Professional Documents
Culture Documents
TEODORSSON (1979) - On The Pronunciation of Ancient Greek Zeta
TEODORSSON (1979) - On The Pronunciation of Ancient Greek Zeta
TEODORSSON (1979) - On The Pronunciation of Ancient Greek Zeta
Sven-Tage TEODORSSON
Department of Classics, University of Giitehwg
Ionic-Attic zeta undeniably represents the reflex of *[dj], *[gjJ, *[j] as well as
original *[zd]. These two sources fused imo a cluster *[3d3] that existed by the time
the alphabet was introduced. Then there was a divergent evolution into [d:], [dz],
and [zd] in different dialects. That ieta did not correspond to [zd] in Attic is demon-
strated by the absence of an orthographic variation <Z> m (Cd) in that dialect.
Phonetic prehistory
(on analogy with -KOVTGI) (?) > Lesb. (Tp&)KOl(TTOS; &V~U (Arcad., Cret.
etc.) > 7&7(x.
PhonoEBgicalititerpretationand graphemiccorrespondence
Together with the fact that there is nothing to show that the reflexes of*
Proto-Greek *[dj], *[gj], *[j], and of *[zd] respectively were ever reprc-
sented by different graphemes when the alphabet was introduced,’ the’
above evidence indicates that these reflexes had amalgamated to a cluster
that was interpreted as a phonological unit. There is no n /~riori reason to
suppose that any one of the Phoenician or other signs was originall!
adopted for the denotation of anything but what was felt as one phon+
logical unit. This is shown by the use of zllJVi2<I> (zeta) instead of st’qut’nco
as ‘ZA>, ~‘AQ, c’ ..L’ZJ~,or the like. Other indications of such monophoncn~ic
anal:isis are the uses of @>, (Y), <X) and (@>, the Carian-Ionian ,T
(sampi), and the Pamphylian CL+.‘).
Indications of a biphonemic analysis is provided by the use in SOIIIL
dialects of grapheme sequences, either consistently or sporadically 01
temporarily, -instead of these graphemes, e.g. Attic (KY), i:;;‘c), for \‘f/ !,
G), literary Lesbian (cj, for (Z), Theran (l7H), (KH) for <cli>, .Y p
Cretan .:lIZ>. (KC)
\ for (Y), (E), etc. These data show that either of’tht:
two phonological inter Jretations was possible, which means th;tt thz
phonetic correspondences of the signs mentioned above differed son~t’h~
from other*, that were never denoted by either one or two signs, SLKII\WYJ
the phonemic sequences [zb] and [zg] which, unlike the originally (i.c’, I!\
onic-Attic dialects
Lesbian literary orthography OS, Boeot.-Thess. OE~C~OTOS (OE+~~OTOS),
Delphic inscriptional 21[0v for &t70v etc. suggest the pronunciation [zd].
But the Lesbian inscriptional orthography (2) (&.&&J) and sporadic
use of (Z> for <A0 (5% and the Boeot.(-Thess.) orthography (A(A))
(5&, qkqb~3Sw)show that in Lesbian the pronunciation was probably not
generally a sequence [zd] but rather some affricative cluster,3 and that in
Boeot.(-Thess.) the pronunciation was dominated by a dental sound.
In other dialects there is no decisive evidence that the reflex of *[dj],
*[gj], *[j]. and the original sequence *[zd], was pronounced [zd]. In Elean,
Laconian. and temporarily in Cretan, the normal spelling is (d(d)), beside
.‘T ( T ) ) (Cret and El.) and (Z) (Cret.). In these dialects, then, the pronun-
i&ion was probably [d(: j], or an affricate, or possibly [a( :)]. The last two!
alternatives are also most probable for Corinthian* and Argolic? That
original *[zd] had been amalgamated with the reflex of the palatalization
in Boric also is only suggested by the gloss &5avdv* &pdv* A&CWVES
(Hesych.)?
Ionic-Attic
The prcnunciation [zd] has been assumed most generally in Ionic-Attic?
Some scholars allow for two phonetic values, [zd] and [dz] (Schwyzer 1939 :
I, 1’79),while a few maintain the latter alternative (Curtius and Windisch
1879: 615; Matthews 1954/5). It has also been suggested that the pronun-
ciation was [3] (Witton 1898). Finally, a fifth solution has been proposed
by Bailey (1968), who argues, on analogy wit Old Church Slavic, that (2)
had the pronunciation [3d3] generally.
2 Lup~$ (1972: 27) suggests that the use of one grapheme (Z) indicates that the sounds of
this particular sequence had combined closely so as to be pronounced [zd], and not [sd].
