Atticist Lexica and the Pronunciation of Greek

Citation with persistent identifier:

Vessella, Carlo. “Atticist Lexica and the Pronunciation of Greek.” CHS Research Bulletin 3, no. 1 (2014).

Atticism and pronunciation

1§1 This paper argues that some of the Atticist lexica written between the second and third centuries CE contain prescriptions that reveal ideas about the correct pronunciation of Greek among the educated elites of the Imperial period. The same individuals who thought there was a pure variety of Greek to write—the Atticists—also thought that there was a pure pronunciation of Greek one should adopt in speaking.

1§2 Atticism in Greek literature has long been studied as a set of linguistic choices in vocabulary, morphology, and syntax[1]. There are certainly very good reasons to privilege these aspects of Atticism. We know ancient literature only in written form, and certainly writers of the Imperial period composed their works knowing that they would circulate mostly in writing. But oratory—despite hard copies of speeches that could be purchased on the book market—was still performed orally.

1§3 Correct pronunciation of Greek was certainly a key component of good delivery for any orator who could claim to have received respectable rhetorical training: literary criticism of the Imperial period rejects bad habits in composition as appalling to the hearers, not the readers[2]. In this respect, Philostratus’ remarks on Pausanias of Cesarea are of the greatest interest:


ὁ δὲ Παυϲανίαϲ ἐπαιδεύθη μὲν ὑπὸ Ἡρώδου καὶ τῶν τοῦ Κλεψυδρίου μετεχόντων εἷϲ ἐγένετο … ἐϲ πολλὰ δὲ ἀναφέρων τῶν Ἡρώδου πλεονεκτημάτων καὶ μάλιϲτα τὸ αὐτοϲχεδιάζειν ἀπήγγελλε δὲ αὐτὰ παχείᾳ τῇ γλώττῃ καὶ ὡϲ Καππαδόκαιϲ ξύνηθεϲ, ξυγκρούων μὲν τὰ ϲύμφωνα τῶν ϲτοιχείων, ϲυϲτέλλων δὲ τὰ μηκυνόμενα καὶ μηκύνων τὰ βραχέα, ὅθεν ἐκάλουν αὐτὸν οἱ πολλοὶ μάγειρον πολυτελῆ ὄψα πονήρωϲ ἀρτύοντα.

Philostratus, Lives of the Sophists 594

1§4 Despite his training in one of the most prestigious rhetorical milieus of his time, and even though his rhetorical skills likened him to his teacher Herodes Atticus, Pausanias had a heavy, unpleasant accent: bad pronunciation could seriously impair the effectiveness of otherwise excellent rhetorical training.

1§5 The link between rhetorical training and proficiency in Attic is explicit in other passages of Philostratus’ work. Apollonius of Tyana, portentous also in many other respects, was even more so because his Attic was flawless even before his actual rhetorical training started—all the more surprising as Apollonius had exactly the same regional background as Pausanias (Philostratus, Life of Apollonius 1.7). Latin speakers, too, would learn (Attic) Greek, and some would even receive praise for their mastery of the language (Aelianus, a Roman, ἠττίκιζε, Philostratus, Lives of the Sophists 624, Favorinus, a Gaul, could ἑλληνίζειν, Philostratus, Lives of the Sophists 489).

1§6 “In centuries of the Roman Empire ability to use the classical language (rather like the use of ‘BBC English’ until very recently) came to be regarded as a conspicuous and exclusive badge of class membership”[3]. The simile is particularly fortunate: as BBC English is distinctive also because of its specific phonological traits, we may expect atticising Greek, too, to be characterised by a special phonology, and literary sources of the period prove that there was one.

