The X-files
The letter X in the local archaic Greek scripts
p. 165-182
Résumés
The letter X stands for [ks] in the Greek “dark blue” scripts, and for /kh/ in the “red” scripts. Moreover, the sequence [ks] is usually spelt with the digraph ΧΣ in the “light blue” scripts, rather than with expected ΚΣ (ΧΣΕΝΟΣ). Several phonetic explanations have been proposed to account for this fact, but none of them proves satisfactory. Furthermore, the spelling ΧΣ is purportedly attested in some “red” scripts (Boeotian ΑΡΓΥΡΟΤΟΧΣΟΙ = ἀργυροτόξωι), in which the letter X or the digraph ▯Σ should be expected. To complicate matters further, a spelling ▯Σ, usually interpreted as a variant of “light blue” ΧΣ, occurs at Naxos (NA▯ΣIO). This paper tries to integrate all these seemingly disparate facts into a principled explanation. “Dark blue” Ξ, Naxian ▯Σ, and “red” Χ are probably all avatars of Phoenician samek, and the use of X for /kh/ in the “blue” scripts resulted from a misanalysis of pleonastic ΧΣ spellings.
La lettre X représente [ks] dans l’alphabet « bleu foncé », et /kh/ dans le « rouge ». En outre, dans l’alphabet « bleu clair », la séquence [ks] est généralement orthographiée avec le digraphe ΧΣ, plutôt qu’avec le digraphe ΚΣ attendu (ΧΣΕΝΟΣ). Plusieurs explications phonétiques ont été proposées pour rendre compte de ce fait, mais aucune ne s’avère satisfaisante. Ensuite, l’orthographe ΧΣ est prétendument attestée dans plusieurs variétés de l’alphabet « rouge » (béotien ΑΡΓΥΡΟΤΟΧΣΟΙ), dans lesquelles la lettre X ou le digraphe ▯Σ seraient attendus. Pour compliquer davantage les choses, une orthographe ▯Σ, généralement interprétée comme une variante « bleu clair » de ΧΣ, est attestée à Naxos (NA▯ΣIO). Cet article tente d’intégrer tous ces faits apparemment disparates dans une explication globale. Le Ξ « bleu foncé », le ▯Σ naxien et le Χ « rouge » seraient tous des avatars du samekh phénicien, et l’emploi de X pour /kh/ dans les alphabets « bleus » serait dû à une réinterprétation erronée des graphies pléonastiques ΧΣ.
Texte intégral
“Il est doux de croire, même à l’enfer” (Anatole France)
1. Introduction
1This paper delves into the unclear origin of the cross-shaped or, more commonly, X‑shaped sign used with different sound values in the archaic Greek local alphabets. In this connection I shall also deal with the much-discussed spellings ΦΣ, ΧΣ for /ps, ks/ found in several regions.1
2Before proceeding into the subject matter, a preliminary caveat is in order: due to the scarcity and elusiveness of the evidence available, many of the issues addressed here will necessarily remain in the realm of the hypothetical and of the speculative. Hopefully, my speculation will be at least plausible and coherent.
3I begin by summarily presenting the relevant data, which are otherwise well-known. In the third edition of his pioneering book on the topic, Kirchhoff 1877 classified the ancient Greek alphabetic scripts into four types, which were displayed on a colour-coded map. A. Kirchhoff based his taxonomy on the presence or absence and on the shape and sound value of the so‑called “supplementals”, the letters added at the end of the alphabet and used to denote the labial and velar aspirated stops (/ph/, /kh/) and the clusters /ps/, /ks/.2
4The green alphabets are the most primitive in lacking all supplementals. The letters Π and Κ (or its allograph Ϙ) indifferently denote plain /p/, /k/ and aspirate /ph/, /kh/.3 The sequences /ps/, /ks/ are spelled ΠΣ and ΚΣ (or ϘΣ).4 Since the green scripts are of little consequence for the issue addressed here, they will be largely ignored in what follows.
5At the other end of the supplementals scale, the dark blue alphabet is the only to include four distinct graphemes for the aspirates and stop + sibilant sequences: Φ, Χ, Ψ, Ξ5 for /ph/, /kh/, /ps/, /ks/ respectively.6
6More central to our issue are the light blue and the red alphabets, which occupy an intermediate position. Unlike the green scripts, both have special letters for /ph/and /kh/: Φ, Χ in the light blue scripts and Φ, Ψ in the red ones. However, unlike the dark blue alphabet, the light blue has no special signs for the clusters /ps/, /ks/, which are spelled ΠΣ, ΚΣ and, much more frequently, ΦΣ, ΧΣ.
7In the red scripts the sequence /ps/ is usually spelled ΠΣ or ΦΣ,7 while /ks/ is spelled Χ (so‑called red xi).8
8Tab. 1 summarizes the relevant spellings in the archaic scripts:
Tab. 1 – Supplementals in the ancient Greek local scripts.
Alphabet | /ph/ | /kh/ | /ps/ | /ks/ |
Green | ||||
Light blue | ||||
Dark blue | ||||
Red |
9Worthy of note are two curious facts that, at first sight, hardly bear any relation to each another. First, while Φ has one and the same sound value in the blue and red alphabets, the signs X and Ψ are assigned disparate values.
10Second, the digraphs ΦΣ in the light blue and red scripts and ΧΣ /ks/ in the light blue scripts include aspirate letters to denote the sequences [ps], [ks].
11In what follows I shall attempt to answer two questions, which, as we shall later see, happen to be closely interrelated.
- How can one explain the unexpected aspirate in the spelling ΧΣ for /ks/?
- Were blue chi and red xi two accidentally identical signs with different sound values from the very beginning as is commonly believed, or were they one and the same symbol (X) which eventually split into two letters with different sound values as I will try to demonstrate?
2. The spellings ΦΣ and ΧΣ for /ps/, /ks/: previous explanations
2.1. Aspiration
12A few scholars take the spellings ΦΣ, ΧΣ at face value: allegedly, Φ, Χ represent real aspirate stops /ph/, /kh/.9 This explanation is straightforward, but collides with the testimony of the ancient grammarians.10 Moreover, there is incontrovertible evidence that /ph th kh/ lost their aspiration before /s/ and, as a consequence, the application of Grassmann’s law was blocked. This accounts for the contrast between e.g. PIE *seǵh‑ > pres. *[hékhɔː] > ἔχω [ékhɔː] and nom. pl. *[thríkhes] > τρίχες [tríkhes] vs. fut. ἕξω [héksɔ] and nom. sg. θρίξ [thríks] (not †ἔξω [ékhsɔː], †τρίξ [tríkhs]).
13W. Kastner and A.L. Sihler compare the spellings ΦΣ, ΧΣ to the aspirates of *eks‑tos > Locr., Epid. ἐχθός and *teksnā > τέχνη, which allegedly went through intermediate stages *ekhstos, *tekhsnā.11 Such a connection is, however, objectionable since the aspirates in ἐχθός and τέχνη patently result from the deoralisation of /s/: *eks‑tos > *ekhthos, *teksnā > *tekhnā.
14E. Schwyzer notes that ξ, ψ in Greek loanwords were adapted in Armenian as /khs/, /phs/.12 W.S. Allen waters down the claim and considers unlikely that full aspiration was involved.13 To all appearances, aspiration is an internal development specific to Armenian, which, whatever its cause, should not be projected back into the donor language.14
2.2. Articulatory weakening
15A more popular option is a hypothetical articulatory weakening of the plosive in /ps/ and /ks/. Under this premise, Φ and Χ are supposed to represent either a fricative of some sort or a “lenis” stop (French occlusive douce) intermediate between voiceless /p, k/ and fully voiced /b, g/.15
16The strongest case for a weakening of the stops in /ps, ks/ clusters has been made by Gomis García, Striano Corrochano 2017. These authors compare the hypothetical weakening in Greek to the pronunciation of /ps, ks /as /s/ in informal registers of Spanish (e.g. examen [e'samen]),16 even if they ultimately shy away from identifying specifically the actual pronunciation of the deviant spellings that allegedly reflect the sporadic weakening.
17Besides the spellings ΦΣ, ΧΣ (on Naxian Σ, see below), they marshal the following pieces of evidence:
- the doubling of blue and red xi (ΞΞ, ΧΧ) attested across all the Greek-speaking world in all periods: e.g. WLocr. δόξξαι, ἀξξιόμαχος, ἐξξόλειαν (IG IX, 12, no. 609, l. 7, 9 and 15, Naupaktos, ca 500?); Thess. ἐξξανακάδ(δ)ɛ̄ν, IG IX, 2, no. 257, l. 8‑9, Thetonion, ca 450‑425?); Att. Δόξξα (IG II2, no. 8160, l. 1, 1st‑2nd c. AD);
- some pleonastic spellings with double blue chi or, more frequently, with xi + superfluous sigma: e.g. Καλλίχχσενος for Καλλίχσενος (Agora XXV, no. 540, ca 485); ϙύλιξς (LSAG2, p. 461 J, pl. 77, no. 8, Selinous, ca 600); ἔξς (I.Ephesos, no. 1, A, l. 3; Ephesos, ca 550);
- ΣΦ, ΣΧ (also ΣΚ) for ΦΣ, ΧΣ, which are relatively frequent in Attic ostraca and dipinti:17 ἔγρασφεν for ἔγραφσεν on eighteen vases signed by the painter Epiktetos (active ca 520‑490); Καλλίσχενος or Καλίσχενος for Καλλίχσενος on seven ostraca dated ca 485 (Agora XXV, no. 319, 340, 411, 455, 487, 537 and 578, ca 480); Ἀνασχίλεō for Ἀναχσίλεō (O.Kerameikos, no. 278, 301, 311); Σχανθίας for Χσανθίας (O.Kerameikos, no. 8600 and 8601); also, Καλλίσκενος (Agora XXV, no. 328);18
- sporadic Σ or ΣΣ for /ks/:19 Καλλισένōι (Agora XXV, no. 440, ca 485); Κ{λ}αλίσενος (Agora XXV, no. 464, ca 485); Φιλοσσένο<υ> (Agora XXV, no. 18, 490‑480); Καλλίσσεν{ν}ος (Agora XXV, no. 439, 490‑480); Καλλίσενος (O.Kerameikos, no. 1102); Σανθιππō (O.Kerameikos, no. 83);
- a few compound personal names with a first element ending in oσι‑ instead of expected oξι‑ in Cypriot syllabic inscriptions: e.g., pa‑ra‑si‑po /Prās(s)ippō/) (Kouklia-Paphos, no. 35, 550‑500) and ta‑si‑o‑ro‑wo /Tas(s)iorwō/ (Kouklia-Paphos, no. 226, 550‑500), which V. Gomis García and A. Striano Corrochano compare to names like Πράξιππος and Ταξίλαος;
- the preposition e‑se /es/ attested at Kafizin in two inscriptions of the late 3rd c. (Egetmeyer 2010, p. 177);20 Lejeune 1972, p. 73 cites the case of ξύν > σύν;
- the name Κόκκυφς (I.Thespies, no. 485, A, l. 12, late 5th c.), interpreted as a misspelling for Κόκκυχς (κόκκυξ ‘cuckoo’), which allegedly attests to the weakening and subsequent merger of /ps/ and /ks/.
