The Use of Gamma in Place of Digamma in Ancient Greek

Home , Apollonius Dyscolus, Digamma, Gamma, Jewish Koine Greek, Modern Greek, Omicron

Mnemosyne (2020) 1-22

brill.com/mnem

Francesco Camagni University of Manchester, UK [email protected]

Received August 2019 | Accepted March 2020

Abstract

Originally, Ancient Greek employed the letter digamma ( ϝ) to represent the /w/ sound. Over time, this sound disappeared, alongside the digamma that denoted it. However, to transcribe those archaic, dialectal, or foreign words that still retained this sound, lexicographers employed other letters, whose sound was close enough to /w/. Among these, there is the letter gamma (γ), attested mostly but not only in the Lexicon of Hesychius. Given what we know about the sound of gamma, it is difficult to explain this use. The most straightforward hypothesis suggests that the scribes who copied these words misread the capital digamma (Ϝ) as gamma (Γ). Presenting new and old evidence of gamma used to denote digamma in Ancient Greek literary and documentary papyri, lexicography, and medieval manuscripts, this paper refutes this hypothesis, and demonstrates that a peculiar evolution in the pronunciation of gamma in Post-Classical Greek triggered a systematic use of this letter to denote the sound once represented by the digamma.

Keywords

Ancient Greek language – gamma – digamma – Greek phonetics – Hesychius – lexicography

© Francesco Camagni, 2020 | doi:10.1163/1568525X-bja10018 This is an open access article distributed under the terms of the CC BY 4.0Downloaded license. from Brill.com09/30/2021 01:54:17PM via free access 2 Camagni

1 Introduction

It is well known that many ancient Greek dialects preserved the /w/ sound into the historical period, contrary to Attic-Ionic and Koine Greek. This sound can be found usually in dialectal inscriptions, represented by the letter digamma (Ϝ). However, after Koine Greek replaced all other varieties of Greek, the sound /w/ disappeared alongside the digamma that represented it. The only exception is represented by those dialectal words attested in poetry and in lexicographical glosses. In these words the sound /w/ is represented in a variety of ways, such as by the letter beta (β), upsilon (υ), omicron-upsilon (ου), or gamma (γ). This paper focusses on the use of gamma to replace the letter digamma, and to represent the /w/ sound in general. This spelling is attested particularly in the Lexicon of Hesychius and certain scholarship believes it is a scribal blunder, originating in the confusion between the shape of capital gamma (Γ) and digamma (Ϝ).1 Contrary to this theory, this paper will argue that from the second century bc onwards, the voiced velar plosive /g/, represented by the letter gamma (γ), occasionally came to sound like /w/, usually in intervocalic position after a rounded vowel. As a consequence, the letter gamma was at times employed to represent the sound /w/—mostly in dialectal Greek and foreign words, but also as a reinforcement to the Greek diphthongs αυ and ευ, when pronounced /aw/ and /ew/. This paper illustrates and defends the theory of this phonetic shift—already advanced by Gignac2—and extends it to explain the philological question of gamma used in place of digamma. This argument is substantiated by new and old evidence of gamma used to denote digamma and the sound /w/ in general, found in Ancient Greek literature, ancient lexicography and grammar, literary and documentary papyri, and medieval manuscripts. The paper begins by sketching the history of the digamma, for the benefit of those readers who are not well-acquainted with the question. Next, it re- views the available evidence regarding the theoretical knowledge around the digamma possessed by the ancient grammarians and lexicographers, from the first century bc until the Byzantine era. This review aims to evaluate the extent of their theoretical knowledge, in order to assess whether grammarians and scribes failed to recognise the sound /w/ as the one once represented by digamma, thus using other letters to denote this sound.

1 E.g. Sophocles 1854, 21-22; Clayman 1978, 396; Sihler 1995, 183. 2 Gignac 1970, 189; 1976, 74.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 3

Subsequently, the paper provides a brief overview of the graphic solutions adopted to replace the digamma, and introduces the core topic of this discussion, that is the letter gamma attested in place of digamma. After discussing the current theories surrounding this question, the paper then proceeds to present the argument for discarding the theory of the scribal mistake. It does so by providing and discussing old and new evidence that indicates that this unorthodox use of gamma is linked to the phonetic evolution of the sound represented by gamma (i.e. /g/), which under some circumstances came to coincide with the sound once denoted by digamma (i.e. /w/), through a process of combined fricativisation (or spirantisation) and labialisation (or rounding) of the velar stop, triggered by its proximity to a rounded vowel. The set of the evidence comprises instances of gamma used to represent the /w/ sound found in: (1) the Lexicon of Hesychius; (2) documentary papyri from Egypt from the second century bc to the seventh century ad; (3) medieval Greek manuscripts; (4) a second century bc handwritten ostracon that contains a fragment by Sappho; (5) a second century ad inscription by poet Iulia Balbilla; (6) Ancient Greek transcriptions of Latin words; and (7) Modern Greek transcriptions of English words. Finally, in order to provide a parallel from historical linguistics, the theory of the development of the velar plosive /g/ into the voiced labiovelar approximant /w/ in Post-Classical Greek is compared with the opposite evolution of /w/ into /g/, which can be observed in some Romance languages, in Armenian, and in some Celtic and Germanic languages.

2 Background: A Brief History of the Letter Digamma

Ancient Greek inherited from Indo-European the sound /w/, the voiced labiovelar approximant also present in Latin: e.g. uicus, ‘neighbourhood’, ‘village’ (pronounced /ˈwi:kus/).3 Mycenaean Greek, the earliest attested form of Greek, still retained this sound at the end of the second millennium bc: e.g. wa-na- ka, corresponding to Attic ἄναξ, ‘lord, master’, and Doric ϝάναξ (pronounced /ˈwanaks/).4 In the mid eighth century bc, when alphabetic Greek is first attested, almost all of the Greek dialects with the exception of Attic-Ionic still retained the /w/ sound. It was represented by a sign commonly called digamma ( ϝ), due to its capital epigraphic shape (Ϝ), which resembles a capital gamma (Γ) with an added horizontal stroke—hence the name: ‘twofold gamma’.

