Professional Documents
Culture Documents
A Linguistic Analysis of The Vase Inscriptions of Sophilos by SHANE HAWKINS, Ottawa
A Linguistic Analysis of The Vase Inscriptions of Sophilos by SHANE HAWKINS, Ottawa
!
Abstract: This study of the vase inscriptions of Sophilos, the Attic
black-figure painter active around 580 BCE, offers linguistic explana-
tions for several problematic spellings that have previously been mis-
understood as illiterate mistakes or products of dialect interference.
The examples in question are better characterized as non-standard,
quasi-phonetic spellings, some of which show precocious changes in
the historical phonology of the Attic dialect. An epigraphic addendum
follows the article.
1. Background
!
! A decade ago Martin Kilmer and Robert Develin examined
the vase inscriptions of Sophilos in order “to discover whether
Sophilos ought to be considered literate, and if so, in what sense
of that word” (2001: 9). The authors touched on a range of inter-
esting areas, including problems of orthography, phonology, and
etymology.1
___________
With thanks to Michael Weiss for comments on this paper, and to Marie-
Hélène Côté for helping me formulate part of the argument on !"#$%&!'.
1
Their contribution is said to be part of a larger project on archaic Attic
vase painters being prepared for access on the Internet. A reliable corpus of
Attic vase inscriptions has long been a keen desideratum for those interested.
See also the short notice by Kilmer/Desrochers 2001. Kretschmer 1894 was
unreliable in many places and is now long out of date. Immerwahr 1990 cata-
logs hundreds of Attic vase inscriptions, but it is incomplete (cf. Immerwahr
1967 [1971]). A large collection under the direction of Rudolf Wachter and
Henry Immerwahr is now underway and will also be available on-line:
http://avi.unibas.ch/home.html. Estimates of the number of Attic pots and
fragments with figured decorations run from eighty to one hundred thousand
(Keuls 1988: 222 = 1997: 300, Immerwahr ‘Prehistory of the Project’ at
http://avi.unibas.ch/prehistory.html), those with inscriptions about ten
thousand (Immerwahr 1971: 53, Wachter 2001: 1–4). Studies devoted solely
to vase inscriptions have tended to be epigraphic rather than linguistic
(Immerwahr 1964, 1990, Guarducci 1974: 456–495, Moret 1979; Kretschmer
1894 and Wachter 2001 are exceptions).
Glotta 88, 122–165, ISSN 0017–1298
© Vandenhoeck & Ruprecht GmbH & Co. KG, Göttingen 2012
A Linguistic Analysis of the Vase Inscriptions of Sophilos 123
2.1.1 &(%)*(+
Sophilos is the first Attic vase-painter whose name is
known.4 Beazley 1956: 37 wrote that it is “[i]mpossible to say
whether the name is ,!%)*(+, ,(%"*(+, or ,#%)**(+.” To these
three we may add ,!%)**(+, because the painter used a pre-
Euclidean alphabet and his orthographic practice was to write
geminate consonants singly.
___________
#
!The signature appears on four vases: Athens Acr. 587, ABV 39.15, Add 2
10; ‘Pharsalos dinos’, Athens 15499, ABV 39.16, Para 18, Add 2 10; ‘Erskine
dinos’, London 1971.11-1.1, ABV40.16bis, Para 19, 523, Add2 10–11;
perhaps also Athens National Museum 15918, ABV 40. 21.
4
Correctly Birchal 1972: 109, Moore/Philippides 1986: 79, Mertens
1988: 426, Schefold 1993: 184, de la Genière 1995: 35, Baurain-Rebillard
1999: 156. It is not true that Sophilos is the first Attic pottery painter known
to have signed his work (contra Kilmer/Develin 2001: 12), nor is he the
earliest vase-painter whose name is known (contra Brownlee 1988: 80, 1995:
363). Those distinctions belong for now, so far as I know, to fragmentary
signatures from a painter of Ischia and to the Athenian Analatos Painter, both
dated to ca. 700. The names of earlier non-Attic painters are also known
(Hurwitt 1985: 141).
A Linguistic Analysis of the Vase Inscriptions of Sophilos 125
2.1.2 !"#$%&!'
The second issue raised by the signature is the pronunciation
of the standard pre-Euclidean spelling %& in the word !"#$%&!'.
