Skip to main content

Full text of "Elision, Especially in Greek"

See other formats


STOP 



Early Journal Content on JSTOR, Free to Anyone in the World 

This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in 
the world by JSTOR. 

Known as the Early Journal Content, this set of works include research articles, news, letters, and other 
writings published in more than 200 of the oldest leading academic journals. The works date from the 
mid-seventeenth to the early twentieth centuries. 

We encourage people to read and share the Early Journal Content openly and to tell others that this 
resource exists. People may post this content online or redistribute in any way for non-commercial 
purposes. 

Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- 
journal-content . 



JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people 
discover, use, and build upon a wide range of content through a powerful research and teaching 
platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit 
organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please 
contact support@jstor.org. 



VI. — Elision, especially in Greek. 
By M. W. HUMPHREYS, 

PROFESSOR OF GREEK IN VANDERBILT UNIVERSITY. 

I propose, in this paper, to discuss the nature of elision in 
Greek ; and, in so doing, I shall first examine the views of 
others, and then present my own. 

Corssen, Westphal, Heinrich Ahrens, and many others, 
hold that elided vowels were not entirely suppressed, but 
merely diminished, and one argument they employ is that the 
Greek name itself, awaXoifi] or a-vys-pta-ie, does not signify total 
expulsion, but implies that the vowels were united in a rapid 
pronunciation, and did not suffer what was called tY-SAni-ie, or 
expulsion. To this I reply: 1. That the ancient writers are 
not always to be interpreted literally ; for, as one vowel, or 
rather one syllable, appeared to result from a combination of 
two, and the elided vowels sometimes were slightly sounded (a 
point to be explained hereafter), there was no reason why 
they should not, in a loose way, designate the process by the 
word (TwaXoifii, or rrvyxpiaic, which does not necessarily mean 
anything more than "conjunction"; and besides, already at a 
tolerably early day they employed the term «-£X<^te to denote 
elision. Even modern writers are not exempt from much 
more inaccurate applications of terms than even awaXoupii in 
the sense of elision, to say nothing of the other words. As, 
for instance, the German grammarians call final m and n in 
French a " Nachklang," or " after-sound," as if they were 
pronounced after the accompanying vowel, while every one 
knows that they merely give the entire vowel a nasal tone. 
Many illustrations of this could be cited, but this one must 
suffice. 

2. Moreover, when the ancient grammarians speak of 
the suppression of hiatus, they frequently fail to distinguish 
between the various processes, or else between the words that 
denote them; and in the very passage cited by Corssen, 
elision is confounded with crasis. The passage is: "E<m 11 



On Elision, especially in G-reek. 85 

<rvi'a\m<l>i) cvo <p<i>)'r)£)Tii>)> hirjprifiiviov e«c ixiav nvXXajifiv tvuxnc, oiov 

to oi'Ojua, Toivofia. If the metrician had not added his 
example, the inference would have been that avvaXoL^i was 
always the combination of two vowels into one, whilst the 
example he gave shows that he had only crasis in mind when 
he cast his definition ; and yet Corssen wishes to apply the 
definition to elision. 

But as it is the custom of many now-a-days to dismiss the 
question of elision with the statement that Ahrens has shown 
it to have been only a partial expulsion, and as Ahrens has 
given about all the arguments for that view, I proceed to take 
up his arguments and examine them one by one. In the first 
place, Ahrens says that if elision is total, the letter immediately 
preceding the elided vowel closes the word as thus modified, 
and he calls attention to the fact that we then find not a few 
unpronounceable combinations, as e<t3\', akpv, etc., and others 
which the Greeks would not tolerate, as wit', wtlS', etc. But 
if, as he asserts, elision does combine two vowels into one, 
then the two words become one ; and why then may we not 
be allowed to combine the words after expelling one of the 
vowels ? And this is exactly what happens, except in some 
instances about which 1 shall presently speak. Secondly, 
Ahrens says that <i.Xye idrj^ei', avrC ijie'io, and similar combina- 
tions, would still have a " hiatus offensionem, quam non inesse 
constat." How does he know? The Greeks did not suppress 
two syllables, by elision, because this would have maimed the 
word too severely, so there was nothing left them but to 
tolerate the new hiatus, as custom required them in poetry 
to remove the original hiatus. And besides, does Ahrcns's 
diminution-theory remove his own difficulty? It seems to 
me to increase it, for who will pronounce &Xye a t^t-tc for us 
(pronouncing the final a of &Xyea, yet making it of inap- 
preciable length)? And in avTi Efieio there is surely a less 
offensive hiatus than in avri a ifiuo. Moreover, it is a well- 
known fact that languages, in removing one hiatus, sometimes 
create another no less offensive ; as, in Sanskrit vanij &sit is 
resolved into vanai ksit, and the i being elided, we have vaua 
asit, where the hiatus appears even worse than at first (Bopp, 
12 



