax 
OH 


ee e 





faotpoltomedq ai aiial> trieoex anpe o' cael .W Bex, ov ale 
$€-£1.1 apmtelupnet Jo Taemet - ptoeds ; re 
af 7 - 


u . . 
46 famoot. .'dremme ted s + \yrost lanipelonaty” oer” . P 
igo 001-€0.$ godbetempait 
nk  -taer Ns 
Of cml .‘aensgsesi orjormag Ore pene wt svidsnitvers?? ~Ve@L =. = ie 
\SOIGGM ‘seermeHott = .53-1D° qo ol wv \tosdii mame at 
ae 
‘PidiehisH .chattrsjosmel’ sation ies pus ogi tA -CERL .figgaot i 
. — a we ia one te - Tad — —) - 
= 
em 


- - ' s La 7 

ay ab stitiee tecyuen -Euetuid sb = ton si* fT@f enkbeelO: » Era : aoe 
’ La oan 

ESTSVE.ES AUN "Rel iovev" ecb “setriete { asonpaakea” eat ae. of 


= 

ni =.00Gl 230 iniiots fixie uprol mpeg 
+ retest ay irl 2 SA honed ; Rist are Ls) SIBey 
easyd 


wmiumcid to mefdoxl extt Cease bas neoevT .St@h 4G precer asst 
nono! 2 Valeisvind oder Nipsy Eo: Sei) ctf << 
mse al 
siptpolonon totmmersin of we eaoitevisedD" .26eL ae 
s6 supttaiean! sles) ait enoied haar eagt-”. ‘earw@oencoD ash” 
apsepnoD lamoitmrisial iat oft Js hetndaarg tate! «S6eL 13 


(CECE Sho) esons. bow es 
@ 


gd oagptuR qadanoits bin! SLest cigs a e ee 
u = 


f 


L _ 7 
“Soinites?. .£80F .olisi 2hrichos, taal .M Tepe .2 > se cost 
") Sen's  .zeet 21.7 a :epb liam) «'. ot asit ” 


229 


Johnson , R. C., N. S. Suzuki and W. K. Olds. 1964. ‘Phonetic symbo- 
lism in an artificial language'. Journal of Abnormal and Social 


Psychology 69.233-36. ti pat 
Josselyn, Freeman M. 1900. Etude sur la Phonétique Italienne. Paris. 


Juca (filho), Candido. 1961. 'O factor psicolégico na mutag&o vocd- 
lica portuguesa’. Boletim de Filologia (Lisboa) 18.143-62. 


Kent, Roland G. 1945. The Sounds of Latin. Baltimore: Waverley Press. 
3rd ed. . 


Kim, Chin-Wu. 1966. 'The linguistic specification of speech'. Ph. D. 
Dissertation, U. C: 1. A. 
- 1968. ‘When liquids become glides'. Chicago Journal of 
Linguistics 2.45-50. 


. 1970. 'Two phonological notes: A and BP'. Mimeographed: 
Indiana University Linguistics Club. 





w= L971. —~'Experimental, phonetics")... In Dingwall (1971), pp. 16-135. 
King, Robert D. 1969a. 'Push chains and drag chains’. Glossa 3.3-21. 


1969b. Historical Linguistics and Generative Grammar. Engle- 
wood Cliffs, N. J.: Prentice-Hall. 


Kiparsky, Paul. 1965. ‘Phonological change'. Ph. D. dissertation, 
M. i iaec. 


. 1971. ‘Historical linguistics’. In Dingwall (1971), pp. 
577-649. 


. 1972. ‘Explanation in phonology'. In Goals of Linguistic 


Theory, ed. by Stanley Peters, pp. 189-227. Englewood Cliffs, 
N. J.: Prentice-Hall. 


Kisseberth, Charles W. 1970. 'On the functional unity of phonological 
rules'. Linguistic Inquiry 1.291-306. 


Klausenburger, Jtirgen. 1972. French Prosodics and Phonotactics. 
Tubingen: Max Niemeyer. 


Kuen, H. 1932. 'El dialecto alguer y su posici6én en la historia de 
la lengua catalana'. Anuari de 1'Oficina Romanica de Lingufstica 
i Literatura 5.121-78. 


Kuipers, A. H. 1968. 'Unique types and typological universals'. In 
Pratidanam: Indian, Iranian and Indo-European Studies Presented 
to F. B. J. Kuiper, pp. 68-88. The Hague: Mouton. 








Kurath, Hans. 1964. A Phonology and Prosody of Modern English. 
Heidelberg: Winter. 












, Paes 


weds ofsonON MOCK .2bf0. (8 Ahn Oe “8 BE sii i xh 
| Egtoe® bas. Semmongit Yo Simran, +" Spekhe " e888 Spe se 


7 ao of in | ate a : 
si ila nin a 008! 4 k oe 

“Roo OF xedem sri pening ae 2 

SOR 84 (soi) te 0 wohl 


ne ua ay ‘ 

seont yelraviw : einTe tet to 'ah baie: a wok . Jot 
-be Bx . SLs a 

~ : _— _ “FOta eal ‘4 an 

.GA .'doeces 20 missoftiosqe Siem eat Pay enn nb 
Ad 3D vpbtatrseaia ~~ 


to Li waa ops... ‘aeitip groped ehcp et yr ge B9CF . 
a ee eres ee Nir 2 Betpald ~~ Da 


Sete yrosanit | lg bas *A” tBacon Pe eOveL <a 3 
{9 selon yhbemewia spatial ¢ 
7 - : e24 


OCeOL co . ({02l) Olswonid m1 eo itonmotiq Satnstireex’ VOL” 7 aes i 


5 


























: 


he 
-ES-€.£ srecoin .'‘eakero path bos eated> gest”  /eeeer el ste as a 


toe) mee 2eES tet +14 3 
f eH es we APS « boo boo = 


,aorteytiseerd .G ind. Sores fecipefonet” (ageE dea elke 
Pi i Fp 2 M : 


~efoud "Sai SOS 





.aq , (NGL) Sfewortd ot .'eoitampetl so mosea® VEVer a ee re a 
-epa-tte 
a ee 


Weal: 


oiteiupnit Zo aisod nt . .' ypofdpolig: rolsess tga" eSVORey 
yr) booveipn® .YSS-sf .qq ,aneted yatngse yo ibe bee | 

foe Sn, eS 

= ata — 


iso rpeLonoric to. tiny asim or - mm! ,O0RE ow 


ONE ALO. f yout 
sp Qostorad? bas pes thiert ".<ver — 










oe Binvoteisé el res jimi Ue YY teupis ajoeleib oe seer - 


» pital: sb se aioe eni9i30'5 ia anacla 
| eee a 


° 
nr 


230 


Labov, William. 1966. The Social Stratification of English in 


te 


New York City. Washington, D. C.: Center for Applied Linguistics. 


1969. 'Contraction, deletion and inherent variability of the 
English copula'. Language 45.715-62. 


- 1970. ‘The study of language in its social context’. 
Studium Generale 23.30-87. 





1971. 'Methodology'. In Dingwall (1971), pp. 412-97. 


- 1972. ‘The internal evolution of linguistic rules'. In 
Linguistic Change and Generative Theory, ed. by R. P. Stockwell 
and R. K. S. Macaulay, pp. 101-71. Bloomington: Indiana Univer- 
sity Press. 


- 41973. 'The linguistic consequences of being a lame'. Language 
Hanecociety Zee lah 5: 





Labov, William, P. Cohen, C. Robins and J. Lewis. 1968. 'A study of 
the non-standard English of Negro and Puerto Rican speakers in 
New York City'. Cooperative Research Report 3288, v. 1 and 2. 
New York: Columbia University. 


Ladefoged, Peter. 1966. 'The nature of general phonetic theories'. 
Reprinted in the Georgetown University Round Table Selected Papers 
in Linguistics 1961-1965, pp. 283-298. Washington, D. C.: George- 
town University Press. 








- 1969. ‘The phonetic framework of generative phonology’. 
Unpublished MS, U. C. L. A. 


Langaker, Ronald W. 1969. ‘Mirror image rules II: lexicon and phono- 
logy'. Language 45.844-62. 


Lapesa, Rafael. 1968. Historia de la Lengua Espafiola. Madrid: 
Escelicer. 7th ed. 


Lass, Roger. 1971. ‘Boundaries as obstruents: Old English voicing 
assimilation and universal strength hierarchies’. Journal of 
Linguistics 7.15-30. 


Lausberg, Heinrich. 1956. Romanische Sprachwissenschaft. Berlin: 
Walter de Gruyter. 


Leite de Vasconcellos, José. 1890a. 'Dialectos Alemtejanos'. 
Revista Lusitana 2.15-45. 


1890b. 'Dialectos Trasmontanos'. Revista Lusitana 2.97-120. 
1896a. 'Dialectos Alemtejanos'. Revista Lusitana 4.215-46. 


- 1896b. 'Dialectos Algarvios'. Revista Lusitana 4.324-38. 


tah + tne 













mm .*xelire ottaigeni 
sable 4A yd be oes ‘ 
“SOyiA) stsibel 9 snooprnS imooka Lf 


& } * 
: : : . ie ; : te en pm : 
SGU PET j*omsl s pried io 2SonsreamoD DET: a ¥ ry _ a & - 
a | 261-18. | ‘ 


20 yiecte A’ .80Cl- .alwal..0 bos eftiod 2 eae goes, ome 
ai @xepidece asoth otiexd bos ope Jo £ one a 
SS Bits | .v 8086 syroget reussseh esse AsoY we i 
ile eon (4180) :20Y w + 
.aeitost aisenmiq [e1sisp lo snes oom! 


= hettaolad. slant frat aa ; 


2.) -¢ \ACponineaW . .Bec-EBS 
tO rowan xt 
fh wh A ® -U (eM { 


“Srodg fxs coultel ‘f1l esl spect sorgim’ Jegeh a ea ae 
SOC PAB SS SRSRAST « ae = 


(HivieM .sloregel supinal si ab nize ae 2 


. *ypolomoiiq aiiictoisp 





grioioy datiprd’ bfO -+sriecrwetio 28 3 
ao femme .‘ asinoisisid epasie: 


abiied.. . Aproartaes|nbege Besieii re 


.aonstetmalhA anxtosfeid’ OO 


Lt mi esol 
OhrEIS 8 ae See 
/BENST.E smadlen) siaiven Ne 


- 1899. 'Phonologia Mirandesa'. Romania 28.598-620. 


- 1901. lLigdes de Philologia Portuguesa. Lisboa. 


caawe 


1970. Esquisse d'une Dialectologie Portugaise. Lisbon. 2nd ed. 


Lightner, Theodore M. 1970. 'Why and how does vowel nasalization 
taker places (Papers in Linguistics 2.179-226. 


Lipski, John M. 1973. 'The surface structure of Portuguese: plurals 
and other things’. Linguistics 111.67-82. 


Lleé, Concepcién. 1970. Problems of Catalan Phonology. Seattle: 
University of Washington Studies in Linguistics and Language 
learning, v. 8. 


Longacre, Robert E. 1969. ‘Hierarchy in language’. In Method and 
Theory in Linguistics, ed. by Paul L. Garvin, pp. 173-98. The 
Hague: Mouton. 


Louro, José Ines. 1952. 'Origeme flexdéo dalguns nomes portugueses 
em -&0'. Boletim de Filologia (Lisboa) 13.37-65. 


- 1954. 'Estudo e classificacéo das vogais'. Boletim de Filo- 
Vogt aie] s215—48). 


Ma, Roxanna and Eleanor Herasimchuk. 1968. 'The linguistic dimen- 
sions of a bilingual neighborhood'. In Fishman et. al. (1968), 
pp. 349-464. 


McCawley, James. 1967. 'Ie role d'un systéme de traits phonologiques 
dans une théorie du langage'. Langages 8.112-23. 


McNeill, David. 1970. The Acguisition of Language. New York: Harper 
& Row. 


Makoto, Hara. 1971. ‘The invariant factors in Spanish diachronic 
phonology'. Gengo Kenkyu no. 58.20-38. In Japanese with an 
English summary on p. 38. 


Malkiel, Yakov. 1960. 'Paradigmatic resistance to sound change'. 
Language 36.281-345. 


- 1962. ‘Weak phonetic change, spontaneous sound shift, lexical 


contamination’. Lingua 11.263-75. 


- 1963-64. 'The interlocking of narrow sound change, broad pho- 
nological pattern, level of transmission, areal configuration, 
sound symbolism: diachronic studies in the Hispano-Latin conso- 
nant clusters cl-, fl-, pl-'. Archivum Linguisticum 15.144-73, 
ANS SMS Joie 


- 1964. ‘Distinctive traits of Romance linguistics'. In Lan- 


guage in Culture and Society, ed. by Dell Hymes, pp. 671-88. New 
York: Harper & Row. 








soda en 
bors .adzhi .gpistats0% ofpeliale 


norms ileasty laws soot-wrt 
pee=er doo 





















alsarilg pasEMUT OTRO ese 

Sea, Sak - 
tefkies? § .ypolonod peter } 

5S spanpomd € bre ettahioais ria 


‘s boijeat ol (ct tlie ar page a Dh tn 
OE CL .q nevis fc Eel ya. is 


BeReULUIIOY ZaNON anipish of*xsht o imsseahar’* 


-Ba-TE.EL (sodas) 2ipeioe 6 migaiog -s- 


“OLiT sb mitelot .'=ttpay sb cb) soitiaaats Ss dbersett* heer 
a -Bb-248.2 ‘sa 
an sisettipail of?’ ‘Baer bas: pay ag re 
WOOL) Le ite nemetT at ‘booriedelp tn & 30. 

Reet ‘7 eee a 
= waren — z % 
eeupipolonodia ejisst ah anfteve ou"h ofor ax’ ie ‘talendh @ 

-ES-S51,.8 anpepet .'spnpcdel ub a sna iy ; 


weqisH itcY welt .speupasl Fo tol eee. IRE. ive sitio Le 


ofnonthbe th detmmead ni -aeicdéost tris hevik err’ Poe 
Mi mifw sascedet al. .8€-08.8e ean a . 


.spnario Exivoe od sonmdaiesr steers 


fepixel .jtinte Bruoe viet 4 Spoed sta 
ba at 


Suites ead spac Rance soexen 86-4 
nottexne £00 


232 


- 1966. '‘'Diphthongization, monophthongi zation, metaphony. 
Studies in their interaction in the paradigm of the Old Spanish 
-ir verbs'. Language 42.430-72. 


1968. ‘The inflectional paradigm as an occasional determmi- 
nant of sound change’. In Directions for Historical Linguistics, 
ed. by W. P. Lehmann and Y. Malkiel, pp. 21-64. Austin: Univer- 
sity of Texas Press. 





- 1969. ‘Morphological analogy as a stimulus for sound change’. 
Minguaceyocllee37305—-2/. 


1970. ‘Sound changes rooted in morphological conditions: 
the case of Old Spanish /sk/ changing to /@k/'. Romance Philology 
2308-2 0Ue 


Malmberg, Bertil. 1955. 'The phonetic basis for syllable division'. 
Studia Linguistica 9.80-87. 


Maltzmann, I., L. Morrisett, Jr. and L. O. Brooks. 1956. 'An investi- 
gation of phonetic symbolism'. Journal of Abnormal and Social 
Psychology 53.249-51. 


Marchand, Hans. 1959. ‘Phonetic symbolism in English word-formation'. 
Indogermanische Forschungen 64.146-68; 356-77. 


Marigo, Aristide, trans. and ed. 1957. Dante Alighiere: De Vulgari 
Eloguentia. Firenze: Félice le Monnier. 3rd ed. 





Martinet, André. 1955. Economie de Changements Phon&tiques. Bem: 
Francke . 


- 1966. ‘Structure and language’. Yale French Studies no. 36-37, 
poe lU—se 


Mattoso Camara, Joaquim, Jr. 1957. 'Erros de escolares como sintomas 
de tendéncias lingiifsticas no portugués do Rio de Janeiro’. 
Romanistisches Jahrbuch 8.279-86. 


Melhem, David H. 1973. ‘Ivan Fénagy and Paul Delbouille: sonority 
structures in poetic language'. Language and Style’.6.206-15. 


Mendeloff, Henry. 1969. A Manual of Comparative Romance Romance Linguistics. 
Washington, D. C.: Catholic University of America Press. 


Menéndez-Pidal, Ram6n. 1964. Orfgines del Espafiol. Madrid: Espasa- 
Calpe. 5th ed. 


- 1966. Manual de Gram4tica Hist6rica Espafiola. Madrid: Espasa- 
Calpe. 12th ed. 





Menyuk, Paula. 1968. 'The role of distinctive features in children's 
acquisition of phonology'. Journal of Speech and Hearing Research 
11.138-46. 


pene oie bprtoteciefntani , ae 
dabrsce BiO ‘add to mpibstec. orth, ar er aa 


5S eaten - 


~innietsh Laaciemeo Ab as tap chenang, nee ac {8208 r 
(BD. Se Lip tT Iso inate ty 10t gotta st ‘apastio Bros Yo tog 

Wage” saris Bo-l. - ay. fore 6 ‘bas nsnried .7) veg be oo y 
) sagt eax? io wie 


. ‘Sermo fenioe Tot siete wide 6 ee ypotens feokpoforgacm”  5@eRQL ees 
R-20€.0 Okive @ kee 


eAsttitcco Esobesict trent ft Bedeps ebopetio Ipyce*’ OFeL 7 


=: erm \ao, oo oiiocedo \le\ rleing? BIO 2o, oaso act 
-OOS-881.ES 

























inoreivih 4fsifye tor ekead pitepodg ort see! .Lbtred ,predniat 


~jtesveni c4' gc8} siecsh .O .0 Bre .at Jeena wi ,el feos Les 
Taesce bas Lencrond to fern . matiodmve sivenotig. to noitsp ~~. 
| [8-868.€8 ypolodoyed ‘ 


“anesanrrol-row detipaa oo tte ame Seicatarsl .C@Qs  eneH , basco rem - 


(C-32r ;S0-860,00 nepiiog et appetimantepabed é 
—_—— eon = - 


nigidptlA ctypd .T2h  .bo bes) enters) yaneee ee era f 
Bice Dt a = ef sarlea romney yo} | : 


<imrist 2RIDIF Sons asrniontspni ) ob STK oa .2@@]  .etboA ,JanitteM 

' soormsrts — > 

WE=-86 .on sotOngs ronsTi Sissy. epeupial rs autre” ,ee@L at 

. ae i AE-OL .Qg oF > 

a0. 4b mbupsob eam) ceodtjam 

t erostetpntt 26 tones eb = 
.o 


(35-P08.8 Auten coche isalneey, 


yviixonos -:effamdied fust Sas vp SA movi’ .E98L ff Sived eae: 
~Cf-008 .5 ‘sf vie Sas opsypasa + ee oktang, o£ ener re.. 


aay Owes —_od 


~igelupnit sogamni avissrsgneo to a & .27eL .yanen oneal sotabeaat 
Seatd BOf<anmA to Wrewvind OLiahes 7.0 v4 abner 


qgenges ;briBaAM .Loliecalt fob aanikpti hae! ——— bec 


a= | ee 


Pa = 


-sas0as -binhe ‘.sloisged solstdeih saissim “a 
a “As Bh ee 

e*nexblino mrvaenicdoelt arisoatsere to, efor set" -2a0l led .zuyn 
Spee eae he Sa ae NE, Ph ntpeebar 3-7 pop ae 


if 
‘ i ay = 7 










pie ae - 






- 1971. The Acquisition and Development of Language. Englewood 
Cliffs, N. J.: Prentice-Hall. 