18Not all these pieces of evidence are of equal weight and value. Some are manifestly uncompelling or irrelevant. Thus, the pleonastic spellings ΞΞ, ΞΣ (ΧΧ, ΧΣ in the red scripts) undoubtedly have more to do with the bisegmental value of xi, a “double” letter, than with the pronunciation of /ks/ itself.21
19No more relevant are the occurrences of ΣΧ for ΧΣ, which are likely to be graphic mistakes of little significance for the pronunciation of the cluster.22 Incidentally, V. Gomis García and A. Striano Corrochano fail to mention the rare occurrences of a spelling Χ for ΧΣ (e.g. Καλίχ<σ>εν<ος>, Agora XXV, no. 542, ca 485; Πρόχ<σ>ενος, O.Kerameikos, no. 8599b), which, under their premises, should be indicative of an otherwise unlikely loss of /s/. An accidental omission of Σ is certainly a more convincing explanation. Misspellings of all sorts abound in “casual” texts like ostraca, graffiti, and vase inscriptions which were executed in a rush and/or with little care by non‑professionals with a limited command of writing.23
20By the same token, careless omission of X can by no means be excluded for Σ for ΧΣ (the type Καλλίσενος), even if in this case an assimilation /ks/ > /ss/ cannot be excluded (see further below) since the writing of single letters for geminate consonants is common practice in archaic inscriptions.
21The equation Κόκκυφς = κόκκυξ allegedly suggestive of a weakening of the stop in word-final /ps/ and /ks/ is not without problems. As far as I know, a personal name *Kόκκυξ fails to be attested. Albeit poorly documented, a few names with an “expressive” ending ‑υψ are recorded, especially in central and northern Greece: Κῖνυψ (Wolters, Bruns 1940, no. 200, Thebes, Kabeirion, ca 500); [Κ]όρυψ (Wolters, Bruns 1940, no. 147, Thebes, Kabeirion, 450‑400);24 Θάρυψ, name of a Molossian (IG IV2, 1, no. 95, Epidauros, ca 365‑311), Γόργυψ (gen. Γόργυπος, IG XII, 9, no. 248, B, l. 7, Eretria, early 3rd c.; also name of a Molossian king, Th., II, 80, 6).25 Even more to the point, Boeot. Κοκκου̣β̣ί̣α̣[ς(?)] (I.Thespies, no. 85, l. 10, Thespiae, 3rd c.) is likely to be a derivate of Κόκκυψ, which, in turn, looks like a hypocoristic built on the compound κοκκυβόας (ms. κοκκοβόας) [S., frg. 791 Radt, Kannicht].26 Alternatively, Bechtel 1921, p. 262, might be right in surmising that *κόκκυψ was actually the Boeotian name of the cuckoo with an alternative suffix.27 Be that as it may, the suffix variation cannot be accepted as evidence for an articulatory weakening.
22Finally, as regards the Cypriot compound names with an unexpected first element in oσι‑, V. Gomis García and A. Striano Corrochano aptly note that pa‑ra‑si‑ could also stand for /Phrasi‑/ (cf. Φρασίδημος).28 If real, the names in ta‑si‑ (/Tasi‑/ rather than /Tas(s)i‑/) for expected **ta‑ka‑si (Ταξι‑) may well have been attracted to the larger class of V1 compounds with a first element oσι‑ (e.g. Δαμασι‑, Παυσι‑, Σπευσι‑, etc.), rather than a phonological change /ksi/ > /(s)i/. For a similar fluctuation cf. Ἁρμοξίδαμος (Petelia, 5th c.), Ἁρμοσίλας (Rhodes, 3rd c.).29
23Altogether, the only cogent proof of an assimilation of /ks/ to /ss/ are the spellings ΣΣ for ΧΣ, but the occurrences are far too scanty to offer real proof of widespread assimilation. In fact, the very idea of an across‑the‑board weakening of the stop in /ps, ks/ sequences that might account for the spellings ΦΣ, ΧΣ, must be rejected as contradictory with what we know of Greek phonology.
24To begin with, the hypothetical “softening” of stops is suspect. We know for certain that, by the general rule of voice assimilation, both aspirate /ph, kh/ and voiced /b, g/ became voiceless /p, k/ when followed by any voiceless obstruent (/p, t, k, s/): cf. abovementioned fut. ἕξω, nom. sg. θρίξ and fut. ἄξω (pres. ἄγω), nom. sg. αἶξ (nom. pl. αἶγες). But there is no reason whatsoever for a voiceless stop to become something in between voiceless and voiced stops before a sibilant, i.e. /p(h)s/, /k(h)s/ > [b̥s], [g̥s]. Granting for the sake of the argument that such a partial voicing is a plausible phonetic process, should one not expect the spellings B and Γ to represent those “softened” stops at least occasionally?
25M. Lejeune’s idea that aspirate /ph th kh/ were lax (lenes) vis‑à‑vis strong (fortes) /p t k/ is a leap in the dark. Admittedly, all other things being equal, aspirates are more exposed to spirantisation than plain stops, but in languages with allophonic aspiration (e.g. English), aspirated allophones tend to occur in strong positions: word and syllable initial position, geminate stops.
26Conjecturing a fricative does not make things any better. To begin with, this should presuppose that /ph th kh/ had already become fricatives in all positions or, to put it another way, that the letters Φ, Θ, Χ represented fricatives like in Modern Greek. While the spirantisation of former aspirates may have been under way in more than one region, such a circumstance is unlikely for the great majority of dialects.
27Be that as it may, a hypothetical change /ps, ks/ > /fs, xs/ is contradicted by the historical evolution of Greek. It is true that syllable-final voiceless stops have become fricatives before stops in all modern varieties. This is, however, not the case for the sequences /ps, ks/, which have resisted change30 with the sole – and negligible – exception of the vernaculars spoken to this day in the Greek pockets of Southern Italy.31
28The reflexes in Modern Greek of the clusters at issue are conveniently summarised in tab. 2:
Tab. 2 – Classical Greek /pt, kt/ vs. /ps, ks/ in Modern Greek.
ClGk /pt kt/ | MnGk /ft xt/ | ClGk /ps ks/ | MnGk /ps ks/ | |
πτάρνυμαι | φταρνύζομαι | ψαύω | ψάχνω | |
ἑπτά | εφτά | ἔκοψα | έκοψα | |
ὑψηλός | ψηλός | |||
κτίζω | χτίζω | ξένος | ξένος | |
ὀκτώ | οχτώ | ἐδίωξα | έδιωξα | |
ἕξ | έξι |
29Admittedly, /ks/ underwent reduction to /s/ in two prepositions: ἐξ became ἐς before consonants in Thessalian, Boeotian, Arcadian, Cretan, and Cypriot (e‑se‑ /es/), and ἐσς before vowels in Boeotian;32 ξύν eventually became σύν in all dialects. But these are to be explained as instances of the irregular phonetic reduction characteristic of high-frequency function words,33 and not as evidence of an extensive assimilation of /ks/ sequences.
30Even more significantly, secondary /f θ x/ have become /p t k/ in Modern Greek fricative + /s/ clusters, as in the examples in tab. 3:
Tab. 3 – Secondary fricative + /s/ clusters in Modern Greek.
Secondary /fs θs xs/ | MnGk /ps ts ks/ |
ἐδούλευσα | δούλεψα |
ἔκαυσα | έκαψα |
κάθισε | κάτσε |
μπαχτσές (< Turk. Bahçe ‘garden’) | μπαξές |
31It has also been pointed out that Greek loanwords in Latin have /ps/ and not /fs/: e.g. clepsydra.34
32According to a widely held view, the most compelling piece of evidence for articulatory weaking in stop + /s/ clusters is provided by a well-known epigram of three hexameters engraved boustrophedon on a female statue dedicated to Artemis at Delos by a woman from Naxos named Nicandre (I.Délos, no. 2; LSAG, p. 303, no. 2, plate 55; Hansen, CEG, no. 430, ca 650).35
33This inscription famously exhibits two special letters similar in shape. One is a bisected rectangle, so‑called “boxed” or “closed” heta (), a primitive shape closest to Phoenician ḥêt (Jeffery’s η1).36 This sign serves a double function. In six words,
corresponds to the outcome of an earlier *ā: Νικάνδρ
(l. 1), ἰοχεαίρ
ι (l. 1), ϙṓρ
(l. 1), Δεινοδίκ
ο (l. 1‑2), ἀλ(λ)
ōν (l. 2), κασιγνε̄́τ
(l. 2). It must represent an intermediate stage in the raising of *ā to /εː/, i.e. something like /æː/ (the vowel of E. sad).37 This /æː/ had not yet merged with /εː/ from ancient *ē, which is spelled E: cf. ἀνέθε̄κεν (l. 1), κασιγνε̄́τ
(l. 2).38
34On the other hand, consistent with the original value of heta, the sign also represents aspiration in two words: in
κ
βόλōι (i.e. h(ε)κηβόλōι) (l. 1), the first
stands for /he/;39 in Φ
ρά
σō, pleonastic
redundantly represents the aspiration of /ph/.