1§7 Recreating the usage of Classical Attic could be relatively easy in writing: a written copy of Demosthenes is helpful for imitating the orator’s word-choice, syntax, or morphology, all of them bestowed with the mark of good Attic usage. But the question of how good Attic pronunciation could be imitated is an open one. Hearing Demosthenes again was certainly not an option. Audiences of Roman times would have had to invent their own criteria of linguistic purity to judge which was the purest accent, in a time when the number of speakers of Greek as a second language was high, and Greek itself was undergoing major sound changes. There are some hints that the Greek spoken in Attica in the second century CE was singled out as a model of linguistic purity for pronunciation (cf. the episode of Herodes Atticus and Agathion in Philostratus’ Lives of the Sophists 553), and this is confirmed by the fact that Attic inscriptions do reflect a more archaizing pronunciation of Greek through the early second century CE (Threatte 1982).

1§8 I believe that the Atticists’ ideas concerning ‘correct pronunciation’ are reflected in a special kind of technical texts, the prescriptive Atticist lexica. These are word-lists composed in the same period when Philostratus wrote his judgment on Pausanias (ex. [1]), and they are a by-product of the same rhetorical training that Pausanias had received. The entries of such lexica prescribe correct linguistic usage contrasting ‘correct Attic’ forms to others that are not acceptable in correct writing, and—as I hope to show—correct speaking.

The lexica and their audience

Reading, education, pronunciation

2§1 Reading skills were part of the elementary curriculum of education, although they were probably acquired and fully mastered only over the course of a number of years[4]. The only passage of the Techne (Ars Grammatica) attributed to Dionysius Thrax which is likely to be authentic lists ‘accurate reading, following the προϲῳδία (ἀνάγνωϲιϲ ἐντριβὴϲ κατὰ προϲῳδίαν)[5] as the first of the six elements that constitute grammar itself. The adjective ἐντριβήϲ may point to an ‘elaborate’ pronunciation, and the προϲῳδίαι that it takes into account are the differences between the three types of accents and long and short quantities (and the breathing, we may add): all phonological traits that the Koine had lost by the Imperial period. These differences needed to be reproduced in pronunciation while reading, in most cases supplying them from memory, and not just reading diacritical signs that most texts would not have[6].

2§2 The lexica were meant for an audience of Atticists, and, as titles like The Training of the Sophist (cf. §2.2 below) suggest, they were designed for orators who had undergone the full school curriculum, and had received higher education. They provide a solution to a problem of the early stages of education, as envisaged by Quintilian:


non assuescat ergo [puer], ne dum infans quidem est, sermoni qui dediscendus sit

Quintilian, Institutio Oratoria 1.1.5

2§3 Quintilian is referring to the speech of the nurses, the first contact of a baby with language, but this kind of ‘unlearning’ is precisely what the lexica provide: they help their audience to get rid of ‘bad habits’. Think for instance of the ‘ignorants’, ἀμαθεῖϲ, attacked by Phrynichus in a gloss of the Ecloga:


πελαργόϲ· οἱ ἀμαθεῖϲ ἐκτείνουϲι τὸ α, δέον ϲυϲτέλλειν· Πελαργὸϲ γὰρ οὐδὲν ἄλλο ἢ Ἐρετριακῶϲ Πελαϲγόϲ.

Phryn. Ecl. 80

2§4 One must expect the same qualities that characterize proficiency in reading to characterize proficient delivery of oratory: a close reading of the glosses can give us an insight into the ‘bad habits’ an educated speaker of the 2nd century may have wanted to unlearn (cf. §5 below).

The corpus

2§5 The earliest examples of extant prescriptive Atticist lexicography date from the second century CE[7], and the latest are probably not later than the third century. The earliest lexica concerned with the usage of Attic authors (λέξειϲ Ἀττικαί) are attested as early as the 3rd century BCE[8]: but these lexica are descriptive, rather than prescriptive; they are collections of special usages or special vocabulary that can be found in Attic authors of the Classical period[9].

2§6 Prescriptive Atticist lexica are peculiar, for the first time in the history of ancient Greek lexicography, in that they do not describe a variety of the language, but prescribe what their authors believed to be the correct usage of Attic Greek. As I hope to show, they also provide their readers with useful information about the best pronunciation to adopt in combination with a pure atticizing style.