3 Sihler 1995, 178 and 182. 4 Attested in glosses (e.g. D.H. 1.20.3-4) and inscriptions (e.g. IG VII 3187).

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 4 Camagni

The /w/ sound and its graphic representation ( ϝ) disappeared first from Attic-Ionic, in which the letter is already absent in the earliest attestations of this dialect.5 Following the conquests of Alexander the Great, Attic-Ionic came to constitute the base for Koine Greek, a trans-regional language that substituted most of the Greek dialects, and became the main variety of Greek. For this reason, Koine Greek and the later forms of Greek lack the /w/ sound, while the letter digamma remained in use to denote only the numeral six, although in a different graphic variant (ŝ). In contrast to Attic-Ionic, the digamma is attested in most of the Greek dialects, primarily in inscriptions. The letter digamma and the sound it represented disappeared first in consonant clusters; afterwards between vowels; and lastly in initial position before a vowel, where it persisted well into the Post-Classical period, as late as the second century bc. In addition, a relic of the sound uniquely survived to date in Tsakonian, a rural Laconian dialect.6 Here, it can be observed for example in βάννε, ‘lamb’, pronounced /ˈvane/,7 where /v/ is an evolution of ancient /w/, attested in Cretan ϝαρήν, ‘lamb’, and Mycenaean we-re-ne-ja, ‘made of sheepskin’.8

3 What Ancient Grammarians Knew about Digamma

In order to understand how the sound /w/ came to be denoted by gamma, it may be useful to observe to what extent the ancient grammarians were acquainted with the digamma, and with the words that retained this sound— because the alternative forms of denoting the sound /w/ are likely to have originated from a gap in the grammarians’ knowledge about digamma. The digamma did not abruptly disappear from the Greek language. Its practical use died out alongside the writing of Greek dialects. As late as the second and third century ad, it is still possible to find—albeit rarely—instances of digamma in literary papyri containing Aeolic and Doric poetry.9 At any rate, from the first century bc until at least the sixth century ad, the ancient grammarians preserved some theoretical knowledge about the digamma. Observations about digamma are made, for example, by Dionysius

5 Sihler 1995, 182; Colvin 2007, 37. 6 On Tsakonian see Pernot 1934. 7 Comparable to βάννεια, ‘lamb meat’ (Hsch. β 195). 8 Sihler 1995, 182-183. 9 E.g. Corinna in P.Berol. inv. 13284 (2nd c. AD); Corinna in P.Oxy. 12.2370 (ca. 200 AD); Sappho in P.Oxy. 1.7 (3rd c. AD).

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 5 of Halicarnassus,10 Apollonius Dyscolus,11 Quintilian,12 Priscian,13 and the Byzantine scholiast to Dionysius Thrax.14 These observations suggest that the use of digamma was already obsolete, mostly a relic of the Doric and Aeolic poetry, hence worthy of explanatory remarks in treatises on historical grammar. A recently discovered passage of Galen’s On Simple Drugs, found in an Italian codex (Vat. Urb. Gr. 67, ff. 239v-240r), contains some philologically correct remarks on the use of digamma, attributed to Galen himself.15 Here, the author states that in his time the digamma was still present in ‘accurate copies’ (ἐν τοῖς ἀκριβέσιν ἀντιγράφοις) of Aeolic and Doric authors; and the sound of this letter corresponds to Latin u (ου in the Greek text), of names such as Valerius and Seuerus. If the attribution of the passage to Galen is correct, the original text should date to the second century ad. Therefore, his mention of book copies containing the digamma is consistent with the instances found in the aforementioned literary papyri. Nevertheless, despite some correct theoretical knowledge about digamma, the ancient grammarians are often mistaken about the practical use of this letter, or about the words that contained it. For example, some authors are persuaded that digamma had applied generically to those words that in their time began with a vowel; and therefore they prefixed a digamma to words that—from an etymological point of view—did not have it, or contained it in internal position. For example, the first century bc historian Dionysius of Halicarnassus states:

σύνηθες γὰρ ἦν τοῖς ἀρχαίοις Ἕλλησιν ὡς τὰ πολλὰ προτιθέναι τῶν ὀνομάτων, ὁπόσων αἱ ἀρχαὶ ἀπὸ φωνηέντων ἐγίνοντο, τὴν ου συλλαβὴν ἑνὶ στοιχείῳ γρα- φομένην. τοῦτο δ’ ἦν ὥσπερ γάμμα διτταῖς ἐπὶ μίαν ὀρθὴν ἐπιζευγνύμενον ταῖς πλαγίοις, ὡς Ϝελένη καὶ ϝάναξ καὶ ϝοῖκος καὶ ϝαὴρ καὶ πολλὰ τοιαῦτα.16

For it was the custom of the ancient Greeks generally to place before those words that began with a vowel the syllable ου (pronounced /u/), written with one letter. This was like a gamma, formed by two oblique

10 D.H. 1.20.3-4. 11 Apollonius Dyscolus (GG I 77) cites a fragment by Alcaeus (fr. 649a Lobel-Page) with ϝέθεν (γέθεν in its manuscript), genitive singular of the personal pronoun. 12 Inst. 1.4.8. 13 Inst. 1.18-23 (2.15-18 Keil). 14 Sch. D. Thr. 34.20 Hilgard. 15 For details and discussion see Martelli 2012. 16 D.H. 1.20.3-4.

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 6 Camagni

lines joined to one upright line, such as Ϝελένη, ϝάναξ, ϝοῖκος, ϝαὴρ and many such words.

Here the letter digamma is not mentioned by name, and the author refers to it as the ‘syllable ου’, clearly on the basis of its sound alone. Furthermore, the author wrongly prefixes the digamma to certain words beginning with a vowel.17 Moreover, the Lexicon of Hesychius defines digamma simply as ‘a letter’, or ‘a sound’.18 We cannot reliably date this gloss, which could have been written at any point between the mid first and the sixth century ad, but the fact that the name appears as a dictionary entry is a sign that digamma was unknown to most Greek speakers, as the Lexicon does not generally include letter names. The only other letter here glossed is lambda (λάβδα), which is probably included because it has also another meaning.19 In addition, in contrast to the succinct definition of digamma, lambda is precisely described as a liquid consonant (στοιχεῖον ὑγρὸν καὶ ἀμετάβολον). This difference of treatment may indicate the limited familiarity with digamma possessed by the compiler(s) of the Lexicon. To summarise, digamma is attested in literary papyri containing Aeolic and Doric poetry until the second and third centuries ad. However, according to Galen, in the second century ad digamma could be found only in ‘accurate copies’ of the lyric poets. This circumscription may explain why attestations of digamma are rare and inconsistent in these centuries, to the extent that some literary papyri contain at the same time words with digamma, words in which the sound of digamma is represented by other letters, and words from which the digamma is omitted altogether.20 After this period, digamma disappears, together with dialectal poetry, which survives only in the form of rare lines scat- tered within the works of grammarians. Then, other letters become the norm to denote the sound /w/. Nevertheless, from the first century bc onwards, ancient grammarians persevere and share some theoretical knowledge—albeit imperfect—about the letter digamma. Within this philological tradition, a few dialectal words are often quoted as examples, such as ( ϝ)άναξ, ( ϝ)οινος, and (Ϝ)ελένα. However, neither the theoretical knowledge nor the words quoted are always philologically correct—at least from our modern point of view. This imperfect knowledge may have led the lexicographers/scribes who recorded

17 E.g. Ϝελένη for Ἑλήνη , ‘Helen’. 18 Hsch. δ 1481. 19 Hsch. λ 8. 20 P.Paris. 71 = P.Louvr. E 3320 (Alcm. fr. 1 Page-Davies): ϝανακτα (l. 6), εργα (l. 35) for ϝέργα, αυηρομεναι for ἀϝηρομέναι (l. 63).