86 M. W. Humphreys, 

Crit. Skt. Gram. § 38). But after all, I am willing to admit 
that there is nothing offensive in the remaining condition of 
things when elision has been made ; for the two words are 
pronounced continuously, and the vocal muscles do not have 
to arrest themselves and then renew the exertion as they do 
in case of real hiatus. 

Thirdly, Ahrens draws his conclusion from the scholia on 
Eukip. Or. 279 : 

'Ek KVfiuTtav yap avSic av ya\i]v' opw. 

The scholiasts on this passage (and also on Aristoph. Frogs, 
304 — not cited by Ahrens) say that Hegelochus, getting out 
of breath, passed rapidly over the elision, and the spectators 
thought he said yaXfjv 6pw, which circumstance gave Aris- 
tophanes (Frogs, 304), Strattis (Anthroporrhaestes) , Sannyrion 
(Danae), and others an opportunity to amuse their audience 
at the expense of Euripides and his great actor. But, as I 
shall show hereafter, when occasion demanded, the Greeks did 
sometimes slightly sound elided vowels, and one of the most 
natural places to do this is where ambiguity might result 
from total elision ; and the statements of the scholiasts show 
that what Hegelochus did was nothing unusual under ordinary 
circumstances ; and as to his breath failing him, that seems 
to be one of the many inventions of the very fertile minds of 
the scholiasts. If he did not have enough breath to utter a 
" diminished" vowel, how could he add opw so as to be heard 
by thirty thousand people ? And if he slopped to take breath, 
then he did not pass rapidly over the synaloephe. The fact 
may be that, getting out of breath he lowered his voice, thus 
making yaXfjv out of yaXifv, and then took a breath and added 
opS>, which would complete the transformation, since the elision 
should not be complete where any pause is made. This, 
however, is a mere conjecture, and ' is not necessary to the 
explanation of the matter. The mistake on the part of the 
spectators was anyhow quite natural, because bpav ya\r)va was 
a forced and unnatural expression. 

Fourthly, Ahrens observes that rtwaXoupri does not denote 
expulsion, and that elision takes place before a pause, and at 



On Elision, especially in Greek. 87 

the end of a verse, and even between two speakers. I have 
already spoken of the meaning of iruraXoi^; but here I shall 
discuss the subject more at length. That the word ek$\ji//ic 
was employed to designate elision is well known, and the only 
question is how early it was so used. I shall not attempt, 
however, to settle this question, for it is clear that it was so 
used sufficiently early to show that whatever it denoted had 
an existence in classic times. Draco enumerates seven kinds 
of synaloephe, among which he places £K$Xi\pie, which he defines 

thus: Kal iicBXixpig fiiv kari tvdc (puff/eyTog d-rrojXe la, and 

illustrates by v* kfiov for l™ kfxov, although he defines syna- 
loephe itself thus : 2vi>aXoup>) Be >i tov wpoupmixivov Kal kvTeXovQ 
irvfxivTviiQ Tt i'«i Ei'wirtc: a definition which shows how 
careless the ancient grammarians could be in their state- 
ments ; and, in my opinion, they had crasis also in mind, or 
even exclusively in mind, when they appear to apply the word 
vvva\ouf)Ti to elision, except that when employing it as a generie 
term, they sometimes apply it specifically to elision, just as 
one may call a temporal sentence a relative sentence. Here 
is another statement of the subject : SuraXoicpjj kari Bio avXXafliiv 

