As Postgate notes (“Short Guide…”, 79), the rule of accent on L of -LS being circumflex does not apply to vowels so lengthened in verse, as in Homer φῑ́λε κασίγνητε. Correption does not change accentuation (AFAIK?). My general working assumption has always been that lengthening or shortening of vowels in verse does not effect alterations in accentuation.*
But something strange seems to be going on with the plural personal pronouns. Smyth (325f) writes:
" ἡμῶν, ἡμῖν, ἡμᾶς, ὑ_μῶν, ὑ_μῖν, ὑ_μᾶς, when unemphatic, are sometimes accented in poetry on the penult, and -ῑν and -ᾱς are usually shortened. Thus, ἥμων, ἧμιν, ἧμας, ὕ_μων, ὗμιν, ὗμας. -ῑν and -ᾱς are sometimes shortened even if the pronouns are emphatic, and we have ἡμίν, ἡμάς, ὑ_μίν, ὑ_μάς. σφάς occurs for σφᾶς. "
These - and from my recollection forms like ἡμίν and ὑμίν occur pretty plentifully - are counter-examples to the above formulated principle that vowel length alterations in verse does not cause the accentuation change.
Can anyone explain this apparent contradiction and what’s going on here, and/or provide some grounding how to think about when (under what circumstances and in which cases) vowel length changes should be accompanied by accentuation changes and when not?
\
I am aware that editorial practice may differ. Where this arises in relation to my question I’d be happy to hear about differing practices, but I’d guess most of what I have written is the uncontroversial consensus of modern editors.
It’s a while since I looked into this, and I may not have it quite right, but my understanding is this. Along with the orthotone forms, ημᾶς ημῶν etc., there are quasi-enclitic forms, which are barytone, ήμας ήμων ήμιν (ῆμιν where ι is shortened). Most modern editors wrongly treat the latter (unemphatic) forms as orthotone, but this makes no linguistic sense and is contradicted by Babrius’ practice (accent on the penult), and by ancient authorities, esp. Herodian, as documented by Chandler, and good manuscripts, cf. West’s Ilias praef. xviii.
Then there’s correspondence with the singular, εμέ σέ etc. vs. με σε enclitic.
Thanks mwh. This is certainly interesting: you mean to say that editors ought to distinguish between the enclitic and orthotone forms of the plurals (ήμας ήμων ήμιν vs ημᾶς ημῶν etc.), on the same basis (I presume) as they do for the singulars (με σε vs εμέ σέ etc.), but instead they simply (and in contradiction to good evidence) print the orthotone plural forms under all conditions and in all cases - if I understand you right? I do wonder why this phenomenon is restricted to poetry, if Smyth is correct in saying so.
Any help with the question of what vowel length changes are accompanied by accentuation changes and which aren’t, and why? You mentioned for instance ῆμιν where ι is shortened, and I’m sure ῆμας when ας is shortened; and we have e.g. the ἡμῖν > ἡμίν change if ι is shortened; but correption as far as I am aware is not represented with an accentual change, and when Homer uses φῑ́λε at the beginning of a verse (e.g. Iliad 4.155, 5.359) we do not change the acute to a circumflex.
Thanks mwh. This is certainly interesting: you mean to say that editors ought to distinguish between the enclitic and orthotone forms of the plurals (ήμας ήμων ήμιν vs ημᾶς ημῶν etc.), on the same basis (I presume) as they do for the singulars (με σε vs εμέ σέ etc.), but instead they simply (and in contradiction to good evidence) print the orthotone plural forms under all conditions and in all cases - if I understand you right? I do wonder why this phenomenon is restricted to poetry, if Smyth is correct in saying so.
Yes, that’s it. Just as σε is either enclitic or orthotone, so is υμας. (Compare the difference between e.g. ημεις ηλθομεν and plain ηλθομεν, emphatic vs. unemphatic.) An editor has to decide which, if the word order is not sufficiently determinative. The distinction is semantic first, accentual second. It makes no sense for it to be restricted to poetry.
Any help with the question of what vowel length changes are accompanied by accentuation changes and which aren’t, and why? You mentioned for instance ῆμιν where ι is shortened, and I’m sure ῆμας when ας is shortened; and we have e.g. the ἡμῖν > ἡμίν change if ι is shortened; but correption as far as I am aware is not represented with an accentual change, and when Homer uses φῑ́λε at the beginning of a verse (e.g. Iliad 4.155, 5.359) we do not change the acute to a circumflex.
