You are here

The Invention of the Greek Accent Marks

Philomen Probert

Oxford University

The invention of the Greek accent marks


Aristophanes of Byzantium is credited with inventing the signs for Greek accents, breathings, and vowel lengths, according to a single source: a passage found in two sixteenth-century Paris manuscripts. The passage has a doubtful history, but the story it tells generates considerable interest (see Prauscello 2006: 33–40, with bibliography). This paper argues that at least for the material on accents the passage had a source that was in Latin, and whose subject was the Latin accent.


One of the manuscripts (Par. Gr. 2102), in the hand of Jacob Diassorinus, presents the passage as part of Book 20 of [Arcadius]’ epitome of Herodian, Περὶ Καθολικῆς Προσῳδίας. Book 20 is missing from all other copies of [Arcadius]; Diassorinus clearly compiled (or as it is often put ‘forged’) this book from other sources, to repair the loss of the original Book 20—a point that does not itself tell us where the material came from in the first place. In the other manuscript (Par. Gr. 2603) the passage appears separately from the text of [Arcadius] (which this manuscript also contains): see Nauck (1848: 12 n. 2); Lentz (1867: xxxviii); Laum (1928: 99); Roussou (forthcoming, § Evidence for a Latin source has not been noticed before, and adds a new wrinkle to debate about the text’s history.

Summary of the argument

Initial evidence for a Latin source comes from features with better parallels in Latin sources than in Greek ones. For example:

•  The term ὀξυβαρεῖα (or ὀξυβαρύς) for ‘circumflex’, otherwise only at [Sergius] iv. 531. 19 Keil.

• The point that the grave accent is capable of spreading itself out over more of the word than the acute or circumflex; the best parallel is [Sergius] iv. 532. 12–14 Keil.

•  The idea that the accented syllable κυριεύει τοῦ ὀνόματος ‘dominates the word’; for this verbal expression cf. Diomedes’ regens uerba ‘ruling words’ (i. 430. 30 Keil).

•  The expression οὐδ’ ἂν ἐπιμήκιστον εἴη τὸ ὄνομα ‘(the accent goes no further back than the antepenultimate syllable) even if the word is very long’. Such allusions to possible word length are standard in the Latin tradition, not the Greek one: quotlibet syllabarum sit dictio (Donatus, Ars Maior 609. 8 Holtz; Diomedes i. 431. 13–14 Keil); neque enim refert plurium syllabarum esse partem orationis (Diomedes i. 431. 11–12 Keil); si quantarumuis sermo sit syllabarum ([Sergius] iv. 482. 20 Keil); in quantouis numero syllabarum ([Sergius] iv. 483. 11–12 Keil).

• A new examination of the manuscripts yields the new reading ὁπότε δὲ αὐτὸς ἐκσταίη τῆς λέξεως ὁ τόνος, τηνικαῦτα περισπώμενος γίνεται ‘whenever this accent (i.e. the acute) abandons the word, then the accent becomes a circumflex’. For the idea that an accent that is not present has ‘abandoned’ its place cf. Cledonius (v. 32. 8–10 Keil): loca quae circumflexus aut acutus dimiserit, grauis possidet... dimissum ab aliis possidet ‘the grave occupies the places that the circumflex or acute has occupies (the place) abandoned by others’. Compare Pompeius’ sibi sermonem uindicat ‘(the acute or circumflex) claims the word for itself’ (v. 126. 16 Keil).

Moreover, the clause ὁπότε καὶ κυριεύοι τοῦ ὀνόματος ἡ παρὰ τὸ πέρας, ἢ τρίτον ἀπὸ τοῦ πέρατος may derive from an expression meaning ‘depending whether the penultimate syllable or the antepenultimate was the dominant syllable of the word’—a complete set of options for the Latin accent, not the Greek one. Later on we read, oddly, that the circumflex is ‘mostly found at the end’ (φαίνεται δὲ καὶ οὗτος τὰ πολλὰ ἐπὶ τοῦ πέρατος). Latin sources tell us that Latin accents (whether circumflex or acute) are not usually found on the final syllable: it will be suggested that such a claim has been ineptly adapted for Greek, with the removal of ‘not’.

A working text will be provided.

Session/Panel Title

Greek and Latin Linguistics

Session/Paper Number


© 2020, Society for Classical Studies Privacy Policy