Kara (pwi'rjevra tvuxric KaTafyoXij t6vu>v. yiyvzrcu Be Kara Tponovg Irra, 
cnrXovg uev rpelg Kara eK&Xixbiv, kir' kfie dvrl tov kwl kfie' Kara 
Kpamv, Tafia o.vt\ tov to. k/xd' Kara avvaipeaiv, vrfprfBeQ airi tov vrjprftBeQ. 
(TvvSeTove Be Terrrrapag, Kara eK$X(.\piv Kal avvaipemv, k LiovwoSiv ei 
ai'Ti tov kiiol viroBivei' cara Kpairiv Kal avvaipemv, 'tiitoXog dvrl tov 
6 a'nroXoe' Kara eK$Xi\piv Kal Kpaaiv, ta'yai dvrl tov Kal kyiy 
KaTa ckSXi^iv Kal Kpamv Kal dfatpemv, kv Ta&wrria dvrl tov kv rr\ 

AlStmriq. Here it is evident that the grammarian by kk-BXixpic 
means total expulsion, for in en' kfie he says we have <;V3Xn//(<.-, 
and in Kayii both 'skSXi^iq and Kp&mg; that is, the < in Kal being 
elided, we have ku' kyw which then suffers update; and no one 
will deny that this i was totally suppressed. Hephaestion 
therefore rightly distinguishes between o-vreKQuvrfaic, by which 
irXkwv (II. A. 183) is reduced to one syllable, and synaloephe 
(generic, including elision), by which a vowel is rejected, as 

&X EKaToyxctpor (II. A. 402), $Tv kf' aXoq (II. A. 350). But 

these two processes would have been the same, if elision had 
only been a diminution. And the scholiast on this passage 
does not err when he says : Aiaipepei 8k o-uvaXoi<pi) avveKfjuovrio-etuc, 



88 M. W. Humphreys, 

ijyovv avi'ifr'itrtuic. j; /.lev yap iu i>a\oi<j>>) lie ypaftTai ovtui tai 
kitty to I'elrai' >/ St ovvi^rjniq ov% wq ypa'tyerai ktupiavEirai, aW kv rid 
fiati'tiv rag Svo fTvXKct[3a(: Ofxov Inrep Ttjc roil fxkrpov Btnairtiar rrvvtKipwi'ti: 
hw Kai (Tvi'l£r]tTic \kyerai KTt. 

§ 2. So far my arguments have been negative. I shall 
now present my own views, and support them with a brief 
discussion of the evidence in their favor. 

In prose, as is well known, the Greeks tolerated hiatus, 
except that some rhetoricians tried to banish it in artificial 
compositions, an illustration of which we have in the orations 
of Isocrates. But then, if they chose, they could elide. 
Hence we draw the important conclusion that the Greeks 
could elide or not elide, as suited their convenience. One 
might assume this as a matter of course, but Cicero, while 
testifying to this peculiarity of Greek, denies that the same 
privilege exists in Latin. Be says (Or. 44, 152): " Sed 
Graeci viderint : nobis ne si cupiamus quidem distrahere 
voces conceditur," etc. But in poetry the Greeks avoided 
hiatus for the most part, and in tragic trimeters banished it 
entirely, except (apparently) after W. as Aesch. Sup. 306, W 
olf, Soph. Philoct. 917, W uwac; and rarely after tv in close 
combinations. But frequently it was difficult to prevent a 
word which ended with a vowel from preceding one beginning 
with a vowel, even when there was a pause between them. 
In that case they did not totally expel the vowel, nor even 
necessarily reduce it to inappreciable quantity. Whenever 
this happened the elision was indicated as if total, while in 
recitation the elided vowel was either pronounced in full or 
merely diminished, just as the sense required or permitted. 
Another instance of partial, or apparent elision is where an 
emphatic monosyllable apparently loses its vowel, as Eurip. 

Tl'O. 94b : ov (T,aW kfiavT)))' tovwi ro>6' kpijtroftcu. So Alcest. 984. 

