From Iliad 18.333:

νῦν δ' ἐπεὶ οὖν, Πάτροκλε, σεῦ ὕστερος εἶμ' ὑπὸ γαῖαν

As best I can tell from the scansion, the σεῦ here is shortened by correption, letting it be the final syllable of a dactyl.

But my understanding of a circumflex accent is that it represents a falling pitch: a high pitch on the first mora of the vowel, a low pitch on the second mora.

How do these two things combine? When a vowel with a circumflex is correpted, what happens to its pitch? I know that both of these phenomena (correption and pitch accent) were written about by ancient scholars, and I'm curious if any ever touched on the combination of the two. Or, alternately, if modern linguists have an idea of how this worked.

  • 1
    I don't know the answer to this, but I wonder if this being Homeric complicates things. Since the accents are Alexandrian, who know if epic poetry hundreds of years earlier kept the circumflex or even pronounced it short (despite the meter)?
    – cmw
    Feb 9, 2022 at 20:28

1 Answer 1


Your question has two levels: 1. How do we understand the underlying reality of the Attic pitch system recorded in our texts and 2. With that general understanding, what would the system do in this particular case of correption.

I think the emerging majority view of the circumflex accent is that it is not different phonemically or phonetically from the acute accent, but just represents the "special" case where the accent fell on the first mora of a long vowel, necessitating that people write a combined high-low accent that developed orthographically into the circumflex accent.

What I have read about diphthongs like εῦ is that they are the orthographic descendants of writing έὺ. Ι think it would have been possible to write just the acute accent on ε and drop the grave accent on υ; however, this would have been particularly confusing for orthographic diphthongs like ει (pronounced identically to ι during the period the accent marks were invented and used), so the simplest practice would be to extend the use of the circumflex to all diphthongs when the first mora is accented.

In the case of correption in phrases like σεῦ ὕστερος, I think most scholars presume the following transformation. The semivowel written as υ that would normally close the syllable of σεῦ is transferred to the beginning of the next syllable, leaving σέ as now a short/light syllable. The question now becomes, what happened to the following drop in pitch normally required by the acute accent?

We have orthographic precedents like γαλήν' ὁρῶ to tell us that from a systemic point of view, words could be written with an acute accent in unusual circumstances (such as the deletion of the last vowel and syllable of γαληνά before ὁρῶ). No one knows how such things were actually pronounced, but there are, of course, proposals that are more probable or more accepted then others.

My own view is that the emphasis on the "musicality" of the "pitch" accent is misleading and that the accent is distinctly closer in nature to stress accents than to true tones, just with different details about what possible components of "stress" were salient in pronunciation (e.g., higher pitch, higher volume, increased length, or other features in which even European languages differ). I know of no oppositions in Greek that hang on the behavior of a single mora that can't be explained by the placement of the accent in the word. This is not true of Swedish, Mandarin, or other languages where stress and tone can have different parameters.

Based on the scholarship I have read, I think the acute accent represents two underlyingly different phenomena: the syllable that mandates a following drop in pitch, as in Japanese, or the operation of a phrasal peak that operates even on single words, as in French. My interpretation is that you rise in pitch in any phrase until you hit a mandatory drop in pitch or reach the syllable closest to the end of the phrase that can carry pitch. (Those are basically the combined rules of Japanese and French).

In σεῦ ὕστερος, I think it is likely that speakers realized a drop in pitch on the now word initial semivowel (pronounced like a consonantal ου), representing a pronunciation like σέ οὺύστερος. I speak a tiny bit of a tonal language that actually can make phonemic and phonetic distinctions according to the pitch of initial sonorants independently of the tone of the rest of the word. The initial tone in this case represents a particular morpheme. Such pronunciations, though perhaps unusual cross-linguistically, are not actually difficult to realize once you can manage tone generally. This is how I personally would pronounce the phrase in question.

Alternatively, σεῦ ὕστερος was pronounced in the same fashion as however γαλήν' ὁρῶ was pronounced (i.e., as σέ ουύστερος, without any special drop in pitch on the ου semi-consonant). Keeping a phrase-internal acute accent at the end of a word violates the normal accent rules, so we must ask what this exception means in this case. For example, γαλήν' ὁρῶ could equally have been written γαλν' ὁρῶ, with a grave accent on the Eta. Our view of the pronunciation will then depend on whether we think γαλήν'and γαλν' represent different phonetic realities or just some orthographic convention.

In the view of many scholars, and in my view, the grave accent does not represent some new and special pitch phenomenon on the mora that carries it, but rather just a suppression of the final pitch according to the intonation rules of Greek phrases. If this is true, then γαλήν' ὁρῶ probably just represents an orthographic convention, perhaps motivated by the normal phrasal rise in pitch and a hesitancy to apply the grave-accent rule to what felt like an internal syllable. Under this reasoning, a pronunciation like σέ ουύστερος would simply represent the normal rise in pitch within any phrase, peaking at ουύσ and having a pronounced drop on the syllable τε.

  • While I like what I read, I'm surprised that you don't explore the question of whether the σεῦ in σεῦ ὕστερος ("after you", right?) is accented at all. From all appearances it ought to be an unaccented clitic; even if it isn't, one would expect it to lose the accent altogether because of accent clash and no prior syllable to retract it onto. Feb 10, 2022 at 19:51
  • @Unbrutal_Russian The accent makes sense in context since there's some focus on you; in any case, since the Alexandrian editors wrote σεῦ and not σευ, they must have thought it was accented, so it's reasonable to ask how they would have pronounced it. I don't think you could have an enclitic right after a vocative in Greek (though I could be wrong about that), and the kind of retraction due to accent clash that you describe in the last sentence isn't something that happened in Greek, as far as is known.
    – TKR
    Feb 11, 2022 at 0:14
  • Unbrutal_Russion, I am not following what you mean by accent clash. Could you explain? I also agree with TKR that this seems to represent a phonetically stressed pronoun in context and that is the form that tradition handed down to us. That is why it did not occur to me to discuss it. Feb 11, 2022 at 0:43
  • I just recalled what is meant by accent retraction due to stress class. I know this occurs in English. I don't think this occurred in Greek, because of how they used the grave accent. No retraction is represented. I think in Modern Greek, stress class is handled by deleting the first stress, but without retraction. Feb 11, 2022 at 0:53
  • If a drop in pitch was realized with σεῦ, then there would be no stress clash. If there was no drop in pitch, the we have the γαλὴν' ὁρῶ situation I discussed, which does not seem to document retraction because of stress clash, but retraction because of the vowel deletion. Feb 11, 2022 at 0:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.