35The second special letter is and empty rectangle () that occurs only before /s/: Να
σίō (l. 2), ἔ
σοχος (l. 2), Φ
ρά
σō (l. 3).
36An identical distribution of ,
, and E occurs in other Naxian texts: Εὐ̣θ̣υκαρτίδ
ς (l. 1),
o (l. 2)40 vs. Νά
σιος (l. 3) vs. ἀ{⁝}νέθε̄κε (l. 2), ποιε̄́σας (l. 3‑4) (I.Délos, no. 1, Delos, 620‑600); Ἀ̣λε
σιτίδ
ς41 vs. ἀν[έ]θ̣ε̣̄́σ̣α̣ν̣ (SEG XVI, no. 477, Naxos, 600‑575?); Ν
σστια[‑‑] (l. 1),
ποίσεν (l. 1), [‑‑]λ̣
ος (l. 3) vs. ἄνα
[ς?]42 (l. 2) [SEG XLVII, no. 1220, Delos, ca 600].
37In line with the idea of an articulatory weakening, many scholars interpret as a variant of “boxed” (h)eta (Jeffery’s η2) representing a fricative allophone of /k/, either [x] or [h]: cf. e.g. “[this spelling] implies therefore that the Naxian ξ was not pronounced as ks, but as something like hs” (LSAG2, p. 291).
38Tab. 4 reflects the supposed values of the relevant spellings in the Nikandre inscription according to Hypothesis A:
Tab. 4 – Spellings in two Naxian inscriptions (Hypothesis A: articulatory weakening).
Letter | Presumed value | |
E | /εː/ | ἀνέθε̄κεν |
/æː/ /h(e)/ | Νικάνδρ | |
/h/ | Φhρά | |
X | /kh/ | ἰοχεαίρηι, ἄλοχος |
39Supposedly, a strong argument in favour of this interpretation is the grave stele of Alexo from Amorgos (IG XII, 7, no. 142, 525‑500?), in which an open (h)eta (H = Jeffery’s η4) allegedly represents both the reflex of an ancient *ā and the first consonant of /ks/: μνHμα (l. 1), ἔστHσεν (l. 2), ἀδελφHι (l. 2‑3), and Ἀλε̣Hσοῖ (l. 3). Yet, the argument is not as trenchant as it might as first seem.
40On the one hand, the interchanges between some allographs of (h)eta and dark blue xi similar in shape are not unheard‑of. Thus, R.D. Woodard notes that the windowpane xi occurs in the (h)eta‑position43 and, in turn, the closed heta occurs in the xi‑position in the controversial copper plates from Fayum (ca 800) as well as in some Etruscan abecedaria (7th c.).44 These interchanges attest to the confluence of the xi‑ and (h)eta‑symbols.
41On the other hand, and more importantly, the reading Ἀλε̣Hσοῖ itself raises serious doubts.45 F. Bechtel’s reading Ἀλξ̣σοῖ with a sidelong xi, i.e. rotated ninety degrees (Jeffery’s ξ9) and a pleonastic sigma is to be preferred.46
42The idea that /ks/ did not undergo any articulatory weakening in Naxos seems to be confirmed by the spelling ΧΣ in the signature of the sculptor Alxenor from Naxos on a marble grave-stele found in Boeotia (IG VII, no. 3225; Hansen, CEG, no. 150, Orchomenos, ca 500‑475?): Ἀλχσήνōρ̣, Νάχσιος.47
43To my mind, Hypothesis A makes little sense. We have already seen that a weakening of /k/ before /s/ is improbable, but still worse, the supposed orthographical system lacks any internal logic. It seems unlikely that one and the same variant of (h)eta (η1) is used for two entirely dissimilar sounds (/æː/ and /h(e)/), while two different allographs of the same letter (allegedly, η1 and η2) are used to represent two purportedly similar if not identical fricatives in different contexts.48
44A much more reasonable hypothesis (Hypothesis B) has been embraced by several scholars.49 The sign must not be considered a variant of heta, but one of xi. This implies that
Σ is equivalent to ΞΣ with a pleonastic sigma (ἔ
σοχος = ἔξσοχος). A redundant heta has already been noted in Φhράξσō.50
45Tab. 5 reflects the distribution of and
according to this view:
Tab. 5 – Some spellings in two Naxian inscriptions (Hypothesis B: dark blue xi with redundant sigma).
Letter | Presumed value | |
E | /εː/ | ἀνέθε̄κεν |
/æː/ /h(e)/ | Νικάνδρ | |
/ks/ | ἔ | |
X | /kh/ | ἰοχεαίρηι, ἄλοχος |
2.3. More recent proposals
46New approaches have been attempted in recent years to overcome the shortcomings of previous explanations. They all start from the initial premise that the spellings ΦΣ, ΧΣ are phonetically significant and were devised to represent some subtle allophonic differences of one sort or another.
2.3.1. Voice-onset delay
47J. Clackson hypothesised that “the graphemes Φ and X were used to represent not aspirated stops as traditionally understood but stops with voicing onset delay”.51 Purportedly, when followed by the voiceless fricative /s/, voicing was delayed until the onset of the following vowel. He buttressed this theory with the evidence of the Styra lead tablets (IG XII, no. 9, l. 56, mid‑5th c.), in which /ps/ is spelled ΦΣ word-internally, but ΠΣ in word-final position, a position in which the hypothetical voicing delay is impossible: cf. Μοφσίδε̄ς (no. 285), Ὀφσ[- - -] (no. 319) vs. Xάροπς = Xάροψ (no. 359 and 435).
48J. Clackson’s explanation has not met with much acceptance for various reasons.52 The concept of voicing delay itself is unclear and lacks a convincing phonetic motivation. Why voice-onset delay was not perceived and reproduced in writing other clusters (e.g. /pt, kt/) remains an unanswered question. The hard‑and‑fast orthographic rule of the Styra tablets with ΦΣ and ΠΣ in complementary distribution is in all probability an illusion, the evidence being too scanty (two occurrences in each case) to draw any reliable conclusions. At any rate, word-final /ps/, /ks/ are regularly spelled ΦΣ and ΧΣ in Attic inscriptions: e.g., κɛ̃ρυχς (IG I3, no. 776, l. 5, 500‑480); Ἀστυάναχς (IG I3, no. 1162, l. 87, ca 447); γρύφς, ὄνυχς (IG I3, no. 776, l. 78 and 85, 419/418). Finally, as S. Hawkins elegantly put it, “it is doubtful whether such a sophisticated apparatus as that assumed under the rubric of Voice Onset Timing was ever recognized by the Greeks, or that even if it were somehow perceived by them in a fuzzy way, they would have incorporated it so conscientiously into the orthography”.53 Actually, speakers of any language are largely “deaf” to context-dependent noncontrastive phonetic detail, and, unlike professional phonologists, they have no real interest in representing such futile trifles in writing. This criticism also applies to most of the hypotheses reviewed in the preceding sections.
2.3.2. Turbulent airflow, ambient phonetic noise
49Dissatisfied with previous hypotheses, S. Hawkins advanced his own explanation that Φ and Χ were an attempt to represent orthographically the turbulent airflow characteristic of sibilants on the stops of /ps, ks/ clusters which, due to articulatory overlap, were perceived as single articulatory events.54 In a similar vein, but, as they make clear, independently from S. Hawkins, J.F. Eska and J.‑F. Mondon concluded that the highly salient turbulence of /s/ was perceived as concurrent aspiration on the stop in /ps, ks/ sequences.55 This perception allegedly explains the use of the “quasi-phonetic” spellings ΦΣ, ΧΣ in the light blue alphabets and ΦΣ (and exceptionally ΨΣ for prevalent X) in the red ones. J.F. Eska and J.‑F. Mondon also maintained that the contrast between word-final ΠΣ vs. word-internal ΦΣ in the Styra tablets reflects “the attenuation of aerodynamic energy typically found in coda position”.56
50In two previous articles, I dismissed S. Hawkin’s explanation as far‑fetched and unrealistic.57 There was no obvious reason for Greek speakers to represent in orthography the noise typical of /s/ since the letter sigma (or, for that matter, san) was more than sufficient to that effect. As stated above, speakers largely ignore allophones, and accordingly, rarely feel the need to represent them in writing.
51In some previous papers, I have tried to provide alternative explanations for some spellings which are generally believed to represent allophones in Greek. Thus, according to a widespread view,58 the letter Ͱ indifferently represents in a handful of inscriptions from Thespiae the long /eː/ resulting from the contraction of the diphthong /ei̯/ (cf. ΤͰσιμένε̄ς, ἈντιγενͰδας for Τεισιμένης, Ἀντιγενείδας, I.Thespies, no. 485, H2, H7, 424?) and an allophone of short /e/ markedly close in prevocalic position: cf. ἈρξικλͰε̄ς, ΚλͰεγένε̄ς purportedly for Ἀρξικλέης, Κλεεγένε̄ς (I.Thespies, no. 485, A1, I1); [H?]ερακλͰος (SEG XXX, no. 541; cf. I.Thespies, no. 276bis, Votonosi, Epirus < Thespiae, 475‑450). However, since all the apparent instances of Ͱ for short /e/ happen to occur in compound names with a first element ΚλͰε‑ or a second ‑κλͰε̄ς (gen. ‑κλͰος), and since the inflection of compounds in ‑κλῆς demonstrably had a prevocalic long vowel in Boeotian,59 the odds are that Ͱ stands for a long vowel in all cases. Therefore, the sign Ͱ is not to be understood as a doublet of epsilon,60 but as a reduced variant of heta known as “half‑H” (Jeffery’s η7), which, much like eta in Ionia and in other regions, was secondarily put into service for representing a long mid front vowel.61
52By the same token, contrary to common belief, the letter qoppa found in archaic inscriptions is unlikely to represent a backed or a rounded allophone of /k/. The distribution of kappa, used before A, Ε, Η, Ι even if a consonant (usually a liquid) intervenes, and qoppa used before Ο, Ω, Υ, under similar conditions,62 is an orthographic rule which reveals more of the phonology of the Phoenicians who taught some Greeks to read and write, than of the phonology of Greek.63
53Uncontroversially, an efficient orthographical system must conform as much as possible to the “phonemic principle”, whose ideal is a biunivocal correspondence between graphemes and phonemes: one grapheme for every phoneme and one phoneme for every grapheme.