This Pmpiies that, by the introduction of the alphabet, the sibilant was voiceless in the
sequences [!*b] and [sg], which is structurally improbable and remains to be demonstrated.
3 For a presentation and discussion of the evidence pro el contra the pronunciation [zd] in
Lesbian, set: Arena 1960: 513-523.
a Cf. the fcrm {iKa (= &a) Hesperia 5 (1936) 235-241, 6th century.
5 Cf. Fan<& i Mt. &l~) (lG IV 506, 5).
GFrom $&I, d{aXos; cf. Old PO]., Czech ozd ‘kiln’. The etymology has been questioned.
7 Blass 1888;: 112-l 19; Brugmann 1913: 42; Buck 1928: 66; Brandenstein 1950/l: 45-46;
Diver 1958: 18-19; Bartonek 1961: 152; Lejeune 1972: 112-114, et al.
326 S.-T, Teodorsson 1 On Ancient Greek zeta
EcThe cases are: ZS& (Kretschmer 1894: 103, 85) 600/575: and perhaps 7oio5’ ht%p[av~os
&5G 10, 404, 3) ca. 47615 ; ndcdas (Audollent 1904: 94, 66, 5) 4th/3rd c.; L&V’S (Graef and
Langlotz 1933: 211) 530/00. The last three instances do not provide any conclusive rnforma-
tion a bout the phonetic correspondence of (2 ,\. The data were collected by Teodorsson
f 1974) in a comprehensive study of orthographic variation in archaic and classical Attic.
’ There is a varied and rather widespread variation between (5, and \!?‘J respectively, and
relevant graphemes or graphemic sequences (Teodorssan 1974: 141-143). The Old Attic use
of the sequences :’XZ and (@Z:. indicates a biphonemic interpretation.
I’ An exhaustive investigation of the orthographic variation in Ionic would give the definite
ansiver.
palatalization, which was most probably an affricate *[d3], was combined
with the original sequence *[zd]. The metathesis theory is rightly questioned
by many phonologists (e.g. Witton 1898: 429; Allen 1957/8: 120,n. 40;
Bailey 1968 : 177-I 78). Even less plausible is the idea that the mutation was
directly into [zd], without the appearance of an affricate at all (Meyer 1896:
369-371; Brandenstein 1950/l : 45-46; Diver 1958: 18). It is therefore
appropriate to adopt with Matthews (1954/5: 75-76) and Bailey (1968:
182; the apparently parallel Old Church Slavic case as an explanation of
the Jzvelopment. I1 Modern Bulgarian [3d] (meidu, cf. Lat. medius, Skt.
mehas) derives from Old Bulgarian *[dj] through *[3d3], by the mediation
of criginal *[zd] (cf. Diver 1955: 229-233). However, Bailey is hardly right
in assuming a lasting existence of a cluster *[3d3] in Greek. Such an
XX lgamation would be unprecedented in Greek and, as lacking any
structural support, is possible only as a short-lived transient stage.
However, as such it may have been possible and, indeed, such a stage offers
a phonetically natural mediation between the affricate *[d3] and original
*[zd], which undoubtedly fell together.
If the stage *[3d3] is posited for the time when the alphabet was
introduced in Greece, then the double reflex reference of I, viz. *[zd]
and *[dj]/*[gj]/*[j], becomes natural. The alveopalatal initial sound [3]
either itself brought about the loss of a preceding nasal (a6u-3dyyos >
*o&3~/3uyor > &~u~os) or was depalatalized early into [z]. The cluster
*[3d3] was invariably interpreted as a phonological unit, unlike the
seqi.ences /sb/ and /sg/, and was thus denoted by one sign ~nyin (I) (zeta).
It is reasonable to assume that for a short period the stage *[3d3] was
common Greek, although the evidence is very meagre for Doric (cf. above,
fn. 6)
of the following Ionic development *[tj’] > [ts] > [s:] > [s] (Koine) is not
established.13
In Boeotian there seems to have been an early development of both the
voiced and the voiceless reflex of the palatalization into [d:] and [t:]
respectively. Attic is intermediary between these extremes, Ionic and
Boeotian. In the early post-Mycenaean period, during the Aegean migra-
tion, Ionic-Attic was probably still undifferentiated.14 The origin of Attic
[t :] l5 as opposed to the Ionic sibilant pronunciation is supposed to be due
to Bocotian influence. However, the difi’erentiatiorl between Attic and Ionic
in this point appears to still go on in archaic and classical Attic? Such
formal categories as xwkaacr: suggest the undecided situation. There ~-my
have existed a (sociolinguistic?) variation in Attic in this point ot- the
phonemic system. l7 The variation was possibly of the type [t :] m [s: 3, the
latter being the Ionic pronunciation, which can be assumed at least ii OIII
the 5th century on, although (T) was still in use besides @‘L’) (cf. fn. 13).
