Sunteți pe pagina 1din 16

Linguistic Society of America

The Traces of the Digamma in Ionic and Lesbian Greek


Author(s): Milman Parry
Source: Language, Vol. 10, No. 2 (Jun., 1934), pp. 130-144
Published by: Linguistic Society of America
Stable URL: http://www.jstor.org/stable/409604
Accessed: 11/03/2010 11:47

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=lsa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to Language.

http://www.jstor.org
THE TRACES OF THE DIGAMMA IN IONIC AND LESBIAN
GREEK
MILMAN PARRY
HARVARD UNIVERSITY

Richard Bentley has won only blame for wishing to change a 29 from
,uvriaro yap Kal-a 6vgbv ayu&uovosAVyiLa'owto ,uvi?aro Pyap Kra vovv
avot5,ovos AltyiaoLo1;whereas his plan of writing the digamma into the
Homeric text2 is still cited as one of his claims to fame. Yet in both
cases he did much the same thing: he was unable to see why the tra-
ditional text was as it was, he was unwilling to grant a simple lack of
understanding on his own part, and so he changed the text. Had he
known Homer better, however, or known more about other early
poetries, he would have seen that the unreasoned use of the fixed epithet
is so common that we must explain it, not try to do away with it. First,
the analysis of Homer's diction might have shown him that the poet
had, to help him in his verse-making, many fixed phrases in which there
was an epithet, and that he used these phrases so often that he forgot
to think about the meaning of the epithets in them.3 Or second, the
study of oral poetries might have shown him that the use of the fixed
epithet is common there, and this would have led him on to the cause of
metrical usefulness.4 It is the same for the digamma. Had Bentley,
or any of all those scholars who have corrected Homer or printed the
digamma in their editions been willing to grant that there might be some
force acting on the Homeric language which they did not see, they would
not have fought so fiercely against the stubborn text. But they had
1 Cf. R. C. Jebb, Bentley 149-154 (London, 1882).
2 Cf. J. W. Donaldson, The New Cratylus3 219-25 (London, 1849).

: Cf. Parry, L'epithete traditionnelle dans Homere 146-217 (Paris, 1928).


4 E.g. rusa 'blond', in Southslavic heroic poetry as an epithet of the head. In
one of the Bulgarian poems (36, v. 45 in A. Dozon, Chansons populaires bulgares
inedites, Paris, 1875) a 'black negro', as he has constantly been called, is killed.
The words of the singer are Otsecemu nemu rusa glava, 'he cut off his blond head'.
The verse is one which occurs often both in the Serbo-Croatian texts of Karadzid
(1823-1833), and in those which I myself collected in Hertzegovina in 1933. In
the same way the 'hyacinthine' locks which Athene had given Odysseus (1 231,
cf. A 158), are later called 'blond' (v 399, 431).
130
DIGANIMA IN IONIC AND LESBIAN GREEK1 131

seen a part of the truth, and they were beguiled by the complexity of
what they had seen. Yet a fuller knowledge of Homer's poetry and of
oral poetry shows us why Homer's language has traces of the digamma,
but not the digamma itself.
The poet who has no writing materials to aid him can make his poetry
only out of fixed phrases, verses, and passages which have come down to
him from the past, and which are the gradual work of generations of
countless poets.5 The phrase which will easily fall into the mould of
the verse in the right place to make the sentence is a hard thing to
create, as is the verse which forcefully expresses a given idea in just its
length. One poet might make a few such phrases and verses, but he
could not make many, let alone the vast system of them needed for
free composition. Really he does not even think of making them. He
has learned the poetic diction by hearing many poems which had been
composed out of it, and his art lies in putting that diction to its best
and fullest use, not in changing it or adding to it.
He does, however, without thinking, change it in one way. If
phrases in the diction have come from the people of another dialect,
or if they have come down from a time when the spoken language was
different, he will tend to change the language of such phrases to suit his
own spoken language. He will usually make such changes only if he
can do so without damaging the rhythm, but he may even do so if the
damage is only slight. Beowulf, for example, has forms asfrea, gan, dos,
where the meter calls for an older or Anglian dissyllabic form.6 Old
Norse poetry has such verses as en at virYi rekaz (Hdvamdl32), ]'ess mun
Vi'Yarrreka (Vafpru'Sismdl53) which depart from the rule of alliteration
through the loss of the older forms vrekazand vreka.7 Krohn has pointed
out such cases in Finnish poetry as that in which the singers of Savolax
have turned the Tavastland form orasta into oraasta, though the result-

C
Cf. Parry, Homer and Homeric Style in Harvard Studies in Classical Phil-
ology 41.77-89, 117-47 (1930); The Homeric Language as the Language of an Oral
Poetry, ib. 43.8-23 (1932).
6 Cf. Beowulf, ed. A. J. Wyatt and R. W. Chambers xxiii (Cambridge, 1925);