Also where the sense would be obscured, as Here. Fur. 972 : 

aWoij aWotr, zq irtVAoue 6 fxiv \xr\Tpoc are. Cf. SOPH. Elect. 1499, 

Euuip. Ion 3, etc., etc. And thus it came about that if 
for any reason they desired it, they felt themselves at liberty 
merely to diminish a vowel, even when there was nothing but 
metrical considerations to prevent its total expulsion. This fact 
is of special importance in determining certain effects of elision 



On Elision, especially in Greek. 89 

in the construction of verses — a subject on which I propose 
to present a paper at some future time. 

But that vowels could be, and actually were, entirely 
expelled by elision, is shown by the following considerations : 

1. When the second word begins with an aspirated vowel, 
then the aspirate affects the final consonant of the first word, 
if it can be aspirated without changing its character, as vi^S' 
o\ri>>, kij> t'l/jiy, 5m,»«x oror, which seems to me impossible if the 
elided vowel was pronounced ever so little, for then it would 
have separated the consonant from the aspirate. This, it is 
true, does not happen in Herodotus; but then in H. it does 
not happen in compound words, like awitjftt, where all admit 
total elision. 

2. If the ultima has the accent, it goes back to the next 
syllable when elision takes place, and enclitics retain their 
accent when elision takes place before them. This would 
hardly have been the case if the elided syllable had only been 
diminished, for the Greek accent was merely an elevation of 
the voice, and not stress. (This recession of the accent from 
an elided ultima is found in some of the examples used by 
Ahrens to prove that the vowel was not entirely suppressed, 
as irrSW o-fjui'.) We have in the Greek language itself an 
instance of the accent remaining on a merely diminished 
vowel (or at least not seeking another syllable), and that is 
in aphaeresis, as ekeivu> '<Wei' or aaVu "Jwhv. Thiersch ridi- 
cules such accents, calling them " accentus azpofiarovrTag," but 
if he had put on his phrontistic spectacles he would have 
detected a k-ptfiatya on which they ride ; in other words, the 
omission of the vowel in this case only indicated its diminution 
to inappreciable quantity, while the accent still remained on it, 
just as in Sanskrit we find an accent (the svarita) partly on v 
and j, although these not only fail to make syllables them- 
selves, but even do not lengthen a short syllable, as in svar, 
kba, nadjas (Bopp, Crit. Gram. § 30, 1); and similarly even 
in Greek where an accented vowel suffers synizesis, as in 
Ah'iac (Rhes. 85), apiarewv (Alcest. 921), rivyiwv (Androm. 
167), oirriioy (Tro. 1177). So OlXtivc, 'Ay^iWiwc, and in Hom. 
Od. \ly\mrious, with hundreds of instances everywhere. In 



90 M. W. Humphreys, 

all these the accented vowels become virtual consonants. 
This view is further sustained by the fact that when the first 
syllable of a word is entirely lost, the accent on it is removed 
to the next syllable, as in the Homeric paWc for I iaWe, where 
no vowel precedes. Corssen, indeed, denies that such forms 
have lost the augment, but G. Curtius more successfully 
maintains that they have. But as it is important to establish 
the position that in aphaeresis the vowel was thus diminished 
and yet retained the accent, I must not leave unnoticed the 
fact that Thiersch, Buttmann, and others deny the existence 
of aphaeresis, and assert that all the apparent instances of it 
really belong to crasis. This view, though, cannot stand in 
the face of the following facts : 

First, such combinations as Ufafiat 'yi>, which are fre- 
quently found in mss., and are not wanting in inscriptions, 
would have to be written Swafiuyi), with omission of < and 
contraction of a with t. Secondly, when the word suffering 
aphaeresis begins with an aspirated vowel, the consonant 
beginning the syllable preceding would become exposed to 
the aspirate as in SoljiaTwv for ™ ifianov, Sfifiipy for rjj jj/me'p^, 
SaTtpg. for rjj hepy (once arepy), whilst in reality we find such 