Kilmer/Develin take the spelling at face value and claim that
such a sequence would have been pronounced [phs], “with some
degree of separation and distinctness for the consonant sounds
represented by the aspirated consonant 4 combined with ,”
___________
8
On the naming practice see Solmsen 1922: 118 and Runes 1925.
A Linguistic Analysis of the Vase Inscriptions of Sophilos 127
Lejeune claims that Greek voiceless aspirates (%, <, 5) were also
lenis. 14 For Lejeune, the %& and 5& spellings were a graphic
solution to the problem of how to represent lenis consonants
before /s/. The signs % and 5 were chosen because even though
the sounds meant to be represented were unaspirated, % and 5
correctly indicated lenis voiceless labials. This was a better
choice, the reasoning goes, than : or /, which would have
correctly indicated the lack of aspiration but would have
incorrectly signaled fortis stops.15
This scenario is problematic. First, the distinctive use of
increased respiratory energy on segments is comparatively rare,
found in a few languages like Korean and (maybe) LuGanda
(Ladefoged/Maddieson 1996: 95). On the other hand, only a
very few languages are thought to have articulatory strength
differences that cannot be explained independently of voicing,
and when this is so the difference can often be attributed to
duration (Ladefoged/Maddieson 1996: 96–99). In other words,
Lejeune’s theory is not impossible, but the rarity of such a
system, for which there is no concrete evidence in Greek, makes
it very unlikely.
Two additional facts weigh against it. It is hard to imagine
that two lenis allophones would for any reason be readily
perceived or orthographically represented by ancient Greek
speakers, whose phonemic inventory shows no systematic
employment of lenis or fortis stops. Or again, we might wonder
___________
14
Lejeune 1972: 54. There is no way to know for certain, I think, whether
this is a correct assumption or not. Pace Lejeune and Threatte 1980: 20, 571,
it is not generally the case that aspirated stops are thought of as lenis; aspira-
tion is frequently coupled with fortis articulation in the literature (see biblio-
graphy in n. 13 and Clark/Yallop 1995: 51–52). In fact, aspiration cannot be
directly correlated to fortis or lenis articulation at all, since the fortis/lenis
distinction (as generally understood) is independent of aspiration (Catford
1977: 200).
15
Lejeune 1972: 69, 72. Devine 1974: 131 n. 45 takes %&, 5& and h& as
either spirant plus s or unaspirated stop plus s, “the aspirate being a possible
graph in this position of neutralization and chosen since the archiphoneme
representative was lenis (but not aspirated),” and adds (132) that “we must be
content with saying that, at least, there is no reason to deny the possibility of
an aspirated voiceless lenis.”
130 Shane Hawkins
why the letters 7 [ps] and 6 [ks] would later have been readily
adopted in Attica if Attic %& and 5& really represented anything
other than [ps] and [ks].16 In fact, Archinus, who put forward the
measure to reform the Old Attic alphabet in Eucleides’ archon-
ship of 403/2, and who appears to have been something of a
language theorist, seems to have argued explicitly that these
were the values of 7 and 6, and that this was one reason for the
reform.17
Finally, Clackson 2004 suggested the explanation lies in the
notion of Voice Onset Timing (VOT). In current phonetic theory
aspiration is generally conceived of as a matter of voice onset
timing, or the period of time between the release of an articula-
tory occlusion and the beginning of vocal fold vibration. In
simplified terms, three possible scenarios are distinguishable. If
voiced phonation begins before or with the occlusion made in
articulating a consonant, the stop is considered fully voiced. If
there is no voicing during the occlusion, but voicing is simulta-
neous with the release of the occlusion (or follows shortly there-
after), the stop is considered voiceless. Finally, if voicing is
delayed significantly after the release of the occlusion, the
voiceless stop will be aspirated because of the unobstructed flow
of air through the abducted vocal folds of the larynx. The timing
is a continuum.18 A number of languages, like Ancient Greek,
make a three-way distinction among voiced (voice onset at or
near the occlusion phase), voiceless unaspirated (voice onset at
or just after the release of the occlusion), and voiceless aspirated
stops (voice onset significantly later than the release of the
occlusion).
According to Clackson’s idea, in sequences of the type [ksa]
the unvoiced sibilant following the stop delayed the onset of!