Messer, Stanley. 1967. ‘Implicit phonology in children’. Journal 
of Verbal Leaming and Verbal Behavior 6.609-13. 


Metz, Clara. 1914. 'Ein experimentell-phonetischer Beitrag zur Unter- 
suchung der italienischen Konsonanten-Gemination'. Vox 24.201-70. 


Meyer-Litbke, Wilhelm. 1890a. Grammatik der romanischen Sprachen, v. 1 
(Lautlehre). Leipzig: Fues's Verlag. 


- 1890b. Italienische Grammatik. Leipzig: O. R. Reisland. 
- 1925. Das Katalanische. Heidelberg: Winter. 


- 1967. Grammatica Storica della Lingua Italiana e dei Dialetti 


Toscani, trans. by Matteo Bartoli and Giacomo Braun. Torino: 
Ioescher. 2nd ed. 


. 1968. Romanisches etymologisches W6rterbuch. Heidelberg: 
Carl Winter. 4th ed. 


Michelena, Luis. 1972. 'Color y sonido en la lengua’. Revista 
Espafiola de Lingtiistica 2.83-102. 


Migliorini, Bruno. 1963. Storia della Lingua Italiana. Firenze: 
Sansoni. 4th ed. 


Mila y Fontanals, M. 1876. '‘'Phoétique catalan: @'. Révue des 
Langues Romanes 10.146-7. 


Miret i Sans, Joaquim. 1915. Antics Documents de Llengua Catalana: 
v. l-reimpressi6 de les Hamilies d' Organya. Barcelona. 





Miron, M. S. 1961. 'A cross-linguistic investigation of phonetic 
symbolism'. Journal of Abnormal and Social Psychology 62.623-30. 


ea 


Moffatt, Lucius G. 1948. ‘Considerations on the interchange of -ou-, 
-Oi- in Portuguese'. In Mediaeval Studies in Honor of Jeremiah 


rn 


Denis Matthias Ford, ed. by Urban T. Holmes and A. J. Denomy, pp. 


—$—— 


161-73. Cambridge: Harvard University Press. 





Moll, Francisco de B. 1952. Gram&tica Hist6rica Catalana. Madrid: 
Gredos. ns oa ae PS 


Morin, Yves Ch. 1972. 'The phonology of echo-words in French'. 
Language 48.97-108. 


Moura Santos, Maria José de. 1964. 'Rémarques sur deux systémes vo- 
caliques anciens du portugais du nord'. Bulletin des Jeunes 


Romanistes 9.5-12. 


233 


bowwslipnad .apsnpred to Jriengateyed 





terol .'maxblids at wpolonatt diokignt' .Voer johbenda ‘sroenett 
panei’ £1-203.3 ohyede a 


“adds ais pew bstt torbe Lisnerig--{faseamiveges cat" bL@L .6x6ID, , 
O\- FOL. aS xv feos Ac OF =P Lark bist @), I rtafpeiaebilets xeb Dp i 


[ow yr DE ge nemoaiteuor wah 2h NINES 7) stot .miladl iw ,eoldiI—reys 
oh il ale aad “usiiay esau :pisqial . (eudsbjom)) 


Basluret 5 .0  zpisqial oitteimew sipetoatiegs dest . 


———— ~ —— 


yedniw: :pretfabioh -orpabasheiet wet, .eSOt 7, 


—e 


idaeterc feb © atetisiI sel ehlad SIE a bist TENSE) eel 





ewradioh nuchetiGN sorberpotlomcis Banos toga .89eL ., 
bs cf8 Stetaiw [x80 


eretve . ‘ein ince y 2o0fo0" ..SVO@L - seal , srelerhM | 


psaneut BELeIT sept Lleb einose .€08L One ,tarrotipia 
; .bo fhyh =. Enoanse 


nai apie ; 7 =D 


cletseo aypitsnont vet .M valenetaot y 6Lia 
V-c51.01 saree eeupoet , 

sls ginmell ab atasmnod soida af¢ mimeo ene i toriM 
esti! ois fyni0'b S83) ion esl oh Gieeamenternf av 

: orks Jo ootraenigaeve: sitesyaiti-eanm A’ .feel .2 5.4 ,.nom 
NE“ES0.89 Yeolotbve? Lerood bas .AttonaA 2o An ‘uct.  . ‘euetLodnys 


“~~. 30 sonsdoxwedh iO. atoitexzsbianc’ JBReL 2 eae (soe race 
felieras TO TON url eos ea us Bi wink SOIT IG ni. -LO- 


ag axed A ono sadlon .T i670 va ibe EXOT es at okie 
S269) Whisvev ine Bieyssh “spb , aa 
HiwbéM .ensisse) soivtieis suitans <Seel .& 6b qpekoaare, ties 


‘rire os ebrowsctis to yoelonodg eff <STGl . 2 geve, mito 
80L-F2.86 spmepast 

-oy goméjave xiteh wwe asus’ hoe! Jab Sank sitvett Ph aka 

zentreb 28h piretivd 'hron ub axcopucteg, ob esaions eaupiiso 


ps ‘i 8 sasehamacdl a 





Gs 





23h 


Mulja“ié, Zarko. 1966. ‘'Due analisi binarie del sistema fonematico 
2 aan inguare ot le ls 265-70. 


- 1969. Fonologia Generale e Fonologia della Lingua Italiana. 
Bologna: I1 Mulino. 


Naro, Anthony J. 197la. 'Directionality and assimilation’. Linguis- 


bt Ce Inquiry @2.o 7207). 
- l197lb. "The history of e and o in Portuguese’. Language 
Ay eoOlo—=45 


Navarro Tomas, Tom4s. 1916. 'Cantidad de las vocales acentuadas'. 
Revista de Filologia Espafiola 3.387-408. 


- 1917. 'Cantidad de las vocales inacentuadas'. Revista de 
Filologfa Espafiola 4.371-88 


- 1966. Manual de Pronunciaci6n Espafiola. New York: Hafner. 
6th ed. 


- 1971. Manual de Pronunciaci6én Espafiola. Madrid: Instituto 
Miguel de Cervantes. 16th ed. 


Newman, S. S. 1933. ‘Further experiments in phonetic symbolism'. 
American Journal of Psychology 45.53-75. 


Newton, Brian E. 1972a. ‘Loss of /r/ in a modem Greek dialect'. 
Language 48.566-72. 


- 1972b. ‘Are drag chains simple?'. Linguistische Berichte 
19.38-44. 


Nietzsche, Friedrich. 1967. The Will to Power, trans. by Walter Kauf- 


man and R. J. Hollingdale. New York: Random House. 





Nobiling, Oskar. 1903. 'Die Nasalvokale im Portugiesischen'. Die 
Neueren Sprachen 11. 129-53. 


Nunes, J. J. 1902. '‘Dialectos Algarvios'. Revista Lusitana 7.33-55. 
- 1945. Compendio de Gramftica Hist6rica Portuguesa. Lisbon. 


Orkin, Mark. 1971. Speaking Canadian French. Toronto: General 
Publishing Company, Ltd. 


Osgood, C. E. 1960. 'The cross-cultural generality of visual-verbal 
synesthetic tendencies'. Behavioral Science 5.146-69. 


Otero, Carlos P. 1971. Evoluci6n y Revoluci6én en Romance. Barcelona: 
Seix Barral. 


Paducheva, E. V. 1963. ‘Information theory and the study of language'. 
In Hays and Mohr (1963), pp. 119-79. 





















“or GS gt ner eo 


Cat 7 _ 7 
7 - ag Le ? + 
3 Ul ak 
) wa : _ 





«a Py Bae) 
ar 
' 






bons CU -2aC LL Site et pemeht CMO Cae 
; : F 4s ew a ar 6 et aie cl is 
EOBGUES sue, ELieh siphons? 2 SERS RT SORE UY LNT 
ae * en >. 


: 
c 
‘J 


4 


, 
speuprs] . ‘sexfnot ai. o bis ¢ to yaaetd sat is as 
2 
: a 
. eeinacicoore. ealsoov asl ab Bsibitned 
; BOR-VBE.E Rect niotadas stpodoeht ob mind 


eb etaivel .'anhasinedanit esieoov eel sb babigns>’ .TLeL a. 


S8-10E .b sloneget 7 
s3E2SH = erOY wos .sLofuecea A MO Ems LOUISE & isunes . pres: ‘ . 


~Biypaid P EA he ees ee , | . Twcee 


Geitisent bie .Slotisqed ice iomees Fy ie _ 


5 , jot mh a 7 : , 
: ‘mas Llodmnye oitenony of etnemirsats sacral £€ei .2 “om i 
) “CV Ee. ct YEsLOnGyeS 20, Aerts Barre is | t 
Ps ‘S5e.0e16 tesa.) mrabom 5 fis \a\ to aaat* SST OL F. 4 nasiza re , : 
SY 002. 8b 

eiijined scoaitainoniit .'Tsiguie entero par sa” LeSver se : 
_+BD-8E. Lt + s 
~306N tatish ah  etteert see SLiw or .TAes rh bheirt , celicsiets 7, 
nem eee eee -Sisbprifion .G 48 bas feat > 3 
SKI. ,'norherestpusxod mi slseiovises eid! £001 pps wa 
2 St Sa a 


-226E.¥ gostiarn stabvet "watintelee iicoanclt pie, Pi 
toda estrus! Bob aaintgeit 0 soci amet eb storage 








fsthov-leveny Io vthisronep Ei “2 
sRO-ebLl, ¢ soneitod 


§ 


230 


Palmer, L. R. 1954. The Latin Language. London: Faber & Faber. 


Parmenter, C. E. and J. N. Carmen. 1932. 'Some remarks on Italian 
quantity'. Italica 9.103-8. 


Parodi, E. G. 1907. ‘Sul raddoppiamento di consonanti postoniche 
negli sdruccioli italiana’. Romanische Forschungen 23.755-75. 


Patterson, William T. 1973. 'A genealogical classification of 
Spanish words'. Linguistics 114.83-8. 


Pei, Mario. 1941. The Italian Language. New York: Columbia Univer- 
Sity Press. 


Pereira, J. Arteaga. 1915. ‘Textes Catalans avec leur Transcription 


ee 


Phonétique. Barcelona: Institut d'Estudis Catalans. 


Petrovici, Emil. 1957. Kann das Phonemsystem einer Sprache durch 
fremden Einfluss umgestaltet werden? Zum slavischen Einfluss auf 


—— ee 








Phelps, E. 1972. 'Catalan vowel reduction--alpha, braces or angled 


brackets?'. Linguistic Inquiry 3.246-9. 
Pilch, H. and R. Hemmer. 1970. 'Phonematische Aphasie'. Phonetica 
22523139. 


Politzer, Robert L. 1954. 'On the development of Latin -1l- to —dd- 
in Romance'. Modern Language Notes 69.325-31. a; 


Pope, M. K. 1934. From Latin to Modem French. Manchester: Manches- 


ter University Press. 


Popper, Karl R. 1959. ‘The Logic of Scientific Discovery. New York: 
Basic Books. 


Posner, Rebecca R. 1961. Consonantal Dissimilation in the Romance 
Languages. Oxford: Basil Blackwell. 


. 1967. ‘Positivism in historical linguistics'. Romance Philo- 
MOgy «202 321-31. 





Prado Coelho, Jacinto do. 1946. 'Para o estudo da pronuncia do portu- 
gues medieval'. Revista de Portugal 10.217-21. 


Pulgram, Emst. 1970. Syllable, Word, Nexus, Cursus. The Hague: Mouton. 


Pylyshyn, Zenon W. 1973. 'The role of competence theories in cogni- 
tive psychology'. Jourmal of Psycholinguistic Research 2.21-50. 


Ramasubramanian, N. and R. B. Thosar. 1973. 'The role of articulation- 
based rules in speech synthesis'. Phonetica 27.65-81. 


Rapoport, Anatol. 1972. ‘Explanatory power and explanatory appeal of 
theories'. Synthese 24.321-42. 


BES | seu 




























att 8 eth staid Sppereett ign gat et 
ad ¢ r : re, ants 28 


ghtisdl fo wanes ame’ .Se@l  .aemesd uw u Bets. 9% mE Pion 


S£0L,8 | Sains Bal 
atintnatecg tinsaoeteo th otremsiegdiber ne’ | 
.BY-Ae) .28 ‘pippraroarot sige air . ‘eriblisti tloisounibe 


to notteci3 = Ppaer Isoipolsanes AT EVOL .T mp0 LEN eae 1G 
One \Omee “ait anttelupe.t -rebsrow dininsee Ao 


—toyin! pidmiod <xroY wat .spempied ostissl act -1et LoixeM ce i 

. a easrtyie a 
naisqie aus 2. mal ats aps hese SBochaett .2f¢l VLeowedtA .t \siiersd 
1 Tada aE jusigeay -anolsousd .euplySoodd 


= 


iii oriosige on to, ‘meseyenonod’ enh aoe .VeeL sSima \inivoxted ; 


lus aal Tig fohcag ele mis eae Seas a90m! 2aut¢n td mabe? 
‘oder’ «=: Serobh on! fipseyEspa) erbeinasamrr @eb: 
ng 


Pelooe TO asoetc’ .angie- -nttoub vt Tavov nei g65" S&VOL ‘a aqiadd 
b--JRC LE inal © iteloupa is . *Tadtososid “i 

Ssigeoott .'ciszsig\ svhatiensheds' .0PCL iaeneeelt of bee: genes a 

— : »8E-fES.$S 





. > 
*hb- ad -{l- nist to, dnenolovebh oft 1° -bCGL Al Seed eee 
ee fE~2cF ee setchl Spaupnsst reboeM . ‘eonamost mi 
weSlibneM :xsteeconsM .cocen) ish of nitnD ae @RERE ee da 
; | seid \iesevidd xe 


satay we .iovogetl oliinmeto® 2o otpal sdf MC. WA dase. teqaet 
.et00d olesi - 7 


soem xis ai sottelinteeid TetosnoganD. .[d@l i @epeden cmeaeet 


_ ——_ ee eg 


Liewissid (fest : ibid - BRBUDME ST ae 


solmts soncemd . ‘eolsaivonis (sn ivotatad on Meivisieed? Teel 2% a 
F ; -1£-ISE 08, ypot 

+ r os « . : 
rtiteg G6 sipminerqg sb chutes o steT' JObeL ..cb eteneiat  onlbecd Share: = 
ARF IS VOL Pe ES salves . fewekben. 280, ae AMT 

oJ ft 


MOGUL  sAwpSsH ofl = .2newD Yeuxe brow’ \efdetive . Over” 
owbapeo ai estupett, sonetegum Ie elas. ant’ exer 
AGAIS.S seemeeeh ey St Cte ae eam 
erat to sior adit’ ever | 

A T8Re8. vs spitotons 


to > Sacre Seachem bas SW NB sf 


ag 


236 


Reichard, Gladys A., Roman Jakobson and Elizabeth Werth. 1949. 'Lan- 
guage and synesthesia’. Word 5.224-33. 


Reichling, Anton. 1961. 'Principles and methods of syntax: cryptana- 
lytical formalism'. Lingua 10.1-17. 


Révah, I. S. 1958. 'L'évolution de la pronunciation au Portugal et 
au Brésil du XVI© siécle A nos jours'. Anais do Primeiro Congresso 
Brasileiro de Lingua Falada no Teatro, pp. 387-99. Bahia. 





- 1959. ‘Comment et jusqu'a quel point les parlers brésiliens 
permettent-ils de reconstituer le systéme phonétique des parlers 
portugais des XVIC-XVIIf siécles?' Actas do III Coloquio Interma- 


eee 


oes 


Robe, Stanley. 1959. '-L y -R implosivas en el espafiol de Panama’. 


ee 


Robinson, Fred C. 1972. ‘Appropriate naming in English literature'. 
Names 20. 13]-7.- 


Rochet, Bernard L. 1970. 'The formation and the evolution of the French 
nasal vowels: a functional-structural accowmt'. Ph. D. disser- 
tation, The University of Alberta. 


- MS. 'A morphologically-determined "sound" change: the merger 
en-an in old French'. Unpublished MS, The University of Alberta. 


- forthcoming. Review of Klausenburger (1972). 


Rogers, Francis M. 1946. 'Insular Portuguese pronunciation: Madeira’. 
Hispanic Review 14.235-53. 


- 1948. ‘Insular Portuguese pronunciation: Porto Santo and 
Eastern Azores'. Hispanic Review 16.1-32. 





- 1949. ‘Insular Portuguese pronunciation: Central and Western 
Azores’. Hispanic Review 17.47-70. 





Rohlfs, Gerhard. 1949. Historische Granmatik der italienischen Sprache 
und ihrer Mundarten. Band I: Lautlehre. Berm: Francke. 


Sooo eee 


- 1950. Historische Grammatik der italienischen Sprache und 
ihrer Mundarten. Band II: Formenlehre und Syntax. Bern; Francke. 


Rohner, Kurt. 1948. 'Um capitulo de fonética dialectal: a inicial 
em Cachapo (Algarve)'. Boletim de Filologia (Lisboa) 9.251-77. 


Rosenzweig, Alta Gail. 1965. 'A spectrographic analysis of consonant 
length in standard Italian'. Ph. D. dissertation, The University 
of Michican. 


Rosketh, Pierre. 1921. ‘La diphtongaison en catalan'. Romania 47.532-46. 































~raat' .PbOt wrbrrow rivedescta brid noedoAst mama Ae 
£LE-8SS.8 Bx 6 patie 


~anstqys <xednye 20 eborttan bis salginniva’ .f0@E 
Ti-E. OL sonmit ‘atten J SOLS 

; ad 

te [aputiet 66 aoiys ion bi 5f & notsgigvs*1' Baer | “38 igtsoires 


SeserpinO orhertixd ob ulsat zivot pong wfobie 21K ub Lietxd 1 hs sear : 
vEsned SCU-VEElqq .oxteg? Gn Absfst etpekl 2b ) 
rs h 


aiehiietrd exalisa ee! tnicy tem 6 epaut te team" “eer icae 7 
wralyeg ash supi parrada °2 nfceye Bl Yeeg,berdes: 2b ell-taedtemrsa “12 Nae 
~Baistn olipoloy (Tl ah sdtoA ‘SesloSte SEIVA-SIVE gab elepetrog ei 


Bo ee Ch sotiellemd-caul eoteset sb femig 
. 'Bmacied ab Toteges fo vo asviediaut a- y I-" Peel lysines: sdot at 
c-STS.S soiabge Gt StoORAC KR ob BIGIVSH SYSUM. op | 
“@icteasci! detlori ni prilmsn sterxcosqgA” EVEL 9 Bexd \openta 
.V-DEL.0$ eens! 


fonexa ait! to soitulove tt bos cottamie? edt’ OVeL~ sl beeesd panto 
eugeaif ./ .d¢ ,‘tnycoos Tewtture-Icnoisoned 6 salewov Leman [os 
.citedié 2o viherevinv ed? ,pories.  . Me 


tepxem sit -gortsrio “hritce" Banos sh-yilsotpeioiasan Ay” 62a 2 
z ‘ iy i ZS p- 4 J 7 ; .. - 

Sfte0 A to vileraind ot! .eM botehicaand ." dons? Gio ak nen | i 
4 7 —— — cm 


(SVOl) yoprentemuslA to waive palmar . 
‘SrseoeM = inditdsioouncig. sesyprayo! sete aE -obe@f 4 efonsxT ,@rspon 
c-efS BI wakved pinkge tH 


° a . . ~ u 3 x 
Bris otiee odted snoltsiomunoiad seeupiciol talgent’ WBser 


S€-1.0f Weilved >ioggeth .. "Sagal mcesh 


eee : j 


¢ 


ftTisseod fie fertns 2neits.com IOUT Se Teale tent e-4 xeloant*® .@beL ~. 
. OF tS. <I woived sinsgeih .*eedosa 


adage? nooo lab bets a vijsmmet) setxei2it .eRCL  JbmereD abides 7 
Soi isd -Sabeitaal, #T besa ~ ogehieBnci Henri Bare . 


oes LO re tte: 


.eonsxi ied. igadtagey bets rice data iy (bage™ 1 of 


See ae) 


‘gees § :ndppleth sce op olyvlieso ity" ‘epee 2 
“80 <hae. 0 (sodadt) Spool aap med LOE "aan rs 


Jasnoaron to Blaylece. ulysyprugsqe: A | .28GL 
“Pimevin Sa molistigsib .d 2d 2 \qsktest 


bf qs? corpetaci let: 16h litem atbelsotete 40eer 


- : at 
7 oY : 
AR-SEDND Singmest "nik Leta no noeisphowigibiad’ .1Sep 
‘ : a P + f 


ENE 


Ruhlen, Merrit. 1973. ‘On the importance of minor rules in a des- 
cription of the Rumanian verb'. Romance Philology 27.37-45. 