examples as atirj; 'repa (AltlSTOPH. Lys. 736) for ai!r?) crcpa, 

which, by crasis, would become abSyTipa (avSarepal) . I am 
willing, indeed, to concede that some instances of aphaeresis, 
as found in the texts, are to be written otherwise, as x') 'yx ovaa 
(Lys. 48), which ought to be written x>)rx ou<ra ' f° r * ne article 
loves crasis, and I suspect that the usual way of writing these 
words is due to the fact that the double crasis seemed rather 
bold, and obscured the words. Felton's a 'Xafiev (Clouds 
1268) with long & is certainly wrong. Thirdly, aphaeresis 
sometimes takes place after a long pause where crasis is 
impossible, as Clouds 1854: lyw typarrw VeiSi) ktL ; Philoct. 591 : 
Xe'yw .V! tovtov ktc. ; Rhesus 157 : fj£,ii>- Vi tovtoiq ktL ; Iph. in 
Aiil. 719: ^t'XX<o- Vi ravrri Kre'., etc., etc. It is sometimes 
regarded as taking place at the beginning of a verse, but a 
careful examination of all the Greek dramatic poetry convinces 
me that this may have been a mere omission of the augment 
of verbs (which frequently occurs in p>i<retg dyyeXiKai'), although 



On Elision, especially in Greek. 91 

in the great majority of cases the preceding verse ends with 
a vowel. This vowel is sometimes short, as in Oed. Colon. 
1605-6, and sometimes we find a consonant, as Oed. Rex 
1248-49. I am not so sure, however, that aphaeresis may 
not take place after a short vowel ; and I shall presently have 
occasion to cite a case of similar aphaeresis in Latin. But to 
return : 

3. Diphthongs are frequently elided, and especially in the 
verbal ending -at. Now can a whole diphthong be reduced to 
inappreciable quantity ? It is difficult to reduce a diphthong 
even to a short syllable ; nor is there any reason why the first 
vowel should be diminished unless the second is entirely 
removed, so that those who assert that elision is mere diminu- 
tion are compelled to affirm that i is dropped entirely and 
a diminished ; but if i in a diphthong can be dropped entirely, 
why cannot any elidable vowel be thus dropped, as a in akyea 
eBriKev, yaXrjia <>pib ? One might reply that the < becomes a. 
sort of consonant or semi-vowel, like y ; and I believe that 
this is what actually happens when a diphthong is shortened, 

as in ovk eat i ov rpvywv, where t = y, and in ahrov iv ve<pi\rf<n, 

and 'i(ev kjxeio, where v = w, since v is never elided, the well- 
known exception in a quoted oracle in Herodotus being only 
apparent. But if this is what becomes of the second vowel 
in case of elision of a diphthong, there is no reason at all for 
the shortening of the first vowel, as there is no longer hiatus. 
In such instances, therefore, as k-oifia<r$' ir w6\ei (for rot^oVSat), 

Ko\aa k^eari (for KokaaaC), lovv 'ivtori for (Bovvai), yrj/x 7 iirrjpe (for 

yjjfiai), iiofi iyoj (for Uofiai), necessarily the second vowel, 
and in fact the first, too, was elided, unless for some' special 
reason it was desirable to make the first audible. 

4. Epicharmus, as quoted by Atlienaeus (vm, p. 338, d ; 
see Ahrens, de Crasi et Aphaer. p. 2) plays upon y' 'ipavoq and 
yipavoQ, from which it appears that the e in y£ was suppressed. 
Aristophanes (Clouds 1273) appears also to play upon aV 
ocou and aVo vov. Further, Dion. Hal. (De Comp. Verb. c. 11) 
calls KrvweiT (for KTVTre'iTt.') "two syllables." I am not disposed 
to make much of this, as an inappreciable vowel might be 
omitted in counting syllables metrically. 



92 M. W. Humphreys, 

5. The Words I'trav, oworav, yap, yuvv, etc., for ore av, more av, 

ye ftp, ye <wr, etc., show that the vowel was entirely suppressed; 
and after they had been a long time in use, the combinations 
began to be regarded as single words. This might happen, it 
is true, merely from long juxtaposition, as we have in Latin, 
(where elision does not appear to have been total) tantopere, 
magnopere , for tanto opere, magna opere. But this is much 
rarer than in Greek, and we have in Latin two vowels united 
into a diphthong, as in neuter, neutiquam, deinde, etc. 