___________
16
On the adoption and use of the Ionic alphabet see Threatte 1986: 26–51,
Immerwahr 1990: 179–182, and the intriguing study of D’Angour 1999.
17
Theophr. ap. Syr. ad Arist. Met. 1093a20, Comm. Arist.Gr., 6.1.191. 29 f.
18
This explanation follows closely Clark/Yallop 1995: 52–53. Voiced
aspirate plosives such as those found in South Asian languages are generally
made with ‘breathy voice’ and a slight delay in onset (Clark/Yallop 1995:
20–21).
A Linguistic Analysis of the Vase Inscriptions of Sophilos 131
3. Nasals
___________
40
For illustrative examples of the process see Ohala 1993: 250–1, who
discusses *penk"e (Gk. :.'0!, Vedic páñca) > (by assimilation in Proto-
Latin) *k"enk"e > Latin qu/nque. In Latin, both stops were formed with lip
rounding, which was probably also found on the vowel between these
segments. Listeners must have perceived the rounding on the initial stop as a
perturbation from the following vowel and ‘corrected’ it by eliminating the
rounding on the first stop, to judge from examples like Italian cinque
/t,i?kwe/, French cinq /s!6k/ and Spanish cinco /<i?ko/. See Katvitskaya 2002:
60–1 on the loss of -n- with compensatory lengthening in Old Latin.
Spellings like cosul suggest that n was converted into a nasalization on the
preceding vowel, and was subsequently restored in the classical period. Here,
too, hypercorrections like th0nsaurus for th0saurus were formed.
41
I note here Athens, Acr. 587 fr. i, ABV 39.15, Add 2 10, a small piece
that preserves three female figures, two left-facing women to the right and
one front-facing female playing the syrinx. They are labeled '.&$). This
concerned Beazley (1986: 17 = the 1951 edition, p. 18), who noted that one
expects Muses in this context (the gods arriving at the wedding of Peleus and
Thetis) and that the plural form of A4&$ is otherwise unattested. Vanderpool
1953: 322 suggested reading 8.&$) (= B(4&$) ‘Muses’) instead, with nu for
mu. This has not been widely accepted (Stewart 1983: 59, Beazley 1986: 17,
Immerwahr 1990: 186, Kilmer/Develin 2001: 16–17 all ignore or reject it).
Immerwahr, following West 1978: 374 and Henrichs 1987: 117, n. 41,
suggests the Nyssai are tree nymphs; however, Beazley noted that the '.&$)
“seem to correspond to the Muses who accompany the chariots on the
François vase and who sang at the wedding”, and Carpenter 1986: 9 pointed
out that the very same depiction of female figures is labeled 8(&$) on the
London dinos by Sophilos (London 1971.11-1.1). He puts this all down to the
idea that Sophilos was “a poor speller” and he appears to suggest that '.&$)
is a confusion of 8(&$) and '.%$) ('(8%$)). I prefer the explanation given by
Haslam 1991: 42, n. 23, that the confusion is not in the spelling, but in the
labeling. That is, the Muses are pictured just as they are on the London dinos,
but the inscription is an inaccurate label that reads ‘Nyssai’.
A Linguistic Analysis of the Vase Inscriptions of Sophilos 139
4.1. *1$%'&
!
The spelling $5)*!+ for the regular /5)**!(+ is noteworthy.
Kretschmer 1894: 191–192 (followed by Immerwahr 1990: 21,
n. 4; Kilmer/Develin 2001: 21, n. 34) argued that -!+ is not an
error but a substitution of the masculine first conjugation ending
in -)+ for the eu-stem nom. in -!(+. This is possible but there is
no clear explanation why this substitution was made. Further-
more, one cannot assume the formation was an isolated mistake
by Sophilos, because the existence of such -es-stems seem to be
present already in Mycenaean (Perpillou 1973: 238ff., Risch
1987, Hajnal 1995: 30–31) and there are Latin and Etruscan
forms from the fourth century that presume the same ancestral
form (e.g., Praenestine ACILES, CIL I2 564, 567, Etruscan aciles,
Latin Achill0s; see further Wachter 1987: 148–9).43
A better explanation recalls the development of the acc. sg.
*-e"m to *-0m > -1' via Stang’s Law (cf. acc. sg. C)' < *di7e"m).