Russell-Gobbett, Paul. 1965. Mediaeval Catalan Linguistic Texts. 
Oxford: Dolphin Books. 


Saciuk, Bohdan. 1970. 'Some basic rules in ie aE phonology'. 
In Studies Presented to Robert B. Lees by his Students, ed. by 


J. Sadock and A. Vanek, pp. 197-222. Edmonton: Linguistic Research. 
Saco Arce, Juan A. 1898. GramAtica Gallega. Lugo. 


St. Clair, Robert N. 1971. 'The Portuguese plural formation'. 
Linguistics 68.90-102. 


Saltarelli, Mario. 1968. 'Marsian vocalism. Intelligibility and 
rules of grammar’. Orbis 17.88-96. 


1970a. A Phonology of Italian in a Generative Grammar. The 


_—- Oe eee 


Hague: Mouton. 


- 1970b. La Grammatica Generativa Transformazionale. Firenze: 
Sansoni. 





Sanchis Guarner, M. 1949. Introduccioh a la Historia Lingtiistica de 
Valencia. Valencia. 


Sankoff, David. 1972. 'A quantitative paradigm for the study of com 
Mmunicative competence'. Unpublished MS, Université de Montréal. 


Sankoff, Gillian and Henrietta Cedergren. 1971. ‘Some results of a 
sociolinguistic study of Montreal French’. In Darnell (1971), pp. 
61-87. 


S& Nogueira, Rodrigo de. 1936. 'Subsidios para o estudo das onomato- 
peias em portugués'. Boletim de Filologia (Lisboa) 4.221-84. 


Sapir, Edward. 1929. 'A study in phonetic symbolism’. Journal of 
Experimental Psychology 12.225-39. 


Saporta, Sol and Donald Olson. 1958. ‘Classification of intervocalic 
clusters'. Language 34.261-6. 


Schdédel, B. 1908. ‘Ie frontiére entre le gascon et le catalan'. 
Romania 37.140-56. 


Schmidt, Gerhard. 1947. 'A sociological theory of language’. Modern 
Language Journal 31.351-8. 


- 1955. ‘Thinking and language’. Orbis 4.66-73. 


Schnitzer, Marc. 1973. Generative Phonology: Evidence from Aphasia. 
University Park, Pa.: Pennsylvania State University Press. 











edie goat apkahh xemrim Bo ¢ 
MEE TS wpolol eet simone 


ri aber Ha 
Biya? Sboetupr gs elaine fore we Ty: 


% 


" wpaLonendey aur peer et Seg oieed hace’ -ovel 


‘end staatoepel © nga aes er ac , 
(iF 


AU J 


‘aeidetso? (swlq sagupysxeT od! Pee weer ie ie a 


a eat RE ee 


; bas yoilidipilietnt ‘metisooy neteteM' Beet oni sti fexsd 
ine J2-88.TI sidsO .!aamap Ip tai 


<r’ «.xhAMSyD ovisgsiers) 5 at Pelisil go eS A woorel. 


eee 

























i . 


Sacer 17 Si BOL GANTOTANS TT EVLISIONISS so lteinet) al dover: ‘ee 
a Te he a 

ab soitvetiiriit sivotath sf 5s Rorsouborini -O88E uM tars ‘eitfons2 

- Sipdoisv et 7 


“Me 20 yorse ait sci mpifsisq svitetiingup A”  .SVeL 
. istagnc sh atievavini! .aM bedecitiqn .."Someteaquae ait 


. ® 


a So actives: snme Vii .cesptebed sg jo banen Sos rong att 
or. (EVOL) ifserso at .‘donext Tsettoot io ye S&S 


“of@ioa eeb. obiises o sxe pre ise! .d€6f ey ee fa 3 
sPS-1S8 Oo (sodest) “sien 1h mis ated  eanpotxog mS 3 


26 igersol , me.ifodnve .o bene it ik VEose A’ ,eoet 
.2€-cS% SF Vpotodsyed 
viiscovusiiit Jo notisofeteasi>" .8cRl _ .noetO bleged a Jee “ 


0088.25 Spedonad ,' a 


~ 4 
are 
“Misieo af 36 nooesp ef shorn exSiinort al’ 


ie 
32-08 L. 


sree Rroeroees to yroany {soipolotoos, 4" Veet is 
-B=Lee JIE rou 


Schourup, Lawrence C. 1973. 'A cross-language study of vowel nasali- 
zation’. Ohio State University Working Papers in Linguistics 
15.190-221. 


Schuchardt, H. 1874. 'Phonétique comparée'. Romania 3.1-30. 


Schuhmacher, W. W. 1973. 'Onomatopoetic resistance to lenition rules 
in Malagasy'. Orbis 22.134-7. 


Shankweiler, Donald and Katherine S. Harris. 1966. 'An experimental 
approach to the problem of articulation in aphasia’. Cortex 
Deel l-92; 


Shapiro, Michael. 1972. ‘Explorations into markedness'. Language 
48.343-64. 


Seelmann, E. 1885. Die Ausprache des Latein nach physiologische 
historischen Grundsdtzen. aie ae 


Silveira Bueno, Francisco da. 1958. A Formacdo Hist6rica da Lingua 
Portuguesa. Rio de Janeiro. 


pletsie steift.. (1959 sale Développement de de 1-et n en Ancien Portugais. 


Etude Fond&ée sur les | Diplomes des Portugaliae M Monumenta Historica. 


Oslo. 


Soares de Azévedo, Celestino Monteiro. 1903. 'Ervedosa: lingoagem 
popular de Ervedosa de Douro'. Revista Lusitana 27.86-197. 


Strevens, Peter D. 1954. ‘Some observations on the phonetics and 
pronunciation of modern Portuguese’. Revista do Laboratorio de 
Fonética Experimental (Coimbra) 2.5-29. 


Tai Whan, Kim. 1965. ‘Description and history of consonant groups 
from Latin to Italian'. Ph. D. dissertation, the University of 
Michigan. 


Tanaka, Ronald. 1972. ‘Action and meaning in literary theory’. 
Journal of Literary Semantics 1.41-56. 


Taylor, I. K. 1963. 'Phonetic symbolism reexamined'. Psychological 
Bulletin 60.200-9. 


Tekavtié, Pavao. 1965. Grammatica Storica Italiana. Parte II: 


Nee ene ey a 


Morfosintassi Italiana. Zagreb. 


Thelin, Nils B. 1971. ‘On stress assignment and vowel reduction in 
contemporary standard Russian’. Acta Universitatis Upsaliensis. 
Studia Slavica Upsaliensa 9. Uppsala: Skriv Service AB. 





Thomas, Earl W. 1969. The Syntax of Spoken Brazilian Portuguese. 
Nashville: Vanderbilt University Press. 





238 


a a Acta cece 
SaMoteti etasiings sellers re Fo: Berm iejake T sel te sibaot shoie 
— oe ola 



























4 


~tleesn low to ybote speupmlosenrs A’ Q88L 3% 
isola ai mses TLIO oC fee kD: o 


-L.€ sine. ‘otncnas SupteSnadtt’ AUBE oe 

antler noisinsl ot consveredh5 ttseqosemgnet «EVEL ott W secbendores 
Sab. S 0 mh .*ysepeteM int : 
Festremni tpt nA ise .claedl .@ onbrertie® Bos blieaod ,tebiewlmer 


geduct) .'stescar ni noigedmas 20 meldong end’ os ar at F 
tai .Se-T7S.5 


SPALL tenhottem abit anoitexokexs' sS0eL . feacoil ortgene 
wi é. wb3-€D£.85 


1 a Tw 4 > CO Le \ 7 fi fT mi Sofa == 8) arte Qeeré > 748 Me) al Pr | ,oanemt 2e8 ' 
er cond! ta PaeaoeorbRD nero bse tt 
i d6 somtdelt omemot& (820! ab copionsy’ ,oneutemtevite 
» asia aml | O1ishst Sp Op  .ppeiperzog 
fa it >) ob He LSarQggcu avert a Omey tial ,eetatale 


———— 


* - = 


MSE Oo: 11 + Ko EDR of rceatne onrcaeied ,obevasA ae esrs0d 
- C- Afist fest eter est ‘ jowed Oz) Ssecharral Ea) s6Lugog is 
Ges coidosoud add ro enniinevresds snod” el Se aedst ,ansverte 
8 oliatesodel ob sjetye . enemies miaber te mots conuniqy 
ia o¢ec.6 (erated) edna abbr Sact 


Som treceansh To voor bas poitgisoes]' «;2R@L saa nemn ist 
to ‘eftayswirt) off .noitecsieeath:.d no . kinase Ghee eee 


“mosis weetet il. ak pritmesr bas aie eb sDlesKe sAscer 7 


Ino tpotorgye? ‘hoa unszasy aellodtive obtenddt' J6aeL 2b. Leable 
2-08.08 Absa ie 


‘Tl aed =. Bnetietl soisot2 shitemisso cael  .Qewed Stivedtet : 


dem ss” Baps leet Segesmkpodeaget 
AaL jakled? 


ts a 


ti ookthube: lewov bos giremipiees ecarie aD! <f7es 
Biensilee stintieerh sigk  ." dereent biebasta 4 


—S sae 


Of acters vias salsacd 2 senoileedtt 


Seetpis a8 aes SERS us I ap to wane mu we 
as Tinea | 


& 1 





7. 
ead 


_ ny eee} 7 


ete) 


Thomas, William. 1968. Principle Rules of the Italian Grammar 1550. 


ee 


Menston: Scholar Press. Facsimile Edition. 


Thompson, Robert Wallace. 1959. 'O dialecto portugués de Hongkong’. 
Actes du 9M Congrés International de Linguistique Romane (Lisbonne), 
View) DOssieso=90° 





ee ee eee 


Alston. Menston: ‘The Scholar Press. 


Trubetzkoy, Nicolai. 1939. Grundziige der Phonologie. Travaux du 
Cercle Linguistique de Prague 7. 


Twaddell, W. Freeman. 1935. 'On defining the phoneme’. Language 
monograph no. 16. Baltimore: Waverly Press. 


Vasconcellos Abreu, G. de. 1887. 'A gradac&éo prosédica de A'. Re- 
vista Lusitana 1-30-4. 


Vasiliu, Emanuel. 1968. Fonologia Istoric# a Dialectelor Dacoromane. 


cc 


Bucharest: Editura Academiei Republicii Socialiste Romania. 


Vennemann, Theo. 1972. 'Phonetic detail in assimilation: problems 
in Germanic phonology'. Language 48.863-92. 


Vennemann, Theo and Peter Ladefoged. 1973. 'Phonetic features and 
phonological features'. Lingua 32.61-74. 


Vitale, Maurizio. 1957. 'Di alcune forme verbali nella prima codifi- 
cazione grammaticale cinquecentesca'. Acme (Universita di Milano) 
102235—75. 


Vogt, Eric E. 1971. ‘Catalan vowel reduction and the angled bracket 
notation’. Linguistic Inquiry 2.233-7. 

Vogt, Hans. 1954. 'Phoneme classes and phoneme classification’. 
Word 10.20-34. 


Wang, William S.-Y. 1968. 'Vowel features, paired variables, and 
the English vowel shift'. Language 44.695-708. 


- 1969. ‘Competing changes as a cause of residue’. Language 
AB. 1=25. 


Weinrich, Harald. 1969. Phonologische Studien zur romanischen Sprach- 
geschichte. Minster Westfalen: Aschendorff. 2nd ed. 


Weiss, Jonathan H. 1963. 'Further study of the relation between the 
sound of a word and its meaning’. American Journal of Psychology 
76.624-30. 


- 1964. ‘Phonetic symbolism re-examined'. Psychological Bulle- 
tin 61.454-8. 


















Wo-tsar spent anklste ox a 
5 A 0d petibe tas Ravaloe 


haces aeree "ts, fin’ 
ub wwevsxT .cipatanods Jap sista REV 
“SOpant. ab ie 
; apeuptis! § . ‘amanorny ah palnize® 2" eer 


rece visews rnmittiet 21 aa pith <a 


= dees 
agi . 1A ob svifddorg omgsberp A’ 760k .6b 2 usi@A gotteancoasy 
| .b-08-L enagiest ayeky 
ane ree) ache 


init» Yolnoien sigstorss 80h “taahaia 3 
SABA Sd tte fboa Ef co Lidge eS sou 3 


aie ine ad ‘ 
amiidony :nolteLiniees qt Disgep siteada’ ver ood —— 
.@-£38 8 Spsupesy 'gpolenorky Sknemrs) al a 


Bits. eoxutsst ol?sdorit veyel bapetatea Jaiel bon omit poeta | | 
Here. cuphid .'autkss? Espdpofonodg 


erties aniza elion tindrev orict smumif it’ Neh ~OntaysM ie A 

bra int it éjienevind) omé °. soecnsosapi 3liols ganszp 
= wi Ov -ces. : 

Yapigexd belpas ord bos qbitouber fewoy \fetesso' «Vee ae aa a 

e (PES IE att Lisl Et bt int pary a. 

. S'nottenitieesin sarmaora Bins poanslo emecont' Beer ens FO + 

erm Wy 


Bos ,eoidsinev beciég rotten Kewell | 8seh tole eoldt oe - 


Tc, PE Spee . tocol lowe 
pring ae 
ae 
i a 





a 
> 





pice} ‘anblesr 16 S@U5S & BS eae pubtogim" - 





1 ae a 

=< i ee al 
7 Le bf 4 
a0 a 


2h0 


1966. 'A study of the ability of English speakers to guess 
the meanings of foreign words'. Journal of General Psychology 
74.97-106. Cc 


- 1968. 'Phonetic symbolism and perception of connotative mean- 
ing'. Journal of Verbal Learning and Verbal Behavior 7.574-6. 

Wertheimer, M. 1958. 'The relation between the sound of a word and 
its meaning’. American Journal of Psychology 71.412-15. 


Wescott, Roger W. 1971. 'Linguistic iconism'. Language 47.416-28. 


Wheeler, Max. 1972. 'Distinctive features and natural classes in 
phonological theory'. Journal of Linguistics 8.87-102. 


Wickelgren, Wayne A. 1965. 'Distinctive features and errors in short- 
term memory for English vowels'. Journal of the Acoustical 
Society of America 38.583-88. 


- 1966. ‘Distinctive features and errors in short-tem memory 
for English consonants'. Journal of the Acoustic Society of 
America 39.388-98. 





Williams, Edwin B. 1962. From Latin to Portuguese. Philadelphia: 
University of Pennsylvania Press. 2nd ed. 





Wittgenstein, Ludwig. 1958. ‘The Blue and Brown Books. New York: 
Harper & Row. 


- 1968. Philosophical Investigations. London: Basil Blackwell. 


Zipf, George K. 1935. The Psycho-Biology of Language: an Introduction 
to Dynamic Philology. Reprinted by the M. I. T. Press, 1968. 


Zwicky, Arnold M. 1972. ‘Note on a phonological hierarchy in English’. 
In Linguistic Change and Generative Theory, ed. by R. P. Stock- 
well and R. K. S. Macaulay, pp. 275-301. Bloomington: Indiana 
University Press. 























bos ected Bo Gawoe ptt reainetert 
PERS yobtortye 30 


kn . 
C35 


. -OS-955 0h epemnnt =. ‘nies obredonhr' .1tef bie 











ot ae 


nt eseasio Deuter bros esy ‘teeter itonivasia" -STOL |.x6M- o 
: .SOL=T8 .E apis pinpais - to Isemyob .~‘yuoedt 
—trada ci gxowxs bas acuetect svisonitel’ -20eL A cxegs i 
Lepivenoo’ siz 3o Leruicit .‘afsyey dekfpnt sok: yarn seg _— 

7 .88-€82.8€ sotxemA Bo! ee = 


hy : 
"ranean terai~trode mi store bun esuitask oviventdeia' opel : 
Tp yisioc? nitavoos sit Io ferwel .‘etasmpenm A 07 - > 


atipng 
| | .Be-GBE RE so kisMis ~ 
iKirighsbabirt .seaitecdted ot cittgl moxt Soe! eS jae 


) .bS bat .Beaxd EineviyeansA 20 vaiteraving a 
OY wai. wihod word bos avis oct .820L pineal erbsten oes EW 
a ee, wae WAS tsqisH th 
2 thawins 18 [fest <coohial .neomtepitesynl fables 80@L wae 7 can 
' IpeBortel ns. .spsipeel 5o yoolol-odsyet off cel «df av10s) agis * ui, 
- =a , : : 





BoP mart T 7 .M Sed oot Geta rigad Abate Liss Sea. 
: : ey * vy 
. Metient nt ata teil leo tpoldénedg s no eto! ter .M Blowté \yo! a i 


“Heats «1 .A yt .be yyrcesl! syitexone® bas spec abetnate et a \) 
ss *hnl idee Sieeth ,fUc-cls Log \ VSIGBORM: 2 yee : 
nag? yistovin «=> 





. rs 
“F P 
r - 
. 2 \ 
ee 180 ’ - 
7 nt as 
- wel 
2 


Ohi 


APPENDIX 
LATIN-ROMANCE ETYMOLOGIES 


(Corrected derivations fram the REW) 