Eduard Sievers, Altgermanische Metrik 122f. (Halle, 1893). On the language of


Beowulf as a poetic language cf. 0. Jespersen, Growth and Structure of the Eng-
lish Language 55f. (Leipzig, 1905);H. Collitz, The Home of the Heliand, in Publi-
cations of the Modern Language Association 16.123-40 (1902).
7 This parallel was first pointed out by 0. W. Knos (De digammo Homerico
9-10 (Upsala, 1872); cf. also S. H. Grundtvig, Om nordens gamle literatur 71
(Copenhagen, 1867); Er nordens gamle literatur 69-73 (Copenhagen, 1869).
132 MILMAN PARRY

ing verse has one too many syllables.8 In the same way he gives the
metrical irregularity as one of the chief means of finding the older verses
in Esthonian poetry.9 The editors of Beowulf and of the Norse poetry
have in many cases followed the same reasoning as Bentley, and changed
the text; but the mere fact that we find this same thing in a number of
early poetries so far removed from one another, and that the thing is
unknown in the manuscript tradition of any poet for whom we can be
sure that writing was the normal means of composition, shows that we
have here a true feature of oral poetry, and no scribe's changes. More-
over the Finnish and Esthonian poetry show that the faults really do
occur.
Similarly the digamma was lost in the diction of early Greek heroic
poetry neither sooner nor later than it was lost in the daily speech, but
the singers who had to compose in a rigorous and therefore highly
conservative verse-form, still used the old phrases and verses because
that was their way of making poetry, because to have given up the tra-
ditional phrase wherever the loss of the digamma now caused hiatus or
failure to make position, would have been to destroy the diction almost
entirely.
The traces of the lost digamma were not maintained simply by the
regular failure to avoid hiatus or to make position within certain often
used phrases, such as the following which show by the number of the
times which they occur in the short space of the Iliad and Odyssey the
fixed and helpful place which they had in the diction: w6e be nS
(F)ea7cEKEV (24 times), av3pa t(F)eKaaoTov or avapl '(F)eK6aSoIUL (12 times),
4'/hara '(F)cTofE or '(F)ecro, etc. (17 times), o6s 'earal ra5E (F)epya (7
times), 1roXcAjLuia(F)erPa (7 times), ,uoep8aX?a (Ff)a4Xcov (8 times), j3OaSKal
(/-)'i<a ,xujXa (10 times), wH kvl (f-)oLKa (11 times), aii9ovaO(f)O'lVOV (16
times), ,ieXt'q&ca(f)olvov (5 times), krl (F)otvora iovrov (8 times), and so
on. It is not enough merely to point out how many phrases, often used,
have kept the trace of the digamma, nor even to show how many differ-
ent repeated phrases a single word of the sort can occur in, such as
(F)l1cET: tai3ga (F) ie(n(at (8 times), v7r6bpa(F)l8&v (23 times), Cii-I (F)&ev
times), rTObe (F)X'cor (F)6i8, etc. (20 times), ot 8e (F) 66vTEs
6,paIX/o.LOLOv(6
(6 times), 6oppa (f)i'7fL or (F)Cu8faL, etc. (17 times), avra (F)lt8v (7
times), oaavTaor elao&ra (f) lwV (7 times), and so on. To understand
fully why the traces of the digamma are so firmly fixed in the epic
8 Cf. K. Krohn, Kalevalastudien, in Folklore Fellows Communications 16.44

(1924).
9 Ib. 56-7.
DIGAMiMA IN IONIC AND LESBIAN GREEK 133

diction we must understand the technique of formulaic verse-making.


In this way alone can we grasp the nature of the fixed phrases which
have just been quoted.
The easiest unit of diction for the poet to handle is the sentence which
fills just a verse. Such a formula is complete in itself both in meaning
and in rhythm; it carries the poem on from the end of one verse, where
most formulaic phrases or groups of phrases end, to the beginning of
another verse, where they mostly begin; and it is the one kind of formula
which can be followed by another of the same kind.'0 The technique of
all oral poetries is more or less simply that of grouping together whole
formulaic verses. The traces of the digamma are found in a large num-
ber of formulaic Homeric verses:
aXX' .y6 LOLrTO6sE(f)EirE Kal a4TpEKcWSKLtadXEOV (17 times)
Efpav (F)otKove
ol fiEv KCtKKELOvrES '(F)eKcaoTos (4 times)
aXX' Oy7E 5cVpo, 7Tr7rOV, KaL (F)68E (F)>pyozY, (P 179
7rap' L' icTLaraao x 233)
'ArpEiS7] Ka16%WTEf
(f)iiva4 avw3piv 'Ayap&euvov (8 times)