54In W.S. Allen’s words, “the ‘phonemic principle’ is an economic principle, reducing redundancy and employing the minimal number of symbols that is consistent with the unambiguous representation of speech”.64 But economy per se is not the rationale behind the phonemic principle. As stated above, speakers are largely indifferent to predictable context-dependent allophones (phonological deafness). For instance, the Spanish (wo)man in the street is largely unaware that the pronunciation of the letters b, d, g as approximants crucially depends on the context: cf. dado ['dað̞o] ‘dice’, el dado [el'dað̞o] ‘the dice’, un dado ['un 'dað̞o] ‘a dice’, forma de dado ['forma ðe'ð̞að̞o] ‘dice shape’. In all cases (s)he only recognises a unitary phoneme /d/.65 For similar reasons, while (s)he naturally pronounces [v] as an allophone of /f/ in afgano [av'ɣ̞ano] ‘Afghan’, (s)he may experience considerable difficulty in reproducing the /v/ of foreign languages in words like e.g. E. vine, Fr. vin, G. Wein, It. vino, etc.
55J.F. Eska and J.‑F. Mondon express their disagreement with my criticism of S. Hawkins’ explanation,66 but they misrepresent my views in one crucial point. I have never denied the existence of allophonic or quasi-phonetic spellings as they call them. This notwithstanding, spellings of this type are extremely rare, and, in the great majority of cases, they are restricted either to easily perceivable extrinsic allophones or quasi-phonemes, or, more frequently, to allophones that happen to be identical to another phoneme. The spelling of preconsonantal nasals in Greek which I have discussed in two previous articles, is a case in point.67 Allophonic, i.e. context-dependent [m] in e.g. ἀμφί or ἐμβαίνω was easily identified with phonemic /m/ in ἅμαξα or μέμνημαι. The spelling γ (agma) representing merely allophonic [ŋ] in e.g. ἄγχος must have been modelled on that of ἄγνυμι or πέπραγμαι (probably pronounced [áŋnyːmi] [pépraːŋmai̯] with anticipatory assimilation to the following nasal). Similarly, the speakers of Castilian Spanish may misspell juzgar [xuð'ɣaɾ] ‘to judge’ as judgar,68 or verdad ‘truth’ [beɾ'ðað] or [beɾ'ðaθ] as verdaz and conversely veraz [be'ɾaθ] ‘truthful’ as hypercorrect verad. All these misspellings crucially occur in contexts where /d/ (spelled <d>) and /θ/ (spelled <z>) regularly merge.
56J.F. Eska and J.‑F. Mondon themselves provide a nice example of an allophonic spelling in Gaulish inscriptions, in which the Greek blue chi (Χ) represents a fricative allophone of /k/ in /kt, ks/ clusters, both in Gallo-Greek and Gallo-Latin texts (Lejeune, RIG 1, p. 445; Lejeune, RIG 2/1, pp. 6‑7): cf. Gaul. Ανεχτλοιαττηος (RIG 1, G‑268, Alise-Sainte-Reine, ca 40‑70 AD), duxtir (RIG 2/2, L‑98.1a.11, L’Hospitalet‑du‑Larzac, 1st c. AD), brixtía (RIG 2/2, L‑100.3, Chamalières, 1st c. AD). This could be a case of auditory (i.e., phonological) hyperesthesia on the part of the Greeks who taught the Gauls how to write their own language. They may well have perceived the Gaulish allophone as a true phoneme analogous to /kh/ or possibly /x/ in their native phonological system.69 Conversely, in the Lugano script, one of the northern variants of the Etruscan script, itself an adaptation of the red Greek alphabet, chi (Ψ) is never used to denote the velar fricative in the cluster [xt]. Pace J.F. Eska and J.‑F. Mondon, Cisalpine Gaulish teuoχtoni[o]n rendering /deːwogdonion/ (RIG 2/1, E‑2.11‑12, Vercelli, ca 89‑49 BC)70 is hardly a case in point since chi is occasionally used for /g/ (and possibly also for /k/) in all kinds of contexts: cf. seχeθu (on coins, from *seg‑ ‘victory’). After an initial phase of experimentation, the distinction between voiced and voiceless stops was given up in spelling.71
57Be that as it may, the indubitable existence of such allophonic spellings hardly substantiates J.F. Eska and J.‑F. Mondon’s claim that the aspirates of “quasi-phonetic” spellings ΦΣ, ΧΣ were meant to represent “ambiance phonetic noise”. To my mind, the scenario imagined by S. Hawkins and J.F. Eska and J.‑F. Mondon presupposes a willingness and a degree of sophisticated expertise in writing and in phonetics on the part of the users of the Greek alphabet that are utterly unrealistic.
2.4. A non-phonetic approach
58Notwithstanding their specific differences of interpretation, all the hypotheses examined in the previous sections share in the view that, to a greater or lesser extent, the spellings ΦΣ, ΧΣ and Naxian Σ were consciously designed to convey relevant phonetic information.
59My explanation of these spellings starts from the opposite premise, i.e. that the “aspirates” were a graphic convention with no phonetic significance.72 Phonology came into play only at a later stage to perform a secondary role, if at all. Moreover, contrary to most of the defenders of previous explanations, I assume that the identity in shape of blue chi () and red xi (
) is not accidental. Both were originally one and the same letter and, together with the Naxian xi (
), were relatives of the blue xi (Ξ).
60As regards this second assumption, there have been earlier attempts to establish a common source for the X‑letters. Thus, according to several scholars, the red xi (X = [ks]) arose as an abbreviation of the spelling ΧΣ of the light blue scripts.73 But this solution relies on a false assumption, i.e. that ΧΣ is equivalent to χσ (i.e. blue chi + sigma) both in the light blue and in the red scripts,74 and ultimately begs the question since the “aspirate” of light blue ΧΣ remains unexplained.
61Woodard 2021, p. 75, n. 5, explicitly states that the spellings ΦΣ, ΧΣ might be a non‑phonetic orthographic phenomenon, which he mystifyingly attributes to the status of phi and chi as “supplemental letters”.
62I assume that, paleographically, all types of xi in the Greek alphabets (dark blue xi, Naxian xi, red xi) may be taken back to one common archetype, which, in turn, derived from Phoenician samek. The ancestor of the letters cited must be the windowpane xi ( = Jeffery’s ξ8), an archaic form of xi, attested in several Etruscan abecedaria, and possibly in a badly damaged inscription from Eretria (SEG LV, no. 980; cf. Kenzelmann-Pfyffer, Theurillat, Verdan 2005, p. 60, no. 3, 8th c.).75 Wachter 2005 also notes the presence of a samek‑like blue xi (Jeffery’s ξ1) in a nonsense vase inscription from Boeotia (Wachter, Non‑Attic Vase Inscr., no. BOI 30, 6th c.).
63This evidence strongly suggests that was the original shape of xi in the Euboean alphabet, which constituted the basis of the Etruscan script.76 In time, this dark blue xi became a dead letter in the alphabets of Boeotia and Euboea and was eventually dropped to the benefit of red xi (X) which was added at the end of the alphabet.
64By eliminating the internal cross of , we obtain the Naxian xi (
) and by dispensing with the external bars of
, we obtain a standard cross, a well-known variant of red xi (+ = Jeffery’s χ2), which, in turn, with a forty-five-degree rotation, became Saint Andrew’s cross (X), the most widespread shape of both red xi (Jeffery’s ξ6) and blue chi (Jeffery’s χ1).77 The hourglass shape (
= Jeffery’s ξ7) documented in some graffiti from Abu Simbel can be interpreted as a variant of X.
65Now the problem is how the sign X developed two different sound values. In my opinion, the key to the mystery may lie in the above-mentioned instances of xi + pleonastic sigma: dark blue ΞΣ, Naxian Σ, red ΧΣ = ξσ. The following examples in tab. 6 illustrate this type of spellings:
Tab. 6 – Redundant spellings of /ks/.
Dark blue scripts | Red scripts |
Cor. ϘόραΞΣ (Wachter, Non-Attic Vase Inscr., no. COR 70 f) Megar. ϙύλιΞΣ (I.dial. Sicile Ι, no. 80, Selinous,7th‑6th c.) | Rhod. ϙύλιΧΣ (Jeffery) [I.Lindos, no. 710, 8th c.] Boeot. ἀργυροτόΧΣōι (Hansen, CEG, no. 326, Thebes?, ca 700‑675) Rhod. ἈναΧΣάνōρ (SEG XXVI, no. 1812, Abu Simbel, ca 590) Thess. [παρ]έ̣ΧΣε̄ (?) (IG IX, 2, no. 1202, l. 2, Magnesia, ca 550) |
66The red xi (X) retained the originally sound value of the letter. Conversely, somewhere in the “blue” domain at some point of time the redundant spellings with X, originally an allograph of xi before a pleonastic sigma (ΧΣ), were misinterpreted – or, alternatively, reinterpreted – as analytic spellings consisting of two letters: X representing a velar aspirate, and sigma or san representing the sibilant of the cluster. The new letter (chi) gained independence from its original matrix to represent /kh/. The order ΧΦΨ (sc. xi‑phi‑chi) of the supplementals in the red alphabets vs. ΦΧΨ (sc. phi‑chi‑psi) in the blue ones might be indicative of the relative chronology of the letters in both types of scripts.78
67Morphology may have also played a role in the reinterpretation of redundant spellings in cases like fut. ἕξει which had an “underlying” aspirate (root ἐχ‑). By analogy, the use of an aspirate must have been generalised to /ps/. Significantly, most red alphabets lacked a synthetic sign to represent this cluster. Likewise, the occasional use of analytic ΨΣ for /ks/ seems secondary.