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 7 the dialectal words to fail to recognise the sound /w/ of these words as the sound once represented by digamma; and triggered the replacement of digamma with other letters, which at that time roughly conveyed the /w/ sound, as will be explained in the rest of the discussion.

4 Graphic Representations of /w/ in Greek (β, ο, υ, ου)

In the history of the Greek language, the sound /w/ has been represented in different ways, owing to the instability of its use and pronunciation. As dis- cussed above, the most ancient sign employed for it in alphabetic Greek was the letter digamma ( ϝ). However, Attic-Ionic and Koine Greek were devoid of this sound, and oblivious of its traditional representation; after the Koine had largely replaced all the other dialects, digamma became obsolete. As a result, in order to write the poetic and dialectal words that retained the /w/ sound, scribes and grammarians employed other letters of the alphabet. Following a phono-acoustic principle, they used letters commonly used to represent a familiar sound to reproduce what was then an unfamiliar one. This principle is followed when digamma is replaced, for example, with beta (β), upsilon (υ), omicron (ο), and omicron-upsilon (ου); and the aim of this discussion is to argue that the same principle underlies the use of gamma (γ) for this purpose. One of the most common choices to replace digamma is the letter beta (β). As early as the third century bc the sound of this letter had evolved from voiced plosive bilabial /b/ to voiced fricative bilabial /β/.21 Therefore, when Attic-Ionic and Koine Greek writers transcribed dialectal words containing the /w/ sound, the letter beta was among the favourite substitutes. This transcription implied a compromise between the sound /β/ of beta, and the sound /w/ of digamma. Thus, for example, the Aeolic name of the ‘rose’, which on etymological grounds is believed to have been pronounced /ˈwrodon/,22 was written by scribes and grammarians of the first and second century ad as βρόδον, pronounced /ˈβrodon/.23 In addition, other letters were employed in place of digamma, such as omicron (ο) and upsilon (υ): e.g. Ὀαξός24 for Cretan Ϝάξος, the city of Axos; Οἴτυλος/

21 Schwyzer 1939, 207-208; Horrocks 2010, 170. 22 In contrast to Attic-Ionic ῥόδον; cf. EDG 1289-1290. 23 E.g. βρόδον in Apollonius Dyscolus (GG I 157); βρόδοισι in P.Oxy. 21.2288. 24 Attested e.g. in Hdt. 4.154.

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 8 Camagni

Βοίτυλος/Βίτυλα,25 the city of Otylon, possibly from *Ϝίτυλος;26 ἀυάτᾱ,27 Aeolic for ἄτη, ‘bewilderment’; ναῦος,28 Aeolic for ναός, ‘temple’, akin to Laconian ναϝôν (gen. pl.).29 Similarly, these letters were also employed to transliterate the /w/ sound in Latin, represented by the Latin semivowel u.30 For example, in the Jewish catacombs of Rome,31 among the inscriptions we can find the word βιξιτ, which transcribes Latin uixit, ‘(he) lived’.32 More extensive evidence is provided by the Greek papyri produced during the Roman rule over Egypt. In these, beta (β) is often interchanged with omicron-upsilon (ου) and upsilon (υ) to represent the /w/ sound of various Latin words: e.g. Σεβήρου, Σεουήρου, and Σευήρου for Latin Seuerus;33 Βίκτωρ for Latin Victor;34 Φλαβία for Latin Flauia;35 βιάτικον for Latin uiaticum, ‘travel money’;36 and beta (β) is used also in combinations with omicron-upsilon (oυ): e.g. Φλαουβίου for Latin Flauius;37 and Οὐβαλέρι(ο)ς for Latin Valerius.38 These examples show that there was no standard for transcribing the /w/ sound, for dialectal Greek words or for foreign ones. Therefore, the choice of the letter employed to represent this sound remained at the discretion of the scribes.

5 Gamma for /w/

A few sources offer instances of another letter used to represent the /w/ sound in ancient Greek, namely gamma (γ). Most of these instances come from the Lexicon of Hesychius;39 a few more are sparsely attested in Greek literature: in

25 Str. 8.4.4; Ptol. Geog. 3.16.22. 26 Ahrens 1843, 46; 51. 27 Attested e.g. in Alc. fr. 70 Lobel-Page, and Pi. P. 2.28, 3.24. Confirmed by Prisc. Inst. 1.36-2.28 Keil, who states that the Aeolians pronounced Greek υ as Latin u, i.e /w/. 28 E.g. Alc. fr. S262 Page; Sapph. fr. 2 Lobel-Page; IG XII2 6. 29 IG V1 1564. 30 Sihler 1995, 182. 31 Built in the 2nd-3rd c. ad, the Jewish catacombs of Rome were used until the 5th c. ad. 32 Allen 1968, 29. 33 E.g. SB 21.9332 (199 ad); BGU 1.145 (213 AD); and BGU 1.98 (211 ad). 34 E.g. SB 12.2155 (481 ad). 35 E.g. BGU 9.1893 (149 ad). 36 E.g. BGU 2.423 (2nd c. ad). 37 SB 2.4122 (120 ad). 38 OMich. 1.203 (336 ad). 39 The Lexicon is believed to date to the 5th or 6th c. ad, consisting of miscellaneous material from earlier word collections as early as the 1st c. ad. The only surviving codex (Venice, Marc. Gr. 622) dates to the 15th c. ad.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 9 a fragment of the poet Sappho,40 in a poetic inscription by Iulia Balbilla,41 and in a couple of glosses from a passage attributed by Eustathius to the grammarian Herodian.42 There is general agreement that gamma represents an alternative to the other letters employed to substitute digamma, although the reason for its use is still unclear.43 The use of the letter gamma in place of digamma has been explained as the result of a scribal mistake.44 According to this hypothesis, in an age in which the digamma was largely unknown to Greek scribes, instances of digamma— most likely in its capital form (Ϝ)—found in manuscripts may have been mis- interpreted as gamma (Γ), due to the similarity of their shapes. In this way, words with gamma in place of digamma would have entered the lexicographical tradition (e.g. the Lexicon of Hesychius) ordered alphabetically among entries beginning with gamma. In contrast to this explanation, Masson believes that this use of gamma is a scribal convention, owing to the systematicity of its attestations.45 This way of transcribing digamma would have developed in a specific ‘school’ of grammarians, and established itself in parallel to the other options, such as beta (β) and upsilon (υ). Masson refutes the hypothesis of the scribal mistake and is persuaded that this type of gamma is not a phonetic transcription of the sound behind digamma. However, he fails to provide any explanation for its use. In the nineteenth century, one hypothesis employed to justify this peculiar- ity was that certain Greek dialects had mutated the pronunciation of /w/ into a velar plosive /g/, citing as a parallel a similar evolution of the Latin semivowel u into the Gallo-Romance languages.46 Similarly, Milman Parry, pioneer of the oral-formulaic composition theory applied to the Homeric poems, suggests that “the easier explanation” for gamma used in placed of digamma “is that the grammarians had found, or rather heard of, some regional Aeolic dialect in which the digamma had survived in a velarised form, and that they concluded that this was the actual sound which had been used in Lesbian poetry”; and