2§7 The corpus of prescriptive Atticist lexica can be found at the end of this paper, together with a list of abbreviations and reference editions. Not all of the nine surviving Attist lexica are transmitted in full, or under the name of their authors. Two of them are the work of Phrynichus, a contemporary of the Emperor Commodus: the Training of the Sophist (Σοφιϲτικὴ προπαραϲκευή, Praeparatio Sophistica, fragmentary) and the Selection of Attic verbs and nouns (Ἐκλογή Ἀττικῶν ῥημάτων καὶ ὀνομάτων, Ecloga, surviving in full). Another is a lexicon ascribed to an otherwise unknown Moeris, likely to have lived in the third century CE, and, very fragmentarily, a lexicon transmitted under the name of Philemon (active around CE 200). The Φιλέταιροϲ (Philetaerus) is a lexicon transmitted under the name of Herodian, but it must be at least a century later than the death of the grammarian. The Antiatticista is—despite the name—an anonymous Atticist lexicon. The works of Aelius Dionysius and Pausanias (early second century CE) are only known by indirect quotations in Eustathius’s commentary on Homer. The lexicon trasmitted under the name of Ammonius is a later reworking of a second century CE lexicon by Herennius Philo. Degani (1995) and Dickey (2007) provide very useful overviews on Greek lexicography, and further bibliography.

Lexica without lectional signs

3§1 Even as late as the second century CE, one must assume that prose texts lacked nearly all accents and quantity marks[10]. The nature and the position of the accent, as well as the breathings and the quantities of the dichrona, were learnt at school, under the unifying label of προϲῳδίαι, comprising χρόνοι, τόνοι, πνεύματα (quantity marks, accents, breathings). The system of diacritical marks associated to προϲῳδίαι considered in the first place a means to distinguish ambiguous words (‘γέγονε πρὸϲ διαϲτολήν τῆϲ ἀμφιβόλου λέξεωϲ’ [Arcadius] 186, 3 ss. Barker)—accents, generally rare, are more common in school exercises when doubts could arise in reading[11]. It is surprising to see that when lectional marks do appear, they may include marks that current editorial practice does not normally include, such as quantity marks. This may actually reflect the special care dedicated to pronunciation in that specific period[12].

3§2 One has to expect the lexica themselves to have circulated without diacritical marks in the Roman period. We are particularly lucky to have a clear example of this practice in P.Oxy. 1012, an erudite commentary on literary works. The papyrus dates from the beginning of the third century CE[13] and shows various affinities with Atticist lexicography of the same—or immediately later—period[14] (which however is only attested in later manuscript tradition on codex). It provides us with direct insight into the usual layout that erudite works, including the lexica, may have had during the first phase of their textual tradition—on papyrus.

3§3 I reproduce here an excerpt of the papyrus, which discusses the accent of αγροικοϲ, αληθεϲ, αχρειον, yet without writing a single accent[15]. I keep the reading and the word-spacing of the modern editor, without introducing accents and breathings, i.e. preserving in this respect the original layout of the work:


[. . . . . . .]ον αγροικοϲ προπερι|[ϲπωμενω]ϲ οι ελληνεϲ τον ϊδιω|[την ομοι]ωϲ      αληθεϲ      ομοιωϲ και | [διτταϲ] δυναμειϲ εχει παρα τοιϲ | [αττικοι]ϲ̣ τω δε τονω διαϲτελλεται | [το ϲημ]αινομενον οταν μεν > | [γαρ ϲυγ]κατατιθηται τοιϲ ϋπο τινοϲ | λεγομενοιϲ αληθεϲ ερει ωϲ ϲαφεϲ | οταν δε κατ ερωτηϲιν προφερ||10ητ[α]ι τοτε τοξυτατονωϲ την πρω|[την ε]ξοιϲει ϲυλλαβην αληθεϲ | [ωϲ εληθεϲ οι γε μην ελλενηϲ | [ομοι]ωϲ ειωθαϲιν λεγειν αληθεϲ | [ωϲ ϲα]φεϲ       αχρειον      και τουτο| [φαϲι]ν διττωϲ οι αττικο[ι] αχρειον | με[ν β]αρυτονωϲ προφερομενοι | ωϲ α̣[.]ρειον τον αχρηϲτον αχρει|[ον δε] προπερ̣[ιϲπωμενωϲ] τ̣ον|[. . . .] . ον οι [δε ελληνεϲ ομο]ι̣ω̣ϲ̣ [