and so on."l
The next easiest formulas to handle are those which fall into the verse
between one of the rhythmic breaks and one of the verse-ends, such, for
example, as the noun-epithet formulas which just fill the last half of the
verse and supply a grammatical subject for any predicate which just
fills the first half of the verse :12
7'ThLEicIeT' eWeLTa | (3 times)
plyroffEv('ap' r(f)&vai 4vlpCv 'A^yacL/J4wv (twice)
rELTGra
&pvvro 'aVc'LK'ETE6LTcJ (r 267)
XVOe Oe '(f)O. OUpOpKa (F)&cav At6s vLos 'AwroXXwv (II 804)
Kal YV K?V EV^' &r6Xotro (f)a&ac av&pJsvAlvEclas (E 311)
This formulaic device of the predicate-subject verse divided at the
trochaic caesura of the third foot regularly supposes that the predicate
will end with a short vowel, as for example, rTO 5'fi,aerLE' 7rELTa,
which is used in 62 verses of this type, or c3s oparo,MeL5ffrE?v
5, which
10 Cf. Parry, Whole Formulaic Verses in Greek and Southslavic Heroic Poetry
in Transactions of the American Philological Association 64 (1933).
11Cf. the number of whole verses showing trace of the digamma which are to
be found in H. Dunbar's list of the verses found both in the Iliad and Odyssey
(Concordance to the Odyssey and Hymns of Honor 393-419 (Oxford, 1880)).
12 Cf. Parry, Les formules et la metrique d'Homere 3, 17-42 (Paris, 1928), on
the fallacy of distinguishing between 'legitimate' and 'illegitimate' hiatus.
134 MILMAN PARRY

is used in 10 verses of this type. The device also supposes that the
subject phrase will begin with a simple consonant, as in roXMrXas tos
'Obvaaevs(38 times), or ed yXtavKwnLrs
'A95vr1(51 times).l13 *F&vyaav&pGv
'A7a,ukvwv served at one time in this way without fault until the di-
gamma was lost, when the poets were faced with three choices: (1)
to find a new subject formula, (2) to give up this type of verse for
Agamemnon, (3) to use the verse in spite of the hiatus. An epithet of
the metrical value of *fpavai a&vpiv could not have been an easy thing
to find; in any case there is no epithet of this sort in Homer.l4 If the
poet were to give up the type of verse he would find himself greatly
hampered whenever he was telling about Agamemnon, and forced to
avoid the phrases which he regularly used for his other characters;
indeed it is doubtful if he had other phrases which he could have used
instead. Therefore he simply followed the age-long habit of the diction
and committed a fault, which was scarcely a fault, since he and his
hearers had been used to it from their earliest days.
In the complex play of the formulaic technique the device is not always
as simple as that of the predicate-subject verse. The oblique cases of
(F)ivat, for example, came down from the older poetry as the word
which could be used to finish the verse after a proper name ending in a
vowel and in the rhythm. .... . - :
. H.. IIpta6uolo ) (8 times)
.... w
'EXcMouo f)avaKTos (3 times)
.... 'Hyaalu-ooJ (twice)
and so on for nine other different names;15
... Ad KpovLCi'v( ,, (4 times)
.... lolaOOECmA JLft" (9 times)
and so on. These genitive and dative phrases in turn have their fixed
uses. Thus, for example, they allow the poet to make a subject phrase
with ,8F]where the metrical value of a name does not allow the poet
to use the nominative of the name with an epithet:
&S T d)pl.O fX Jupyas
TeXa,uiotos Aias (twice)
[lLiitaTEVKpotO(f)&vaKros (' 23)16
13 On the Homeric formulas for the predicate-subject verse cf. Parry, L'epi-

thete traditionnelle dans Homere 11-9, 47-69, and Tableau I (Paris, 1928).
14 Ib., Tableau III (opposite 112), measure XIV.

'5 Cf. L'epithete traditionnelle, 108-9.


16 Cf. H 194,
7 43, 54, 1412, X130, Q 277.
DIGAMMA IN IONIC AND LESBIAN GREEK 135

The two dative phrases quoted, on the other hand, are used chiefly in a
type of verse which expresses the essential idea of praying or sacrificing
to a god:
(s par' OLl (H 200)
OiY\ fasKpoVY
Ka,LW
sEVXOP1 f{(F)aa'VaKratl (v 185)17

Least of all should we think that the smaller words are freer from the
traditional fixity of the diction than the longer words. If anything, it is
the other way. The vast number of fixed phrases in Homer in which we
find 8e '(F)Oi, 6oppa'(F)Ol, yap '(,)oH, 7rCp '(f)ot, apa '(F)Ot, and so on,
show that the traces of the digamma are probably more firmly fixed
in this word than in any other. There are not only the many longer
formulas, such as
bY 6' apa (f)ot (V
XqLpi
(f)XE ros r' ear' (K T' 6vbcate (10 times)