40 Sappho lived between the 7th and 6th c. bc. The fragment in question dates to the 2nd c. bc. 41 130 ad. 42 The 12th c. ad scholar Eustathius (Commentarii ad Homeri Odysseam 2.281.10-11), possibly quoting the 2nd c. ad grammarian Herodian (ὁ τεχνικός, GG III 2.429), explicitly refers to γρῖνος, ‘skin’, and γρίντης, ‘tanner’, as equivalent to Attic ῥινός, ‘skin’, ‘hide’, written with the addition of the letter gamma. Both words derive from an original with initial /w/ sound, comparable to Mycenaean wi-ri-no, ‘cow-hide’ (EDG 1287 s.v. ῥινός). They are also attested in the Lexicon of Hesychius (γ 918, 919). 43 DELG 785 s.v. οἶνος; GEW 365 s.v. οἶνος; Masson 1990; Dickey 2007, 108 n. 1. 44 E.g. Sophocles 1854: 21-22; Clayman 1978, 396; Sihler 1995, 183. 45 Masson 1990, 208-211. 46 Ahrens 1843, 52.

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 10 Camagni that Iulia Balbilla also employed gamma in place of digamma “in accordance with such doctrine”.47 In partial agreement, on the one hand, with Masson’s belief that this use has its origin in grammatical or lexicographical practices, and, on the other hand, reversing the phonetic shift theory that imagines a velarisation of /w/, I will present here my argument for discarding the theory of the scribal mistake. I will also provide evidence indicating that this unorthodox use of gamma is linked to the phonetic evolution of the sound represented by gamma (voiced velar plosive /g/), which under some circumstances came to coincide with the sound once denoted by digamma (voiced labiovelar approximant /w/) through a process of combined fricativisation (or spirantisation) and labialisation (or rounding) of a velar stop, triggered by the proximity with a rounded vowel.

5.1 The Glosses of Hesychius The main source for words in which gamma is used to represent /w/ is the Lexicon of Hesychius.48 Here, words characterised by this feature are recorded throughout the whole collection, and their identification is not always straightforward. The easiest to spot are those words that begin with gamma and are matched by their standard Greek equivalent without this initial letter, such as γοῖνος· οἶνος, ‘wine’,49 and γέαρ· ἔαρ, ‘spring’.50 Nevertheless, some of these equivalences should be treated with care, and analysed case by case, especially if they have no parallel elsewhere. Other instances need some phonetic adjustment, such as γῖπον· εἶπον, ‘(I) said’,51 γάδεται· ἥδεται, ‘(he) enjoys’.52 Still others display this kind of gamma in internal position, which makes them less obvious in the plethora of the glosses of the Lexicon, such as ἀγατᾶσθαι· βλάπτεσθαι53 and ἀγάτημαι· βέβλαμμαι,54 ‘to disable’, ‘to baffle’, ‘to distract’, from the same root ἀϝατ- as Attic ἄτη, ‘bewilderment’.55 In certain cases some etymological work is needed to reveal this feature: for instance, there is a series of

47 Parry 1934, 142. 48 One of the most complete lists of Hesychian glosses that bear gamma in place of digamma is provided by Ahrens 1843, 53-56, whose account remains one of the most relevant studies on the question. 49 Hsch. γ 784; cf. Doric ϝοῖνος, ‘wine’. 50 Hsch. γ 224; cf. Latin uēr, ‘spring’. 51 Hsch. γ 570; cf. Cretan ϝεῖπαι (IC IV 72 8.15), and Elean ϝέπος (IvO 9). 52 Hsch. γ 22; cf. Cretan ϝάδυος, ‘pleasant’ (IC I xvii 10), and βάδομαι, ‘I am pleased’ (Hsch. β 37). 53 Hsch. α 361. 54 Hsch. α 362. 55 Vide supra Aeolic ἀυάτᾱ.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 11 glosses with gamma related to ‘sun’, ‘sunlight’: γέλαν· αὐγὴν ἡλίου, ‘sunlight’,56 γέλας· αὐγάς, ‘(sun) rays’,57 γελεῖν· λάμπειν. ἀνθεῖν, ‘to illuminate’, ‘to shine’,58 and γελοδυτία· ἡλιοδυσία, ‘sunset’;59 these are linked to a series of glosses with beta, around the same semantics:60 βέλα· ἥλιος καὶ αὐγή, ὑπὸ Λακώνων, ‘sun, and sunlight (in Laconian)’,61 and βελάσεται· ἡλιωθήσεται, ‘(he) will be illuminated’.62 In these, the original /w/ sound is represented by gamma (γ) in one series, and by beta (β) in the other, and all of them would stem from a Proto-Greek form with original digamma. The source of the Hesychian glosses with gamma in place of digamma is unknown. The most recent edition of the Lexicon63 ascribes most of them to the second century ad grammarian Diogenianus of Heraclea,64 author of Παντοδαπὴ λέξις, All Types of Sayings, an abridged version of the lexicographical works of Zopyrion and Pamphilus, which are considered the main models of the Lexicon of Hesychius.65 The fact that the glosses with gamma in place of digamma are placed in the correct alphabetical order, under gamma, could be in itself considered a reason not to consider this a scribal mistake, as it would involve a twofold error: mistaking digamma for gamma, and re-alphabetising all these entries, including those that display gamma in the word. Nevertheless, this paper will not focus on this aspect to substantiate its argument, and will provide more compelling evidence.

5.2 Spelling Mistakes in Greek Papyri and Manuscripts The first and most crucial reason not to consider the use of gamma in place of digamma as a mistake is its use to represent the sounds /w/ and /u/ in Greek

56 Hsch. γ 285. 57 Hsch. γ 290. 58 Hsch. γ 298. 59 Hsch. γ 309. 60 Ahrens 1843, 46; GEW I 458; DELG 320; EDG 385. 61 Hsch. β 476. 62 Hsch. β 486. 63 Latte and Cunningham 2018. 64 Masson 1990, 210. 65 The main core of the Lexicon is believed to be a word collection called Περι γλωσσῶν και ὀνοματων (according to Ath. 14.63) or Περι γλωσσων ἤτοι λεξεων (according to the Suid. π 142). Zopyrion is said to have compiled the first part (Α–Δ), while the 1st c. ad grammarian Pamphilus completed it (Ε–Ω) (Suid. δ 1140, π 1, 142). This material was later abridged by the 2nd c. ad Roman lexicographer Lucius Iulius Vestinus in the Ἑλληνικα ὀνοματα (Suid. ο 835), and then reused in the late 2nd c. ad by Diogenianus in the Λεξις παντοδαπή (Suid. δ 1140).