P.Oxy. 1012, fr. 16, col. I

3§4 Τhe papyrus discusses accentuation only by comparing words with identical accentuation (for instance, pointing out that ἄληθεϲ is accented like ἔληθεϲ) or by employing technical terms as ὀξυτόνωϲ[16], βαρυτόνωϲ, προπεριϲπωμένωϲ. Still, in the only papyrus fragment which is likely to be an Atticist lexicon, P.Oxy. XV 1803, there are no accent marks and only one rough breathing[17], despite the much later date (sixth century CE).

3§5 This strongly suggests that glosses concerning anything that modern conventions would signal in writing with a diacritical mark have in fact to do with the pronunciation of those words, and not with their spelling.

Descriptions of pronunciation in the lexica

4§1 Atticist lexica may describe (or prescribe) pronunciation in a number of ways. Sometimes pronunciation is inferred only through spelling, most of the times shown with ‘διά’ and the grapheme(s) that distinguish the recommended term:


ἀπέϲβηϲε πῦρ· ὁμοίωϲ καὶ λύχνον. διὰ τοῦ η, οὐ διὰ τοῦ ε.

Phryn. PS 26, 9-10


γρυλίζειν διττὴν ἔχει ἁμαρτίαν, ἔν τε τῇ προφορᾷ καὶ τῷ ϲημαινομένῳ· ἐν μὲν τῇ προφορᾷ διὰ τῶν δύο λλ, ἐν τῷ ϲημαινομένῳ ὅτι παρὰ τοῖϲ ἀρχαίοιϲ τὸ γρυλίζειν ἐϲτὶ τιθέμενον ἐπὶ μὲν τῆϲ τῶν ὑῶν φωνῆϲ, οἱ δὲ νῦν τάττουϲι ἐπὶ τῶν φορτικῶϲ καὶ ἀϲχημόνωϲ ὀρχουμένων. ἐρεῖϲ οὖν γρυλίζειν καὶ γρυλιϲμόϲ, οὐ γρυλλιϲμόϲ.

Phryn. Ecl. 72

When the lexicographers resort to specialized terminology, the usual terms are:

  1. the verb meaning ‘pronounce’, προφέρω, and its derivative προφορά ‘pronunciation’;
  2. terminology referring to the προϲῳδίαι: to vowel length (as μηκύνω, ἐκτείνω, μακρόν ποιῶ, βραχύνω, ϲυϲτέλλω, μακρόϲ, βραχύϲ, etc.), to the nature and position of the accent (as ὀξυτονῶ, βαρυτονῶ, περιϲπῶ, προπεριϲπῶ, etc.), to the nature of the breathing (as δαϲέωϲ, ψιλῶϲ, etc.);
  3. general verbs of saying, as λέγω, φημί say’.

One may object that even technical terms are sometimes used ambiguously, and with good reason. There are many glosses in which ‘saying’, ‘reading’ clearly do not refer to pronunciation, as for instance:


ἀφηλικέϲτεροι: […] πλὴν κατὰ ϲύγκριϲιν ἡ λέξιϲ προφέρεται καὶ ἐν ὑπερθέϲει, ἀφηλικέϲτατοϲ καὶ ἀφηλικέϲτατοι. οἱ δὲ ἀπολελυμένωϲ λέγοντεϲ ἀφῆλιξ, ἀφήλικεϲ ἀμαθέϲτατοι.

Phryn. PS 1, 1–6

Similarly, verbs addressing pronunciation as ‘lengthen’ and ‘shorten’ can be used alongside γράφω ‘write’:


αἱματοπώτηϲ· οἱ Ἀττικοὶ μηκύνοντεϲ τὸ ο προφέρουϲιν τὴν λέξιν, ὥϲπερ καὶ τὸ οἰνοπώτηϲ καὶ ὑδροπώτηϲ, ἐπεὶ καὶ τὸ πόμα οἱ μὲν ἄλλοι διὰ βραχέοϲ τοῦ ο γράφουϲιν, Ἀττικοὶ δὲ ἐπεκτείνοντεϲ.