68The double value of Χ turns out to be a trivial case of two allographs of one letter splitting into two or more independent graphemes like (h)eta (/h/, then /εː/) vs. “half‑ache” (usually /h/; /eː/ in Boeotia), digamma (w/) vs. upsilon (/y(ː)/ or /u(ː)/), and omicron (/o/) vs. omega (/ɔ(ː)/) in the Greek alphabet, or C vs. G and, more recently, i vs. j and u vs. v vs. w in the Latin alphabet and its descendants.
69As I stated at the very beginning, I am perfectly conscious that my proposal remains largely conjectural, but not more so than previous explanations. For the time being, I have no convincing proposal as regards the source of the supplementary signs Φ (phi) and Ψ (blue psi, red chi).79 The challenge remains for others to address and settle once and for all this complex issue.
Bibliographie
Abbreviations and works of reference
Abbreviations of editions and of works of reference for alphabetic Greek epigraphy are those of the list of the AIEGL published online: GrEpiAbbr, version January 2022, https://www.aiegl.org/grepiabbr.html (accessed 06/07/2023) and abbreviations of ancient authors of the Diccionario griego-español, http://dge.cchs.csic.es/lst/lst4.htm (accessed 06/07/2023).
Kouklia-Paphos: O. Masson, T.B. Mitford, Les inscriptions syllabiques de Kouklia-Paphos, Konstanz, Universitätsverlag, 1986.
RIG 1: M. Lejeune, Recueil des inscriptions gauloises I. Textes gallo-grecs, Paris, CNRS, 1985.
RIG 2/1: M. Lejeune, Recueil des inscriptions gauloises II 1. Textes gallo-étrusques. Textes gallo-latins sur pierre, Paris, CNRS, 1988.
RIG 2/2: P.‑Y. Lambert, Recueil des inscriptions gauloises II 2. Textes gallo-latins sur instrumentum, Paris, CNRS, 2002.
Sophocles, Fragments. Tragicorum Graecorum fragmenta 4. Sophocles, ed. S. Radt, R. Kannicht, Göttingen, Vandenhoeck & Ruprecht, 1999.
References
Allen 1987: W.S. Allen, Vox Graeca. The pronunciation of Classical Greek, Cambridge, Cambridge University Press, 1987 (3rd ed.).
Bailey 1985: C.‑J.N. Bailey, “Towards principles of governing the progress and patterning of phonetological development”, in C.‑J.N. Bailey, R. Harris (dir.), Developmental mechanisms of language, Oxford, Pergamon Press, 1985, pp. 1‑49.
Bakalakis 1964: G. Bakalakis, “Notes cycladiques”, BCH 88, 1964, pp. 539‑558.
Bechtel 1921: F. Bechtel, Die griechischen Dialekte, vol. I, Der lesbische, thessalische, boeotische, arkadische und kyprische Dialekt, Berlin, Weidmann, 1921.
Blass 1891: F. Blass, “Zu der naxischen inschrift der Timandre (sic)”, Neue Jahrbücher für Philologie und Paedagogik 143, 1891, pp. 335‑336.
Blümel 1982: W. Blümel, Die aiolischen Dialekte: Phonologie und Morphologie der inschriftlichen Texte aus generativer Sicht, Göttingen, Vandenhoeck & Ruprecht, 1982.
Brixhe 1991: C. Brixhe, “De la phonologie à l’ecriture: quelques aspects de l’adaptation de l’alphabet cananéen au grec”, in C. Baurain, C. Bonnet, V. Krings (dir.), Phoinikeia grammata. Lire et écrire en Méditerranée, Namur/Louvain, Société des Études Classiques/Peeters, 1991, pp. 313‑356.
Buck 1955: C.D. Buck, The Greek dialects. Grammar, selected inscriptions, glossary, Chicago/London, The University of Chicago Press, 1955.
Clackson 2002: J. Clackson, “The writing of χσ and φσ for ξ and ψ”, Glotta 78, 2002, pp. 22‑35.
Egetmeyer 2010: M. Egetmeyer, Le dialecte grec ancien de Chypre, vol. I‑II, Berlin/New York, De Gruyter, 2010.
Elvira Astoreca 2021: N. Elvira Astoreca, Early Greek alphabetic writing. A linguistic approach, Oxford/Philadelphia, Oxbow, 2021.
Eska, Mondon 2018: J.F. Eska, J.‑F. Mondon, “Phonological spreading, voice-onset delay, or phonetic noise? Orthographic <φσ> and <χσ> in Greek epichoric inscriptions”, in D. Gunkel, S.W. Jamison, A.O. Mercado, K. Yoshida (dir.), Vina diem celebrent. Studies in linguistics and philology in honor of Brent Vine, Beech Stave Press, Ann Arbor/New York, 2018, pp. 35‑42.
Gallavotti 1986: C. Gallavotti, “Il timbro vocalico nella grafia di Nicandra e in inscrizioni arcaiche delle Cicladi”, in A. Etter (dir.), O‑o‑pe‑ro‑si: Festschrift für Ernst Risch zum 75. Geburtstag, Berlin/New York, De Gruyter, 1986, pp. 231‑240.
Gomis García, Striano Corrochano 2017: V. Gomis García, A. Striano Corrochano, “Los grupos de consonantes oclusivas labiales y velares seguidas de silbante en los dialectos griegos”, in P. Poccetti, F. Logozzo (dir.), Ancient Greek linguistics. New approaches, insights, perspectives, Boston/Berlin, De Gruyter, 2017, pp. 19‑34.
Gruben 1997: G. Gruben, “Naxos und Delos”, JDAI 112, 1997, pp. 261‑416.
Hawkins 2012: S. Hawkins, “A linguistic analysis of the vase inscriptions of Sophilos”, Glotta 88, 2012, pp. 122‑165.
Hermann 1898: O. Hermann, Die griechischen Dialekte in ihrem historischen Zusammenhange mit der wichtigsten ihrer Quellen, vol. III, Der ionische Dialekt. Quellen und Lautlehre, Göttingen, Vandenhoeck & Ruprecht, 1898.
Janko 2017: R. Janko, “From Gabii and Gordion to Eretria and Methone. The rise of the Greek alphabet”, in J. Strauss Clay, I. Malkin, Y. Tzifopoulos (dir.), Panhellenes at Methone. Graphê in late geometric and protoarchaic Methone, Macedonia (ca 700 BCE), Berlin, De Gruyter, 2017, pp. 135‑164.
Kastner 1989: W. Kastner, “ἔσχατος ‑ ἔγκατα”, MH 46, 1989, pp. 9‑14.
Kenzelmann-Pfyffer, Theurillat, Verdan 2005: A. Kenzelmann-Pfyffer, T. Theurillat, S. Verdan, “Graffiti d’époque géométrique provenant du sanctuaire d’Apollon Daphnéphoros à Erétrie”, ZPE 151, 2005, pp. 51‑83.
Kirchhoff 1877: A. Kirchhoff, Studien zur Geschichte des griechischen Alphabets, Berlin, Dümmler, 1877 (3rd ed.).
Kretschmer 1896: P. Kretschmer, “Die sekundären Zeichen des griechischen Alphabets”, MDAI(A) 21, 1896, pp. 410‑433.
Lambert 2002: P.‑Y. Lambert, Recueil des inscriptions gauloises, vol. II/2, Textes gallo-latins sur instrumentum, Paris, CNRS, 2002.
Lejeune 1971: M. Lejeune, “La dédicace de Νικάνδρη et l’écriture archaïque de Naxos”, RPh 45, 1971, pp. 209‑215.
Lejeune 1972: M. Lejeune, Phonétique historique du mycénien et du grec ancien, Paris, Klincksieck, 1972.
Masson 1984: O. Masson, “Noms de femmes rares en attique”, Onomata 9, 1984, pp. 33‑38 (reprinted in Masson, OGS I‑II, pp. 421‑426).
McGowan 1997: E.P. McGowan, “The origin of the Athenian Ionic capital”, Hesperia 66, pp. 209‑233.
Méndez Dosuna 1991‑1992: J. Méndez Dosuna, “<ει> por <ε> ante vocal en griego, el valor del signo <Ͱ> en Tespias y otras cuestiones”, Veleia 8‑9, 1991‑1992, pp. 309‑330.
Méndez Dosuna 1993a: J. Méndez Dosuna, “Metátesis de cantidad en jónico-ático y heracleota”, Emerita 61, 1993, pp. 95‑134.
Méndez Dosuna 1993b: J. Méndez Dosuna, “Los griegos y la realidad psicológica del fonema: κ y ϙ en los alfabetos arcaicos”, Kadmos 32, 1993, pp. 96‑126.
Méndez Dosuna 2017a: J. Méndez Dosuna, “Methone of Pieria. A reassessment of epigraphical evidence (with a special attention to pleonastic sigma)”, in J. Strauss Clay, I. Malkin, Y. Tzifopoulos (dir.), Panhellenes at Methone. Graphê in late geometric and protoarchaic Methone, Macedonia (ca 700 BCE), Berlin, De Gruyter, 2017, pp. 242‑258.
Méndez Dosuna 2017b: J. Méndez Dosuna, “Once again on allophonic spellings in Ancient Greek”, in I. Hajnal, D. Kölligan, K. Zipser (dir.), Miscellanea Indogermanica. Festschrift für José Luis García Ramón zum 65. Geburtstag, Innsbruck, IBS, 2017, pp. 487‑498.