6. Finally, if elision had been only a diminution of the 
vowel, as in Latin, it would not have been subject to so strict 
limitations, but would have been as universal as it was in 
Latin. But, as is well known, elision in Greek was strictly 
forbidden under certain circumstances. For instance, «, <, 
and o in monosyllables were not elided (except a in <r«) ; 
and v was never elided at all. The seeming exception in 
Herodotus (vin, 220) rW' eoiKvck should most probably be 
aerv 'ptKvceg (a sort of aphaeresis after a short vowel), or perhaps 
the oracular poet or priest was at his wit's end for a verse, 
and admitted diminution where expulsion was not tolerated. 
When v closes a diphthong, it does, indeed, seem to be elided ; 
but in that case, as I have already said, it was probably 
pronounced somewhat like w, just as 6 in Sanskrit before a 
vowel becomes av, where v was, no doubt, pronounced like w. 
Further, on and nepi do not suffer elision, possibly because 
they would then sound like ore (with its e elided) and nep, 
which would not be the case if their vowels were sounded 
ever so little. Some words, however, with long vowels, did 
suffer a partial elision ; but this is one of those exceptions 
that prove a rule ; for if all elisions were only partial, then 
fit) oh (as one syllable) should be written /*' oh, and would be 
an ordinary case of elision. Nor is it crasis, for then it would 
be /t<i (cf . fifor for fit) olv~) ; and moreover the combination may 
occur when a slight pause intervenes, as Oed. Tyr. 944 : 
T&vr)Kev el U fit), (throe d^iat Savelv, and also where crasis 
would utterly obscure the sense, as Trach. 85 : »W'' n aeaiia- 

ueBa | tceivov fiiov awtravros, ») o'f% 6/J.e a$' iifia. The contraction 

of fin ob into fiM would itself be rather obscure ; but if so in 



On Elision, especially in Greek. 93 

writing, then certainly also in speaking ; and that contractions 
of the sort, when, made in speaking, were also indicated in 
writing, is shown by %ohv for koX 6 iv, Eurip. Heracl. 173 ; 
§&wX airiivai for ra SirXa awuvai, Birds 449 ; and even Kal for 
rac al, Lysist. 1105. We have, however, an instance of crasis 
not indicated in writing, in Eurip. Orest. 599 : el /xt] 6 KeXevaae 
pitrerai /j.e fiij Saveh', unless with Witzschel and others we omit 
o, or admit synizesis of a long vowel with a short one, result- 
ing in a long syllable. 

Further instances of non-elidable vowels are found in the 
genitive ending -ow,-ao, and to a great extent in the dative 
ending -t; and third singular endings in -e are not elided 
before av, unless we admit elx'a'' ptrpov (Ion 354); for awka^ 
civ (Alcest. 901) is an impossible conjecture, and IXavSuv' &v 
(Soph. Elect. 914), though desirable as to the sense, is not 
the mss. reading. Other instances of forbidden elision might 
be cited. All this proves conclusively that elision was recog- 
nized as having power to remove a vowel entirely ; for, 
otherwise, there was no reason why elision might not have 
been as general as it was in Latin, where the restrictions, as 
far as they exist, merely have reference to too great a mutila- 
tion of the word, and were a refinement of artificial writers. 
The vowels which could* not be elided entirely in Greek, were, 
for the most part, not even allowed to suffer diminution to 
inappreciable quantity (Latin elision), because this was a 
modified form of ordinary elision, and was written as elision, 
i. e. the vowel was omitted in writing, except in those few 
cases where the elision was never total, as /u>) oh. An inves- 
tigation of these latter cases would lead to a discussion of 
synizesis and synaeresis, which is foreign to the object of this 
paper. 

§ 3. These arguments seem to prove that vowels could be 
and frequently were entirely expelled by elision. It now 
remains to be shown that not unfrequently they were, for 
special reasons, only partially elided ; and indeed they some- 
times had nearly or quite their full time, although they counted 
for nothing in the structure of the verse. 