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 12 Camagni papyri, dating to between the second century bc and the seventh century ad, and in some medieval manuscripts. This spelling is motivated by one specific shift—first noticed by Gignac66—of the voiced velar plosive /g/, which became a voiced labiovelar approximant /w/ in certain contexts, mostly in intervocalic position after a rounded vowel. This shift of the voiced velar plosive/g/ is not isolated, as it also became (1) voiced velar fricative /ɣ/, in most positions; (2) voiced palatal fricative /ʝ/, before high front vowels /e/ and /i/;67 while it remained /g/ exclusively after a nasal.68 As the sound of /g/ changed, the letter gamma that denoted it was employed to represent this set of allophones. The shift of gamma to denote the /w/ sound is substantiated by instances of this letter wrongly omitted before or after the /u/ sound (e.g. ὁμολοῦντα69 for ὁμολογοῦντα, προαούσης70 for προαγούσης, σεβαστοῦ Ερμανικοῦ71 for σεβα- στοῦ Γερμανικοῦ); and before or after the /w/ sound of the short diphthongs αυ and ευ (e.g. in the transcription of the Roman names Augustus and Augusta: Ἀούστου,72 Αὐούστα73 instead of Αὔγουστος). On the other hand, gamma is sometimes wrongly inserted after the /w/ sound, particularly to reinforce the diphthongs αυ and ευ, which were pronounced /aw/ and /ew/ at least since the mid fourth century bc.74 For example, Φλαυγίος75 for Latin Flauius (pronounced /fla:ˈwjus/ or /fla:ˈwius/), whereas somewhere else the same name is transcribed as Φλαουβίου,76 Φλαουίου,77 and Φλαβίος.78 Moreover, Εὐγεργέταις79 is attested instead of Εὐεργέταις, εὐγεργε- τημένος80 instead of εὐεργετημένος, εὐγαπόδεικτον81 instead of εὐαπόδεικτον,

66 Gignac 1970, 189; 1976, 74. 67 In documentary papyri, gamma is often wrongly added/omitted before front vowels (e.g. εοργῶι for γεοργῶι), and also wrongly inserted to denote the voiced palatal approximant /j/ (e.g. ἱγεροῦ for ἱεροῦ; Τραγιανοῦ for Latin Traianus), Gignac 1976, 72. 68 Horrocks 2010, 170. 69 P.Fam.Tebt. 13 (113-114 ad). Et similia: BGU 21.2044 (46 ad); P.Lond. 3.1166 (41-42 ad); P.Vind.Worp. 10 (143-144 ad); P.Muench. 1.14 (594 ad); P.Oxy. 12.1430 (324); BGU 11.2121 (81- 96 ad); P.Mon. 14.14 (594 ad); etc. 70 P.Oxy. 10.1265 (336 ad). 71 Misspelt twice in the same document, which is P.Flor. 3.353 (83-95 ad). 72 P.Oxy. 14.1716 (333 ad). 73 Stud.Pal. 20.68 (3rd c. ad); cf. Gignac 1976, 74 for further details. 74 Schwyzer 1939, 197; Horrocks 2010, 165. 75 Thrice in P.Oxy. 3.504 (early 2nd c. ad); SB 1.4179 (ca. 1st-3rd c. ad). 76 SB 2.4122. 77 BGU 1.288 (144-147 ad). 78 PSI 12.1254 (237 ad). 79 P.Tebt. 1.11 (119 bc). Similarly P.Tebt. 1.26 (114 bc). 80 P.Oxy. 3.487 (156 ad). 81 P.Cair.Isidor. 16.62 (296 ad).

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 13

σκευγη82 instead of σκεύη, and προσαγωρεύγω83 instead of προσαγορεύω. These examples show that αυγ/ευγ are common misspellings for αυ/ευ (pronounced /aw/ and /ew/), like αου/εου,84 αυου/ευου,85 αβ/εβ,86 and αυβ/ευβ.87 In addition, examples of this use of gamma are attested in some Byzantine manuscripts: e.g. πεζεύγειν for πεζεύειν, τοξεύγοντες for τοξεύοντες, ναυγάγια for ναυάγια.88 Moreover, the aforementioned codex containing a passage from Galen’s On Simple Drugs displays σέγερο for σεϝήρῳ, ‘Severus’.89 In all of these misspellings—contrary to the instances found in the Lexicon of Hesychius—gamma is not used in place of digamma, but rather to represent the /w/ sound. Its incorrect omission/insertion can only be motivated by the confusion between the sound of gamma and the sound of the second ele- ment of diphthongs αυ and ευ (pronounced /aw/ and /ew/), of ου (pronounced /u/, but also /w/ after a vowel), and of Latin u (pronounced /u/ and /w/).

5.3 Sappho The same type of misspelling attested in documentary papyri and manuscripts can be found in an ostracon from Egypt dating to the second century bc, which preserves a fragment of the Lesbian poet Sappho (fr. 2 Lobel-Page).90 The piece is believed to be either a personal copy of the poem or a school exercise, characterised by some dictation mistakes not unusual for the period and the region.91 At the end of the first line of the poem, located on the third line of the potsherd, we can find the word ναυγον. This is amended by the editors of the fragment as ναῦον,92 which is the Aeolic equivalent of Attic ναός, ‘temple’, and of Laconian ναϝôν. In this case, however, the letter gamma does not denote the Aeolic digamma, which is already represented by the letter upsilon (υ); it is a pleonastic reinforcement of the /w/ sound of the diphthong αυ, of the same kind attested in non-literary papyri. In addition, in the same fragment (l. 6) we can find βρόδοισι for ῥόδοισι, ‘by roses’, where beta is also employed to denote the sound /w/, and—from an

82 P.Berl.Sarisch. 20 (4th c. ad). 83 PSI 14.1430 (7th c. ad). 84 E.g. P.Mich. 5.261 (35 ad); BGU 16.2590 (25 bc). 85 E.g. BGU 6.215 (2nd c. ad); P.Oxy. 43.3093 (217 ad). 86 E.g. P.Lond. 4.1610 (709 ad). 87 E.g. SB 12.10841 (4th c. ad). 88 Crönert 1903, 91 n. 3. 89 The codex also has γελίνην for Ἑλήνη, ‘Helen’ (vide supra Ϝελένη for Ἑλήνη). 90 PSI 13.1300 (2nd c. bc). 91 Norsa 1937, 8-10. 92 This word is also attested twice in another Lesbian poet of the same period, Alcaeus (fr. S262 Page, and in Str. 9.2.29).