Méndez Dosuna 2021: J. Méndez Dosuna, “The pronunciation of upsilon and related matters. A U‑turn”, in R. Parker, P.M. Steele (dir.), The early Greek alphabets. Origin, diffusion, uses, Oxford, Oxford University Press, 2021, pp. 119‑145.
Miller 1994: D.G. Miller, Ancient scripts and phonological knowledge, Amsterdam, Benjamins, 1994.
Miller 2014: D.G. Miller, Ancient Greek dialects and early authors. Introduction to the dialect mixture in Homer, with notes on lyric and Herodotus, Boston/Berlin, De Gruyter, 2014.
Minon 2021: S. Minon, “Letter forms and distinctive spellings. Date and context of the ‘New Festival Calendar from Arkadia’”, in R. Parker, P.M. Steele (dir.), The early Greek alphabets. Origin, diffusion, uses, Oxford, Oxford University Press, 2021, pp. 146‑184.
Minon, Dubois, Le Feuvre 2017: S. Minon, L. Dubois, C. Le Feuvre, “Introduction”, in A. Alonso Déniz, L. Dubois, C. Le Feuvre, S. Minon (dir.), La suffixation des anthroponymes grecs antiques (SAGA). Actes du colloque international de Lyon, 17‑19 septembre 2015, Université Jean Moulin‑Lyon 3, Geneva, Droz, 2017, pp. 1‑30.
Powell 1991: B.B. Powell, Homer and the origin of the Greek alphabet, Cambridge, Cambridge University Press, 1991.
Probert 2010: P. Probert “Phonology”, in E.J. Bakker, A companion to the Ancient Greek language, Chichester, Wailey-Blackwell, 2010, pp. 85‑103.
Rohlfs 1977: G. Rohlfs, Grammatica storica dei dialetti italogreci (Calabria, Salento), Munich, Beck, 1977.
Ruijgh 1997: C.J. Ruijgh, “La date de la création de l’alphabet grec et celle de l’épopée homérique”, BO 54, 1997, pp. 534‑603.
Ruijgh 1998: C.J. Ruijgh, “Sur la date de la création de l’alphabet grec”, Mnemosyne 51, 1998, pp. 658‑687.
Schwink 1991: F.W. Schwink, “The writing of Ancient Greek consonant clusters”, Kadmos 30, 1991, pp. 113‑127.
Schwyzer 1939: E. Schwyzer, Griechische Grammatik, vol. I, Allgemeiner Teil, Lautlehre, Wortbildung, Flexion, Munich, Beck, 1939.
Sihler 1995: A.L. Sihler, New comparative grammar of Greek and Latin, New York/Oxford, Oxford University Press, 1995.
Slings 1998: S.R. Slings, “Tsadē and hē. Two problems in the early history of the Greek alphabet”, Mnemosyne 51, 1998, pp. 641‑657.
Solmsen 1904: F. Solmsen, “Eigennamen als Zeugen der Stammesmischung in Böotien”, RhM 59, 1904, pp. 481‑505.
Threatte 1980: L. Threatte, The Grammar of Attic Inscriptions, vol. I, Phonology, Berlin/New York, De Gruyter, 1980.
Threatte 2016: L. Threatte, “Attic Dipinti: a trove of evidence for the Attic dialect”, in R. Wachter (dir.), Potters – painters – scribes. Inscriptions on Attic vases (proceedings of the colloquium held at the Universities of Lausanne and Basel from 20th to 23rd September 2012), Zurich, Akanthus, 2016, pp. 131‑139.
Thumb, Scherer 1959: A. Thumb, A. Scherer, Handbuch der griechischen Dialekte, vol. II, Heidelberg, Winter, 1959.
Tribulato 2015: O. Tribulato, Ancient Greek verb-initial compounds. Their diachronic development within the Greek compound system, Berlin/New York, De Gruyter, 2015.
Vaux 1998a: B. Vaux, The phonology of Armenian, Oxford, Clarendon Press, 1998.
Vaux 1998b: B. Vaux, “The laryngeal specifications of fricatives”, Linguistic Inquiry 29, 1998, pp. 497‑511.
Vottéro 2002: G. Vottéro, “Boeotica Epigrammata”, in J. Dion (dir.), L’épigramme de l’Antiquité au xviie siècle ou du ciseau à la pointe, Nancy, A.D.R.A., 2002, pp. 69‑122.
Waal 2018: W. Waal, “On the ‘Phoenician letters’. The case for an early transmission of the Greek alphabet from an archaeological, epigraphic, and linguistic perspective”, Aegean Studies 1, 2018, pp. 83‑125, https://www.aegeussociety.org/en/studies/on-the-phoenician-letters-the-case-for-an-early-transmission-of-the-greek-alphabet-from-an-archaeological-epigraphic-and-linguistic-perspective/ (accessed 21/07/2022).
Waal 2020: W. Waal, “Mother or sister? Rethinking the origins of the Greek alphabet and its relation to the other ‘western’ alphabets”, in P.J. Boyes, P.M. Steele (dir.), Understanding relations between scripts, vol. II, Early alphabets, Oxford, Oxbow Books, 2020, pp. 109‑124.
Wachter 1991: R. Wachter, “Abbreviated writing”, Kadmos 30, 1991, pp. 49‑80.
Wachter 2005: R. Wachter, “Annex zu Fragment 3 der Graffiti von Eretria”, ZPE 151, 2005, pp. 84‑86.
Wachter 2021: R. Wachter, “The genesis of the local alphabets of archaic Greece”, in R. Parker, P.M. Steele (dir.), The early Greek alphabets. Origin, diffusion, uses, Oxford, Oxford University Press, 2021, pp. 19‑31.
Wolters, Bruns 1940: P. Wolters, G. Bruns, Das Kabirenheiligtum bei Theben, Berlin, Deutsches Archäologisches Institut, 1940.
Woodard 2012: R.D. Woodard, “Labiovelar development in Greek and an alphabetic repercussion”, in R. Sukač (dir.), The sounds of Indo-European, vol. II, Phonetics, phonemics, and morphophonemics, Munich, Lincom Europa, 2012, pp. 299‑313.
Woodard 2014: R.D. Woodard, The textualization of the Greek alphabet, Cambridge, Cambridge University Press, 2014.
Woodard 2017: R.D. Woodard, “Alphabet and phonology at Methone. Beginning a typology of Methone alphabetic symbols and an alternative hypothesis for reading Ηακεσάνδρō”, in J. Strauss Clay, I. Malkin, Y. Tzifopoulos (dir.), Panhellenes at Methone. Graphê in late geometric and protoarchaic Methone, Macedonia (ca 700 BCE), Berlin, De Gruyter, 2017, pp. 182‑218.
Woodard 2020: R.D. Woodard, “Vowel representation in the Archaic Greek and Old Aramaic scripts. A comparative and phonological examination”, in P.J. Boyes, P.M. Steele (dir.), Understanding relations between scripts, vol. II, Early alphabets, Oxford, Oxbow Books, 2020, pp. 91‑107.
Woodard 2021: R.D. Woodard, “Contextualizing the origin of the Greek alphabet”, in R. Parker, P.M. Steele (dir.), The early Greek alphabets. Origin, diffusion, uses, Oxford, Oxford University Press, 2021, pp. 74‑103.
Notes de bas de page
1 This research is part of the project “Onomastics and Linguistic Contact in Ancient Greek” (PID2020-114162GB-I00, MCIN/AEI/10.13039/501100011033). I am greatly indebted to Alcorac Alonso Déniz, Enrique Nieto Izquierdo, Blanca Prósper, and Philippa M. Steele for their enlightening comments on earlier versions of this article. Needless to say, all remaining flaws and misconceptions are my own responsability. Unless otherwise indicated, the dates of the inscriptions cited are BC. For convenience, I refer to Jeffery’s typology of the local variants of the Greek letters (LSAG2, pp. 23‑40). The following abbreviations are used: Att. = Attic, Boeot. = Boeotian, Calabr. = Calabrian Greek (Greko), ClGk = Classical Greek, Cor. = Corinthian, Cypr. = Cypriot, Delph. = Delphian, dial. = dialectal, Dor. = Doric, Epid. = Epidaurian, Eub. = Euboean, Fr. = French, fut. = future, G. = German, Gaul. = Gaulish, gen. = genitive, Ion. = Ionic, It. = Italian, Locr. = Locrian, Megar. = Megarian, MnE. = Modern English, MnGk = Modern Greek, ms. = manuscript, nom. = nominative, OE. = Old English, PIE = Proto-Indo-European, pl. = plural, pres. = present, Rhod. = Rhodian, Sal. = Salentine Greek (Griko), sg. = singular, Thess. = Thessalian, Transalp. Celt. = Transalpine Celtic, Turk. = Turkish, WLocr. = West Locrian.
2 For an updated reassessment with diverse viewpoints on some of the contentious issues concerning the origin and diffusion of the Greek alphabet, see Waal 2018; Waal 2020; Elvira Astoreca 2021; Wachter 2021.
3 To avoid ambiguity, the digraphs ΠΗ, ΚΗ (ϘΗ) were used for /ph/ and /kh/ in Thera and Melos.
4 For simplicity, here and thereafter the letter Σ stands indifferently for sigma or san. Similarly, K stands for K and Ϙ as used in the relevant contexts.
5 Strictly speaking, the letter xi does not feature among the supplementals since its shape clearly continued that of Phoenician samek and it retained its original position in the alphabetical order. Samek represented a voiceless alveolar fricative /s/. For reasons not fully understood, Phoenician ṣade (/sˤ/ or /ts/) and shin (/ʃ/) were adopted as san and sigma to denote Greek /s/, while samek ended up as the spelling of [ks].