Ael. D. α 53

4§2 Yet if audiences could be biased against speakers whose command of Greek pronunciation was less than perfect (as Pausanias in text [1] above), we should wonder whether these prescriptions actually addressed only spelling, and not also the pronunciation that came with it. Proficient speakers of Attic would have made clear in their pronunciation, too, that they used αἱματοπώτηϲ and not αἱματοπότηϲ.

4§3 The flexibility with which the lexicographers combine terminology referring to pronunciation with terminology referring to reading attests to the overlapping of these two areas in the fruition of the texts. For instance, προφέρω and ἀναγιγνώϲκω occur together with prescriptions about accents, and quantities (cf. texts nn. [8] and [9]) and even involving spelling itself (text n. [6] above).

4§4 In a world where written texts circulated without reading marks (cf. §3 above), prosody was described (or prescribed) mostly through specialized vocabulary. In this light, we have to think of pronunciation, not spelling, when we read an entry like:


γενέϲθαι, λαβέϲθαι παροξυτόνωϲ μᾶλλον προ{ϲ}ενεκτέον.

Philet. 252

which we could probably just read as

γενεϲθαι λαβεϲθαι παροξυτονωϲ μαλλον προενεκτεον.

We can apply the same principle to all glosses involving lectional marks in modern writing/printing conventions, as the ones dealing with breathings:


ἅθυρμα δαϲέωϲ Ἀττικοί· ψιλῶϲ Ἕλληνεϲ

αθυρμα δαϲεωϲ Αττικοι· ψιλωϲ Ελληνεϲ

Moer. α 11

or to the position of the accent, and/or its nature:


χαμᾶζε ἀεὶ <προ>περιϲπᾶται, τὸ δὲ χαμᾶθεν ὡϲ ἐπὶ πλεῖϲτον.

χαμαζε αει προπεριϲπαται, το δε χαμαθεν ωϲ επι πλειϲτον.

Ael. D. χ 3


γέλοιον βαρυτόνωϲ Ἀττικοί· γελοῖον προπεριϲπωμένωϲ Ἕλληνεϲ.

γελοιον βαρυτονωϲ Αττικοι· γελοιον προπεριϲπωμενωϲ Ελληνεϲ.

Moer. γ 4

Habits to unlearn

5§1 Some glosses suggest that hypercorrections were common among the audience of the lexica. The gloss to πελᾱργόϲ (ex.[3]) raises a number of questions. Note that (a) the scansion of pe.lār.gós does not change if the second syllable has a short vowel, and that (b) the length of the alpha is not relevant here for the purpose of writing an accent. Why would a lexicographer care about the length of a vowel that is never accented? Why would a lexicographer care about the length of a vowel that is entirely indifferent for the prosody of its word?

5§2 The most likely answer is that there were speakers who wanted to sound ‘more Attic’. Distinctive vowel length was lost in some varieties of Greek possibly as early as the beginning of the Hellenistic period. Attic, as the official inscriptions attest, may have preserved it just until the second century CE[18]. The writing system had always been ambiguous when it came to the three dichrona, ‹ι› /i ~ iː/, ‹α› /a ~ aː/, and ‹υ› /y ~ yː/, and it seems that sometimes even the lexicographers had contrasting ideas on their exact quantities. See for instance:

πνῖγοϲ μακρῶϲ Ἀττικοί· βραχέωϲ Ἕλληνεϲ.

Moer. π 38


πνῖγοϲ· ἀμαρτάνουϲιν οἱ βραχύνοντεϲ τὸ ι· ἐκτείνουϲι γὰρ τοὔνομα καὶ τὰ ἀπ’αὐτοῦ.