The creation of such an accusative may have led to a nom. sg. -1+
on the analogy of the first declension masculines, viz., /5)**)' :
-)' :: X : -)+, where X = /5)**)+, in the so-called “Achaean”
dialects, where it is preserved as an archaism.44 Possibly, then, a
trace of this development is attested in a handful of Attic
examples with nom. sg. -!+ (i.e., -1+), such as $5)*!+. Rau
suggests that the Attic names in -!+ were influenced by a non-
Homeric epic tradition current in archaic Athens. This is a
possibility for most of the names, including $5)*!+, though some!
___________
42
‘Pharsalos dinos’, Athens 15499, ABV 39.16, Para 18, Add 2 10.
43
Compare Ulix0s for 89.&&!(+, which may have a Greek equivalent
Olixes (i.e., 8*)6.+, vel sim.), if its attribution to Ibycus at Diomedes ars
gramm. i (PMG 305) is reliable.
44
The analogy would have to be to the first declension and not to the
,-/#$01+ type, since the original acc. sg. there was ,-/#$01. Acc. sg.
,-/#$01' only became normal by the later fifth century (Threatte 1996: 138,
173–4). See further Risch 1987, Hajnal 1995: 30–31 (who suggest an analo-
gical proportion masc. acc. sg. -on : nom. sg. -os : : Acc. sg. -0n : nom. sg. X,
where X = -0s), Rau 2003: 12–13. On Arcadian 9!#)+ ‘priest’ see Buck 1955: 92.
140 Shane Hawkins
!
5. Inscriptions from the Erskine Dinos61
5.1. h$%'54*
started from the verbal root &*!.<- ‘to come’ and argued that the
form is a unreduplicated participle in -.)$ with recessive accent
of the type '".)$, $C<.)$, D#:.)$), E#".)$. The name means
something like ‘who comes (to help with the birth)’ or ‘who
makes the birth come.’ Heubeck also suggested the possibility
of a hypothetical *0*!(<!)' ‘to be free’ or ‘to make free.’ This
etymology is sound phonologically and semantically, and it also
allows for plausible explanations of the divergent forms attested
in our sources.
According to Threatte 1980: 342, there are no less than nine
different spellings in Attic: L+*E<.)$, 62 L+*!"<.)$, L+*"<.)$,
L+*(<!)$, ?*!"<.)$, hE*!"<.$, (h))*!"<.$ (to which perhaps
h)*"<.$ also belongs), F*"<.)$, F*(<!)$. Threatte does not
mention Sophilos’ h)*!<.$, but one assumes he might group it
with (h))*!"<.$. Non-Attic examples include L+*!"<.)$ (Il.
11.270), ?*!"<.)$ (Pindar O. 6.42), L+*)<.)$ (Callim. 53.1 Pf.),
?*!(<.)$ (Crete), ?*!.<"1 (Paros), ?*!.<"$ (Laconia), Myce-
naean e-re-u-ti-ja = ?*!.<"$ (Cnossus). According to Heubeck
the original form *eleuthu#a is preserved in the Cretan inscrip-
tions. The Mycenaean form is the result of dissimilation of eu -
u# with loss of u in the third syllable. A similar dissimilatory
process of !. - .) to !) - .) occurs in Pindar’s ?*!"<.)$, and most
other forms can be explained as the result of dissimilations,
assimilations, and metrical lengthenings (Homer’s L+*!"<.)$ ).
Sophilos’ h)*!<.$ is interesting in four respects. First, the
initial aspiration is unetymological. Threatte 1980: 342, 456
suggests that it is the result of assimilation from the following
theta, a common phenomenon and not limited to dipinti. The
aspiration may also be due in part to (a folk-etymological)
contamination with G18) ‘send forth’; cf. Il. 11.270, 9#)8(, 0# 0!
:#(M!1&) 8("(&0#/() L+*!"<.)$) ‘the bitterness that the hard!
___________
62
By E Threatte indicates graphic <!>, which represented 1) /e/, the
inherited short front mid-vowel, 2) /N%/, the long upper mid-vowel later
represented in Attic after the adoption of the Ionian alphabet by the ‘spurious
diphthong’ !), and 3) /O!!/, the long lower mid-vowel later written 1.