Latin-Italian 


Latin Italian Latin Italian 
abacus abbaco aesculus ischio 
abatissa (a) badessa affectare affettare 
abbattére abbattere afféras affuori 
abé1lana avellana aggravare aggravare 
abhorescére abhorire agulia aguglia 
abdctilis avocolo alabastrum alabastro 
abortare aortare albarus albero 
abstergére astergere albumen albume 
abiila avolo alicunus alcuno 
accadére accadere allactare allattare 
accingére accingere alliminare alluminare 
acere acero altitia altezza 
acetum ace to amaricare amaricare 
acquisitare aquistare amatore amatore 
acutiare aguzzare anbulare ambiare 
adaptus adatto amicus amico 
add6nare addonare amptilla ampolla 
adjacére aggiacere ancflla ancella 
admonés tare ammones tare ancora ancora 

ad pressum appresso anéllus anello 
advénire avvenire angélus angelo 
aequalis iguale anguilla anguilla 





oireat 
Pistoiis 
pete (hees 
eipiiieices 
) ailpyps 
ortasdels 
Giscihs 


amd 


sé a 
erfooaese - 


gipdoSstis 
esidtits 
SIBVSIpPs 


5 Ligps 


muitzedsis 


arses 
recudis 
enmoils 
srpso0s lis 

Wome 
eysrtumi lis 
pisidis, 

Vv 

NBO TEs 
exe {vans 
auD ims 
sf items 
siltons 


SIBTS LPS 






















are - : 
= a 
<* > 
b “* 


Latin 


anglistia 
annéctére 
anguina 
anténna 
apérire 
appactum 
appareére 
appodiare 
aprilis 
acquarium 
aranea 
arboretum 
arcuballista 
ardore 
argentarius 
argentum 
armarium 
aromaticus 
arripare 
articilus 
asdrum 
ascalonia 
asinus 
aspar&gus 
assaltus 
assécurare 


assicélla 


Italian 


angoscia 
annettere 
anchini 
antenna 
aprire 
appatto 
apparere 
appoggiare 
aprile 
acquaio 
aragna 
alboreto 
arcobalestra 
ardore 
argentaio 
argento 
armario 
aromatico 
arrivare 
artiglio 
asero 
scalogno 
asino 
sparago 
assalto 
assicurare 


asSicella 


Latin 


astutus 
atriplice 
audire 
augustus 
auridiare 
avarus 
avena 
avulus 
baccalaris 
baccéllu 
baculum 
bactila 
bajtila 
bajtlus 
ballare 
ballista 
balsamum 
barbatus 
barbitium 
barbtilus 
basire 
batillum 
battuactiLum 
baucalis 
betiilla 
bicongius 


bicornia 


Italian 


astuto 
atrepice 
udire 
agosto 
oreggiare 
avaro 
avena 
avolo 
baccalaro 
baccello 
bacchio 
bagola 
baila 
bailo 
ballare 
ballestra 
balsimo 
barbato 
barbigi 
barbio 
basire 
badile 
batacchio 
boccale 
bidollo 
bigoncio 


bigornia 


2he2 


oliseced 
oOLroosd 

ns 
aBLlopad 


pLlisd 


asian 


(Ree 


sis IbiIw6 
SUtEVE - 

an Lins 

2 ix sisoosd 
ul lsnord 
mu fuosd 
pigosd 
sintsd 
auttped 
gisifsd 
stetiied 
mamnaisd 
attsdisd- 
mg i ichied 
ass ectred 
si.tesd 
mulittsd 
mo Konustted 





Latin 


bifurcus 
bilaneia 
birotium 
bisaccium 
biscoctum 
bitumen 
blandire 
blattila 
boletus 
bonitas 
botéllum 
botina 
bovinus 
brachiale 
bubalus 
bubtiLare 
buccata 
bucélla 
bucc@11atum 
bucciila 
buctilus 
bufone 
bullicare 


buriccus 


Italian 


biforco 
bilancia 
biroccio 
bisaccia 
biscolto 
bitume 
blandire 
piattola 
boleto 
bonta 
budello 
borni 
bovino 
bracciale 
bufalo 
bubbolare 
boccata 
bucella 
buccellato 
borchia 
bucchio 
bufone 
bullicare 
bEUGGO 
burla 
bottiglia 


cavalla 


Latin 


caballicare 


caballus 
caducus 
caeléstis 
caementum 
calamellus 
calcare 
caldaria 
calura 
camella 
caméra 
caminus 
campana 
cambarus 
canaba 
candela 
canile 
caninus 
cannabis 
cannamllis 
cantharella 
canticum 
capillatura 
capistrum 
capitanus 
capitellum 


Vou. 
capitulum 


23 


Ttalian 


cavalcare 


cavallo 
caluco 
cilestro 
cimento 
caramella 
Calcare 
caldaia 
calura 
gamella 
camera 
camino 
campana 
cambero 
canova 
candela 
canile 
canino 
cannapa 
canname le 
canterella 
cantico 
capellatura 
capes tro 
cattano 
cattella 


capecchio 





~ ae 
a =e) 


PETE, Hye te 
a 7 








eiisb LED 
seal 


a Etensp 


sisbise 
saw a5 
SrLIMAD 
ortimn 
oyarine 
BVOMmSaD 
Sh 
SLSMBO 
OFLLita> 


‘oL 


sfietains> 





Latin 


cappella 
caprificus 
Yu 
capulum 
carbtinctilus 
cardellus 
in, 
caricia 
carnalis 
carpinus 
casalis 
Sh 
casicare 
catafalcum 
catalectus 
cattlLus 

Y 
caudica 
AS 
colictilus 
centenarium 
ceptilla 
cerebellum 
cermictiLum 
cerretanus 
cicada 

A wv 
cicercula 
cicindéla 
ciconia 
en, 
cimice 
cimssa 


cinctura 


Italian 


cappella 
caprifico 
cappio 
carbonchio 
cardello 
carezza 
carnale 
carpino 
casale 
cascare 
catafalco 
cataletto 
cacchio 
cocca 
colecchio 
centinaio 
cipolla 
cervello 
cernecchio 
cerretano 
cicala 
Cicerchia 
cicindello 
cicogna 
cimice 
cimossa 


cintura 


Latin 


Mew) 
cingula 
. J. 
clirculare 
cisterna 
cithéra 
Civitas 
clavellus 
clavicula 
clerica 
coactila 
VJ 
coagulum 
cognatus 
cognitus 
collibertus 
colostrum 
colticila 
columbus 
comboros 
~ = 
commissura 
conchila 
confidare 
conjungére 
wr oY 
conquirere 
conspectus 
cons tipatus 
contrata 
convenium 


Oe a 
cooperculum 


eu 


Ttalian 


cinghia 
cerchiare 
cisterna 
cetera 
citta 
chiavello 
cavicchia 
chierica 
quaglia 
caglio 
cognato 
conto 
culverto 
colostro 
conocchia 
colombo 
ingombrare 
commessura 
concola 
confidare 
congiungere 
conquidere 
cospetto 
costipato 
contrada 
convegno 


coperchio 


= is eee 


a in 
we 


7 
“De ¢ « 
a 





esxey iu artitedi ifoo 
mrit20oo 
sitrotifao 


aucherloo 





| +") 
» oF 4 
* 
«# 
‘ 
f & 
p's 
a Ae 
At 
t 





Latin 


cooperimentum 
we aa 
cooperire 
ey 
cophinus 
Vv. 
copula 
coquina 
” 
coracinus 
cordélium 
cornutus 
corredare 
corruptum 
costatum 
citttilus 
Soe ed 
craticula 
ws 
crepitare 
ir d 
cribellum 
Ss 
cristatus 
ves 
crus tosus 
cUbile 
cucuLus 
ww, wv 
cultellus 
etd 
cuniculus 
cupula 
curculio 
cursorius 
4 
cutina 
cotoneum 


dalmatica 


Italian 


coprimento 
coprire 
cofano 
coppia 
cucina 
coracino 
cordoglio 
cornuto 
corredare 
corrotto 
costato 
ciotola 
gratiglia 
crettare 
crivello 
crestato 
crostoso 
covile 
Ccucco 
coltello 
coniglio 
cupola 
gorgoglio 
corsoio 
cotenna 
cotogno 


dalmatica 


Latin 


dardanus 
débilis 
debi ta 
décibilis 
décima 
declinare 
deodllare 
dééxci tare 
défalcare 
defésum 
déforis 
déléctare 
délimare 
dementare 
dentice 
déerenare 
désignare 
déventare 
digitale 
diltvium 
diversus 
divertire 
domesticus 
dominedeus 
dommictlare 
dubitdre 


ducenti 


25 


Italian 


dardano 
debile 
detta 
dicevole 
decima 
dichinare 
dicollare 
destare 
diffalcare 
difesa 
difuori 
dilettare 
dilimare 
dimentare 
dentice 
direnare 
disegnare 
diventare 
ditale 
diluvio 
diverso 
divertire 
domestico 
domineddio 
dormicchiare 
dottare 


dugento 





PENS CLD 





Latin 


duod&cim 
duracinus 
v 
doémina 
J 
ebénus 
4bulum 
encaeniare 
Y 
erica 
ay 
ericuila 
eruca 
Wis 
ericius 
etd, 
evanescere 
exagium 
fabéllare 
Vv 
fabrica 
facella 
factione 
Y 
factOrium 
factura 
falcone 
Y 
fallita 
¥ yw 
familia 
farina 
farinéus 
fascina 
fatiga 
cy 
femina 


fenéstra 


Ttalian 


dodici 
duracine 
donna 
ebano 
ebbio 
incignare 
erica 
grecchia 
ruca 
GECCIO 
svanire 
saggio 
favellare 
fabbrica 
facella 
fazzone 
fattoio 
fattura 
falcone 
falta 
famiglia 
farina 
ferigno 
fascina 
fatiga 
femmina 


finestra 


fenile 
fenticulum 
férramentum 
ferula 
fervére 
festuca 
fibiila 
fidélis 
filatura 
filice 
filiaster 
fissiira 
fistulare 
flammtila 
floccosus 
focacea 
foliatus 
fontana 
foramen 
forasticus 
forfice 
fortuna 
fractura 
fragtfila 
fricaméntum 
frondosus 


frontale 


2h6 


Italian 


finile 
finocchio 
ferramenta 
ferla 
fervere 
festuca 
fibbia 
fedele 
filatura 
felce 
figliastro 
fessura 
fischiare 
fiammola 
fioccoso 
focaccia 
fogliato 
fontana 
forame 
forastico 
forfice 
fortuna 
frattura 
fragola 
fregamento 
frondoso 


frontale 





outros 

















9 fEn3t 


<a 
axevist 
poieteo2 
cibd 
2ifepn 
ever 
sit 
sateen Ett 
piers 
oa lis2it 
5 times 17 


ayecnno lt 


Latin 


fagita 
fullone 
forcilla 
furtinctilus 
fusticSllus 
gallina 
gangraena 
geme 1lus 
genitus 
gentiana 
geniictilum 
gertlus 
gingiva 
blandarius 
glandtila 
glitimia 
gramineus 
gramtila 
granarium 
grumiilus 
gubémum 
guldésus 
gurgtlia 
habitus 
haereticus 
halice 


harpagone 


Italian 


futa 
follone 
forcella 
foroncolo 
fuscello 
gallina 
cancrena 
gemello 
gemito 
genziana 
ginocchio 
gerla 
gingiva 
ghiandaia 
ghiandola 
ghiottornia 
gramigna 
gramola 
granaio 
gromolo 
governo 
goloso 
gorgozza 
abito 
eretico 
alice 


arpagone 


Latin 


hastile 
hédéra 
hérbosus 
4 wi 
hirpice 
horrore 
hdspite 
humidus 
humilis 
gee RY 
1dioticus 
ilYcetum 
vf ee 
imaginatus 
imbracare 
imbriceus 
imbutum 
impactiare 
imparare 
impétus 
Y 
inante 
. . Vv 
incisculare 
incontra 
incude 
indice 
indutiae 
inéscare 
infimmus 
infaltus 


ingénium 


OT 


Italian 


astile 
edera 
erboso 
erpice 
orrore 
ospite 
unido 
umile 
ZOtico 
lecceto 
Maniato 
imbraccare 
breccia 
imbuto 
inPazzare 
imparare 
empito 
inante 
cischiare 
incontro 
incude 
endice 
indugia 
inescare 
infermo 
fulto 


ingegno 





ousfhoetont 








_/ 
- = _ 
7 i ' 
- - A =a 
- ~_ - > 
} - L. 
a” Pe J _* 7 J 
7 
- - » PD 2 as 
7 . € 
« a W 
= 7 5 
° / : 
- . 
7 
= 4) = 
‘ 
» . f p 
¥ 
> Ss 7 * 
y i 
e _ : 7 >) 
: oh 
a 


i 






_ 7 





A 
—— 
_ 


» 
7 : 
way 7 


as 
7 r 
7 






oe 


ce) 

a 
—_ 
Ss 


Ge. 






— 


a 
= 
a 










~~ 
a 


ay 


~~ Pires - 


t 
- 


_ 
Cort 


re 


7 


7 


i : ' 


oma! 
> 
_ 
a 










7 a iL 
a, 
: 


al 
7 


mr ve 


- - 
—_ 


we 


i 
she, 
: = 
1) onal 


7 





ous 
= 


= 


Lilith 


7 
a 
or 


i 





Latin 

a7 Me 
inguine 
inoctilare 


insomium 


instriméentum 
interm&dium 


intervallum 


intéxére 
intimicare 
. Y _ 
invitare 
iratus 
jacile 
joctilare 
jUdaeus 
jadice 

. ~ 
Jumentum 
junctura 
. Vv 
jJuramentum 
jiivencus 
labina 
lacerta 
lacrima 
lactutca 
lacuna 
laetamen 
lamine 
lampada 


lampréda 


Italian 


inguine 
inocchiare 
insogno 
stromento 
intemmezzo 
intervallo 
intessere 
intonaccare 
invitare 
irato 
giaciglio 
gioculare 
giudio 
giudice 
giumento 
giuntura 
giuramento 
giovenco 
lavina 
lucerta 
lagrima 
lattuga 
laguna 
letame 
lamina 
lampada 


lampreda 


Latin 


lanosus 
laqueolus 
larice 
latinus 
latratus 
lavatorium 
lectica 
lectiGne 
legalis 
legénda 
lentictila 
leone 
leporarium 
1épérarius 
1évitum 
licinium 
Ligaméntum 
ligatira 
Lignamen 
limaceus 
lindsus 
Lintedlum 
littéra 
logicus 
licarinus 


lucerna 


Italian 


lanoso 
lacciuolo 
larice 
latino 
latrato 
lavatoio 
lettiga 
lezione 
legale 
lienda 
lenticchia 
lione 
leporaio 
lepraio 
lievito 
lezzino 
ligamento 
legatura 
legname 
lumaccia 
limoso 
lenzuolo 
lettera 
loico 
lucherino 
lucerna 


lucore 


248 


spisial 


anoiss! | 


, Siépel 
shoot 


“e@beroge! 


Oisigss 





1h. oa 


aloes! 
sf 


eyciial - 


ee 
 eutprtel 


nutsotevel 


- 


Bertoel - 


aBidosl 
aiispel 
sbrispel 


Bao Rtael 


ernoal 
om treroqal 


ay iezogel 
















i 
ae 


- 
. id a , 
, ; ad - 7 
o ®t ; rs - 7 7 
‘ q IX, - ; 
2 - 2 
¢ on 
G ou ; . ne 7 
J } : " 
: , 2 7 
'- ° 
\ > 
. 4 J 
1a 
; p 
7 tae 6 PT) ad A 
3 
: i er 
- 4 ~ | “~~ 
i ao . out “— Ny 
: cr» ge? 
- =~ : 
- 





t 
[eit 


& 
= 
ye 
: é 
y 
a 


Latin 


—_——— 


limbricus 
luniila 
ltpinus 
1uptiLus 
lurtdatus 
liscinidlus 
1utdsu 
mac€llarius 
mace llum 
machina 
machina 
macula 
mactilare 
maeniila 
majalis 
majore 
malifatius 


mami lla 


mandragoras 


manianus 
manibella 
Y 
manica 
mantiptilus 
mantaica 
mantellum 
mantica 


margarita 


Italian 


lombrico 
lulla 
lupino 
luppolo 
lurdare 
lusignuolo 
lotoso 
macellaio 
macello 
macina 
macchina 
macchia 
macchiare 
menola 
maiale 
Maggiore 
malvaggio 
mammella 
mandragola 
Magnano 
manovella 
manica 
mannocchio 
manteca 
mantello 
mantaco 


margherita 


Latin 


maritus 
marrubium 
martellum 
masctilus 
Ww 
mastice 

we 
mataris 

v 
matrigna 
mattéola 
- 
matutinum 
y) 
maximus 
medicus 
memdrare 
mendicitate 
mesura 
mentastrum 
mentitore 
mercatus 
wv 
mrula 
micina 

. wa 
miculina 
milliarium 
minacia 

“Ww 
minimus 

ree 
mnutus 

oe Was 
mirabilia 
ra y 
miraculLum 


Vv =— 
mixticius 


2h9 


Italian 


marito 
marrobbio 
martello 
maschio 
mastice 
mattero 
matrigna 
mazzuola 
mattino 
massimo 
medico 
membrare 
mendicita 
misura 
mentastro 
mentitore 
mercato 
merla 
miccino 
miccolina 
mMigliaio 
minaccia 
menomo 
minuto 
meraviglia 
miracolo 


mestizzo 





Latin 





molinarius 
mllica 
mnastérium 
wv — 
montanus 
morgéllus 
mortalis 
movita 
miacidus 
minditia 
v— 
murena 
musculus 
mus tosus 
ifcpunl Of aa oe 
muti lione 
muttlus 
myrtella 
myrtinus 
nakéra 
naricae 
nasutus 
natalis 
natica 
Ya 
navicella 
nebula 
Sie oe 
necessitate 
népote 
nibulus 


nidore 


Italian 


mulinaio 
mollica 
monastero 
montano 
morello 
mortale 
motta 
mucido 
mundezza 
morena 
muschio 
mos toso 
modiglione 
mucchio 
mortella 
mortina 
mnuacchera 
narice 
nasuto 
natale 
natica 
navicella 
nebbia 
nicista 
nipote: 
nibbio 


nidore 


Latin 


nigélla 
nigritia 
nivosus 
notarius 
nubilus 
oboedire 
obscurus 
obsidium 
Sccidére 
officina 
Opacus 
Spéra 
Ordinium 
ordine 
ovile 
paeonia 
palara 
palatium 
péllidus 
palpébra 
pampinus 
panné1lus 
panthéra 
paradisus 
parénte 
parictilus 


/ 
parricus 


250 


Italian 


nigella 
negrezza 
nevoso 
notaio 
nuvola 
ubbidire 
scuro 
assedio 
uccidere 
fucina 
Opaco 
opera 
ordigno 
ordine 
Ovile 
peonia 
palaia 
palazzo 
pallido 
pelpebra 
pampino 
pannello 
pantera 
paradiso 
parenti 
parecchio 


parco 


areb Loos 


BLANDG 
618 f5q 
osssleg 
obits FAI 
Bideqloq 
on tams¢ 
ob Lennsg 


tw 


sLlSein 


ative 
BLNOSSY 
BX1B isa 
mits ica 
vw Ng 
eubiifeg 
sidéqisg 
Nw 
SULlanEg 


en LLeoag 





Latin 


pasté1lum 
patélla 
paupéro 
paviméntum 
peccatum 
pédale 
pedic&llus 
pédilis 
péndtilus 
peniciilus 
pepone 
pérégrinus 
perniore 
pertica 
pictore 
pictura 
piculus 
pilamen 
pildsus 
pilila 
pinnactilum 
Sapnane 
piscarius 
pisellum 
pistrinus 
pipita 


placibilus 


Italian 


pastello 
padella 
povero 
pavimento 
peccato 
pedale 
pedicello 
pedule 
pendolo 
pennecchio 
popone 
pellegrino 
perniore 
pertica 
pittore 
pittura 
picchio 
pelame 
peloso 
pilola 
pennachio 
pignone 
pescaia 
pisello 
pistrino 
pipita 


placevole 


Latin 


planitia 
plumacium 
poditilum 
pollire 
pollice 
porcarius 
portatore 
postcinium 
posula 
praebenda 
préhensidne 
principe 
protimdus 
publicus 
pugnale 
pti licénus 
puilméne 
pulvisctilus 
pungé11us 
purptira 
quadiatus 
quatérnus 
quindécim 
quiritare 
rabiosus 
radidélus 