In view of this Attic variation in the voiceless reflex of the palatalization,
it seems reasonable to suppose that the development of the voiced
affricative cluster, i.e. *[3d3], produced a similar intradialectal variation in
Attic and a similar relation between the three dialects, Ionic, Attic, and
Boeotian. If Ionic-Attic (L’Z>/(~ > corresponded to [ts] ( c *[tj]) from
some early point of time on, it seems probable for structural reasons that
Ionic-Attic :Z\ corresponded to [dz] ( < *[d3] < *[3d3]) during the same
period.18 Structural arguments also seem to account for the lack of
orthographic variants as QLZ) or ld2Y for \;[Z\ in Attic. tJr;Iikc phonetic-
phonemic sequences as /s/ -t /d/, /s/ + /b/, /s/ + /g/, Ldi.1 _ is felt as a
I’ The grapheme s’V \ was still in use in Iomc during the 5th century, which, howeb :r, dots
not mean that the change [ts] > [s:] had not yet taken place.
I4 The development of original *[s:], *[!.s], *[tth’j] (homoll~orphemic, cf. Allen 1957,8: 124,
n. 56) > [s] was common Ionic-Attic.
l5 That the dewlopment was actually into [t :] is demonstrated by Koine and Modern Greek
forms, 7++LCU, ~TLTT&OV, *ypjTTdS.
I9 The Ionic(-At tic) development may perhaps have been *[tJ] > [j’:] ( > [s :]) and *[3d3] >
r3:] (> [z:]) instead (cf. Witton 1898). But the evidence for this, e.g. Skt. hariya dplt(w,
spellings with uu before a stop, E&S$, cliuo/3ov etc. (Bailey 1968: 179), is weaker than the
contrary evidence (cf. the material in Matthews 1954/5). There is also Pers. takabmi Yaumi,
oatm#dp01 “IWV~S < *twak- (cf. Skt. tuac-)which would indicate a preserved f-element even
initially in the reflex of *[tw], which is represented by U- generally in Greek dialects (Pisani
1964). - It cannot be proved that the transliteration was from Ionic, but there is no evidence
against this.
2o There are only two instances of the variation (d) N <Z> (cf. above, fn. 8), and six
instances of variation between (7’) and (Z) (Teodorsson 1974: 140).
2l “You know that our elderly people used the iota and the delta very much, and not least
the women, who more than others preserve the old pronunciation. But now people change
the iota into epsilon or eta, and the delta into zeta, because these sounds are more impressive.
. . . And you know that the old said &uoydv instead of &~ydv.. . . But now we say Svd~. And
there are a lot of other words of that sort”.
330 S.-T. Teodomon / On Ancient Greek zeta
the population, in contrast to the [dz] spoken by the majority. By the ‘new
pronunciation’, in any case, the Koine pronunciation [z:], which is known
to have spread in Attic by about 350 B.C., is most probably meant This
change [dz] > [z: ] was parallel to Ionic [ts] > [s : ] which, as sugb:%ed
above, may also have been Attic to a certain extent. Both changes may be
supposed to have taken place earlier in Ionic than in Attic. In either
[ts] > [s:] probably preceded [dz] > [z:] by a considerable space of time.
It follows that there is no need for the theory proposed by Nagy (1970:
127) and accepted by Allen (1974: 155) that Attic [t :] and the supposed
Attic [zd] were replaced by [s : ] and [z: ] respectively through influence from
the Koine. If most speakers of classical Attic used the pronunciation [dz],
as in Ionic, the change into [z:] was a normal Greek regressive assimilation,
probably furthered by influence from Ionic. Ionic influence during the
classical period is also the most probable cause of the victory of [s: ] over
[t:] in Koine.
Regressive assimilation
he range of [zd]
historical times. This unique value provoked the attention of the first
grammarians and was thus integrated in the learned tradition.23
Transcriptions of foreign words into Greek were certainly seldom made
by people lacking elementary insight into grammatical teachings. It is
therefore only natural that Old Persian Artavazd~ and Semitic Aidtid
appear in Herodotus as A~$~cc~os and A~c~)ToP, or that the name of the
od Auratttazdii is rendered ‘li?p~~c?$ in Plato.
eferences
Allen, WS, 1957/L Some problems of palatalization in Greek. Lingua 7, 113-133.