There are also the many simpler series, such as the following, which gives
the poet a complete sentence to finish his verse after the verse-break of
the third foot:
VeKpOS (A 493)
6TOOV (twice)
8aX6s KEE
r , (0 421)
ZOC \F)O. ?KTT?(T? X^IROS .?
OKlrTTpOP i (t 31)

(TKVfOS | (t 34)

ekwas J (X 17).
When one multiplies this one series of formulas by all the others in
which '(F)oL seemingly causes hiatus, or seemingly fails to make posi-
tion, we can see how firm a hold the older use of the word has upon the
diction.
The few examples which have been analyzed in the foregoing para-
graphs differ in no way from any of the other phrases in the Homeric
poems where the writing in of the digamma would make the rhythm
more regular. When the phrase is often used, or is of a type which is
often used, we can most easily see the part which it plays in the formu-
laic technique; but even a phrase which Homer uses only once, and which
has no close counterpart elsewhere in the diction, is also to be taken as
traditional. Either the phrase as a whole is traditional, and it is only
the scantiness of the remains of Greek heroic poetry which does not allow
17 Cf. N 758, 770, 781; ' 836, t 499; e 409, Q 77, 159.
136 MILMAN PARRY

us to show its place in the formulaic diction;l8 or it is a later creation by


analogy,'9 which is much the same thing. It is probable that in some
cases a poet who himself did not know the digamma made a new phrase
in which he used a word as though it still had the sound; this means
merely that he is following the traditional use of the word, though he is
not using one of the older phrases which maintained that use. Cases,
however, in which any one poet, such as the poet of the Iliad, would use
one of the words we are considering in a new phrase, are so few as to
be negligeable: with the oral poet the making of the new phrase is very
rare, and almost always due to the chance play of traditional phrases.20
We find constantly that the better understanding of Homer's formu-
laic diction confirms the soundness of the traditional text. Thus even
as we should not try to restore the digamma where the rhythm is seem-
ingly irregular, so we should not touch those places in which, while
there is no irregularity, the digamma can be very easily restored. There
was, however, a certain soundness in the emendation of the text which
was practiced for two centuries: the editors in most cases really did
establish the older form of the phrase. For example our texts have
rYv y' aure TrpoceiL7Ev ava!i ava piv 'AyapeJlVvW (5 times)
c&sZ(par' o6' actriLr7aEvava av5pcv 'Ayaiepv,uvv (twice)

By removing the v-movable of WPOpE?EPT?Vand a4rLitio^v, and restor-


ing the digamma we get what must have been the older form of the
verse. Certainly the sense ofQ 154 is better when we read*os f' ateL,
and M?LXLXloLs klEOcTw (10 times) was surely made from *IElX?tXloLf
FbrT7LCPY. These changes, however, were made by the later singers
themselves, drawn on by the habit of their spoken language, but held
back by the constraint of the formulaic diction.
This does not mean that we can use the digamma to establish an older
form of the Iliad and Odyssey. We can say that certain phrases and
verses are the creation of an older time when the digamma was pro-
nounced (though even here we must admit that certain forms may have
been made later by analogy), but such phrases and verses will be only
the older parts of the diction which the singers used side by side with
later phrases and verses. Just as we can show the metrical usefulness
18 Cf.L'6pithete traditionnalle 125-30.
19Cf. ib. 85-94:'La diction formulaire etle jeu del'analogie'.
20 Cf. Homer and Homeric Style 144-7. I have been able to observe in South-
slavic heroic poetry, in its actual practice, this complete absence on the part of
the oral poet of any thought of making original phrases.
DIGAMMAIN IONIC AND LESBIANGREEK 137

of the older phrase, and the fixed place which it holds in the diction, so
can we do for phrases with newer forms.20a For instance, before the
digamma was lost, the singers were able to use the following verse only
in the masculine:
*Kal Fw,'v<pwvl'iTaS fEfrea 7rTEpoEVTa lrpoor?pJca

But by Homer's time the verse could not only be used in the masculine,
as it appears 30 times in the Iliad and Odyssey, but also in the feminine:
Kat u.Lv jvr?rTLaa1
ei7ea IrTepoevra 7rpoaruv&a (9 times)

In the same way, after the example of uEcXt?lca(f)otvov (5 times), the


singers made pueAXiq70osO'Vov (twice); after the series given above,
'HIqafLroto (f)avaKTos (twice), etc., they made HoaTeLacwvos avaKros
(twice). They could now begin verses with
(Alveias (E 168)
TOv' L ev (l8ov) 'AvTLiXOXOS(E 565)
1#VvjaWOL (S 556)

The following verse should be particularly noted, since it shows the


early use of (F)E:TElw (or ELEtrEv), and the late use of ov side by side in the
same formula:

o6X-riaas 5' apa (F)Ec'rE rrpds ov ,eoyaXLXropaiv96,v

(A 403, P 90, S 5, T 343, 4 53, 552, X 98; e 298, 355, 407, 464).21 We

20a Cf. D. B. Monro, Homeric Grammar? 401 (3) (Oxford, 1882).