6 Both dark blue Ξ and red X (along with the syllabograms <xa>, <xe> in Cyprus) were anomalous graphemes in that they represented two phonemes. Schwink 1991 concluded that the clusters /ps/, /ks/ were perceived as single structural units (pseudo-affricates) due to a certain degree of co‑articulation of their components, but another factor mentioned by Schwink 1991, p. 124 almost in passing must have played a more decisive role in the creation of these “double” (i.e. bisegmental) letters: /ps/, /ks/ were the only clusters that could occur both in word-initial and in word-final position: e.g. ψυχή, ξένος, φλέψ, φύλαξ, σάρξ, λύγξ (Allen 1987, p. 59; Hawkins 2012, p. 132; Wachter 2021, p. 28, n. 22). The idea championed by Woodard 2020, p. 107; Woodard 2021, pp. 78‑79 that the use of psi and xi can be accounted for as an instance of the transference of syllabic Cypriot scribal practices to alphabetic Greek, is as unverifiable as it is improbable.
7 The red psi (Jeffery’s ψ5), probably a variant of blue psi and sparsely attested in Arcadia and other regions, is barely anecdotal (LSAG2, pp. 37, 207, 212‑213 and 428).
8 Obviously, red xi is the same letter as the x of the Latin alphabet, which ultimately descended via Etruscan from the red alphabet of Euboea. The combination ΨΣ is uncommon in the red scripts. The rare instances of ΧΣ occurring in inscriptions from regions where otherwise the red alphabet was used have been often misinterpreted as combinations of blue chi + sigma identical to Attic ΧΣ (LSAG2, pp. 32 and 351; Powell 1991, p. 167, n. 125; Ruijgh 1997, p. 576; Vottéro 2002, p. 71). Actually, they must be instances of a red xi with pleonastic sigma (Guarducci, Epigrafia greca I, pp. 83 and 156; Woodard 2014, pp. 43‑45; Woodard 2017, p. 196; Méndez Dosuna 2017a, pp. 251‑257; Elvira Astoreca 2021, p. 121): e.g. Boeot. ἀργυροτόξσōι (not *ἀργυροτόχσōι) with ΧΣ besides χαρίϝετταν with Ψ (Hansen, CEG, no. 326, l. 1; Thebes?, ca 700‑675?). On pleonastic sigma, see infra §§ 2.2, 2.4.
9 Schwyzer 1939, p. 211 with some doubts; Bailey 1985, p. 47; Allen 1987, p. 60; Kastner 1989, p. 10; Sihler 1995, pp. 203 and 219; Slings 1998, p. 655. Threatte 1980, p. 21 and Schwink 1991, p. 121 suspend judgment, and Buck 1955, p. 17 ignores the problem altogether.
10 Allen 1987, pp. 59 and 164.
11 Kastner 1989, pp. 9‑10 for ἐχθός; Sihler 1995, p. 219 for τέχνη.
12 Schwyzer 1939, p. 211.
13 Allen 1987, p. 60.
14 Vaux 1998a ; Vaux 1998b; Clackson 2002, p. 25; Hawkins 2012, p. 128; Eska, Mondon 2018, pp. 36‑38.
15 See especially, Lejeune 1972, pp. 54 and 72, who paradoxically seems to base the difference between the occlusives fortes (the plain voiceless stops) and the occlusives douces (the aspirate stops, which eventually became fricatives), on the degree of occlusion rather than on voice proper. For a fricative, see Powell 1991, p. 60; Ruijgh 1997, pp. 566‑568; Slings 1998, p. 655; Elvira Astoreca 2021, p. 120.
16 In the same vein, Threatte 2016, p. 136 tentatively concludes that “maybe Greek Ξ and Ψ were not as vigorously and distinctly pronounced as the x in English axe or the ps of Ipswich”.
17 See further Lejeune 1972, p. 73; Threatte 1980, p. 21; Threatte 2016, pp. 135‑137. These spellings are considered by Allen 1987, p. 60 to be evidence of subphonemic aspiration [phs, khs], but this is clearly an ad hoc hypothesis.
18 Threatte 1980, p. 21; Threatte 2016, pp. 135‑136.
19 See also Lejeune 1972, p. 73; Threatte 1980, pp. 551‑553; Schwink 1991, p. 121.
20 Note that in both cases the preposition precedes the article: es‑to /es tō/.
21 For pleonastic sigma in dark blue ΞΣ and red ΧΣ, see further below in this section and infra § 2.4. Note also provocative spellings like e.g. Maxx, Foxx in Modern English.
22 This notwithstanding, the transposition is not entirely irrelevant for phonology. On the one hand, the relative frequency of these misspellings is additional evidence that somehow the clusters /ps, ks/ were perceived as single structural units. On the other hand, a metathesis of stop + sibilant and sibilant + stop sequences is not unheard‑of as a real phonetic change: e.g. ψωμίον > [spo'mi], ἔγραψα > ['egraspa] widespread in the modern vernacular of most Greek-speaking villages of Calabria (Rohlfs 1977, p. 53; for other outcomes in the Greek of Apulia and Calabria, see infra n. 31); cf. also MnE. ask > dial. aks, OE. wæps > MnE. wasp. In any case, whether a real phonetic phenomenon (e.g. Schwyzer 1939, p. 211; Lejeune 1972, p. 73; Kastner 1989, pp. 9‑10), or merely a graphic one as I believe, metathesis is not an index of a weakened stop in stop + sibilant sequences. Kastner 1989 imagined an improbable evolution *eks‑atos > *ekhsatos > eskhatos, in which metathesis is viewed as a remedial means of averting the danger of a supposed tendency to weakening of postconsonantal /s/.
23 The total count of the spellings of [ks] in the ostraca from the Agora (Lang apud Agora XXV, p. 14) before ca 450 is as follows, in descending order: ΧΣ (129 occurrences on Kallixenos ostraca), ΣΧ (7, all Kallixenos), Σ (5), ΣΣ (3), Χ (2), Ξ (1); in the ostraca from the Kerameikos (O.Kerameikos, pp. 111‑112): ΧΣ (61 + 8 damaged instances), ΣΧ (5), Σ (5), Χ (1), Ξ (1).
24 Wolters, Bruns 1940 read Κῖνυφ[ς] (no. 200), Κῖνυ[φς] (no. 201).
25 Masson 1984, p. 34; Minon, Dubois, Le Feuvre 2017, pp. 21‑23. For an analysis of Κόρυψ along similar lines, see Solmsen 1904, p. 486, n. 2.
26 I owe this suggestion to Alcorac Alonso Déniz.
27 Bechtel 1921, p. 262. The suffix variation can be compared to e.g. Att.‑Ion. ὄρνις ‑ιθος, κέλης ‑ητος vs. Dor. ὄρνιξ ‑ιχος, κέληξ.
28 Egetmeyer 2010, pp. 177‑178; Gomis García, Striano Corrochano 2017, p. 28.
29 For the diachronic and synchronic analysis of the V1 compounds of the τερψίμβροτος type, see Tribulato 2015. For Cypr. e‑se see below.
30 Hawkins 2012, p. 127.
31 Ancient [ps], [ks] become [fs] or [tts]: ψωμίον > Sal. [f(t)so'mi], Calabr. [(t)tso'mi]; ἔγραψα > Sal. ['egrafsa], Calabr. ['egrattsa]; ξένος > Sal. ['fseno], Calabr. ['(t)tseno], ἕξ > Sal. ['efse], Calabr. ['ettse]; see Rohlfs 1977, pp. 43 and 53. For [ps] > [sp], see supra n. 22.
32 Buck 1955, pp. 81 and 108. In most dialects including Attic-Ionic, preconsonantal ἐξ became ἐκ.
33 The same phenomenon accounts for the widespread apocope of prepositions (Buck 1955, pp. 81‑82; Lejeune 1972, p. 224), unexpected contraction in gen. pl. *τᾱ́ων > Boeot. τᾶν, Ion. τῶν (Buck 1955, p. 43; Lejeune 1972, p. 268), irregular psilosis in the article and in some conjunctions: WLocr., Delph. ὀ, Boeot. ὠς, Delph. ἆς (Buck 1955, p. 53), etc.
34 Hawkins 2012, p. 127.
35 Thumb, Scherer 1959, p. 265; Guarducci, Epigrafia greca I, pp. 154‑156; Lejeune 1971; Lejeune 1972, pp. 72‑73; Threatte 1980, p. 20; Gallavotti 1986, p. 235; LSAG2, p. 291; Powell 1991, pp. 60 and 171; Ruijgh 1997, pp. 565‑566, 568, 570 and 586‑587; Ruijgh 1998, p. 681; Slings 1998, p. 655; Miller 2014, pp. 150‑151; Woodard 2014, pp. 37‑38; Woodard 2020, pp. 102‑103; Gomis García, Striano Corrochano 2017, p. 25; Elvira Astoreca 2021, pp. 73, 94 and 120; Minon 2021, p. 169, n. 151.
36 LSAG2, p. 28.
37 The idea that indifferently represents both long and short /æ(ː)/ is incorrect. The spelling
O in Δεινοδίκ
ο and ἀλ(λ)
ōν corresponds to the lento pronunciation of the vowel sequence with no quantitative metathesis (/æːo/), while scansion reflects an allegro realisation [æ̯ɔː] with synizesis and quantitative metathesis (Gallavotti 1986, pp. 232‑233; Méndez Dosuna 1993a, pp. 100‑101; Miller 2014, p. 151).
38 This value of , documented in other texts, is in all likelihood a loan from psilotic Ionia, where superfluous heta had been recycled as eta to represent /εː/ whether from earlier *ā (e.g. κούρη) or *ē (e.g. ἀνέθηκε); see e.g. Allen 1987, p. 73; LSAG2, p. 28.
39 For heta with a syllabic value [hε(ː)], cf. Buck 1955, p. 19; Lejeune 1971, p. 214; Wachter 1991; Miller 2014, p. 150.
