Skip to main content

Accenting sequences of enclitics in ancient Greek: rediscovering an ancient rule

Philomen Probert, Oxford

This paper results from work done together with Dr Stephanie Roussou. It presents an ancient rule about sequences of Greek enclitics that modern debate has missed, although Lehrs (1837: 129) had correctly interpreted the crucial text.

Previous scholarship

It is traditionally taught that every enclitic in a sequence, except the last, gets an acute accent on its last syllable. Some modern scholars have doubted this rule and proposed alternatives (Göttling 1835: 405; Donaldson 1862: 43; Vendryes 1904: 88–9; Barrett 1964: 426–7).

The traditional doctrine is normally said to be that of the ancient grammarians; the main argument against it has been a vague notion of implausibility (against which see Devine and Stephens 1994: 373–4). It has been claimed that some manuscripts apply a different system, but it is not clear whether a consistent alternative is really found (cf. Chandler 1881: 281–2). But a passage transmitted in [Arcadius’] epitome of Herodian’s Περὶ καθολικῆς προσῳδίας has also been seen as relevant (Göttling 1835: 405; Vendryes 1904: 89):

Εἰ δὲ παραλλήλων ὄντων ἐγκλιτικῶν, ἐν τῷ μεταξὺ περισπωμένων1, ὡς ἐπὶ τοῦ ΠΟΥ ΠΗ ΠΩΣ, καὶ μετὰ τοῦτο ἐπιφέροιτο ἕτερον ἐγκλιτικόν, τοῦτο τὸ περισπώμενον <.....> 2 οὔτε περισπᾶται διὰ τὸ ἐγκλῖναι τὸν ἴδιον τόνον· ἐπεὶ μὴ πέφυκεν ἡ περισπωμένη κατὰ τὸ κοινὸν ἔθος συστολῆς μὴ παρακολουθούσης εἰς ὀξεῖαν μετατίθεσθαι, οἷον· οὔ πως ἔστ’3 ἀγέλαε (ἐν γὰρ τῇ ΟΥ διφθόγγῳ μόνῃ ἡ ὀξεῖα) ἤ που τίς σφιν εἶπεν (ἐν τῷ Η καὶ ἐν τῷ ΤΙΣ ἡ ὀξεῖα) ἄνθρωπόν τινά που φησὶ μελῳδεῖν (πάλιν ἐν τῇ ΠΟΝ συλλαβῇ καὶ τῇ ΝΑ ἡ ὀξεῖα τίθεται). ([Arcadius] 168. 3–13 Schmidt)

1 So Schmidt (1860), following the manuscripts, but we propose the correction περισπώμενον.

2 Schmidt sees a lacuna here and suggests <οὔτε ὀξύνεται>. With Lehrs we prefer τοῦτο τὸ περισπώμενον οὔτε περισπᾶται διὰ τὸ ἐγκλῖναι τὸν ἴδιον τόνον, <οὔτε ὀξύνεται> ἐπεὶ...

3 So Schmidt, but with Lehrs we think the logic demands ἐστ’.

Göttling thought the idea was that an enclitic whose ἴδιος τόνος was a circumflex (on this notion see below) should receive an acute, not a circumflex, when accented because of a following enclitic. Vendryes thought the idea was that an enclitic with a long vowel should be left unaccented in a sequence of enclitics.


The interpretation to be defended is instead that of Lehrs, illustrated via the following translation:

‘If there are enclitics side by side, and one in the middle is [underlyingly] perispomenon, like [indefinite] ΠΟΥ, ΠΗ, and ΠΩΣ, and after this there is another enclitic, this perispomenon one neither gets a circumflex (because it has thrown its own [underlying] accent off) nor gets an acute (because a circumflex is not normally exchanged for an acute unless a vowel is shortened). For example: οὔ πως ἐστ’ ἀγέλαε (with an acute only on οὔ); ἤ που τίς σφιν εἶπεν (with an acute on ἤ and τίς); ἀνθρωπόν τινά που φησὶ μελῳδεῖν (again the acute goes on the syllable -πόν and on ‑νά).’

In other words, enclitics that the grammarians considered to have a circumflex on the final syllable as their underlying accent (ἴδιος τόνος), and only those enclitics, fail to receive an accent when they stand in a sequence of enclitics, with at least one more enclitic following. The ancient concepts that every word had an ἴδιος τόνος (effectively an underlying accent), and that some enclitics (including πη, που, πω, πως) had a circumflex as their ἴδιος τόνος, provide crucial background to be explained and illustrated.

Further support comes from Sch. Il. 20. 464a1 (A), one of the Homeric scholia derived from Herodian. Since [Arcadius’] sections on enclitics have a debateable history, the scholion helps to confirm that this rule was espoused already by Herodian.

Finally the paper will report briefly on a search for typological parallels: need the rule be a grammarians’ invention or could it be real?