21Knos (215) admits the impossibility of emending this verse. Bentley wished
to read ?? wrpos'f6, but such a use of tprfis not Homeric. Fick corrected to
esEL1rEirpos &o, but the contracted tov supposes the earlier loss of an inter-
vocalic digamma in a form *aFFos. There was probably never any form *c4^or,
but only *aF6s. The form k&sis the classic modification of toS which in Lesbian
is itself an artificial creation, on the analogy of &oz'and wos, to compensate for
the loss of position in arsis. Thus TOtSi i&V ovs i2picKaKE p,Iwvvxas tir'ovs (E 321) is
probably for an older *robs ,zv 0o7os t!piucaKe; ractos eoZois for an older *TaL86s
rFolo, etc. In those cases where *UF6Shad made position for a final short vowel
in thesis, however, the metrical fault could not be corrected, and Homer uses
i}vuyarTpa rjv (4 times), 7r&ei Ct) (E 71), TkKeL(c (twice), etc. The actual existence
at some time of a form *ff6S may be doubted. The dissylabic forms of the word
which are cited from Aeolic (Bechtel, Griechische Dialekte 1.277), are all poetic
and are evidence that the Homeric 6s dates from the Aeolic period of the poetry,
when it was adapted either from an old Aeolic or an Arcado-Cypriote *UF6S. Cf.
Homeric Language as the Language of an Oral Poetry 29-33.
138 1MILMANPARRY

can see how this verse was made. The singers had inherited on the one
hand such verses as (1)
,..,~6vos
6 ,lia f , JcEis oipavov evpiv (T 257)
EVc 8 Vosi p ELr^EV
(F l{k(F)oL^ora irOvrov ('F 143)
oX cL?as J r' far' ?K T' ovocE
[eroS (P 248)
and on the other hand such verses as (2)
cs pauav p.6O V
a(XX' o0 WEdjOV /LEyaXflTopa avjF.Ov (( 500)
ir6XX' oTOJvoirj o 6o T&( L2EyaX7T0pL VLiCL (0 674)
(T

They also had such verses as (3)


EvCa,4Evo0sa'apa (F)ElTrEV EK?Of36XO.t 'Air6XXWvz (I 513)

And finally, one of their most common words for 'speak to' was rpoFLwrrtlv.
By the natural play of phrases which is the essential process cf oral
composition, the words of the first two types of verses combined to
give a verse which is syntactically like the third type; lrpoc4(F)ELrte
gave (F)eireI7rpos, and the third person 6v, losing the trace of the
digamma, took the place of the first person k,bvin such a verse as l 500,
or of the second person cOAin such a verse as 0 674. The verse thus
made proved so useful that it won the fixed place in the diction which
led Homer to use it 11 times. It is an example of the stability of the
diction as a whole and of the fluidity of the diction in the grouping of its
elements, and as such it shows that while there are older and newer
phrases in Homer they are not necessarily the mark of older and newer
passages in the poems. It should be fully understood, however, that
the number of outstanding formulaic phrases which refuse the digamma
is small;22this merely shows how little the Greek heroic style changed
over a long period of time.23
The traditional formulaic diction must have trained the ears of the
singers and their hearers to feel the traces of the lost initial digamma
much in the way that the French feel the traces of h-aspire, so that,
while they say l'herbe, l'homme, they say la hache, le hetre. In the
French it is the feeling for the rhythm of the close group of the word
with the article which has kept the feeling for the lost sound; in the
Greek heroic poetry it was a feeling for the equally fixed word group.
It has been objected to this that the keeping of h-aspire is largely due

22 Cf. Knos 50-146.


23 Cf. Homeric Language as the Language of an Oral Poetry 43-5.
DIGAMMA IN IONIC AND LESBIAN GREEK 139

to formal education,24and to the fact that it occurs in regular combina-


tion of article and substantive, whereas in Homer hiatus occurs between
all sorts of words.25 The answer in both cases is that the feeling for the
digamma was maintained by the altogether artificial constraint of the
diction. Nevertheless the parallel of h-aspire should not be pushed too
far, because the singer was also habituated by his spoken language to
the use of the words in question without any trace of the digamma: he
had not one, but two conventions. For this reason we should not insist
too much on the feeling for the lost consonant as an acting factor, but
only as a factor which justified the irregularities which the diction
imposed.26 The ever increasing use of the words without regard for
the lost digamma, which we can trace in Hesiod and the Homeric hymns,
shows on the contrary that the habit of the spoken language was con-
stantly wearing down the habit of the song. Nevertheless the latter
habit should not be overlooked, since we find it lasting well on into Greek
poetry. Thus we have in Archilochus 'EvvaXloLo (F)ivaKTos (1.1
Diehl), in Solon Ka\a (f)'pya (1.21), in the epic poet Antimachus
,eXavos (f)o'zvolo (19.1 Kinkel), and in Apollonius of Rhodes KaractL-
.kvoLo (F)aivaKros (1.411), av3pa '(F)cKaCTroL (1.399), etc.27
It has also been objected that while hiatus resulting from the lost
digamma might be bearable, the fault of the short syllable in the first
part of the foot is too grave to be possible.28 McXavos(f)oolvoLoof
Antimachus, which was just quoted, shows that this is not so, but of
much more weight than this is the great number of short syllables in
the first part of the foot in Homer which come not from the loss of the
digamma, but from the faulty putting together of formulas.29 For
example, verses such as