In the first place, elision takes place before a strong punc- 
13 



94 M. W. Humphreys, 

tuation, as (Birds 990) ovk ti §vpa£; se Kopaicas kte. Eurip. 
Androm. 459: 

KTlb'tlQ fl . ClTtOKTUV' (OQ (iSwITEVTOV y£ <T£. 

Soph. Elect. 662 : 

rati iariv, w £e'v . avrog tjnacrag koXwc. 

Cf. also 671, 1041, 1112, 1470, etc. 

I am not disposed to attach much importance to elision at 
the end of a verse, as in Birds 1716 : 

■^tapei, KaXbv Beafxa' Bvfuafj.aTwi' 2 
avpai Sia\paipov(ri ktL, 

and Oed. Col. 1164 : 

<7oi (fiadiv ahrbv ee Xoyove tXSelv jxoXovt 
alrtiv aireXStlv t a<rcj>a.Xw£ rrje Sevp bcov: 

for I doubt whether this ever happens unless the sense requires 
the verses to be closely connected together ; and that being 
the case, the two verses can be read continuously as one long 
verse, and the vowel can be dropped. 

Again, elision takes place between two speakers, as Birds 
846, 1015 : 

EYEAI1. oifiw^e Trap' ifi'. III26. '(■&', &>ya3', o! Trzfnru) a eyw. 
ITIS0. fia rbv At' ov Srjr . MET. aXXa ttwq; 

So Soph. Elect. 1431 : 

OP • elaopare ttov 

rbv avdp'; EA. i<p' riixiv oiiroe e/c -npoaoriov — . 

Ibid. 1502 : 

OP. dXX' ep(p'. Air. vcj>riyov. OP. trot (3aSi<TTtov Trapog. 

Elisions of the sort just now mentioned — those at a full stop 
not between two speakers — are comparatively rare ; for there 
was something harsh about them ; and although we may use the 
interrogation point or the period, still the pause is really short 
in most cases. That they were in some measure unpleasant 
is shown by the fact that Isocrates, who did not tolerate hiatus 
in his orations, also banishes this sort of elision — a thing 
which he could do more effectually than the poets, who were 
somewhat trammeled in the arrangement of their words by 



On Elision, especially in G-reek. 95 

metrical considerations. But elision between two speakers 
does not appear to have been avoided at all ; for, in fact, it 
was not a real elision. The second actor began to speak just 
as the first one struck his last syllable. To this it may be 
objected that the same thing could have happened just as 
well, if the first speaker closed with a consonant ; and this is 
certainly true. But there would have been nothing to indi- 
cate that it was to be so recited, and, as I have already said, 
verses were so composed that, when written, they looked 
perfect, which could not be done if the first speaker's final 
syllable had been disregarded when it was closed by a conso- 
nant. We find something analogous to this effort to make 
the verse appear perfect in the classic French drama, where, 
without affecting its pronunciation, the mere spelling of a 
word is sometimes altered, so as to make it look like the 
word with which it rhymes, as Le Cid, v. 771, where voi (for 
vois~) rhymes with toi, and 851, where voi rhymes with moi. 
Somewhat analagous is also the method of indicating a pause 
at the end of a piece of music when the last measure is 
incomplete. 

In view of all this it is safe to assert that elision between 
two speakers was relatively more frequent than at a full stop 
in a speech of one person. (I say relatively, because this sort 
of elision only has a chance to occur when a verse is divided 
between two persons.) In fact it was not avoided at all, but 
sometimes appears even to have been sought, as it gave one 
actor an opportunity to fall in before the other had entirely 
finished his last word — a thing to be desired when the dialogue 
is animated, or for any reason rapid. If any one doubts this 
let him examine such passages as Burip. Orest. 1598-1612, 
where in fifteen lines this elision occurs seven times. 

These arguments prove conclusively that elision was some- 
times only partial, and sometimes even only apparent, the 
vowel omitted in writing being pronounced in full, but counting 
for nothing in the structure of the verse. 