The shortening in ημιν and ημας I take to be metri gratia, and hence confined to verse. The accent is a separate issue. The anomalous lengthening in φίλε is also metri gratia, in the opposite direction. Since it’s merely a prolongation of ι (unless the λ was doubled as in e.g. ἔλλαβε) there’s no reason to suppose that the accent was changed. Correption is likewise metrically motivated. I don’t quite understand your question there. Is it ever the case that a circumflexed vowel is corrrepted? I shouldn’t think so, but if it is, the accent would have to change along with the quantity. I don’t understand the meaning of Postgate’s “rule” that you mention, “the rule of accent on L of -LS being circumflex.” If he’s talking of words such as λῦε or δῶμα, where the vowel can’t be circumflex if the form is a longer one (e.g. λύειν, δώματα), of course that doesn’t apply to φίλε with elongated iota, and of course the vowel quantity is unaffected by the accent.
The rule in question is simply that if you have a long penult followed by a short final syllable (I denoted this ‘-LS’), and the penult is accented, that accent will be a circumflex.
Why - when shortening appears to cause a loss of 1 mora for accentual purposes too - would lengthening not have the opposite effect? After which the addition of 1 mora I would expect to produce a circumflex (as the rise in pitch must come down). The retention of an acute here isn’t obvious to me.
Perhaps not (?). But what about the scenario where we have a paroxytone word ending in two longs, of which the final syllable is correpted, so now the word ends ostensibly in a long followed by a short? If editorial practice changed to a circumflex here I feel I’d have noticed, but I can’t understand the logic behind this consensus.
I took correption and the lengthening of φίλε as just two examples. Really I want to understand what is going on for any case of metrical lengthening or shortening that happens in verse.
The overall pattern seems to be that acute > circumflex does not ever occur due to metrical shortening/lengthening, but circumflex > acute occurs in more or less any scenario where the accentuation rules would seem to commend it. (That is to say, the accentuation of the word is ‘re-evaluated’ in terms of new vowel lengths if it is a circumflex that may be made an acute, but not the other way around.) (Though I do wonder about say what happens to a properispomenon if its final syllable is metrically lengthened…) Is this correct? If so, the question remains, why is this the case?
I think we have to approach this more broadly, in the context of the fit between language and meter, which in Homer is very imperfect (a fact habitually disguised by prosodic and orthographic fudging; I mentioned ἔλλαβε). The language—what Aristoxenus well calls τὸ ῥυθμιζόμενον—has to be accommodated to the demands of the meter (τo ῥuθμιζον), which always claims priority. It’s impossible to say for sure how φίλε at verse beginning was accommodated. It’s an exceptionally strained license, on the extreme edge of tolerability, and just how it was realized in performance is beyond recovery.
As to correption, I don’t know that there’s ever any cause for accentual modification.
As to lengthening and shortening in general, all we can do is observe the phenomena. The conditions of epic correption are pretty well understood (the only question is whether it entailed a yod), and metrische Dehnung has been extensively studied too—it happens under metrical pressure.
Slightly better link, as the image in the above was not legible to me, and “pg. 246” of Thesaurus Graecae linguae might be helpfully expanded to “pg. 246 of the appendix on Attic of the Thesaurus Graecae linguae.” (It took me some searching to find the below.)
As terminal -οι and -αι are metrically long, but treated as short by accentual rules, I’d be surprised if metrical lengthening ever moved accent. Some explanation like enclitic forms makes a lot more sense.
Quite startling to learn that there’s strong evidence for a contrast between orthotone and quasi-enclitic forms of these pronouns, but it doesn’t get notated as such in standard modern editions of prose works. What about the third-person pronouns? To the best of my knowledge the dative οἱ is always written enclitic when a personal pronoun (though the demonstrative, which takes an accent, is almost the same as a stressed personal pronoun), but the oblique cases of αὐτός are always written as oxytone/perispomenon. There are certainly many cases where I have felt that an αὐτῶν or αὐτῷ should be enclitic, or unstressed in a way that is hard to reconcile with the circumflex accent, but it never occurred to me that there might be evidence to weigh in on such a matter.
I don’t think αυτον etc. are ever enclitic, but I don’t think they’re ever emphatic either, at least in archaic and classical times; demonstratives (εκεινον, τουτον) are available for that. ἑ οὑ(ἑο) οἱ and σφας etc. in their various forms can be either enclitic (unemphatic) or orthotone in archaic Greek, but it’s all a bit of a mess, since the use of these forms, especially the orthotones, is mostly excluded from Attic and from subsequent prose, plus they came to be exclusively reflexive. Once enclitic οὑ οἱ etc are no longer available for unemphatic 3rd person, Greek is left stuck with αὐτοῦ αὐτῶν etc. along with αὐτόν etc. There seems no way of avoiding the circumflex, though perhaps in practice some adjustment was made; by way of weakening rather than shifting? I’m suspicisous of the rigidity of the conventional rules.