iahL?pov KaLTaSaKpV XkovoaL (Z 496)


evrpowraXLPo/.Lvv?,
/.LXa cTTEVaI Xcs
flVKV&L T?rXis iV7VEclOS (2 318)

may combine to make such a verse as

vTpo7traXoPo/.SEVoscs TE Xis 1viyePELos (P 109)

24 Monro, Homeric Grammar ?402.


25 A. Meillet, Apercu d'une histoire de la langue grecque3154-5 (Paris, 1930).
26 This was the mistake of G. Curtius (Grundzuge der griechischen Etymologie5
560, Leipzig, 1879), who first adopted this explanation.
27 Cf.
Argonautica, ed. G. W. Mooney 416-421, Dublin, 1912.
28 Meillet, loc. cit.
29Cf. Parry, Les formules et la metrique d'Homere 34.
140 MILMAN PARRY

Similarly the poet may change the case of a formula with a resulting
failure to make position, as when after uepo6urwv
aYpcr4wv (9 times) he
makes ,ukporcs 'vppw.iroi (2 288).30 It should also be remembered
that we have to do here with song, and not with speech, so that the poet
would be allowed a freedom in treatment of vowel or consonant which
would not be possible in spoken verse.3'
Those cases in Homer in which the use of an uncontracted form might
seem to show a survival of intervocalic digamma are to be explained
in the same way as the seeming use of initial digamma. A single exam-
ple will suffice. The uncontracted e(F)Etreoccurs in a whole series of
fixed phrases, from whole verses such as
ur] 6' bpz6s Kalti ov(V 'ApYECoLoLwE(f)e.TwEY (7 times)
alvtrare KpoviL8, iro TOV
r?v P.OV <(f)E?iTes (6 times)

to shorter phrases such as KaTra Mopav E(f)et7Es (10 times), or olov


d(f)ELtES (7 timnes). One notable series is the following:
>
PlKEpSES (f 509)
|vjuaXyes| (twice)
Ciros<oboovrvov6v e(r)eti7re(s) (twice)
vr].EpT?ES (r 204)
Kara ,loZpav (4 times)

The problem of the digamma in Lesbian poetry is very similar to that


of this sound in Homer: we find cases where the digamma would correct
a seemingly faulty rhythm and others where its introduction on the
contrary would harm a rhythm which has no fault. The only difference
is that whereas in Homer the number of words which refuse the sound is
small in comparison with those which call for it, or take it by an easy
change in the text, this proportion is reversed in Sappho and Alcaeus.
The natural conclusion is that the Lesbian poets, like Homer, were
following a tradition of poetic diction in which certain phrases had kept
the trace of the digamma, but that they were further away than Homer,
either in time or in evolution of style, from the time when this sound
had its natural place in the poetic language. Now this is not the pre-
vailing critical opinion. Modern editors, with the exception of Lobel,32
follow the line of reasoning which, we have seen,33 is usual in such cases,
30 Cf. ib. 11.
31Cf. Krohn, 56.
2a7roovs MkXn (Oxford, 1925), 'AX,cciouMekXi(Oxford, 1927).
32 E. Lobel,
33Cf. above, p. 132.
DIGAMMA IN IONIC AND LESBIAN GREEK 141

and restore the digamma, and even emend the text to take it, and they
support their view, which is also that of the writers on Greek dialects,34
by a certain amount of ancient evidence. That evidence, however, is
less sure than is generally supposed.
The form FoLtaL in a papyrus text (Sappho a' 3.6 Lobel, reproduced
in photograph in OxyrhyncusPcapyriI, plate II), and fOV in a text from
Herculaneum (Alcaeus 120.235Lobel) actually show us the digamma, but
aside from these two cases there is no other example of initial digamma
in the papyri or in the literary quotations of Sappho and Alcaeus; all other
evidence comes from the grammarians or from Balbilla, a Roman lady
poetess of the second century A. D.36 The correction of r' eireiv to
Ep7LJEIYin the fragments of Sappho and Alcaeus quoted by Aristotle
is uncalled for, since elision of the dative of the pronouns is regular in
Lesbian poetry, and we have ra4' Ew7r7v elsewhere in Alcaeus (48.2
Lobel), and r65' Ei7rl[ elsewhere in Sappho (a' 8.12)37. Similarly axri
6 EK 7EiraXaw racE &yvTTTLt quoted by Proclus has been corrected to
Fa&EaTcrrTt, but the reading of the text is too bad to be the work of
any but the latest time and of a scribe with the haziest ideas of classical
syntax whose ear was led astray by the r's of the following word. The
Fbefore aiEa is no surer than the v after it, and Lobel is right in reading
simply 7-E?TaXcw a5ea TETTLt.38