§ 4. Although it was more especially designed to investi- 
gate G-reek elision in this paper, it will not be irrelevant to 
append a few remarks on elision in Latin. It is conceded by 



96 M. W. Humphreys, 

nearly all that elision in this language was only partial. 
Hence it was subject to less strict laws than in Greek, and 
hiatus is more rarely admitted, it being easier to avoid it. 
Indeed Cicero (Or. 44, 152), as quoted before, says that the 
Romans were not allowed to neglect elision even if they 
desired to ; and then finds fault with poets for doing it. But 
this very fact that poets did sometimes allow it, shows that 
the law Cicero announces was not without exception, and of 
course it was no physical necessity, but merely convenience 
and usage. Yet Cicero's remark shows that in prose it was 
practically universal. Such words as neuter, deinde, etc., show 
that elision was not total, at least, in some cases where there 
was nothing to prevent its being total if it ever was ; and 
those cases where, the vowel was entirely lost (tantopere, 
magnopere, animadvertere, etc.) are mere results of long 
usage, the vowel having -been slightly pronounced at first, just 
as we say " exirordinary " instead of " extra-ordinary." In 
tantopere and magnopere this process was hastened by the 
identity of the two vowels brought in contact, " tantu-opere," 
as elision and contraction are more necessary under these 
circumstances. This is illustrated by the Greek second 
declension in the genitive plural, which contracted before the 
existence of the law that a long ultima should prevent the 
accent from falling on the antepenult ; while the same con- 
traction did not happen in the first declension until after this 
period, the vowels not being so similar ; thus \6yoiDv became 
\6yuv, while fiovaauv remained. Afterwards the long ultima 
removed the accent, and then they said \oywf, nova-awy (an 
extant form) ; and finally fiovaaiav contracted into /uouow. 
Similarly nihil became nil, and mihi, mi; but we must not 
carry the illustrations too far ; for phenomena from within 
words, simple or genuinely compound, will not always hold 
for separate words ; and tantopere and magnopere are not to 
be regarded as genuine compounds, such as cogere, degere, 
in which erasis seems to have been employed. (Corssen, by 
the way, Ausspr. Voc. Beton. n, 889, writes " tantopere," or 
possibly his printer did it for him.) And so in mihi, nihil, 
the process was different from that in tanto opere, but they 



On Elision, especially in Greek. 97 

illustrate the aversion of the vocal organs to a consecutive 
repetition of a vowel. But of genuine crasis between two 
words, not combined into a genuine compound, 1 know of no 
example in Latin. In fact, elision being only partial, and so 
being allowable under almost all circumstances, there was no 
need of crasis ; and the nearest approach wc have to it is what 
we find in degere, cogere (just mentioned), further examples 
of which are deesse (two syllables), deerrare (three syllables), 
in which Velius Longus (p. 2227) says the e or ee was long 
(by nature), which of course we should expect, as the prepo- 
sition sometimes formed a syllable to itself. Contractions 
such as amatast, integratiost, are not crasis, but a species of 
aphaeresis, as is shown by Tibullus (i, 9, 53, and 77) : 

at te qui puerum donis corrumpere's ausus — 
blanditiasne meas aliis tu vendere's ausus. 

The later Roman grammarians speak of elision as if it were 
a total expulsion of the vowel ; but their authority is not of 
any importance. The name " elision," it is true, strictly 
interpreted, would imply total removal; but the Roman gram- 
marians employed terminology that was adapted to Greek, 
and sometimes even mistranslated Greek terms. So we now 
speak, and I have just been speaking of "elision" in Latin; 
and while doing so, I have been trying to show that it is not 
elision, but diminution. 

But there are good reasons for believing that the particles 
-que, -ve, -ne lost their vowels entirely through elision ; and 
-ne is sometimes written without its vowel even before a conso- 
nant; just as /ace, duce, dice lost their e, and even cave (being 
much used) lost its e sometimes, as shown by Cicero's well- 
known remark implying similarity of sound between cauneas 
and cave ne eas. The elision of these particles will come up 
in my next paper. 

Briefly, then, to sum up the whole matter: 

1. In Greek, elision was the total suppression of a vowel; 
but it could be only the partial suppression, and sometimes was 
required to be only partial, or even merely apparent. 

2. In Latin, elision was the partial suppression of a vowel ; 
but in a few special instances it was total. 

14