Vv 
radice 


Ttalian 


planezza 
piumaccio 
poggiuolo 
pollire 
pollice 
porcaio 
portatore 
pusigno 
posola 
prebenda 
prigione 
principe 
protondo 
puvico 
pugnale 
pulcino 
pulmone 
polvischio 
pungello 
purpora 
quadiato 
quademo 
quindici 
gridare 
rabbioso 
razzuolo 


radice 


Zoe 


emit lox 
seiLiea 

“ Hrodedtog 
Of LaL 
 elowog 
abnsdeaxc 
emminisg 
aq foning 
Hadoia 
ooFvuG 
Ssfenpud 
armani 
ofrina ty fag 
oft peters 
BIOPLEG 
ods cbernp 


orrreabanp 


wae 


sit taal 
miioemiia 
mer LLOQ 

$ ont Lisg 

: on oe; 
BULIBOIOQ 
node tnog 
MES LNiIFZog 

| BLeOy 

Pt osyic al phe | 
ari ansxieig 
xjionixg 
eibavtcra 

etre Lidia 

af Soma 
exe £1 Big 
arom gq 

ary ime be Lio 
es Ef Speci 


siicasy 


SUN p 
mio sbrizen 

| SES Eclsp 
axecides 


ets SBeup | 


ol iso thay 


olabriag 


axOttigq 








no 


7 


ib 


A 


Wert 


.) 


it 


S? 
— 












ri 
~~ 
1 oe 7 
ae 
2h 
aD | _ 
a 
{ . _ 
2 
¥; 
~< 


ei 
wre 
- 





Latin 


ragulare 
ramentum 
ramosus 
ramusc?1llum 
ranictila 
~~) 
rapidus 
rasorium 
rasura 
v here 
recapitare 
recadere 
regione 
remdlum 
rémiilctm 
renidone 
receptare 
recdcére 
réfrangére 
refatare 
régulare 
rélégare 
remtindare 
répausare 
Vv —_— 
resécare 
Se A ed 
respondére 
cea Ae he 
retincina 
rétina 


rétdrtus 


Italian 


ragliare 
ramento 
ramoso 
ramoscello 
ranocchia 
rapido 
rasol1o 
rasura 
Yicapitare 
ricadere 
rione 
remolo 
rimorchio 
rognone 
ricettare 
ricuocere 
rifrangere 
rifutare 
regolare 
rilegare 
rimundare 
riposare 
risecare 
rispondere 
ritrecine 
redine 


retorta 


Latin 


Pi Aweet 
retrorsus 
Se, a 
revisitare 
revolitare 
tT Ne 
rimare 
Very 
ringulare 
ap. we 
rivulus 
roboretum 
=v 
robore 
romana 
rosmarinum 
Sf 
rotella 
roticinus 
aS 
rottiLus 
rottindus 
rabétum 


wT 


ruina 
rtmice 
runore 
vy ay 
runcinus 
rosctilum 
—_ df 
rusticus 
sabb4tum 
ee 5 
sabulone 
saetacium 
v 
Ssaggita 
salicastrum 


Saliva 


Italian 


ritroso 
rovistare 
rivoltare 
rumare 
ringhiare 
rivolo 
rovereto 
rovere 
romano 
ramerino 
rotella 
roticino 
rocchio 
ritondo 
roveto 
rovina 
rom(b) ice 
rumore 
ronzino 
ruschio 
rustico 
sabato 
sabbione 
staccio 
saetta 
salicastro 


Saliva 


Boe 


OBOE iy 
DILTSLVOX 
ge tlov ix 

onenr 
Ors tripe: 
OLOv LT 

G3 sIsV iT 
STEVOI 


, OMEAKS. 


OFLlTSans x: 


ad 
Br lt Ssrci 


onio ctor 
OLnmoocy 
Obrat 17 
ofavor 
BaIvos 
a0, (ed) mor 
enh Mt ea 
ONESTOX 
oirbens 
oo LAtaorr 
ocacdse 
epoiddse 
obonad 2 
—« BITS6E 
. ortascilse 


auatet 
ausrettsx 
syst Lalvey 
Esato) ig feet 
oemet 
\ ee " 
sip iifoniy 
vy. 
eg Line ct 


5° Be ope ty yy 
MOdat 

Boemox 
TLOLante 


ret. 
Bi lsSoor 


a = ‘ * 
~JisOt 


bag 


sep lawiont 
exrinertor 


musacits 


wT) 
ae > 
OL | 


= pi rar t 


PE ROTSTe SG 


TEs Fie) ear 
BUOE ] EEX. 
mutiddse 
enplrdse 
fu Loste62 


&t Yoosa 


mraosasot ise: . 


abisoacmss 
sip oons 
ols iy 
OL0aBr 
SwWe2s1 
stat Lanois 
enoieots 
an Ls 
Lomas 
 Oftaraners 
ADOT 
S163 3293 
FisQOUa Is 
STspnsut Ix 
sissy ti 
aE Lops 


sispal be 





Latin 


salute 
sanctitate 
sanguinéus 
sapidus 
sarcina 
sartore 
scabe1lum 
scarpellum 
scarlatum 
schedtila 
schélaris 
scoffina 
scopilia 
scorpione 
scriptiira 
scutélla 
sécale 
sécundum 
sellarius 
séméntia 
sénidre 
séntore 
sémmone 
serpente 
sérvitium 
siccaneus 


v pay ee 
siccator1us 


Italian 


salute 
santita: 
Sanguigno 
sapido 
sarcina 
Sartore 
sgabello 
scarpello 
scarlatto 
cedola 
scolare 
scuffina 
scoviglia 
scorpione 
scrittura 
scotella 
segola 
secondo 
sellaio 
semenza 
signore 
sentore 
sermone 
serpente 
Servizio 
seccagna 


seccatoio 


Latin 


siggillare 
silictila 
sincerus 
sinister 
sittla 
smargadus 
societata 
solarium 
solidus 
sorditia 
sororcula 
spatula 
spectilum 
spélunca 
spictiLum 
spinula 
spiritus 
spdortella 
spumdsus 
squamds us 
stamineus 
stermttare 
stémutus 
stiptila 
stirpétum 
st6lone 


stramentum 


aa 


Italian 


suggellare 
salecchia 
sincero 
Sinestro 
secchia 
smeraldo 
societa 
solaio 
sodo 
sordezza 
sirocchia 
spatola 
specchio 
spilonca 
spigolo 
spilla 
spirito 
sportella 
spumoso 
Squamoso 
stamigna 
stamutare 
stemuto 
stoppia 
Ste. pelo 
stolone 


stramento 















_ 


oe 
Ly: 
a 






- 
7 
7 _ 
yee 
7 > : 
7 Pp, 
- : 
, F * i) 
' 7 an os 
. Pt i 
ee 
i ) j i 
_ Ly aay ; 
H aan = 
re 
: i 
7 
7 ty ; 
.- ’ { 7 4 
a - i oa 7 
— r> 
. & 
7 


Ss 


_ 
_> 
ol a) 


ae ak : 
tte 


a, so 
~~ 


- 5 


<— 
ae 
O 


7 


a @ 
_ 
_ 


Hii 


E 
: 
P a 
fa 
— a 









i 


> 
' 






_ 


ic 








= 
7 
. 
_ 
; aay. 
ray 
7 —_— : 


\legras 
ve a : 


2 2 
a 


9%, 


ay : 
7 . 
ae 
nian 
ain 
7 g ; 
€ ; 
= 


Latin 


vs 

Strictura 
sobér 
sucos us 
stculare 

— 
suculus 
sudore 

V 
suffradcta 

Les 
supéranus 


stperctilus 


Y Vv 
suppedaeneus 


stirctlLus 
stsina 
tabanus 
tabella 
tabemarius 

JA 
tabulatum 
taléntum 
talone 
tamarice 
tapetum 
tardivus 

aed, 
tegula 
témone 
tenace 
af ote 

teneritia 
tenore 


téntidone 


Italian 


strettura 
sughero 
Sugoso 
succhiare 
succhio 
sudore 
soffratta 
soprano 
soperchio 
Soppidiano 
sorcolo 
susino 
tafano 
tavella 
tavernaio 
tavolato 
talento 
talone 
tamerice 
tapeto 
tardivo 
tegghia 
timone 
tenace 
tenerezza 
tinore 


tenzone 


Latin 


téptdus 
térébéllus 
térraneus 
terri torium 
testaceum 
tindre 
tinedla 
titillus 
tittilus 
tdoretiLum 
tomméntum 
torpédine 
tortorium 
téxicum 
traditione 
tradtice 
trajectus 
travérsa 
tredécim 
tremaculum 
termuLus 
tribulare 
tribulum 
trifSlium 
tristitia 
tumtilus 


Ww 
tunica 


Italian 


tepido 
trivello 
terragno 
territorio 
testaccio 
timore 
tignuola 
ditello 
titolo 
torchio 
tormento 
torpedine 
tortoio 
tosco 
tradigione 
tralce 
tragetto 
traversa 
tredici 
tramaglio 
tremolo 
tribbiare 
tribbio 
trifoglio 
tristezza 
tomolo 


tunaca 


25) 


mmoostand 


auf cha 
aus LES 
ams (teonad 
muchos 
eniBgrct 


res rsosict 











Latin 


ee 


turbidus 
turdéla 
ttirttire 
turturélla 
ulmétum 
Ulpicitlum 
tilticcus 
urtlare 
imbractilum 
tmbrélla 
tncinatus 
incinus 
undatus 
undosus 
tnivérsus 
Urcedlus 
urtica 
ustrindre 
usira 
wité11lum 
uvula 
vaccina 
vagabtindus 
vanitate 
vapore 
vapulo 


Wel aden 
varicare 


Italian 


turbido 
turdela 
tortora 
tortorella 
olmeto 
upiglio 
allocco 
urlare 
ombracolo 
umbrello 
uncinato 
uncino 
undato 
ondoso 
universo 
orsivolo 
urtica 
strinare 
usura 
utello 
ugola 
vaccina 
vagabondo 
vanita 
vampore 
vapolo 


varcare 


Latin 


varidla 

J 
varice 
vasoé1llum 
vasctiLum 
vectura 
vénditore 
venéenum 
ventilabrum 
ventosa 
verbena 
verbenaca 
veronice 
veruca 
vertigine 
vésper 
vy 
vessica 

vw ov 
vestimentum 
ym 
vicata 
viosnda 

Ci gh 
viculus 
ed 
vilitate 
oA my 
vincilia 
WM al’ 
vindicta 
(oe 
virgatus 
SJ. 
virtute 
vitéllus 


viticula 


255 
Italian 


vaiuolo 
varice 
vascello 
vascolo 
vettura 
venditore 
veleno 
ventolaio 
ventosa 
vermena 
verminaca 
vernice 
verruca 
vertigine 
vespero 
vescica 
vestimento 
vicata 
vicenda 
vicolo 
vilta 
vinciglio 
vendetta 
vergato 
vertu 


vitello 


viticchio 















Seosnw 
sserav 
 SDErSrroy 
~~ 
Sof wis 
SuISV 


~~ — 
Sipiyrav 


tsqesv" 


Ys 
SU fsa 


mudora’ 


sipolv 
sbatolv 
aupittoty 


aaa bliv 


: i arty 


yon usr 


olongaucdao 


eticddee 
oteniotey 
erours 
othr 
QEnbNO 
onto Lat 


olov tere 


SO LTW 
: . 
KLOgL 


Latin 


v ‘ 
vitrorium 
vittilus 

be ed 
vivula 


volaticus 


Latin 


wv 
abacus 
abéllana 
abismus 
abrétSnum 
absurdus 

Vv. 

acceéia 
accedia 

J 
acere 
aciarium 
acrore 

Ww — 
aculeatus 
adducere 
adirare 
armordium 
adoctlare 
ad retro 
ad versus 
aequalis 
aestivus 


afféctus 


Italian 


vetraia 
vecchio 
vivole 


volatico 


Latin-Spanish 


Spanish 


abaco 
avellana 
abismo — 
brotane 
zurdo 
arcea 
acidia 
arce 
acero 
agrur 
aguijada 
aducir 
airar 
almuerzo 
aojar 
arredro 
aviesos 
igual 
estio 


afecho 


Latin 





volviilus 
zabérna 


zelosus 


Latin 
affSras 
agninus 


agulia 


alabastrum 


albaris 
albura 

— -_ 
alicunus 


altSnus 


alluminare 


altitia 
ws 
amaricus 
amatore 
amistate 
amicus 
ampulla 
anaticula 
anéllus 
wv 
angus tus 


Vv 
annictiLus 


Italian 
volgolo 
giberna 


geloso 


Spanish 
afuera 
aflino 


aguja 


256 


alabastro 


albar 

albura 
alguno 
Aaltano 


alumbrar 
alteza 


amargo 
amador 
amistad 
amigo 
ampolla 
nabaja 
anillo 
angosto 


afiejo 


BISIs 
BrUPS 
orssdsis . 
: “a 

- pind is 
’ orem is 
onesie 
xerdmy Ls 
ssails 


. 


Posten ki 
ayniaps 
situps 
frit escdsl B 
orpdie 
etndis 

eyeh Bot ts 


aries 16 


extaiimlis | 


sigitis 
eum IEMs, 


SLOSEMS 


(Lite 
en 





< 


fair tittl 


‘7 


ai 


a 


— 


= 





of 


Latin 


anteannum 
apuge 
apertura 
apdostS1lus 
aparescére 
apposticiiis 
aprilis 
aquarium 
aranéa 
arbitus 
arditus 
arena 
aridus 
armatura 
Css 
arrugia 
SW Se 
artemisia 
Vy 
articulus 
ww 
asinus 
astella 
as trosus 
audi tus 
v . 
augurium 
auglis tus 
fetid 
auri fice 
autumus 
avena 


ow 
aviolus 


Spanish 


antajo 


abertura 
apéstol 
aparecer 
postizo 
abril 
agueiro 
arafia 
Alborto 
ardite 
arena 
arido 
armadura 
arroyo 
altamiza 
artejo 
asno 
astilla 
astroso 
oido 
aguero 
agos to 
orezbe 
Otono 
avena 


abuelo 


Latin 


bagassa 
baccalaris 
baccinum 
bajtilus 
ballista 
balneatore 
bals&mum 
baptismus 
barbatus 
batillum 
battuacttlum 
battualia 
battilus 
baucalis 
benedictus 
benignus 
beriila 
beryllus 
bestictlum 
bibi tore 
bicornia 
biférus 
bifidus 
bilfsa 
biregeta 
bisaccium 


biscoctum 


eon 
Spanish 


bagasa 
bachiller 
bacin 
baile 
ballesta 
bafiador 
balsamo 
bautismo 
barbado 
badil 
badajo 
batalla 
balde 
bocal 
bendi to 
benigno 
berro 
vericle 
vestiglo 
bebedor 
bigornia 
breva 
befo 
belesa 
vericueto 
bezazas 


bizcocho 





Pes 


: 
i 


ff 


To 





5S. 





i 


7! 


i 





Re | 


} 


ys 






Y 
oo 
f 


' 









U 





— 





i: 
7 
ah 


re 


* 
> 
ie ey 
U 


5 ad 


a 
- 
oe 


7 
- Pa 7 
> : : : 
- aa _ : 
4 ~ ~ . F ie 

ee ; 7 , , 
od 7 m ‘ 

. > ee ; 

: ihe. i ; 
Pet -_ Qe ) 

a ) ; ; 

7 t*% a ; | 
> a oe a 
: = 7 o- ra am a 
: ay. 
ae 4 i 
ee : : a ns : 
: oueea) A . 
: = 
; 
' 






aa. 
S) pie 


ot 
rae 


es 






7 
al 






q 


~ 
[ 
aq 









Sew 
- 7 
Phe: 


— 
x 
77 





ra 


_ 
— / 





— 


7 
j 


1 






Latin 


bitimen 
ponitate 
brachiale 
bucctila 
bucina 
bulgaréne 
bord6ne 
burrago 
burila 
birricus 
buttictla 
caballa 
caballus 
cacc&bus 
cadaver 
caem2ntum 
caesé1lum 
cespite 
cal &mus 
calcaneum 
calceatus 
calligo 
caltimia 
caltira 
calavaria 
camabaeus 


canbita 


Spanish 


betfn 
bondad 
brazal 
bucle 
buzén 
bugarnén 
bordonero 
borraja 
burla 
borrico 
botella 
caballa 
caballus 
cacho 
calabre 
cimiento 
cincel 
césped 
calmo 
calcayjo 
calzado 
calina 
calofia 
calura 
calavera 
camafeo 


camba 


camella 
camra 
camerarius 
cammarus 
camminus 
camuce 
campana 
campec#11us 
campestris 
canale 
cancellarius 
candela 
caninus 
canitia 
cannabis 
cannamel1lis 
canélla 
cantharus 
canticum 
capanna 
capillus 
capitanus 
capitium 
cappella 
caprunus 
capttlum 


carabus 


258 


Spanish 


gamella 
Camara 
camarero 
gambaro 
camino 
gamuza 
campana 
campesillo 
campes tre 
canal 
cancellar 
candela 
canino 
caneza 
Cafiamo 
cafiamiel 
canilla 
cantaro 
cantiga 
cabefia 
cabello 
capitan 
cabeza 
capella 
cabruno 
cable 


caraba 


siiaasp 


Et TSG 

oll taaqns. 
SIILSQSD 
fens: ) 


ys iisonmso 


si sons 


ONLMED 
SSoi5D 

— 
OfS46o 
loaimeis> 


spfilinsns 


CaBaTihs> 


PULLED 
eit fr 


SfpBqs 
au Ti tease 
BLNESIG5S 
tes iF Tags 
ed Poqqeo 
@alniriqao 
multass 
audsxe5 