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 14 Camagni etymological point of view—it substitutes an original digamma. This shows that there was no standard for denoting /w/, and that different ways (i.e. upsilon and beta) could coexist in the same text once digamma had become obsolete. It is difficult to estimate whether the spellings employed by the scribe to replace the digamma were the result of a scribal convention learned before- hand, or an impromptu rendering of the sound /w/, possibly heard during a dictation. The fact that the upsilon-spelling of ναῦον and the beta-spelling of βρόδοισι are elsewhere attested may induce us to believe that they both con- formed to a set of rules, perhaps taught in grammar schools. However, given the spelling mistakes contained in the fragment, among which the pleonastic gamma of ναυγον, we may hypothesise that the scribe was transcribing the sounds as he was hearing them, perhaps conforming to such rules as best as he could. Moreover, we may also hypothesise that he was not aware that upsilon and beta here replaced an original digamma, as this would have required fairly deep knowledge of the Ancient Greek historical grammar that only profession- al grammarians seem to hold. If the use of gamma to denote the /w/ sound resulted from the impromptu transcription of the Aeolic word ναῦον dictated to a Koine Greek speaker, then we may imagine that a similar phenomenon may be the cause of the replacement of digamma with gamma in the dialectal glosses of the Lexicon of Hesychius.

5.4 The Inscription by Iulia Balbilla Two instances of gamma in place of digamma, found in a second century ad inscription from Egypt, are particularly compelling, because the type of document on which they are recorded is immune to the scribal corruption that may have affected the Lexicon of Hesychius. Iulia Balbilla was a Roman noblewoman, the daughter of the exiled prince of Commagene Gaius Iulius Archelaus Antiochus Epiphanes, a Roman client monarch and influential first century ad politician. She was born in Rome and had a cosmopolitan upbringing. She lived in Athens and in Alexandria, where her mother came from; she was a well-educated poet and a friend of Vibia Sabina, the wife of the emperor Hadrian. She arrived in Egypt within the en- tourage of the empress when the entire imperial court was touring the country. In November 130 ad they were near Thebes and payed a visit to the two gigan- tic statues of Pharaoh Amenhotep III, known as the colossi of Memnon. One of these statues was renowned because the wind, blowing through a crack that had been caused by an earthquake, used to howl in a particularly impressive way, making it a popular destination for foreign tourists, who were accustomed to mark their presence there by engraving their name or some lines of verse on

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 15 them. For the occasion, Iulia Balbilla composed four poems (of a total of forty- five lines), and had them inscribed on the left ankle and foot of the howling colossus.93 The elegiac distichs by Iulia Balbilla are composed in literary Lesbian, in the style of Alcaeus and Sappho. Aeolic was a fashionable dialect for poetry at that time, and Sappho a common model for female authors.94 In the first of these texts (inscr. 28 Bernard and Bernard), two words display a gamma instead of the expected digamma: γοί (v. 4) and γέ (v. 12). The former, the third person singular dative of the personal pronoun, is also attested in Hesychius (γ 778) glossed as αὐτῷ. It survives in Attic and in Epic Greek as οἷ (e.g. Od. 11.433), and as ϝοί in various dialects (e.g. GDI 2561). The latter, the accusative singular of the same pronoun, corresponds to ϝέ (Alcm. fr. 1 Page), more commonly attested as ἕ (e.g. Il. 1.236, 4.497). Forms of these pronouns with digamma preserved throughout the declension were known from literary Doric and Aeolic varieties—as we know from the second century ad grammarian Apollonius Dyscolus (Pron. 106a, 136b-c), who cites lines of Alcman, Sappho, and Corinna as examples.95 In addition, Iulia Balbilla employs other Aeolic words that retain the /w/ sound: εὐϊδε,96 aorist of *εἴδω, ‘I see’, to be confronted with ευιδον, ‘(I/they) saw’, attested in another Egyptian inscription,97 with γοίδημι, ‘I know’,98 and akin to Latin uideo, ‘I see’; and Αὐως,99 for ᾽Hώς, ‘Dawn’, also attested in Sappho,100 akin to Latin aurora, ‘dawn’. Once again, the coexistence of gamma and upsilon to represent the /w/ sound shows the lack of an orthographic standard in Aeolic poetry once the digamma had become obsolete. Balbilla’s poem is “sophisticated, mostly unrepetitive” and “metrically flawless”;101 the accuracy of the composition and the careful choice of rare Aeolic words, mostly from Sappho, Alcaeus, and Homer, bear witness to her quality as a writer;102 the poet was relatively proficient in the use of literary

93 Bernand and Bernand 1960, inscr. 28-31. 94 Brennan 1998, 219-220. 95 Alcm. fr. 7 Page; Sapph. fr. 164f. Lobel-Page; Corinn. fr. 7 Page. Codex Parisinus Graecus 2548 (1oth c. ad), which preserves the works of Apollonius Dyscolus, represents the letter digamma in these dialectal Greek fragments in various ways: with digamma ( ϝ), epsilon (ε), gamma (γ), and mu (μ) (see Apollonius Dyscolus Pron. 106a, 136b-c; GG I 82, 107). 96 Bernand and Bernand 1960, inscr. 28, v. 11. 97 IGChrEg 353. 98 Hsch. γ 781. 99 Bernand and Bernand 1960, inscr. 29, v. 1. 100 E.g. Sapph. fr. 123 Lobel-Page. 101 Brennan 1998, 221. 102 Cirio 2011, 79-80.

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 16 Camagni

Lesbian, although some of her Aeolisms are artificial creations and not actual dialectal words (e.g. παῖσι instead of πάντεσσι), probably due to an imperfect command of this dialect,103 which was no longer a living language but a Kunstsprache employed for a specific poetic genre, and accessible only through the study of a canon of ancient Aeolic authors. We may therefore imagine that her choice of representing the sound /w/ of an original digamma with upsilon (εὐϊδε and ὐΑ ως), and with gamma (γοί and γέ) was based on previous literary studies, and not the result of an impromptu decision—as it may have been the case for the aforementioned fragment of Sappho, possibly written under dictation. Nevertheless, it is difficult to ascertain whether Iulia Balbilla devised the orthography of these words herself. She could have adopted the orthography she learnt in school and from edited poetic texts, possibly being unaware that they represented an original digamma.104 Masson believes that this use of gamma is a scribal convention developed by grammarians to transcribe dialectal and poetic words.105 This would explain why the pronoun γοί employed by Iulia Balbilla is attested also in the Lexicon of Hesychius, since it is unlikely that its compiler(s) had seen Balbilla’s inscription. At any rate, the use of gamma in place of digamma by Iulia Balbilla cannot be explained by a misreading of the letter digamma, and the reason for this orthographic solution must reside in her grammatical education, and her knowledge of the Aeolic authors.