Allen, WS., 1974. VOXGraeca. 2nd ed. London: Cambridge Univ. Press.
Arena, FL, 1960. II valore di 5 nei vari diaietti greci. Rendiconti Istituto Lombardo, Classe
di Lettere e Scienze Morali e Storiche 94, 513-553.
Audo!!ent, _4., 1904. Defixionum Tabellae. Lutetiae Parisiorum: Fontemoing.
Bailey, C.J.N., 1968. The pronunciation ofZeta in Ancient Greek. Papers from the Regional
Meeting. Chicago Linguistic Society 4, 177-193.
Bartonitk, A., 1961. Vyvoy konsonantickeho systemu v ieckych dialektech. (With an
exhaustive summary in English.) Praha: Statni Pedagogicke Nakladatelstvi.
Blass, F., 1888. Ueber die Aussprache des Griechischen. 3. Aufl. Berlin: Weidmann.
Brandenstein, W., 195015.. Phonologische Bemerkungen zum Altgriechischen. Acta
Linguistica 6, 31-46.
Brugmann, K., 1897. Vergleichende Laut-, Stammbildungs- und Flexionslehre der
indogermanischen Sprachen I : 1 (Grundriss der vergleichenden Grammatik der
indogermanischen Sprachen I). 2. Bearb. Strassburg: Trubner.
Brugmann, K., 1913. Griechische Grammatik. 4. Aufl. von A. Thumb. (Handbuch der
klassischen Altertumswissenschaft II : 1.) Miinchen : Beck.
Buck, C.D., 1928. Introduction to the study of the Greek dialects. 2nd ed. Boston:
Athenaeum Press.
Curtius, G., 1879. Grundziige der griechischen Etymologie. 5. Aufl. von E. Windisch.
Leipzig: Teubner.
Diver, W., 1955. The problem of Old Bulgarian s’t. Wet
Diver, W., 1958. On the prehistory of Greek consonantism. Word 14, l-25.
Gonzalez de la Calle, P.U., 1948. Francisco de Vergara y la pronunciacibn de la z griega.
Boletin de1 lnstituto Caro y Cuervo, Bogota, 4, 249-320.
Graef, B., and E. Langlotz, 1933. Die Antiken Vasen van der Akropolis zu Athen II. Berlin:
De Gruyter.
Kretschmer, P., 1894. Die griechischen Vaseninschriften, ihrer Sprache nach untersucht.
Giitersloh: Bertelsmann.
Lejeune, M., 1972. Phonetique historique du mycenien et du grec ancien. Paris: Klincksieck.
Lupa$, L., 1972. Phonologie du grec attique. The Hague: Mouton.
Matthews, W.K., 1954/5. The pronunciation of Attic Greek 5 in the sixth and fifth centuries
B.C. Lingua 4, 63-80.
Meyer, G., 1896. Griechische Grammatik, 3. Aufl. Leipzig: Breitkopf & H&-tel.
23 The theory of 5,+a as equivalent to 08 is first found ;A:Aristotle, Metaph. I 9, 993 a 4,
where it is 21~0suggested that this letter corresponded to an ‘unknown’ sound
332 S.-T. Teodorsson 1 On Ancient Greek zeta
Nagy, G., 1970. Greek dialects and the transformation of an Indo-European process.
Cambridge, Mass. : Harvard Univ. Press.
Pisani, V., 1964. Zu griech. +a und zy zF. Glotta 42, 183-185.
Schwyzer, E., 1939. Griechische Grammatik I. (Handbuch der Altertumswissenschaft
II: I : 1.) Munchen: Beck.
Teodorsson, S.-T., 1974. The phonemic system of the Attic dialect 400-340 B c. (Studia
Graeca et Latina Gothoburgensia 32.) Giiteborg: Acta Universitatis Gothoburgensis.
Teodorsson, S.-T., 1977. The phonology of Ptolemaic Greek. (Studia Graeca et Latina
Gothoburgensia 36.) Goteborg: Acta Universitatis Gothoburgensis.
Teodorsson, S.-T., 1978. The phonology of Attic in the Hellenistic period. (Studia Graeca
et Latina Gothoburgensia 40) Giiteborg: Acta Universitatis Gothobxgensis.
Teodorsson, S.-T., 1979. Phonological variation in Classical Attic and th? development of
Koine. Glotta 57 (in press).
Witton, W.F., 1898. On -z%- and -Z-. The American Journal of Philology 19, 420-436.