Latin 


— 


caractiLum 
carbone 
af 

carcere 
cardénus 
card6ne 

sie 
carlicia 
carina 

WJ . 
caristia 
carnacius 
carnarium 
carrasca 
carvalya 
casalis 
casarius 
casearia 
casélla 
castanea 

ww 

castellum 
castitate 
catafalcum 
catalanus 
catellus 
catena 
cellariarius 
cellarium 
centenum 


centipede 


Spanish 


carajo 
carbén 
carcel 
cardeno 
card6n 
careza 
carena 
carestia 
carnaza 
carmero 
carrasca 
carvajo 
casal 
casero 
quesera 
casilla 
cas tajja 
castillo 
castidad 
catabalco 
catalan 
cadillo 
cadena 
cillerero 
cillero 
centeno 


cienpies 


ceptilla 
cereola 
cernictilum 
cervice 
cervunus 
chirugia 
Cicada 
Ciconia 
cimice 
cinttra 
cingtla 
circé1llus 
circtilare 
cithéra 
civitate 
clavellus 
clavictla 
coactila 
coctura 
codice 
coemeterium 
cognatus 
cogitatus 
cold&phus 
coleone 
collacteus 


collecta 


2oy 


Spanish 


cebolla 
ciruela 
cermeja 
cerviz 
cervuno 
sirujia 
Cigarra 
ciguefia 
zisme 
cintura 
cincha 
ZALCINLO 
cerchar 
citara 
ciudad 
clavel 
clavija 
coalla 
cochura 
codigo 
cimenterio 
curiado 
cuidado 
colpo 
cojén 
collazo 


cosecha 


eit 


BiviIso 


sifuaie 

7 
SrIseto 
sRetpD i 
*  ‘gmeis 
eIsn to 


Brioni 


tarisxso 
atetto 
Heian 
ievslo 
BEivBIS 


silsoo 


_ aasttageseo 


, 


shorn | 


sincoi5 


coimto 


Stet 


sitprits 
ani ison 
ent thers to 
srsritio 
atitivic 
aulfave {> 
efipivnts 
6 tia BOD 
&20ID00 


ad 


Sr IDOO 



















Latin 


collibertus 
collina 
colore 
colostrum 
colobra 
coltmellus 
comi'te 
comesti6ne 
commissum 
communicare 
compania 
compdsitus 
comptitus 
conchiila 
cénctiba 
consflium 
consécer 
consdlida 
consitiira 
contentione 
contractus 
contrata 
convenium 
oéphinus 
copula 
coquina 


corbita 


Spanish 


culber 
colina 
color 
calostro 
culebra 
colmello 
conde 
comes t6n 
comiso 
comulgar 
compafia 
compues to 
cuento 
concha 
quéncuba 
consejo 
consuegro 
consuelida 
costura 
contensidén 
con trecho 
contrada 
convenio 
cusvano 
copla 
cocina 


corbeta 


Latin 


cordolium 
coriamen 
comfctila 
cOrniitus 
corona 
coronisa 
corticea 
cortina 
costatum 
cottirnice 
creattra 
credencia 
are 
carbéllum 
crudélis 
criistdsus 
cubictlum 
cuculla 
cuculliata 
cictillus 
cuctilus 
culcita 
cultéllus 
cuniciilus 
cupélla 
curctilidne 


Vv 5 
cuscullium 


Spanish 


cordojo 
corambre 
comeja 
cornudo 
corona 
cornisa 
corteza 
cortina 
cos tado 
cordomiz 
criatura 
creencia 
quebradura 
grabillo 
cruel 
cros toso 
cobija 
cogulla 
cogujada 
cogollo 
cuquillo 
colcedra 
cuchillo 
cone jo 
cubillo 
gorgojo 


coscojo 


260 


BCORTEOO 


silitHti teed 


oe 
apc 


eae } 
- > 










— 





tit 


we 

- 
rs _? 
- 


Latin 


cutina 
daculum 
debilis 
débita 
debitdre 
decima 
delicAtus 
denarius 
derotatus 
dentale 
dentatus 
diabtilus 
digitus 
dominicus 
dubi'ta 
ebriacus 
elemosna 
fabtilare 
factila 
fallita 
februarius 
fentictilum 
feraméntum 
filacia 
filatum 
filiaster 


flébilis 


Spanish 


codena 
dalle 
débil 
deuda 
deudor 
de Ima 
delgado 
dinero 
denodado 
dental 
dentudo 
diablo 
dedo 
domingo 
duda 
embriago 
limosna 
hablar 
faja 
falta 
febrero 
hinojo 
herramienta 
hilacho 
hilador 
hijastro 


feble 


Latin 


focacea 
foetibtindus 
follicare 
foramine 
formaceus 
fortuna 
fricamentum 
fr1ctura 
fronddsus 
frontale 
fundamentum 
orgetaitna 
gabtilum 
ballicus 
garriilare 
gemellicus 
genesta 
gentictiLum 
glandtila 
globéllus 
grammatica 
habitactiLum 
habitus 
hastile 
héd&ra 
hemicrania 


hérbosus 


Spanish 


fogaza 
hediondo 
holgar 
horambre 
hormazo 
fortuna 
fregamento 
fritura 
frondoso 
frontal 
fundamento 
horcilla 
gable 
galgo 
garlar 
emelgo 
hiniestra 
hinojo 
landre 
ovillo 
gramatica 
bitacora 
habito 
astil 
hiedra 
migrafia 


herboso 


261 


esix3 
eeobrott 
Esinort 
cctmansbrws 
ak Li me ; 
gicinp 
opdisp 

ss inap 

op lars 
sitaorniri 
oporirt 


errbyts.! 


Brito 
passiremboibsd 
ane txt 
apZzabnot? 
olstnoxt 
metreamebot 
st Lioxo2 
muitisp 
suai fisd 
a6 Lirrep 
ayo rilanse 
exasnsp 
mu Lions 
aibasIp 
eu ilSdolp 
soLtepmsrp 


mms Stross kde 


and tdert 





Latin 


hereditare 
hibérnus 
hinnittlare 
hocanno 
homicidium 
honore 
horologium 
horrore 
héspite 
humiditate 
humilis 
hutica 
ilice 
imagine 
imbrece 
inéubus 
indicum 
infante 
infémum 
inféstus 
ingenium 
inimicus 
insania 
insignia 
insGmium 
intéger 


intestina 


Spanish 


hered:a'r 
oa 
relinchar 
ogafio 
mohecillo 
honor 
reloj 
horror 
huésped 
humedad 
humilde 
hucha 
ilce 
imagen 
embrice 
éncobo 
indigo 
infante 
infierno 
infiesto 
engefio 
enemigo 
safia 
ensefa 
ensue ho 
enterco 


estentina 


Latin 


intybus 
intonce 
inversum 
invitus 
ironia 
januarius 
jejuna 
jocalia 
judaes 
jumentum 
juramentum 
junctiira 
lebré1lum 
labrusca 
lacertus 
lacrima 
lactuca 
lacuna 
lancearius 
lanosus 
lappaceum 
laridum 


latrone 


latrosinium 


laudator 
latinos 


lavatorium 


Spanish 
envidia 
entonces 
inveso 
amidos 
ronia 
enero 


ayuno 


262 


chocallacho 


judfo 
jumento 
juramento 
juntura 
lebrillo 
labrusca 
lagarto 
lagrima 
lechuga 
laguna 
lancero 
lanoso 
lapaza 
lardo 
ladrén 
ladrocinio 
loador 
latino 


lavadero 


Onmsys 
offbeliscod> 
ORD C 
QaaML (. 


ognetssut 


" of Lines 


spears | 


audivned 


~~ 


“p rert i 
SU PkAUNG 
Ser 

5 LI59Of 

‘ geskerr 
mutes fe 
mysceams iy [ 
estore 
msi Serxded 
so-urrdst 
assxsosl 
clean 
eoicnoal 
srirosi 
euirasornet 
aus | 
musspsqgal 
mip isl 
arts 


L 


il 


. 


4 


L 


a 
# 
¥ 


FEEL 


»> 








iti 


; 






Latin 


legalis 
legumine 
18ndine 
lentictila 
leprosus 
1épdre 
levarius 
levitum 
libélla 
licinium 
1ignamen 
1Ygnarius 
limaceus 
limite 
liminaris 
limosus 
limpidus 
lineola 
Linteolum 
littéra 
1Yxivum 
lécalis 
loc@1lus 
lucerna 
lunbricula 
lumbricus 


— y 
lumine 


Spanish 


leal 
legumbre 
liendre 
lenteja 
leproso 
liebre 
ligero 
levdo 
nivel 
lechino 
lefiame 
lefiero 
limaza 
linde 
limnar 
limoso 
limpio 
linura 
lenzuelo 
Letra 
legia 
lugar 
iuGiL LO 
lucema 
lanbrija 
lonbriz 


lumbre 


Latin 


luparius 
luscinidus 
machina 
mactila 
madore 
majore 
malifatus 
malignus 
malleélus 
mamilla 
manciola 
mancipium 
manica 
manictila 
manicus 
mantptlus 
mantaica 
mantele 
mant@1lum 
manualis 
manuaris 
Manuopera 
marcuLus 
margarita 
maritimus 
martellum 


J 
masculus 


263 
Spanish 


lobero 
rosignol 
naquina 
malla 
mador 
mayor 
malvado 
malina 
majuelo 
mamella 
manzuelo 
mancebo 
manga 
manija 
mango 
manojo 
manteca 
mantel 
Manteo 
manual 
manera 
manobra 
macho 
margarita 
marisma 
martillo 


macho 


robe 
soo 


obay Lar 


sotien | 


oles pein 


eliaren 


ofzersitem 


ApEn 
st insm 
open 
ofonem 
BOSH 
Legian 
oad nem 
Cereteec 
sian 


SIGE 


ortossat 

Bo Psp ian 

al Lists 
ogeent 


Sees 


aan bumegul 

ita tome 

aud atoel, 
~~ S 

See as a, S| 


‘Si Moam 


"ad 


Sstedem 
syof BM 
v. 

puter Lam 
aun £Dsor 
eulteliian 
ef Uimean 
' sLoLfprian 
») tl 
mu be Foren 

Y 
SO inten 
6 ftioinem 

Vv 
2uD2nsm 
ex Liiogieein 
£5.i63 nem 
of Set crest 
mht St asa 
2iisuaan 


ees 


STaau nem 


an Liowem 

: 
aunts Lem 
a 


tm 


omnes. 


3 


OF 


~*~ 


exemt 


if 


’ 


olignit 


of 


Geese: 


vill 


at 


a 


“5 





Latin 


matexa 
materia 
matrastra 
matrisilva 
matrice 
mattianum 
maturus 
maxilla 
maximum 
medietate 
mediilla 
melancolia 
melimetum 
memorare 
mendicus 
mesura 
mercatore 
mercatus 
mercéde 
merénda 
mérula 
mespitus 
metipdnus 
metitore 
mettla 
milimindrus 


millarium 


Spanish 


mateja 
madera 
madrastra 
madreselva 
madriz 
Manzana 
maduro 
mejilla 
majimo 
mitad 
peolla 
malenconia 
membrillo 
lembrar 
mendigo 
mesura 
mercador 
mercado 
merced 
merienda 
mierla 
mispolo 
mimo 
medidor 
myja 
milmandro 


mijero 


Latin 


4 . 
minacia 
v= 
minimare 
ministrium 
Ww . 
minutia 
v= 
minutus 
. wo 
miraculum 
miscilare 
miscellus 
eae geen eo) 
m_xt1cius 
mixtura 
modtilus 
molestia 
molinarius 
molinum 
mollitia 
moneta 
montanea 
montensis 
monticello 
monumentum 
J . 
murdacia 

v 
mordicus 
morellus 
mortalis 

Vvi— 
mortalitate 
mucrone 


mutrone 


264 


Spanish 


menaza 
mermar 
mester 
menuza 
menudo 
milagro 
mezclare 
miesiello 
mestizo 
mistura 
molde 
molestia 
molinero 
molino 
molleza 
moneda 
montafia 
montes 
monticello 
monumento 
mordaza 
muerdijo 
morella 
mortal 
mortaldad 
mugron 
muflon 


Tiled 


: PMisiosan 
olfei2sim 
os itfeam 
wuss ain 
sblor 
sitesion 
oreanifon 
ori fom 


ssetfion 


sitstnom 
2ectirom 
oLfseistrmam 
odcenncRm 
gEshrom 


of ifrraum 


iii 


stoantn 

«= & + 
myiutetiim 
6 thunkn 


Owe 


- Ww 
SISore a 
imu Lee fm 

- Ww - 
exys THOS iM 


apt Fanain 


an Tibom 
sitpaion 

au fe rek fom 
murie fom 
sit ition 
st5nom 
SeRSING 
efanasnon 
offen strom 
Mudra 
sBiosbutim 
ato tbytim 


ay Liezom 


H 





Latin 


musculus 
mustidus 
mtitilione 
nasturcium 
natatore 
natica 
nativitate 
navicella 
navigium 
nebiila 
negotium 
nigellus 
nitidus 
Vv A 
nivaria 
M5 

nivosus 
nddé1lus 
nodosus 

v 
nomine 

yr yews 
nominare 
novale 
novémber 
novicius 

v 
nubilus 
nucalis 
nucetum 
obscuritate 


obscurus 


Spanish 


muslo 


mustio 


modillon 


mas tuerzo 


nadador 
nalga 
navidad 
nacella 
navio 
niebla 
negocio 
niel 
neto 
nevera 
nevoso 
nudillo 
nudoso 
nombre 
nombrar 


noval 


noviembre 


novicio 
nublo 
nogal 


nocedo 


escuridad 


OSCUrO 


Latin 


absidium 
octavos 
octuber 
octilata 
octilu 
oliva 
olivus 
onocrotélus 
opéra 
operarius 
opinione 
ordine 
orgaénum 
orig&num 
Orphanus 
ortilare 
ovictla 
pagensis 
palatium 
pallidus 
paltmbinus 
palimbus 
pampinus 
panaricium 
panarium 
pantice 


vif 
panucula 


Spanish 


osedio 
ochavo 
octubre 
oblada 
ojo 
oliva 
olivo 
ocroto 
obra 
obrero 
opinion 
orden 
organo 


oregano 


huérfano 


orlar 
oveja 
pais 
palacio 


palido 


palomino 


palomo 


pampano 


panadizo 


panero 


panza 


panoja 


265 


OVE<bO 
sxcist 55 
Shs ido 


okos Leg 
onnlag 
onimofsg 
OMOLEG 
os thanag 


C8SNSq 
BSMSG 
BrDaRE 


eared 


muy ib sacs 
SOVBIDO 
wetisso 
. Bisiivo 
uirce 
sy Elo 


a a ae 
aivito 


oY 
SS LOS OFIONO 


TUTTO 
mInsplxc 
pes titkyIG 

oxmlino 

sli tro 
BLeASp5q 
mritai Tat 
abi lisa 
nae, 
auc romleg 
wv 
euctariag 
\ 
SxTIQMSq 
my fs i608 g 
wetinerisg 
W oe 

omer ree! 

siliagnag 


sileosn 


olven 


OLD0PpaiT 
) isia 


SISNET 
OROVON 
of Lthun 
oun 


ocmon 


oipiven 


Legos 





Latin 


parabdla 
paradisus 
paratella 
parcellus 
parentatus 
paricila 
pardchus 
parricus 
partecélla 
pastura 
patélla 
patraster 
patrinus 
pauperitate 
paviméntum 
peccatum 
pedaneus 
pedicus 
pelagus 
pelliceus 
perfectus 
periculum 
persona 
perticus 
pictura 
pigritia 


pinnactilum 


Spanish 


palabra 
paraiso 
paradella 
parcello 
parentado 
pareja 
pArroco 
parque 
partecilla 
pastura 
padilla 
padrastro 
padrino 
pobredad 
pavimiento 
pecado 
peafia 
piezgo 
piélago 
pebliza 
perfecto 
peligro 
persona 
piértica 
pintura 
pereza 


penacho 


Latin 


piscatore 
placitum 
plancdla 
pluviosus 
polenta 
pSpulus 
porcarius 
porcellus 
portarius 
posticius 
podéstate 
potidne 
practica 
praepositus 
presepe 
primicarius 
principe 
probaméntum 
privatus 
proftindus 
proximus 
pugnale 
pullitione 
pullitru 
polmone 
polpdosus 


puritate 


266 


Spanish 


pescador 
pleito 
lancha 
lluvioso 
pulienta 
pueblo 
purguero 
porcel 
portero 
postiz 
podestad 
pozon 
platica 
preboste 
pesebre 
primicero 
principe 
probamiento 
privado 
profundo 
projimo 
pufial 
pollaz6én 
poltro 
pulmén 
pulposo 


poredad 


= 


(a 


= 

7 _ 7 
ee, 

ae : 


ee nae 


_ eeeypxg srreRoseg ae 
Este ass Lf wi08K _ supirieg 
oxstixeg euiediod s Libosdisq 


sitact: an to bdeog sxiktesg 
en edstesbog — slittsg 
aii enBi-toa isdessq 
ease in apttosxg aunkiteq 
agpodsts ans t2oqaerq ett ixagusg 
“gadeesd awesrq mus merakveg ) 
ono imix ey iiso Inrsg mite309q 
eqrontig eqinabrg ’ axronsbsy 


QOnSiMeEdog 
obay Iq 
emtirorg 

ip fer 
(fBxsiiog 


oxtiog - 


Sabgied 


ari iniyeey 


nutrienedorse 
ajay irq 
aubaiiorg 


Hi 





pik 


ee 


ee 





Latin 


purplira 
plistélla 
putatore 
putatoria 
putearius 
ptuitone 
quadratus 
quassicare 
quatemus 
quindécim 
quimddo 
rabiosus 
racemus 
radicalis 
radice 
ramosus 
rancidus 
ranoore 
rapace 
rapidus 
rasicare 
rast41llus 
rasura 
rebucinare 
recoctus 


refindictla 


Spanish 


purpula 
postilla 
podador 
podadera 
potero 
podon 
cuadrado 
cascar 


cuademo 


quince 


rabioso 
racimo 
raigal 
raiz 
ramoso 
rancio 
rangor 
rabaz 
rauda 
rascar 
rastillo 
rasura 
rebuznar 
recocho 
rebendija 


real 


regésta 

aS 
regina 
regula 

11] 

Vv 
renicus 
renione 
respectus 
restticulum 
retinaculum 
letrorsus 
robine 
robéretum 
robus tus 
romana 

v 
romanice 
rotélla 
rottilus 
retondus 
rubinus 
ruina 
rumore 

vv 
ruptura 
ritabellum 
sabanum 
sabbatum 
sabina 


sabuga 


267 
Spanish 
registro 
reina | 
regla 
remolque 
renco 
rifién 
respeto 
restojo 
rendape 
redoso 
robin 
robledo 
robusto 
romana 
romance 
rodilla 
rollo 
redondo 
rubi 
ruina 
rumor 
rotura 
rodavello 
sabana 
sdbado 
sabina 


sabuga 


RAL S ie ALi t Stahepog 
SOIT . soo 
i. ee ree } 

7 my 


ODS pal 


ye J f : 4" 
os CL SF 
Oz ~ 


SsoOrnmipp 


pecs: MILO CMS 


LCST TLDS 


Sonat 
Lopes 
LOS ASI 
(Sos 
beret 


SCH Carr HDS! 