When the ancient grammarians give a certain form as 'Aeolic' we


cannot be at all sure that they mean Lesbian. Their studies of dialect
were rudimentary; they never grasped the rigorous notion of dialectal
subdivisions; and the forms they quote may come not only from Lesbian
poetry, but also from Boeotian poetry or the spoken dialects of Thes-
saly.39 Thus we have the 'Aeolic' glosses 7yaXXot (rXot), Pys,tara
34 0. Hoffmann, Griechische Dialekte 2.454-9 (Gottingen, 1893); F. Bechtel'
Griechische Dialekte 1.11-5 (Berlin, 1921); R. Meister, Griechische Dialekte
1.103-6 (Gottingen, 1882).
35Cf. A. Vogliano, Herculanensia, in Atti della reale accademia delle scienze di
Torino 47.91 (1911).
36 Lobel alone of all the editors has seen the weakness of the evidence for initial
digamma before a vowel (SawXpous McX,1xxx). For a contrary view see J. M.
Edmonds in Cambridge Review 47.211 (1926).
37 Rhetoric 1376a9, 12; cf. F. Bechtel, Griechische Dialekte 1.12; Edmonds,
Lyra Graeca2 1.266 (London, 1928); on elision in Lesbian poetry cf. Sa&r'pousMekX
lxi.
38 Cf. Edmonds, Cambridge Review, loc. cit.
39Cf. Hoffmann 2.223. Edmonds (loc. cit.) is wrong in saying that John the
Grammarian identifies 'Aeolic' with Sappho and Alcaeus. He merely gives his
confused description of the 'Aeolic' dialect, and then says that Alcaeus and Sap-
pho used it. The correct statement could be that the grammarians associated
everything 'Aeolic' with these poets.
142 MILMAN PARRY

(.l4aTra), yolv]fj (7rloTaMAaL),


y6XaMuos
(ta7^y6s), ypiYoS(6epgua),in which
the initial 7 is supposed to be a miswritten F.40 We have, however
another gloss: rO yap ol8a ol574u (cpaSlv ot AioXETs,41
and our texts of
Sappho and Alcaeus show us that where Fp- survived it became Bp-,
as in /3po6o8aKTvXno and fpoba of a papyrus of Sappho (e' 5.8 and
13). The easier explanation is that the grammarians had found, or
rather heard of, some regional Aeolic dialect in which the digamma
had survived in a velarized form, and that they concluded that this
was the actual sound which had been used in Lesbian poetry. It
must have been in accordance with such a doctrine that Balbilla, as we
know from the inscription of her poem on the colossus of Memnon in
Egypt, actually wrote 7yo0and ye as what she thought was Sappho's
language,42and that our texts of the grammarians almost always show
us the digamma as 7 in the words which are given as examples. Bpoao-
aaKTvXoS, etc., however, show that where the sound did survive in
Lesbian it was labialized. If John the Grammarian really states that
poivov was Aeolic for olvov,43 we nevertheless know from the fact
that Aulus Gellius quoted the phrase of Alcaeus (108.1) as rvev,uova
olVOc, while Athenaeus and Plutarch quoted the obvious correction of
this reading irXev,iovaso^vwc, that he did not find this form in the
text of one of the Lesbian poets. Terentianus Maurus says: 'Quamque
ITUV dicunt Achaei, hanc vitym gens Aeolis'44which goes no farther
than the vague term 'Aeolic'. Trypho the Grammarian really does
refer by name to a Lesbian poet: 'a7rat e 7rap' 'AX\KLcl TrO p7lJS
oUp?tlts X76Ta(L,45 where the spelling seems to be the survival of
some attempt at a phonetic explanation. Finally Apollonius Dyscolus
says:
Tfa0cs 6TL Kal TO AlOXLKOV e1&ya/uLa rats KaTa TO rptrov Tp6cTs.wov
7rpoOeJPe/ITCLL,KCLtO KaL L 76
airr (pf?VTOs apcoLpiEvaL aa7vvovTaL. 'AXcKaos
(110) X
5OTE 6awv J.u6 (v' 'OXVpTiVrw
\vfaai ar?p 7c??e.46