40 Jeffery’s facsimile (O) and transcription (Ⱶo; see LSAG2, p. 304, no. 3, p. 412, plate 55) are inaccurate. This fact ruins Lejeune 1971, p. 211 “argument décisif” against F. Blass’ interpretation (see below) since
never stands for h‑ in the inscriptions from Naxos. The sign
is clearly discernible in a photograph taken in 1989 (Bakalakis 1964, p. 543, fig. 3).
41 Τhe transcription Ἀλεξιτίδης (Kontoleon, SEG) is mistaken (see McGowan 1997, p. 224, n. 72).
42 For the spelling ἄνα[ς] see Hansen’s commentary to CEG, no. 402: ex h hasta directa servari videtur. Τhe reading ἄναχ̣[ς] (Gruben 1997, pp. 282‑285; SEG XLVII, no. 1220; LSAG2, p. 292, n. 4) is mistaken.
43 Woodard 2014, pp. 41‑42. On windowpane xi, see infra § 2.4.
44 LSAG2, pp. 18‑23, 240‑241 and pl. 48 (8th‑6th c.). Clearly, the sign was a dead letter in Etruscan. The alphabet incised on a bird-shaped bottle from Viterbo (ca 550‑500) has a three-stroke sigma instead of xi, but this must be a mere error.
45 The reading Ἀλε̣Hσοῖ or Ἀλ[ε]Hσοῖ is advocated by Hermann 1898, pp. 26‑27, no. 43; Delamarre, IG XII, 7, no. 1908; and LSAG2, p. 304, no. 20, p. 412, pl. 56.a.
46 Bechtel, SGDI, no. 5358 (published in 1905). He explains the name Ἀλξώ as a feminine hypocoristic (Kosename) of Ἀλξήνωρ. I thank Alcorac Alonso Déniz for calling my attention to F. Bechtel’s edition, which has been entirely overlooked by later scholars.
47 The script, the dialect, and the content of the text indubitably point to a Naxian origin and exclude the hypothetical readings *Ἀλξσήνōρ̣, *Νάξσιος with the local red xi and pleonastic sigma.
48 Elvira Astoreca 2021, p. 120, speculates that the Naxian stone engravers deliberately chose the allograph before an /s/ “to make sure that [the sign would] be read as a consonant”, but this is unconvincing if only because
κηβόλōι and Φ
ράhσō could be easily misread as /æː/ *[æːkεːboloːi̯], *[phæːrahsoː].
49 Blass 1891; Buck 1955, pp. 189‑190; Woodard 2014, pp. 42‑46; Woodard 2017, p. 196; Méndez Dosuna 2017a, pp. 250‑251, n. 39; Janko 2017, p. 196.
50 We will come back to the issue of pleonastic sigma in due course later in this essay (see infra § 2.4).
51 Clackson 2002, p. 30.
52 Probert 2010, p. 94, cites it favourably. Contra Hawkins 2012, p. 131; Gomis García, Striano Corrochano 2017, pp. 30‑33; Eska, Mondon 2018, p. 40.
53 Hawkins 2012, p. 131.
54 Hawkins 2012, pp. 131‑132.
55 Eska, Mondon 2018.
56 Eska, Mondon 2018, p. 41. For the reliability of the spellings in the Styra tablets, see supra § 2.3.1. Neither Clackson 2002, nor Hawkins 2012 and Eska, Mondon 2018 deal with the spelling Σ in Naxian inscriptions.
57 Méndez Dosuna 2017a, p. 249, n. 29; Méndez Dosuna 2017b, pp. 488‑489.
58 Buck 1955, pp. 18, 22 and 31; Thumb, Scherer 1959, pp. 19 and 21; Blümel 1982, pp. 41 and 63; LSAG2, pp. 89 and 94.
59 Buck 1955, p. 90; Thumb, Scherer 1959, p. 37: cf. e.g. Δαμοκλεῖος (I.Thespies, no. 312, l. 4, Thespiae, 400‑350); τῶ Ἡρακλεῖος (I.Thespies, no. 48, l. 5, Thespiae, ca 250).
60 Buck 1955, p. 18, explained the shape of this Ⱶ as “a compromise between E and I […] used for the close ε, later ι”.
61 See Méndez Dosuna 1991‑1992 on this issue.
62 Cf. e.g. WLocr. δίκαν, ϝοικέοντος, hἀπιϝοικία, κρατεῖν vs. πρόδιϙον, ἐπιϝοίϙον, Λοϙρός (IG IX, 1, no. 718, Chaleion, 500‑450).
63 Méndez Dosuna 1993b; Méndez Dosuna 2017b, pp. 493‑496. The use of qoppa before Υ in cases like Ion. λήϙυθος (I.Smyrna, no. 799.2; Smyrna, 6th c.) or Eub. Ϙλυτṓ (Wachter, Non-Attic Vase Inscr., no. CHA 3b, Vulci, ca 530) does not presuppose a back pronunciation [u(ː)] of Y (Méndez Dosuna 2021).
64 Allen 1987, p. 8.
65 Even so, native speakers immediately perceive as foreign a pronunciation ['dado].
66 Eska, Mondon 2018, p. 41, n. 15.
67 Méndez Dosuna 1993b, pp. 97‑99; Méndez Dosuna 2017b, pp. 492‑493.
68 This accidentally coincides with the etymologically accurate spelling judgar, prevalent in Old Spanish, but now fallen into disuse.
69 J.F. Eska and J.‑F. Mondon also cite Transalp. Celt. cuet[ic] (RIG 2/2, L‑98, 1b.1, L’Hospitalet‑du‑Larzac, 1st c. AD) vis‑à‑vis coetic ‘moreover’ (RIG 2/2, L‑98, 1b.3,) but P.‑Y. Lambert reads epotiniosco · etic and co etic. At any rate, J.F. Eska and J.‑F. Mondon’s coe/cue variation is amenable to a straightforward explanation: U represented /u/, /uː/ and /w/, as well as [w] as an allophone of prevocalic /o/. The case would be not dissimilar to Spanish misspellings like almuada (substandard [al'mwaða]) for almohada (formal [almo'aða], standard [al'mo̯aða]) ‘pillow’ or cuete (substandard ['kwete]) for cohete ([ko'ete], ['ko̯ete]) ‘rocket’; for u representing [w] in the standard orthography, cf. lengua [len'gwa] ‘tongue’, cueva ['kweβa] ‘cave’.
70 The reading of the final letters is disputed.
71 I am grateful to Blanca Prósper for her assistance with the Gaulish data.
72 Cf. already Méndez Dosuna 2017a, p. 251, n. 41; Méndez Dosuna 2017b, pp. 488‑489.
73 Kretschmer 1896, p. 420; LSAG2, p. 36; Powell 1991, p. 60; Miller 1994, p. 51; Ruijgh 1997, p. 576; Ruijgh 1998, p. 681. Contra Janko 2017, p. 157. According to Johnston (LSAG2, p. 427), Guarducci, Epigrafia greca I, pp. 83 and 96 was of the same opinion, but he probably reads too much into her presentation of the different spelling for [ks].
74 As we shall immediately see, ΧΣ in the red scripts is a pleonastic spelling equivalent to ξσ.
75 The fragment reads [‑‑‑]ΟΠ̣[‑‑‑], which Kenzelmann-Pfyffer, Theurillat, Verdan 2005; Wachter 2005; and Woodard 2014, p. 157 interpret as an alphabetical sequence ξοπ. Elvira Astoreca 2021, pp. 45 and 48‑49 questions this interpretation and suggests that the sign could also be interpreted as a squared theta, but her circumspection is probably excessive.
76 It is not clear whether the windowpane xi is the predecessor or, what seems more likely, derives from the three‑bar xi (Jeffery’s ξ1‑3), which is closer to its Phoenician model.
77 Alcorac Alonso Déniz calls my attention to the case of the crossed theta (Jeffery’s θ1) rotating forty-five degrees to become θ2. Cf. also above-mentioned “sidelong xi” (Jeffery’s ξ9), characteristic of Argos.
78 Ruijgh 1998, p. 573.
79 I do not feel particularly attracted to Brixhe 1991, pp. 340‑344; Miller 1994, pp. 52‑53; and Woodard 2012; Woodard 2014, p. 303, n. 289, hypothesis, albeit with some differences of detail, that phi (Φ) was a doublet of qoppa (Ϙ), which denoted /kwh/ in the Greek Uralphabet (e.g. ΦΟΝΟΣ < *[kwhónos]). By a regular sound change, /kwh/ became /ph/ in most contexts which allegedly induced a reinterpretation of Φ, i.e. ΦΟΝΟΣ [phónos]. According to Ruijgh 1998, pp. 575‑576, the supplementary aspirate letters phi and chi were made up by combining sinistroverse and dextroverse variants of the corresponding plain plosives pi and kappa. This hypothesis is much more imaginative than it is convincing.
Auteur

Le texte seul est utilisable sous licence Creative Commons - Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International - CC BY-NC-ND 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
Dire la ville en grec aux époques antique et byzantine
Actes du colloque de Créteil, 10-11 juin 2016
Liliane Lopez-Rabatel, Virginie Mathé et Jean-Charles Moretti (dir.)
2020
Dérivation nominale et innovations dans les langues indo‑européennes anciennes
Alain Blanc et Isabelle Boehm (dir.)
2021
Stylistique et poétique de l'épigramme latine
Nouvelles études
Daniel Vallat et Florence Garambois-Vasquez (dir.)
2022
Contacts linguistiques en Grèce ancienne
Diachronie et synchronie
Alcorac Alonso Déniz, Julián V. Méndez Dosuna, Enrique Nieto Izquierdo et al. (dir.)
2024
Les représentations du désir féminin dans la poésie latine, de Catulle à Juvénal
Contraintes et libertés de l’espace poétique
Giacomo Dimaggio
2024
Définir l'épopée en Grèce ancienne
Contribution à une histoire de l'idée d'épopée d'Hérodote à Tzetzès
Flore Kimmel-Clauzet
2024