& A a - 
eat le oll ites: 
ioe : as SIWeS% 
rie i 


TEASE 
52 ODIs 
sriidse pc idsa 8 Loredox 


snide BpucBe feos 


suBotds:s 
ELIMSOST 
ailso rhs 


. 


eoi.he7 





Latin 


sabuldone 
saectlum 
saetacium 
sagitta 
Sagina 
salicarius 
salice 
saltbris 
sanctitate 
Sanguineus 
sapidus 
Saracenus 
sarcillum 
sartore 
scabellum 
scarpellum 
scand4la 
schedtila 
scholaris 
scintilla 
scopilia 
scorpione 
scutélla 
secundum 
securus 


sellarius 


seméntis 


Spanish 


sab1lé6n 
Siglo 
sedazo 
saeta 
Saina 
salguera 
sauce 
salobre 
santidad 
Sangue fio 
Sabio 
Saraceno 
sarcillo 
sastre 
escabello 
escarpello 
escanda 
cedula 
escolar 
centella 
escobilla 
escorpién 
escudilla 
segin 
seguro 
sillero 


Simiente 


Latin 


seniore 
septimana 
w/ 
serica 
serpénte 
serviente 
sibilare 
v — 
siccaturus 
. wv 
sigillum 
we 
Signale 

vw 
singellus 
ue 
Sinister 
smaragdus 
sobrinus 
societatis 
solanus 
sollacium 
y, 
solidus 
solitate 
Ad 
sorice 
sortictla 
Vv 
spatula 

~~ 
speculum 
ow 
spiculum 
spinosus 
spirauca 
rv 
spiritu 


splendore 


268 


Spanish 


senor 
semana 
jerga 
serpiente 
sargento 
silbar 
segadero 
sello 
sefial 
sencillo 
Siniestro 
esmeraldo 
sobrino 
sociedad 
solano 
solaz 
sueldo 
soledad 
sorce 
sortija 
espalda 
espejo 
espligo 
espinoso 
espiriga 
espiritu 


esplendor 


Otabspee 
efise 
{si 92 
oli — 
OPI 2OL11 ie 
abisrenes 
OLixuoe 
bahs — 
ons.Loz 
S6i02 
ob faye 
behis (oz 
SID: 
Efiszoe 
sb Leq2s 
ofsxKes 


opriges 


ORXMLEGZS | 
spiriges . 


ies 


sioinee 
srremisgqese 
eo htee 
eirimtense 
Safiak rise 
frefrdce 

; 
ATI BOLDL E 


a 
MU LLL pre 


err rioe 


stE i Loe 
estioe 
sivoLtros 
sittsqe 
mw itosge 
muito ige 
Aeon hge 


SQUVETIGA 


olladsves 


ol 


cil mdoues 


OSoBTS2 


of Liouss 


Laqisoss 
sbnsoas ” 


’ 


~ 


6 Luhen 
15Losas 


si fortis 





Latin 


sportella 
squalidus 
statione 
stabilis 
sterilis 
stemutus 
stom&chus 
stranurria 
strictus 
sub%lla 
subterraneus 
sucidus 
sucosus 
sudore 
suffictus 
stpéranus 
stperctilus 
stpernus 
supérare 
slispectus 
tabanus 
tabilla 
tabemarius 
tabtiLa 
talentum 
tamarice 


tapetum 


Spanish 


esportilla 
escalio 
estacién 
etadal 
estéril 
estomudo 
estémago 
estranurria 
estrecha 
cubilla 
subterr4neo 
sucio 
SuCOSO 
sudor 
sofito 
sobrano 
sobejo 
soborno 
sobrar 
sospecho 
tabano 
tabilla 
tavernero 
pie 
talanto 
tamariz 


tapete 


Latin 


tegila 
temOne 
tenace 
tesione 
tentione 
tepidus 
terebinthinus 
terraneus 
territorium 
terrosus 
tertiarius 
tertiolus 
talamus 
timore 
tinctura 
tineosus 
titione 
tittilus 
tomentum 
tormentura 
torpedo 
torace 
tortione 
tosione 
trabictlum 
traditione 


traditore 


269 


Spanish 


tecla 
timdn 
tenaz 
tesén 
tenzon 
tibio 
trementina 
terreno 
terrotorio 
terroso 
tercero 
terzuelo 
talamo 
temor 
tintura 
tifioso 
tizén 
tilde 
tomiento 
tormento 
torpedo 
toraz 
toroz6n 
tuson 
trabejo 
tracién 


traidor 


ad 


Aeris 


Faye 
ta 


mr 
of, Sass 


no inet 


Itobss13 


id el Hetomage 
_ ~ emet o 


BLURS sli sstoges 


y Ys 
{ = ETT. 3 
Pee 
Steg 
ee as 
z 
Fats 
7 
; 
E 
Oz 
emi = 
tO AS 120 
v Fad 
= obo U er 
rey : 
fe Zi TORT 
ms oe! SATE 
CONE pF eat nirtK? 
~ : - 5 ew 
oy, ae eer P 
PLL SOS Y oesjsa 


OFOLT DEE? < By 


10s Le Steqs? 


auensristdye 


etfidsie 
ati ixate 
ausumeade 
anriodmode 
sirmneie 


eutolid 2 















6f Tcive 


sub rove 
SxrBCoUe 
srouue2 
axtortioa 
et) ‘on Scnve 
sultorsqte 
aurrrsqye 
SIBISQua 
actoeqare 
eunsdst 
ailicdes 


ey tvearnrrods3 


Latin 


tramite 
transversa 
transvérsus 
tredécim 
tribtilum 
trillice 
trimensus 
trimodio 
tristitia 
triticum 


trunctilus 


Spanish 
tramite 
traviesa 
travieso 
trece 
trillo 
terliz 
tremes 
tramoya 
tristeza 
trigo 
trancho 
tobillo 
tolva 
tollo 
tonga 
turbio 
torbisco 
tortolilla 
urce 
olmedo 
loco 
ombligo 
untura 
ondoso 
una 
unicornio 


orina 


Latin 


a 


tirtica 
usura 
vadosus 

— 
vagina 
vaginélla 

v 
venabulum 
venatore 
venatorius 
venatus 

ee 
venditare 

~ = 
venditore 
venéno 
veneria 
vendosus 
ventoso 
ventrisca 
ventrisctila 
veranum 
verbéna 
verecundia 

w 
vermiculus 
verruca 

ad 
versicus 

vw. 
veruculum 
verbactum 
vessica 


vestitura 


270 


Spanish 


ortiga 
usura 
vadoso 
vaina 
vainilla 
venab1lo 
venadore 
venadero 
venado 
vendar 
vendidor 
veneno 
venera 
venoso 
ventosa 
ventresca 
ventrecha 
verano 
verbena 
verguenza 
bemme jo 
verruga 
bizco 
berrojo 
barbecho 
vejiga 


vestidura 


a ¢ ’ 
feinsge . cte.t Se Lane 


— 


© ohne ec snsi? 


uy Se oO 
xy BEeaivsiz 
Lu TUE 7 
wOPSV ous lyst 
‘ = xas 
Bi SV 
} coll 
, off 
LLL CV 
e an Be 
yi Cal A 
miGq 
a> 
‘ 
Ss 
i LJ 
. oe 
7 1-? ‘ Ot => , 
~ fei » 
’ 
32%! x 
- 4 st o> 
= ery its wo! 
eit ate, dF wxod 
C¥ib 2 DA ACI 
= eH 5 
7 ¢ 8 ; { F i. aA 
¥: f j ot 
wri 2 ed Cx 
: Low ea peti 
“tse rey Lert EK 
winechisc iret oecroea Eig 
HS Lee COTtOO Lia} 
=p 
eye 


Rasy s1asiJ2ov BILEIG 


od Tmerst 
Seis ens 
srextvensit 
minshsxt 
mLidisd 
eOril ing 
avensnint 
ofborir 
5it dein 
mui ie 
eu Lftorars- 
mf [Sdicrt 
alc 

ex Tirotiis 
solic 

aub ici 
puoe Echitet 
si fsosusc 
sobfg 


moctrrente 


.¢ 


Ww» 
4 ae 8 





vestitus 
Yiv 
vetulus 
viburmnum 
vicata 


Vr 
VLCilnus 


ab&cus 
abadesa 
abbate 
abbatia 
abé11ana 
abéte 
abigone 
abismo 
abusus 
acacia 
acceptore 
accessus 
acedia 
acétum 
acierium 
acuctila 
acularius 


aculeatus 


Spanish 


vestido 
viejo 
viomo 
vegada 


vecino 


Latin 


villanus 
vindemia 
vindicta 
virgine 


viscosus 


Latin-Portuguese 


Portuguese 


abaco 
abadesa 
abade 
abadia 
avellé 
abeto 
abigaso 
abismo 
abuso 
acacia 
eh 
acesso 
acidia 
azedo 
acero 
agulha 
agulheiro 
agulLhado 


Latin 


acutus 
acutia 
acumen 
aditus 
admordium 
adoctilare 
ad, retro 
ad satis 
adversarius 
advocatus 
aequalis 
aeramen 
aestivus 
aevitate 
affora 
affricus 
agina 


~~ 


agninus 


ake 
Spanish 


villano 
vendimia 
vendecha 
virgen 


Viscoso 


Portuguese 


agudo 
acuca 
gume 

eido 
almoco 
aolhar 
arredo 
assaz 
adversario 
(a) vogado 
igual 
arame 
estio 
eitate 
afora 
Abrego 
asinha 


aninho 


; “y Vv 
} > F Se ITi To Ly 
y * 
» 
> Oe Sa 
— ‘ 
4 STROe 
,% te 
; 
CLE 
KX 5 
Se retice 
6BiD506 
‘ NESE 
Blie2eenOsS 
a ‘ 5 LOSOE 
te 
(iFISOsS 
TE LISIDE 
= ' oo Te —" 
, sltiouns 
re - me 


Ci oe eu imeluos 


“<= 


“f BEIT LLYN Lops atctsalios 





Latin 


——— 


agrimonia 
alabastrum 
alamanus 
alatema 
albura 

(Ub as 
alicunus 
altitia 

w 
amaricoous 
amatore 

» Wi —- 
amicitate 
amptillo 
ancora 
anéllus 

v 
anima 
anniciilus 
anténna 

v 
apertum 


wv 
apicula 


apposticum . 


aprilis 
aquarium 
aranea 
arbore 
ardore 
aridus 
armmenius 


ww . 
arrugia 


Portuguese 


aramenha 
alabastro 
alamanda 
ae 
albura 
algum 
alteza 
amargoso 
amador 
ami zade 
empolla 
ancora 
anillo 
alma 

ane lho 
antenna 
aberto 
abe lha 
postico 
avril 
aqueiro 
aranha 
Arvore 
ardor 
arido 
armenho 


arro1io 


Latin 


artemisia 
artictilus 
asinarius 
asinus 
aspide 
astilla 
astrosus 
audentia 
auguarium 
augus tus 
aurictilum 
autinus 
avena 
avisthrusus 
badius 
balléna 
baln#éatore 
balsa&mum 
bancale 
barbutus 
barbtitus 
batualia 
battilus 
bucalis 
benignus 
besttiLus 


bibytus 


Portuguese 


artemista 
artelho 
asneiro 
asno 
aspide 
astella 
astroso 
ouveen¢ a 
aguouro 
agosto 
orelha 
atuno 
aveia 
abestruz 
baio 
baleia 
banhador 
balsamo 
bancal 
barbudo 
barvos 
batalha 
balde 
bocal 
benigno 
bicho 


bebado 


SDE ATI BOL nite 





SetictieTe SLamirrps 
cagyaadale vuryj esds fs 
eunemeals 


rot is smrstsis 


au Lf tre 


“ 
FUTLI Ite 


\ 


en {ies tern 

sniStris 

mus ISGE 

eLitot a 

mutt eoqas 

ei frags 

f my Ct AUDS 
Pert, BOOBS 


— 


More eTonhs 


~ 


SIObis 


be 


Shy eub ius’ 


oat oriiaeirzs aus Livers 


Lic olor sitprss 





Latin 


bicarium 
bicornius 
biferus 
biscoctum 
boarius 
bonitate 
brachiale 
bufulcus 
bucctila 
burd6ne 
borragine 
burricus 
burtila 
buttictila 
cavalla 
caballicare 
caballus 
caeremonia 
calamellus 
calamo 
calcaneum 
calceamentum 
calceatus 
caldaria 
calaveria 
camahaeus 


carmella 


Portuguese 


pichel 
bigorna 
bébera 
biscato 
boeiro 
bondade 
bra¢cal 
bifolco 
broca 
bordao 
borragem 
burrico 
burla 
botelha 
cavalla 
cavalgar 
cavallo 
comminha 
claramela 
calamo 
calcanho 
calc amento 
calg ado 
caldeira 
caveria 
camafeio 


gamella 


Latin 


caméra 
camerarius 
= 
caminus 
camisia 


campania 


campicellus 


campestris 
canalictla 


candela 


cande labrum 


canictila 
canname lus 
canonicus 
cant4rus 
canticum 
cantione 
capacium 
capanna 
capi llatim 
capi llus 
capistrum 
capitanus 
capone 
capella 
capellus 
caprarius 


caprone 


Portuguese 


camara 
camareiro 
caminho 
camisa 
campanha 
campece lo 
campes tre 
guelha 
candeia 
candeiro 
caneja 
carame lo 
conego 
cantaro 
cantiga 
cane 80 
cabaz 
cabana 
cabelladim 
cabelo 
cabrestro 
capitéo 
cabao 
cabela 
capelo 
cabreiro 


cabra0 


BHLSOD , 
fisb055 
On ass 

Sore} 
alemss pa 

oven 30 


Or63°usD 


oletigso 
orjs2axcs> 
GBI iais> 


Oda 


ZOLA 
~~ 
sieimen 


mad POR 


6 tas 
ap LemneriyeS 
¢ 
Sol rariso 
SUYEDS OBS 
NEKO rpg! <S) 
OAL rus 
ini SIBOBD 
SfULSQEo 
; ~ 
mise Ll toss 
a 
ELi1t, SXISD 
v 
Mey TI SLED 
au nes ias> 
ames 
si leqs: 
_ sel Leggo 


a 


BE 
gtiteded 


el isvso 








Latin 


caprunus 
=, 

captivus 

Ys. 
capulum 
carabus 

wv 
caraculum 
carbdne 
carbonaria 
carcére 
carcerarius 
Sf 

cardinus 

wv . 
caricia 

— 
carina 

~/ . 
caristia 
carmacius 
carnalis 
carmarium 
camutus 
caronia 
carpentarius 
carraria 
carrasca 
carvalya 
casalis 
casarius 
cascabellus 
castanea 


cast#1lum 


Po. se 


cabrum 
cautivo 
cabre 
caravo 
caralho 
carbao 
carvoeira 
carcere 
carcereiro 
cardeo 
careco 
querena 
carestia 
carnaz 
carnal 
carmeiro 
carmudo 
carronho 
carpinteiro 
carreira 
carrasca 


carvalho 


caseiro 
cascavel 
castanha 


castelo 


Latin 


castitate 
cas tore 
castila 
catafalcum 
caté1llus 
catena 
catinus 
cellarius 
centenum 
ceptlla 
cermictilum 
certarus 
cervice 
Cicerone 

IVA . 
ciconia 
cicula 

. ov 
cimice 
cinctura 
cingula 

. LS Fed 
Civitate 
clavictiLus 
cloaca 

vw. 

coagulum 
cochlearium 
cognatus 
cohorte 


collatione 


euen 


Portuguese 


castidade 
castor 
casula 
catafalco 
cadi lho 
cadeia 
cadinho 
celeiro 
centeio 
cebula 
cernelha 
certaro 
cerviz 
ciziraio 
cegonha 
cejuda 
chisme 
cintura 
cilha 
cidade 
chavelha 
colaga 
coalho 
colhara 
conhado 
corte 


colacao 


ofeciso 


piodso 


sitteeo 


muDisisteo 


g 
( jae 
“Suiibteo 
Brats 
eNO LIED 


ag tint feo 


merisct ries 


SEA IVIO 
anltsareio 
Bo950LD 


Na 


mu.Lupsao 





Latin 


collina 
collecta 
colore 
colttbra 
comes tiGne 
comitisa 
commater 
commisum 
companione 
compater 
w 
comptitus 
Ww . 
concilium 
conclausum 
condtictum 


conséger 


constipatus 


consitira 
contractus 
contrarius 
contrata 
convenium 


conventus 


View 
cooperimentum 


coptila 
cocina 

of 
corbicula 


corbita 


Portuguese 


colina 
colheita 
cor 

cobra 
comicha.o 
condessa 
comadre 
comiso 
companh 40 
compadre 
conto 
concelho 
conchouso 
condoito 
consogro 
constipado 
costura 
contreito 
contrario 
contrada 
convenio 
convento 
cobrimento 
cobra 
cozinha 
corbelha 


corveta 


Latin 


comutus 
corona 
Wins 
corrigia 
corrugus 
cortitcea 
cortina 
costatum 
cortumice 
creatura 
credencia 
crepatura 
crus tesus 
cubile 
cuctila 
culcita 
cultéllus 
wv 
cumtilus 
ww 
cunictiLus 
wv . 
curctilio 
wv A 
cursorius 
Vv e 
cusculium 
Ww 
cycinus 
oo) —_— 
debitore 
decania 
decémber 
dinarius 


dendtatus 


Portuguese 


cornudo 
coroa 
corveia 
corre go 
cortec a 
cortina 
cos tado 
codorniz 
criatura 
creenca 
quebradura 
crus toso 
covil 
cégula 
colcha 
cutello 
combro 
coe lho 
gorgulho 
cossouro 
cascolho 
cisne 
devedor 
deganha 
dezembro 
dinheiro 


dentado 


ele 


esTIs 


& pein 


che-tacn 

$ fersahos . 
REEISIO 
By fiGsID 
vliaaiitegss 
oactaue 
Livao 
Sip. 


stio.los 


xi latlin 
\y 

scr lita 
aria tin 

» ty ws 
oOHgeup 
Bt EOE 

» . 
mu Ceres 
Nad 

auitiayD 


ox6s Feb 





despectus 
diacénus 
digitale 
diluvium 
domina 
ducatus 
duritia 
emplastrum 
episcoOpus 
escariola 
examen 
exceptis 
facénda 
factiSne 
factisius 
factila 
falcone 
fallita 
farina 
fenéstra 
fermentum 
forrarius 
fibélla 


fidelitate 


Portuguese 


jusante 
dese jo 
deserto 
despeito 
diago 
dedal 
diluvio. 
dona 
ducado 
dureza 
emplastro 
bispo 
escarol 
esame 
exetes 
fazenda 
fazao 
feiti¢o 
falha 
falcgo 
falta 
farinha 
frestra 
fommento 
ferreiro 
fivela 


fieldade 


Latin 


filancia 
filiatin 
filiaster 
flebilis 
focaris 
fucilis 
£61iatus 
£61idsus 
follictilus 
formica 
formosus 
frictura 
frigidus 
friscura 
frondosus 
frontale 
fumarium 
fumosus 
fundamentum 
furcilla 
furnarius 
furrinctilus 
gabella 
galera 
gallicus 
gallina 


gallinaceus 


276 


Portuguese 


filanca 
fiadin 
filhastro 
febre 
fogueira 
fuze 
folhado 
folhoso 
folhelo 
formiga 
formoso 
fritura 
frio 
frescura 
frondoso 
frontal 
fumeiro 
fumoso 
fundamen to 
forquilha 
forneira 
foroncho 
gavela 
galema 
galdo 
galinha 


galinhaca 






‘ > 
f 7 - 7 
f \ be = 
’ is a 
al a2 
4 meee ne Se 
a 
= 
f “a ~ 
- x 
2 
4 7 
- E aa ; 
en 4 , 
te nee 
7 = 
7 
“ 
~ a 7 
a 


PTE 
itabeal 





tidittt | | 
t 






J 


: 
Tiittilnid 


ne 
~ 


- : 


Latin 


gemmanus 
gemmana 
gertila 
gingiva 
globellus 
gltitone 
gractilus 
granarium 


gtbémmum 


habitactl um 


haerente 


hereditarius 


hibérnus 
homicidium 
hominus 
honore 
horologium 
horrore 
hdspite 
httmérus 
hutica 
imagine 
inbrictilus 
imbutus 
imperatore 