Phryn. Ecl. 77

5§3 Quantity in the derivatives of πνῖγοϲ can only be detected in scansion, as none of them bears an accent on the iota. The ‹ι› is always long—or almost always, cf. πνιγεύϲ at the end of an iambic trimeter in Aristophanes, Clouds 96 (ἄνδρεϲ, οἳ τὸν οὐρανὸν / λέγοντεϲ ἀναπείθουϲιν ὡϲ ἔϲτιν πνιγεύϲ)[19]. Were the Atticist generalizing vowel length, eventually hyperatticizing just like the ἀμαθεῖϲ who said πελᾱργόϲ (text n. [3] above)?

5§4 A few glosses recommend forms that do not exist in Attic.


ξυρόν μακρῶϲ Ἀττικοί· βραχέωϲ Ἕλληνεϲ.

Moer. ξ 5

5§5 ξυρόν has an etymologically short first vowel (cf. the cognates Skr. kṣurá-; Lat. nouācula), and it is never attested with a long root vowel. Interestingly, there is one surviving line in Aristophanes (Women at the Assembly 65) where an imaginary scansion ξῡρόν can fit, but in an indifferens: only the inexpert (the ἀμαθήϲ of text [3]) could think of it as evidence for a long vowel.


μανόν· τὸ ἀραιὸν οὕτω λέγουϲιν ‹οἱ› Ἀθηναῖοι τὴν πρώτην ϲυλλαβὴν ἐκτείνοντεϲ.

Phryn. PS 89,6-7

5§6 An adjective in –u, of which μανόϲ could be a later thematized form, is attested in a gloss of Hesychius (μ 250 Latte, μανύ· μικρόν Ἀθαμᾶνεϲ). If μᾱνόϲ comes from *manu̯-ós, then it owes its long stem vowel to a compensatory lengthening based on a heterosyllabic treatment of the cluster /nw/ in *man.u̯ós. This evolution is not at home in Attic (cf. Att. /kalós/̯ós ~ Epic and Ionic /kaːlós/ < kal.u̯ós). In fact, Attic must have had μᾰνόϲ: the atticist Oros (5th c. CE, fr. A 62 Alpers) possibly polemicizing with this view, attests Attic μᾰνόϲ in two lines of Attic comedy, Teleclides fr. 23 K.-A., and Plato com. fr. 178 K.-A.: the latter reads ‘καὶ ταῦτα μανάκιϲ, μυριάκιϲ τῆϲ ἡμέραϲ’ and requires a scansion μᾰνάκιϲ. μᾱνόϲ is attested in the fragments of authors who may have been influenced by epic diction, Empedocles (fr. 75,1 D.–K. τῶν δ᾽ ὅϲ᾽ ἔϲω μὲν πυκνά, τὰ δ᾽ ἔκτοθι μανὰ πέπηγε) and Epicharmus (fr. 185 K.–A. οὔτε πυκινὰϲ οὔτε μανάϲ).

5§7 A tendency to identify long vowels as Attic can be seen also in some ancient scholarship. Apollonius Dyscolus explains the long alpha in forms like ἀεί κλάω φανῶ ῥανῶ as the outcome of the suppression of an iota from αἰεί κλαίω φαίνομαι ῥαίνω, because ‘speakers of Attic are given to lengthening’:


ἀληθέϲἐϲτίν, ὣϲ Ἀθηναῖοι ἐκτατικοί εἰϲι τῶν φωνήεντων.

Apollonius Dyscolus, On Adverbs (GG ΙI,1 187 20-1)


τὸ ἄρα πόρρω ἐκτέταται ὡϲ Ἀττικώτερον, καθὸ καὶ τὸ προπέρυϲιν πρωπέρυϲίν φαϲι καὶ τὰ τούτοιϲ ὅμοια, καὶ ὅτι μᾶλλον μηκυντικοί εἰϲι κατὰ τὰ φωνήεντα.

Apollonius Dyscolus, On Adverbs (GG ΙI,1 166 24-6)

5§8 Atticist lexica, and some of the ancient scholarship associated with them, preserve an archaizing pronunciation of Greek, and take care to signal it with the appropriate technical terminology. In some instances the recommended pronunciation is genuinely conservative, but on other occasions it is artificial. The phonological traits that are unduly extended seem to be precisely those that were being lost in the second century CE, such as distinctive vowel length: the tendency to attribute long vowels to Attic is a particularly clear example of this attitude.