5.5 Use of Gamma to Represent the Latin Semivowel u In addition to the aforementioned instances found in Greek papyri of the Roman era, there are some further noteworthy cases of the letter gamma (γ) employed to transcribe the Latin semivowel u, which represents the sound /w/. Two of these instances come from the Lexicon of Hesychius: σεργοί· ἔλαφοι, ‘deers’,106 and γέντερ· ἡ κοιλία, ‘the belly’.107 The former, σεργοί, is a transliteration of Latin cerui, ‘deers’, where the initial sigma (σ) may be the reflex of the assibilation of Latin c, in its evolution from the classical Latin pronunciation /k/ towards a more Romance development, such as French cerf (pronounced /sɛʁ/), Catalan cérbol (pronounced /ˈsɛrbul/), Spanish ciervo (pronounced

103 Bubenik 1989, 98. 104 Thumb and Scherer 1959, 82-83. 105 Masson 1990, 209. 106 Hsch. σ 422. 107 Hsch. γ 382.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 17

/ˈθjerβo/), Tuscan Italian cervo (pronounced /ˈʃɛrvo/)—all from Latin ceruus, ‘deer’. The latter, γέντερ, is the transliteration of Latin uenter, ‘belly’.108 In addition, one more example is provided by the grammarian, lexicographer, and patriarch Photius, in the 9th century ad:

οἱ Ῥωμαῖοι … σέργους δὲ τοὺς δούλους ἐκάλουν, τοὺς δὲ μὴ δορυκτήτους, δου- λεύοντας δ’ ὅλως, φαμούλλους ἐκάλουν.109

The Romans … called the slaves serui; while they called famuli those that were not ‘won by the spear’ (i.e. war-slaves) but acted generally as servants.

Here we can observe the different treatment in Greek of Latin u, as a semivowel and a vowel: first, the Latin semivowel u (of serui) transcribed with Greek gamma (σέργους); secondly, the Latin vowel u (of famuli) transcribed with omicron-upsilon (φαμούλλους). Although it is difficult to determine whether Greek gamma (γ) here represents the sound /w/ or /v/, this example is a further piece of evidence that the use of gamma to represent a labiovelar consonant— rather than a velar—is not a mistake, but is motivated by its phonetics.110

5.6 Gamma for /w/ in Modern Greek The use of gamma to denote the sound /w/, first attested in some second century bc papyri, can still be seen in Modern Greek in the transcription of the English semivowel w. As it has been said before, the sound of gamma evolved into different sounds in the transition from Ancient to Modern Greek: /g/, /ɣ/, and /ʝ/. However, as this paper argues, in certain contexts the sound of gamma happened to coincide with the voiced labiovelar approximant /w/, once represented by digamma. This is eloquently shown by the Modern Greek transcription and pronunciation of many English names and toponyms with initial w, such as William ~ Γουίλιαμ, Wesley ~ Γουέσλεϊ, Wilson ~ Γουίλσον, Watson ~ Γουάτσον, Wayne ~ Γουέιν, Wolverhampton ~ Γουλβερχάμπτον, Winchester ~ Γουίντσεστερ.

108 Curtius 1879, 600. 109 Phot. Epistulae et Amphilochia 323. 110 The first century ad Roman orator Quintilian states that Greek gamma (called Aeolicum digammon) is equivalent to the Latin semivowel u, and uses as examples seruus and cer uum (Inst. 1.4, 12.10).

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 18 Camagni

5.7 Affinity of /w/ and /g/ in the Romance Languages A shift in the opposite direction, namely from the sound /w/ to /g/, a case of de- labialisation of a labiovelar consonant, can be found in a few Indo-European languages.111 Most notably, in the evolution of Latin words into some Romance languages and dialects of the Gallo-Roman branch. For instance: Latin uespa, ‘wasp’, becomes guêpe, ‘wasp’, in French; Latin uastare, ‘to make empty’, ‘to laid waste’, becomes gâter, ‘to spoil’, in French, and guastare, ‘to spoil’, in Italian; Latin uagina, ‘sheath’, becomes gaine in French, and guaina in Italian, in contrast to Spanish vaina, and Portuguese bainha, all with the same meaning. Moreover, from Latin uulpes ~ uolpes, ‘fox’, derive Old French goupil, Tuscan Italian golpe ~ golpa, and Lombard Italian gulp, ‘fox’, in contrast to standard Italian volpe, and Romanian vulpe, all ‘fox’.112 In addition, there are further examples of the shift of /w/ to /g/ in Armenian (e.g. garun, ‘spring’, akin to Latin vēr, from Proto-Indo-European *wes-r̥;113 and gaṙn ‘lamb’, akin to Skt. úran-, from Proto-Indo-European*wr̥h1ēn);114 in the Celtic languages (e.g. Welsh gwin, ‘wine’, from Latin uinum); and in the Germanic languages (e.g. Swedish lager, ‘laurel’, from Latin laurus). These examples show that the voiced velar plosive /g/ and the voiced labiovelar approximant /w/ can evolve, and come close to the point of coincid- ing in their articulation.

6 Two Hypotheses on the Disappearance of the Digamma from the Manuscripts

Before concluding this paper, I would like to share two hypotheses on the practicalities of how the letter digamma disappeared from the manuscript tradition. As stated by Galen (Vat. Urb. Gr. 67, ff. 239v-240r), up to the second century ad the letter digamma was still present in ‘accurate copies’ of the lyric poets. This means that some expert scribes and grammarians were familiar with the phonetic value of digamma to this date. The decline of this expert

111 The evolution /w/ → /g/ goes against the linguistic principle of directionality, well attested for /g/ → /w/; cf. Campbell 2013, 113-114. 112 Latin /w/ becomes also /b/, e.g. Venetian bolp ~ bulpe, ‘fox’; remains /w/, e.g. Lombard of Valtellina ulp, ‘fox’, (pronounced /ˈwulp/); and at times disappears, e.g. olp, ‘fox’, in Solaro language, a Rhaetian dialect spoken in a few alpine valleys around Trento, in the north of Italy. 113 Martirosyan 2009, 201 s.v. 114 Martirosyan 2009, 198-199 s.v.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 19 knowledge, and the ultimate disappearance of digamma from the manuscript tradition may have happened in two ways. (1) The existence of ‘accurate copies’ implies that also non-accurate copies were in circulation. We may imagine that these were produced for the benefit of those readers who were unfamiliar with the digamma and did not know how to read it. In such copies all digammas were substituted on purpose with other non-obsolete letters that all readers could recognise (e.g. ου, υ, β, γ). Finally, once the use of these alternative letters was institutionalised, it was prescribed in grammar texts and lexica (e.g. Hdn. GG III 2.174). This may also explain why certain words recur with the same letter in place of digamma. (2) Texts containing words with digamma (e.g. lyric poetry) were dictated to scribes or students who either were completely unfamiliar with this letter or occasionally failed to recognise its sound as such in specific words. An example of the former case may be Sappho’s fr. 2 Lobel-Page, in which the sound of digamma is written with either beta (βρόδοισι) or gamma (ναυγον). On the other hand, an example of the latter case may be Alcman’s fr. 1 Page-Davies, in which digamma is either written ( ϝανακτα), omitted (εργα), or substituted with upsilon (αυηρομεναι).