¥ 
Lneensum 


Portug uese 


irmao 
imm 
jerva 
gengiva 
novello 
glotao 
gralha 
granel 
governo 
bitacolo 
rente 
herdeiro 
ivermo 
omizio 
homen 
honor 
relogio 
horror 
héspede 
hombro 
hucha 
imagem 
brelho 
embude 
emperador 
encenso 


endecha 


Latin 


indicum 
infante 
inférnum 
infestus 
inffrmitate 
inflatiGne 
infermus 
ingennus 
{nsipidus 
insapidus 
instla 
tntéger 
interanea 
intermedium 
Yntertignium 
¥ntroitus 
Y{ntrosum 
intybus 
Ynvérsum 
invidia 
iratus 
januarium 
jejunus 
jocularis 
judaeus 
judicium 


jugitLum 


Portuguese 


indigo 
infante 
inferno 
enfesto 
enfermedad 
incha.¢ 40 
enfermo 
engeo 
encibido 
enxabido 
ilha 
enteiro 
entranha 
tremenho 
entertinho 
entrudo 
entrosa 
endivia 
envez 
enveja 
irado 
janeiro 
jejum 
jogral 
jjudeo 
juizo 


Jugo 


onasins 
ebidiors 
obidaxns. 


enti 


By 


omiedno 
Selous ies 
orneneat 
ofigicrratens 
etirtias 
BaOTIINS 
BLY thvia 
sous 
etsvns 


obaxt 


Gries, 


psifOt 





opnetnt 


wae 


en biotin 
auMmroty 
aririsiert.t 
aubigtan’ 
aris rqseni 
sidernt 
coped 
sateroahnt 
my ibenrretet 
mu teipisradtst 
aut bout at 

miacrwtal ’ 
aticdiysrrt 
meet vel 
eat Dees 
= Og ae 8 
| Tp braunst 
aeniipst 

et 


murto toast 


ah 
ras ein 











aw 
oe 
= _ — 


i, _ 
Foal 


- i. 
RG, 4 
_ ‘a a - 





Vv 
= 
_ 


~ 
I 
a 
ca 


7 ay : 
| 
C 
7 - 
a ’ 


ran 


fe 
al 
—_ 












aa 


oe 


sSrxod 
eerie - _F 
7 7a ae r 


g 
Caan © 
al 
t 
see 


ae: 

hd . : _ 
- a5 a S| 
4 rE. —~4 


7 re e 


Latin 


junctura 
juramentum 
juvénis 
labore 
labrusca 
lacertus 
lacrima 
lactna 
laesiGne 
lampréda 
lancareius 
lanoéa 
lantema 
lapathYum 
laridum 
larYce 
latratus 
lauditdre 
latéralis 
latinus 
latrone 
lavatorium 
lavatira 
lectione 
legalis 
legenda 


legumen 


Portuguese 


juntura 
juramento 
jovem 
lavor 
labrusca 
lagarto 
1l4grima 
lagoa 
aleijao 
lamprea 
lanceiro 
lancga 
latema 
lapaca 
lardo 
larice 
ladrado 
loador 
ladraes 
ladinho 
ladrao 
lavadouro 
lavadura 
leicéo 
leal 


lenda 


legure 


Latin 


lendine 
lentictila 
lentiscus 
leprosus 
lep&re 
leviarius 
levitum 
libella 
licinium 
ligamentum 
1f{gnarius 
1igntsus 
legtila 
limite 
liminarius 
limitare 
lindsus 
limpidus 
Linarius 
linedla 
linteoleum 
litania 
littéra 
localis 
lacosta 
lucifer 


linbricus 


278 


Portuguese 


lendea 
lentilha 
lentisco 
leproso 
1ebre 
ligeiro 
1évedo 
nivel 
lichino 
liamento 
lenheiro 
lenhoso 
legra 
linte 
limiar 
lindar 
limoso 
limpo 
lenheiro 
linh6é 
lencgol 
ladainha 
létera 
lugar 
lagosta 
luzeira 


lombriga 



















slledit 
miirteost 
mudremspit 
evitempit 
mur pinp Lt 

| sidpel 

ot init 

auc wentmet 
ethan i 
eyeamit 

au top! 
eutrecct 

| Fides L 
miaioctrrit 


Sings LL 


Latin 


ltparius 
liridus 
1utdsus 
machina 
mactila 

aS 
magicus 
magida 
mMajestate 
majore 
malandria 
malfeactoria 
malignus 
malléus 

Vv 
Manicipo 
manibella 
manitca 

JS 
manicus 

Lo 
manuptlus 
mantaica 
mantéles 
manté1lum 
manualis 
manuarius 
marctilus 
margarita 

Y 

margine 


= 
mariLnus 


Portuguese 


lobeiro 
lordo 
lodoso 
maquina 
malha 
meigo 
malga 
majestade 
maor 
malandres 
malfeitoria 
maligno 
malho 
mancebo 
manivella 
manga 
mango 
manolho 
manteiga 
mantaes 
mantilla 
manoal 
Maneira 
macho 
margarita 
margem 


marinho 


Latin 


maritus 
marritbium 
marstippa 
martellum 
martyrium 
masctilus 
mataxa 
materia 
matraster 
matrina 
matrisilva 
matrona 
malticinum 
maturus 
medimus 
medietate 
medulla 
melanchdlia 
melimelum 
mellacium 
memoria 
mendicus 
mesura 
mentastum 
mentitore 
mercatus 


merénda 


eT9 
Portuguese 
marido 
marroyo 
marsopa 
martelo 
marteiro 
macho 
madeixa 
madeira 
madrastra 
madrinha 
madreselva 
matroa 
maceira 
maduro 
meigo 
mitade 
miola 
menencoria 
mame lo 
me laco 
memoria 
mendinho 
mesura 
mentastra 
mentidor 
mercado 


merenda 





Sey en 


sinedtaa 
tTstesxien 
sokrsen 
spe saltcsiens 
Shere 


meer: Lear 


tk 
mae . | 


> 
> , 


bil 





7 


mae 


oy 
i 
‘awe < 


i 





rt 


7 f b&b 
rat oe is 
: ; 

Zz 
vi 
- 


- 7 
La 6 - 


1 
J 


‘- 


yy 


® 


7 


- 7 


ce. 


ae 


— 


ie 
a 


‘vo a BE Paes 
so 


a 
mon 
7 


Pt 


= 
- 


mr 


1 


=” 5 
> 
= 


_ 
aaa 


_ 
ae fe 


Be re 


Posy 
oe Cla 
ae | 

> 


‘ 


a 


_ 


: 





Latin 


merula 

W 
nespilus 
miliaria 
ministerim 
es 
minore 

NWle caaetaie 
minutias 
Vu 
minutus 
merabilia 
miractilum 
meséllus 
~~ w 
mixticius 
mixtura 
modtilus 
molinarius 
Vi 
mulina 

Say 
mollitia 
CP i: 
manuculus 
monastarium 
montanea 
monumentum 
mordacia 
mortalis 
mortalitate 
mortarium 
mortitcinus 
yv-_ 
mucos us 


a 
murcidus 


Portuguese 


melro 
néspera 
milheira 
mester 
menor 
minu¢cas 
miudo 
maravilha 
milagro 
mesello 
mesti¢o 
mistura 
molde 
moleiro 
moinha 
molleza 
monge 
mosteiro 
montanha 
monumento 
morda ¢a 
mortal 
morteidade 
morteiro 
mortezinho 
moncoso 


murcho 


Latin 


murena 
musaranea 
mus tione 
muti lidne 
myrtinus 
nasturcium 
natatore 
natéca 
natiGne 
navigim 
nebitila 
necromancia 
nepéeta 
nibtilus 
nigella 
nigellus 
nit%dus 
nivosus 
noctivdlus 
novactila 
novale 
novellus 
novémber 
novicius 
nubilis 
nucalis 


nticarius 


280 


Portuguese 


mureia 
musaranha 
muchde 
mode lha&o 
mortinho 
mastruco 
nadador 
natega 
nacaéo 
navio 
névoa 
nigromancia 
neveda 
nebri 
nigella 
nigello 
nédeo 
nevoso 
noitib6 
navalha 
noval 
novel 
novembre 
novico 
nubio 
nogal 


nogeira 


sishemosbis 
Bbsvon 
fidan 
siflepin 
oliepin 
oeher 
QOfo7a1 
Od fon 
silsven 
L8von 
Lavon 
Sereno 


Oyrven 


nident 


Brissrem 
Sane IBS 
eget bean 
cai 14 Bas 4! isi 
mu Iarsteso 
aratatsn 
BDegrsn 
eno.tan 
’ ~~ 
Lt Piven 
& Eide: ] 
5 LOT Sst 
siseen 
eur Lerclirt 
bLispin 
axifoo rn 
wy 

eiibiltia 

— yw 
Queoy fer 
eu Poritson 
sitcsvon 
aiavon 
etyt beor 
tacivon 
eciotver 
ekbicon 

* he 
aLlioun 





Latin 


oblata 
obscuri tate 
obscurus 
obsidium 
octavus 
october 
octilus 
oliva 
olivus 
opéra 
opinione 
oraciilun 
ordine 
ovictila 
ovile 
paganus 
palatimus 
pallYdus 
pallore 
pampYnus 
panaricium 
panarium 
pandora 
panellus 
papilidne 
parabtiLa 


paradisus 


Portuguese 


obreia 
escuridade 
escuro 
assedio 
Oitavo 
outobre 
olho 
oliva 
olivo 
obra 
opiniao 
orago 
ordem 
ovelha 
ovil 
pagao 
paladim 
palido 
bolor 
paémpano 
panariz 
paneiro 
bandora 
painel 
pavilhéo 
palavra 


paraiso 


Latin 


paramus 
parentatus 
parictla 
parictilus 
parréchus 
parricus 
parviilus 
pastellum 
pas tore 
patraster 
patronus 
pavimentum 
pavore 
peccatum 


peccarius 


pectinctilus 


pedale 
pedftum 
pedone 
pedtictilus 
pellamen 
pellarius 
perfectus 
pergaminum 
perictilum 
persGdna 


pertica 


Portuguese 


2 


paramo 
parentado 
parelha 
pare lho 
parroco 
parque 
parvo 
pastel 
pastor 
padrastro 
padrao 
pavimento 
pavor 
pecado 
pegueiro 
pentelho 
peal 
peido 
pedao 
piolho 
pelame 
pelleiro 
perfeito 
pergaminho 
perigo 
pessoa 


percha 








ees dl ca . 
ee ree 


py fri U 


Ve 


yi eee a 








efabea / 
mes Ftc; 
srmibad 


fT SIG 


vuisooed 
an Monttosd 
reais Lf 

; an irs flag 


oribadnsq 
iseq 
obi; 
Oatr5 
aslo 
amp fq 
ce daey 
sagas 
— 
: 





Latin 


pestellum 
petrarium 
pigmentum 
pigritia 
pildsus 
pinétum 
pinnactilum 
pfscarius 
piscatore 
piscosus 
plumacium 
pluvidso 
portarius 
portatore 
postrarius 
potestate 
potidne 
practica 
praebenda 
praepositus 
presépe 
precaria 
prmarius 
principe 
profectus 
proinde 


propingquus 


Portuguese 


pestilo 
pedreiro 
pimenta 
preguica 
peldoso 
pinhedo 
penacho 
pesqmue iro 
pescador 
pescoso 
chumaco 
chuvoso 
porteiro 
portador 
postreiro 
podestade 
pogao 
pratica 
prevenda 
prebos te 
preseve 
pregaria 
primeiro 
principe 
profeito 
porende 


provinco 


Latin 


proximo 
publicus 
pugnale 
puléjum 
pulice 
pulpdsus 
puritate 
pustella 
putatdre 
putone 
quadratus 
quaremonia 
quatemus 
quercinus 
rabiesus 
racemos 
ramosus 
rancore 
rap&no 
rapstrum 
rastellus 
rasura 
recisione 
recoctus 
rectore 
refertus 


regalis 


Portuguese 


proximo 
pruvico 
punhal 
poejo 
pulga 
polposo 
puredade 


pastela 


podador 
podao 
quadrado 
caramunha 
cademmo 
cerquinho 
ralvoso 
racemo 
ramoso 
ranoor 
rabao 
labresto 
restello 
rasura 
reciséo 
recoito 
reitor 
referto 


real 


282 





arotidugq 
ofkrgug 
maf sing 


A = v 
soi five 


ae 





ausogulg 
ohsbsnics adst Psy 
Shetesct | Sifetaug 
eBabor ott eft 
Obie ) snextiig 

obs iiasp ay cheese 
sritz meLES Sinuieysup 
heen autrasaien 
evadetea tsa anEtioaerp 
A ‘OBOV (62 BIE Chey 
ansost aft g oe Sat sp! 
Oeane1 SPREE. 
tootex oe oxtore 

| oscdert _orteps 
otapuds [ merchaige x 


ofiates atst Ferkses 


my 


BIwaEt Biest 


obeings oeLteroe: 


oe 


~ 


Latin 


———__ 


regestra 
yv 
remulicum 
wv 
renicus 
resapium 
a 
resina 
vi) 
retina 
rigidus 
rogetione 
romana 
ww 
romanice 
rosmarium 
rottlus 
rottindus 
rumore 
runcinus 
(es — 
ruptura 
sabbatum 
sabina 
sabtirra 
sacratum 
saeculum 
WwW . 
saeticium 
ov 
sagitta 
i — 
sallicare 
salicetum 
od 
salina 


saliva 


Portuguese 


regestra 
reboque 
rengo 
resaibo 
resina 
redea 
reigo 
rogacses 
romana 
romance 
romaninho 
rolho 
redondo 
rumor 
rossim 
rotura 
sabado 
sabina 
saborra 
sagrado 
sego 
sedaco 
seta 
salgar 
salzedo 
sainha 


Saiva 


Latin 


ee 


salice 
Ssalmaria 
salvitum 
saltbris 
salvatore 
sanctitate 
sanguineus 
sapidus 
sapone 
sarcalum 
sardina 
sarmentum 
statione 
scabe1llum 
scanddla 
scintilla 
scopilia 
scopulus 
escorpione 
scriptura 
scutella 
secundus 
securis 
sellarius 
sementis 
semita 


senara 


283 


Portuguese 


souza 
salmoura 
salvitre 
salobre 
salvador 
santidade 
Sanguinho 
saibo 
sabao 
sarcho 
sardinha 
sarmento 
sazao 
escabello 
escandea 
centelcha 
escovilha 
escolho 
escorpi ao 
escritura 
escudela 
segundo 
seguro 
seleiro 
semente 
senda 


seara 


_ 


— - - = ne he 


a : — ue 
— eee 5 aa - ae Pie 7 Ct jg : 


» _ ea - 
ee, ne 


HET nie 







: | rf 382 & a 3 a > ; oy) Z 4 q) a3 a & @ 2 ie 

OL APEELESLEREEREC OS EL LS 3 

a ee ee a a ee ee 
C 


Bil Evgpss 
arlfoors 


O8ioiozsS 


Btu poss 
a 


Sehasses: - 


CSiarrsKeS 
* (O8sise 
efbodsaes 


aepioedaet 
Bsus 
aguemise 
setivine 
onxdoiee 
aebiins 
Ott iprie= 
odirs2 
GgEcise 
oforse 
ertiinse 


Latin 


septimana 
sarica 
sericila 
serpénte 
serptillum 
servactilum 
servitium 
sicitate 
salvatium 
singlus 
sittla 
sobrinus 
solacium 
solagine 
solitate 
sortictila 
speculum 
spinosus 
spiritus 
stavilis 
stabtilum 
stomAchu 
strumento 
strictu 
subella 
subito 


subterraneus 


Portuguese 


semana 
sarja 
sarilha 
serpente 
serpol 
servalho 
servico 
sequedad 
salvejem 
sendos 
selha 
sobrinho 
solaz 
soagem 
saudade 
sortilha 
espe lho 
espinhoso 
espiritu 
estavil 
estabro 
es tomago 
es trumento 
estreito 
sovela 
suto 


soterr4neo 


Latin 


subtilis 
suci dus 
sucosus 
sudolento 
superctilus 
stipernus 
supinus 
suspitiGne 
tabanus 
taberma 
tabermarius 
tabtila 
tabtilatum 
talone 
tamarice 
tamagnus 
tapitium 
tarantum 
tardivus 
tegtila 
tegulatum 
timone 
tempes tate 
tenaculum 
tenace 


teneritia 


28h 


Portuguese 


sutil 
sujo 
SUCOSO 
souvento 
sobe jo 
sobomo 
sobinho 
suspic ao 
tabao 
taberna 
tabermeiro 
taboa 
taboato 
talao 
tamariz 
tamanho 
tapiz 
taranta 
tardio 
tartago 
telha 
telhado 
tim&o 
tempestade 
tenalha 
tenaz 


termeza 





errmsqhe 


apne - 


sco igare 
auptedest 


serisds. 


a Sustresds) 


elites 
mahi 
spist 

20 Etat 


[Ure mst 


muisttost ° 


mestiste+ 





atigsloa 7 


— toa —- 
a i oo 


4 
-_. 


S ' - 
é 





4. 
‘= 


— 
| 


- o 7 
Fee ae, ee 


; - <=) 5a - 
ae SPAS - 


* 





Latin 


tendone 
tenébrae 
tesione 
tentidone 
tepidus 
terebinthina 
territorium 
tertiarius 
tertialus 
testimonium 
thal amus 
thesaurus 
timdre 
tincttra 
tineosum 
titione 
tormentum 
tomicia 
tosidcne 
tomentum 
torpedo 
tortiSne 
toxicum 
trabiculum 
traditione 
traditdre 


tragula 


Portuguese 


tend&o 
treva 
tesao 
ten¢ao 
tibio 
termentina 
territorio 
terceiro 
trec6 
testemunho 
tamo 
tesouro 
temor 
tintura 
tinhoso 
ticao 
tormento 
tomiza 
tosao 
tomento 
torpedo 
tore ao 
tOxico 
trabelho 
traic&o 
traidor 


tralha 


Latin 


tramite 
tranvérsa 
transvérsus 
trecenti 
tremactilum 
trepestulare 
tribtlum 
trimensis 
trepetia 
tristitia 
triticum 
todtilus 
tubiscus 
Ulice 
ulmetum 
umbilicus 
uncinus 
unctiira 
tndosus 
ungula 
urina 
urtica 
vadosus 
vagabundus 
vanitare 
venatdore 


(Oj eee 
veritate 


285 


Portuguese 


tramite 
travessa 
travesso 
trezentos 
tramalho 
trebelhar 
trilho 
Crenes 
Erepeca 
tristeza 
trigo 
tolho 
trovisco 
urce 
olmedo 
ombigo 
encinho 
untura 
undoso 
unha 
ourina 
urtiga 
vadoso 
vagamundo 
vantar 
veador 


verdade 


wenfodsrs 
Onl rx 


soagesr 
asaderit 
opited 
orilo4 

: coeivots 


Sot 


Opis 


omiions 


ftir 


aa 


otinsxs 
satwieszt 
CEE aELS 
if foext 
sta litteaqe:3 
migdixd — 
ekenemias 
BLisgeait 
eititsrst 

. pita rs 
ae itibod 
posits 

ee L1G 

mMISn {5 





286 


Latin Portuguese Latin Portuguese 
vermactilus vermelha vettilus ve lho 

wv ; : 
versicus vesgo virgine virgem 


vervactum barbeito 


Oy 


te 


odten 


ime Ev 


ritet 


eerliston 


efttox iv 


ideal 

au limantey 
aur we 
’ a cB , 








ey 


fe