Editions of Atticist lexica

Ael. D. Aelii Dionysii Atticistae fragmenta = Erbse 1950:95–151.
Amm. K. Nickau, Ammonii qui dicitur liber De adfinium vocabulorum differentia, Leipzig 1966.
Antiatt. Ἀντιαττικιϲτήϲ (Antiatticista), in I. Bekker, Anecdota Graeca (= An. B.), vol. I:75–116, Berlin 1814.
Heren. Herennius Philo, De diversis verborum significationibus. Testo critico, introduzione, commentario e indici a c. di V. Palmieri, Napoli 1988.
Heren. de propr. Herennius Philo, De propria dictione, in Palmieri 1988 238–42.
Moer. D.U. Hansen, Das attizistische Lexikon des Moeris, Berlin and New York 1998.
Paus. Pausaniae Atticistae fragmenta = Erbse 1950:152–221.
Philem. L. Cohn, “Der Atticist Philemon”, Philologus [NF] 11 (1898):353–67.
Philem. R. R. Reitzenstein, Geschichte der griechischen Etymologika, Leipzig 1896:392–6.
Philet. A. Dain, Le Philétaeros attribué a Hérodien, Paris 1954.
Phryn. Ecl. E. Fischer, Die Ekloge des Phrynichos, Berlin and New York 1974.
Phryn. PS I. de Borries, Phrynichi sophistae Praeparatio sophistica, Leipzig 1911.
Phryn. fr Phrynichi sophistae fragmenta, in Phryn. PS (de Borries 1911) 130–180.


Arrighetti, G. 1987. Poeti, eruditi e biografi: momenti della riflessione dei Greci sulla letteratura. Pisa.

Cribiore, R. 2001. Gymnastics of the mind: Greek education in Hellenistic and Roman Egypt. Princeton.

Colomo, D. Forthcoming. Quantity Marks in Greek Prose Texts on Papyrus. In Signes dans les textes, textes sur les signes. Actes du colloque international. Liège, 6-7 septembre 2013, ed. G. Nocchi Macedo – M.C. Scappaticcio, Liège [Pap.Leod. VI].

Degani, E. 1995. La lessicografia. In Lo spazio letterario della Grecia Antica, II, La ricezione del testo, ed. G. Cambiano, et al., 505–27. Rome.

Dickey, E. 2007. Ancient Greek Scholarship. New York and Oxford.

Erbì, M. 2006. P.Oxy 1012 come testo d’erudizione. In Esegesi letteraria e riflessione sulla lingua, ed. Arrighetti, G. – M. Tulli, 127–157. Pisa.

Erbse, H. 1950. Untersuchungen zu der Attizistischen Lexica [Abhandlungen der Dt. Akad. d. Wiss. zu Berlin, Phil.-Hist. Kl. 1949, 2]. Berlin

GG = A. Hilgard, A. Lentz, R. Schneider, G. Uhlig (1867–1910) Grammatici Graeci, 4 voll., Leipzig [repr. Göttingen 1965].

Horrocks, G.C. 2010. Greek: A History of the Language and its Speakers. Chichester and Malden, MA.

Janse, M. 2002. Aspects of Bilingualism in the History of the Greek Language. In Bilingualism in Ancient Society. Language Contact and the Written Text, ed. Adams, Janse, and Swain, 332–390. Oxford.

Kim, L. 2010. The Literary Heritage as Language: Atticism and the Second Sophistic. In A Companion to the Ancient Greek Language, ed. E.J. Bakker, 468–482. Malden, MA. and Oxford.

Lallot, J. 2003. La grammaire de Denys le Thrace, 2nd ed., Paris LIV = H. Rix, Lexikon der Indogermanischen Verben, Wiesbaden 20012.

Probert, Ph. 2006. Ancient Greek Accentuation. Oxford.

Schmid, W. 1887-1897. Der Attizismus in seinen Hauptvertretern, I-V, Stuttgard [rep. Hildesheim 1964: Olms].