7 Conclusions

This paper has offered a review of the philological problem of the letter gamma (γ) used in place of digamma ( ϝ) in Ancient Greek. After sketching the history of digamma and its use in Ancient Greek, it has provided a sum- mary of the current theories on the question. Refuting the idea that this spelling is a scribal blunder developed in the medieval manuscript tradition of the Lexicon of Hesychius, the paper has gathered and analysed the occurrences of the phenomenon, and provided evidence to substantiate a new solution to the question. Through the analysis of old and new evidence, found in Ancient Greek literary and documentary papyri, lexicography, literature, and medieval manuscripts, the paper has concluded that this use of gamma is not a scribal blunder. It is rather motivated by a specific phonetic development of the sound represented by the letter gamma. This sound (voiced velar plosive /g/) underwent a twofold process of fricativisation (or spirantisation), extensively documented in Greek (/g/→/ɣ/), and of labialisation (or rounding), triggered by its proximity to a rounded vowel (/ɣ/→/w/). The result is a shift from voiced velar plosive /g/ to voiced labiovelar approximant /w/, which occurred as early as the

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 20 Camagni second century bc in certain contexts, above all in intervocalic position after a rounded vowel. Consequently, the sound represented by the letter gamma came to coincide under some circumstances with the sound that was once represented by digamma, and was thus employed by scribes once digamma had become obsolete. This is demonstrated by instances of gamma (γ) wrongly omitted or added next to a rounded vowel or semivowel (/o/, /u/, /w/), in non-edited docu- ments dating to between the second century bc and the seventh century ad. Moreover, gamma in place of digamma is found twice in one of the second century ad inscriptions by the poet Iulia Balbilla, which are by their nature immune to the scribal corruption that has been hypothesised for the medieval codices of the Lexicon of Hesychius. Furthermore, there are instances of gamma employed to transcribe Latin u (/w/) in Ancient Greek, and English w in Modern Greek. To summarise, the letter gamma in place of digamma is a phonetically- motivated spelling solution employed to represent a letter that had become obsolete, and a sound that was no longer present—as an independent phoneme—in the phonetic repertoire of Post-Classical Greek speakers. This use of gamma, first attested in some second century bc papyri from Egypt, is coherent with other Post-Classical scribal practices that employed various letters (e.g. β, υ, ου) in place of digamma. The sound of all these letters, at different stages and in different contexts, happened to coincide with the sound /w/ once represented by digamma.

Bibliography

DELG = Chantraine, P. (1968-1980). EDG = Beekes, R.S.P. (2010). GEW = Frisk, H. (1954-1972). IEW = Pokorny, J. (1959). LSJ = Liddell, H.G. et al. (1940).

Ahrens, H.L. (1843). De Graecae linguae dialectis, Vol. 2: De dialecto dorica. Göttingen. Allen, W.S. (1968). Vox Graeca. A Guide to the Pronunciation of Classical Greek. Cambridge. Beekes, R.S.P. (2010). Etymological Dictionary of Greek (I-II). Leiden/Boston. Bernand, A., and Bernand, É. (1960). Les inscriptions grecques et latines du Colosse de Memnon. Paris.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access The Use of Gamma in Place of Digamma in Ancient Greek 21

Brennan, T.C. (1998). The Poets Julia Balbilla and Damo at the Colossus of Memnon. CW 91, pp. 215-234. Bubenik, V. (1989). Hellenistic and Roman Greece as a Sociolinguistic Area. Amsterdam/ Philadelphia. Campbell, L. (2013). Historical Linguistics. An Introduction. 3rd ed. Edinburgh. Chantraine, P. (1968-1980). Dictionnaire etymologique de la langue grecque. Paris. Cirio, A.M. (2011). Gli epigrammi di Giulia Balbilla. Ricordi di una dama di corte e altri testi al femminile sul Colosso di Memnone. Lecce. Clayman, D.S. (1978). The Meaning of Corinna’s Ϝεροῑα. CQ 28, pp. 396-397. Colvin, S. (2007). A Historical Greek Reader. Mycenaean to the Koiné. Oxford. Crönert, W. (1903). Memoria Graeca Herculanensis. Leipzig. Curtius, G. (1879). Grundzüge der griechischen Etymologie. Leipzig. Dickey, E. (2007). Ancient Greek Scholarship. A Guide to Finding, Reading, and Understanding Scholia, Commentaries, Lexica, and Grammatical Treatises, from Their Beginnings to the Byzantine Period. Oxford/New York. Frisk, H. (1954-1972). Griechisches Etymologisches Wörterbuch. Heidelberg. Gignac, F.T. (1970). The Pronunciation of Greek Stops in the Papyri. TAPhA 101, pp. 185-202. Gignac, F.T. (1976). A Grammar of the Greek Papyri of the Roman and Byzantine Periods. Phonology. Milano. Horrocks, G. (2010). Greek. A History of the Language and its Speakers. 2nd ed. Chichester. Latte, K., and Cunningham, I.C., eds. (2018). Hesychii Alexandrini Lexicon, Vol. 1: A–Δ. Berlin/Boston. Liddell, H.G. et al. (1940). A Greek-English Lexicon. 9th ed. Oxford. Martelli, M. (2012). Galeno grammatico sui nomi stranieri e il digamma. Un passo in- edito dal IX libro del trattato sui medicamenti semplici. Aion 34, pp. 131-147. Martirosyan, H.K. (2009). Etymological Dictionary of the Inherited Armenian Lexicon. Leiden. Masson, O. (1990). Remarques sur la transcription du w par bêta et gamma. In: H. Eichner and H. Rix, eds., Sprachwissenschaft und Philologie. Jacob Wackernagel und die Indogermanistik heute. Kolloquium der Indogermanischen Gesellschaft vom 13. bis 15. Oktober 1988 in Basel, Wiesbaden, pp. 202-212. Norsa, M. (1937). Versi di Saffo in un ostrakon del sec. II a.C. ASNP 6, pp. 8-15. Parry, M. (1934). The Traces of the Digamma in Ionic and Lesbian Greek. Language 10, pp. 130-144. Pernot, H. (1934). Introduction à l’étude du dialecte tsakonien. Paris. Pokorny, J. (1959). Indogermanisches etymologisches Wörterbuch. (3 vols). Bern/ München.

Mnemosyne (2020) 1-22 | 10.1163/1568525X-bja10018 Downloaded from Brill.com09/30/2021 01:54:17PM via free access 22 Camagni

Schwyzer, E. (1939). Griechische Grammatik. Vol. 1: Allgemeiner Teil. Lautlehre. Wortbildung. Flexion. München. Sihler, A.L. (1995). New Comparative Grammar of Greek and Latin. New York/Oxford. Sophocles, E.A. (1854). History of the Greek Alphabet and Pronunciation. Cambridge. Thumb, A., and Scherer, A. (1959). Handbuch der griechischen Dialekte. Heidelberg.

10.1163/1568525X-bja10018 |Downloaded Mnemosyne from Brill.com09/30/2021 (2020) 1-22 01:54:17PM via free access