148 Shane Hawkins
5.2. 5'54&
Sophilos writes <!<.+ for the usual P1<(+, the name of the
sea goddess, consort of Oceanus (Il. 11.201 = 11.302). To my
knowledge, four different explanations of the name have been
put forward.
1) Since antiquity the name has been associated with 0)<1
‘grandmother.’ This association may be latent already in the
words of Hera, 810.#$ P1<(' (Il. 11.200–2 D 301–3). The
connection between P1<(+ and 0)<1 seems to have been made
by Cratinus (483 PCG), who attests the word :#(0)<H+ ‘born
before Tethys’, a comic name of an old woman, and Callima-
chus (194.52 Pf.), who uses the name P1<(+ as a type of very
old woman. Eustathius (978.51f.) also explained P1<(+ from
0)<1 (among other possibilities).
___________
63
A reader has also suggested the possibility of a folk-etymological
association with G*$8$), 9*$&/(8$); cf. Hom. imperative G*1<) and Att. !G*1<)
(Hsch.).
A Linguistic Analysis of the Vase Inscriptions of Sophilos 149
___________
68
Kilmer/Develin 2001: 35 appear to view Grassmann’s Law, inaccu-
rately in my opinion, as a process of aspiration throwback rather than aspirate
dissimilation.
69
For the post-Mycenaean date of Grassmann’s Law see Plath 1987. It
must be noted, however, that a post-Mycenaean date is not accepted by all
scholars; e.g., BartonJk 2003: 147–8.
70
Schwyzer I 219ff., 257, 261f., Buck 1955, § 65, Threatte I 455–457,
Lejeune 1972, §§ 45, 47, Miller 1977: 143–4.
71
Kilmer/Develin 2001: 30 n. 54 invoke the inscription on Berlin R115/2
by Peithinos, <!<)+, i.e., Thetis, in their discussion of <!<.+. The comparison,
however, is not apropos since <!<)+ is clearly an assimilation in aspiration of
the second dental to the first.
152 Shane Hawkins
5.3. 1$)"+
before 8 is the word ('8$0)(', but this has also been explained as
the result of assimilation (Lejeune 1972: 238). The most
straightforward conclusion to make is that in environments like
those exhibited in 5(&*)(), the raising was early and general, while
after * or before *, #, ' or 8 the raising was perhaps later and
then used only by some Attic speakers.
The sound change in question, therefore, is more or less
regular and phonetically natural; 5)#(' exhibits an early example
of the change )* > K. Other scholars have entertained the idea that
the sound change may have arisen in a specific social stratum.
According to Milne 1947: 227, “This tendency in pronouncing
such syllables seems to have been particularly strong in the
Athenian potters’ quarter of the sixth century B.C. Why this was
so we can only guess. There were many resident aliens in the
potters’ quarter, and the spelling we are discussing may reflect
non-Athenian habits of speech.” Ruijgh 1978: 87, on the other
hand, suggested that both 5(&*)() and ('8$0)(' were borrowed
through merchants from a rural dialect where, as in Boeotian,
the raising had already occurred. For Wackernagel 1909: 330, a
word like ('8$0)(' was “der lässigen Sprache des Hauses, dem
Munde von Frauen und Sklaven maßgebenden Einfluß zuzu-
schreiben; man beachte die deminutive Form.”
6. Conclusion
The old man in the painting, who is a seer (so also Bakır 1981:
75), is a “fixture” of such scenes according to Brownlee, and the
inscription will then belong to Amphiaraos’ son, Amphilochos.
4. Kilmer/Develin’s discussion of the Menidi louterion (ABV
42.36, Athens 2035.1, frag. 15918) is badly confused. On the
left side of the fragment are the remains of the heads of two
reined horses, one white and one black, before and under which
appears an open-jawed snake. Three lines of very damaged text
are written between these animals and the far right edge of the
fragment, where a woman’s foot and the hem of her garment can
still be seen. The inscription is mostly illegible, as it was
damaged before discovery and suffered further damage soon
thereafter.78
Wolters was the first to publish a careful study of this piece
from autopsy, and since his work seems to suffer neglect, it is
worthwhile to repeat here what he wrote. He read the inscription
as three lines to be read from bottom to top.
top R <! . ( . ) . ( . ) . !# . / . * … [
middle S ])*(+ 8! … &!'
bottom S ]!# …
Bibliography