Swain, S. 1996. Hellenism and Empire. Language, Classicism and Power in the Greek World AD 50–250. Oxford.

Threatte, L. 1980. The Grammar of Attic Inscriptions, I, Phonology. Berlin and New York.

———. 1982. “The Alleged Conservatism of Attic Epigraphical Documents: A Different View.” Hesperia Suppl. XIX:148–56.

Turner, E.G., and P.J. Parsons. 1987. Greek manuscripts of the Ancient World. 2nd ed. London.

Whitmarsh, T. 2005. The Second Sophistic. Oxford.

[1] Stylistic analysis of Atticism in modern scholarship was initiated by the pioneering work of W. Schmid, Der Atticismus (Schmid 1887 – 1897), and in the last decades a similar approach has informed the works of—to mention but some—Swain (1996), Whitmarsh (2005:41-54), Kim (2010).

[2] Cf. Anon. Rhet. Spengel 1853 [I], 322, 27–9, ‘αἱ καινοτομίαι τῶν ϲυντάξεων καὶ τῶν ἐγκλίϲεων καὶ τῶν διαθέϲεων καὶ αἱ περὶ τοῦϲ χρόνουϲ μεταλλάξειϲ καταϲείουϲι τὰ τοῦ ἀκροατοῦ ὦτα’, a terminology shared in the literary work of Flavius Philostratus (on Quirinus of Nicomedeia, Philostratus, Lives of the Sophists 621), ‘ἐρρωμένοϲ μὴν καὶ ϲφοδρὸϲ καὶ καταϲεῖϲαι δεινὸϲ ἀκροατοῦ ὦτα’.

[3] Horrocks 2010:135.

[4] Cribiore 2001:189.

[5] Dionysius Thrax, Ars grammatica 1.4 Lallot. The Techne then (2.1) defines reading (ἀνάγνωϲιϲ) as ἀδιάπτωτοϲ προφορά.

[6] Lallot 2003:76.

[7] A precursor could be the Λέξειϲ Ἀττικαί οf Irenaeus (Degani 1995:519).

[8] With the work of Philemon of Aixon, a contemporary of Callimachus (Degani 1995:509).

[9] In the 2nd century CE, this descriptive sort of lexicon is still represented by the work of Harpocration on the ten Attic orators, and in a different fashion, by the Onomasticon of Julius Pollux, cf. Dickey 2007:94 and 96.

[10] Even in Byzantine times, copying accents (and thus, expecting to find them in books) was not the rule: ‘[i]n the ninth century, the Stoudion Monastery imposed on a copyist the duty of copying punctuation (PG 99.1740: penalty prescribed εἰ μή … παρατηρεῖται τά τε ἀντίϲτιχα καὶ τοὺϲ τόνουϲ καὶ τὰϲ ϲτιγμάϲ)’ Turner – Parsons 1987:10 n. 43; cf. also Probert 2006:18–21.

[11] Cribiore 2001:191.

[12] Colomo, forthcoming.

[13] Cf. Hunt in P.Oxy. VII:84 on the dating.

[14] Erbì 2006:145. There was contuinity in the development from commentaries and scholia to lexica, cf. Arrighetti 1987:200–1(with n. 108).

[15] The only lectional sign in this section is the trema (ϊδιω|[την ll. 2-3, ϋπο l. 7). The papyrus has only one instance of an accent, on πότε (fr. 13, 32), possibly to differentiate the interrogative from the indefinite adverb.

[16] τοξυτατονωϲ at line 10 is a trivial misspelling of ὀξυτόνωϲ.

[17] In a position, moreover, where later writing practice would have the rough breathing anyway, the word-initial hypsilon ὑ|ποζυγίων (lines 42–3). The breathing occurs at line-end, and it could be there to signal that the word continued in the next line: it seems to function much more like the trema in P.Oxy. 1012 than as a signal of initial aspiration.

[18] Horrocks 2010:160 – 170; Threatte 1980:385–387 and 1982.

[19] Cf. also Aristophanes, Birds 1001.