40 Hoffmann 2.236-7.
41 Choeroboscus, Scholia in Theodosium, 342.1 (ed. Hilgard, Leipzig, 1889).
42 Hoffmann 2.125, vv. 7, 15.
43 Hoffmann 2.217, ?21. The text has sowov which is itself not an impossible
form, but may actually have come into the poetry as a form to correct the rhythm,
as ios (see above, note 21) was made from *Fo6s.
44 De Syllabis, v. 658, text vitem (Keil 6.344).
45Id^q &4ecov11.
46llepL'AvTcvJZas 761.32, ed. Schneider and Uhlig (Leipzig, 1878); text XvoYearep
7eLt?ey.
DIGAMMA IN IONIC AND LESBIAN GREEK1 143

His remark explains all the confusion of forms under which the gram-
marians refer to the digamma, and the discrepancy between their
remarks and the tradition of the poetic text itself: Apollonius is not
basing his view upon the evidence of the texts, but is arguing for a read-
ing of the text from metrical evidence. The last part of his statement
refers to a view set forth more fully by Velius Longus,47 according to
which the sound h was a consonant and could make position in such a
verse as
X 6\lyov o0 rataca iOLKOT6a'yELVaTO
Tvs?us (E 800).
The ancient critics had observed the similarity between Lesbian and
the other Aeolic dialects, and they had found somewhere in these other
dialects in some form or other the digamma which would correct what
they thought were the metrical faults of Sappho and Alcaeus. They
thus practiced emendation on exactly the same grounds as the modern
critics do. Indeed, the source of the modern practice lies entirely
with the ancient grammarians, for it was by reading the Latin gramma-
rians that Bentley got his whole notion of the restoration of the digamma.
In view of this it is reasonable to suppose that the two cases where the
papyri show us the digamma are also due to the work of ancient editors.
That they seem to have limited their correction to the third personal
pronoun and possessive adjective would be indicated by the fact that
Apollonius Dyscolus, in the three passages where he deals with the
'Aeolic' digamma,48 is concerned only with these words, and it would
be explained by the large number of cases in which these words must
have invited emendation, even as they do in Homer. The fixity of
the older use of these words in the diction is further indicated by the
fact that we find even in Ionic poetry 7 be '(f)oi in Archilochus (25.2
Diehl), and ov&e'(F)Ot in Semonides (7.79 Diehl).
When we have thus removed from the problem the confusing theoreti-
cal forms of the grammarians and editors, both ancient and modern, we
find that the Lesbian lyric poetry conserved traces of the digamma in
just the way that the Ionic heroic poetry did. There are some fourteen
cases in which Sappho and Alcaeus elide before words which had once
had an initial digamma, and some seventeen cases in which the intro-
duction of the digamma would lengthen a vowel which must be metri-
47 Keil 7.52.
48 The second caseis qpaLverat (f)ol K77VOS (p. 82, 1. 17). The third case, in which
the text of Apollonius reads r5v eov 7,-aa KaXEt (p. 107, 1. 12), cannot be corrected
to rov F6Wbecause Apollonius is discussing precisely forms in -so-.
144 1MILMANPARRY

cally short. Side by side with this neglect of the digamma we find five
cases where the phrase shows the traces of the digamma: arep (f)cEV
considered above; (paLveraL (f)oi (Sappho, inc. lib. 48), also quoted
as an example by Apollonius; rXEcF4ova (f)olY?L; &7r6 (F)ipyov (Alcaeus
108.1) beside artiuorL6S (F)epyov (Alcaeus 36.19); and the famous
yXwraaa (F)t(g)acE (Sappho a' 2 App., 9).49 Similarly we find the
traces of the digamma in such forms as e(F)eL1rE (Sappho E' 3.3) beside
etrov (Sappho e' 4.8), a]7rv(f)EL7r7l[ (Alcaeus 64.4), and (F)avaaae
(Alcaeus 118). We have not enough of the early Lesbian poetry to be
able to show, as we can for Homer, fixed phrases in the diction which
must be traditional. 4aLverai ,uoi K?YOS (a' 2 App., 1) and <palvera
ot K?jVOSof Sappho may, or may not, be such examples. We do know,
however, that the two poets lived at the beginning of Greek poetry, and
that if they were not following an oral tradition of lyric poetry as com-
pletely as Homer followed the oral tradition of heroic poetry, they must
anyway have been more or less closely bound up with the popular
tradition which had brought down from the past phrases which had in
them traces of the digamma.50

49 Confronted by these last three cases Lobel, who by rigorously keeping to the
evidence of the texts had saved himself from the error of restoring the digamma,
commits an even worse fault: he marks two of the three passages as corrupt and
attempts emendation of them, while he reads irXed,oPas olv&t after the quotations
by Plutarch and Athenaeus. One can understand how 7rX\eovca (F)ootvwt became
plural but not how the plural could ever become the singular.
60I have dealt in Harvard Studies in Classical Philology 43.30-3 (1932) with
those cases in Aeolic and Ionic poetry where the loss of the digamma has modified
the form of the word itself.

S-ar putea să vă placă și