I have the book, but no - it’s not discussed. There’s other interesting stuff in the first half of the book, though, like a discussion on the name Odysseus. The latter half of the book, which is a running commentary of the text and discusses the process of creating the epic, isn’t as good or thorough as the one on the Iliad.
I think the argument is that first (in prehistory as mwh puts it) it wasn’t unmetrical, then by the time of Homer (or Melesigenes) it was, and he would have somehow had to make it metrical again, presuming that unmetrical lines are no-no.
How would the line have sounded in Q’s¹ ears? Did Q somehow hear an echo of the ancient digamma even if he didn’t utter it (i.e. à la Paul’s h aspiré analogy?)? And would that echo of the digamma been enough for him to accept the line?
West answers this in Glotta 44 (1967), pages 135—139. Unfortunately, my institution gives no access to those old volumes. Woe these penniless institutions!
¹West’s monicker to the poet of the Odyssey.
But there are lines that are unmetrical no matter what, such as the ones I quoted earlier:
Il. 5.539 φίλε κασίγνητε κόμισαί τέ με δός τέ μοι ἵππους
Il 23.2 ἐπεὶ δὴ νῆάς τε καὶ Ἑλλήσποντον ἵκοντο
Shouldn’t we be talking about West’s (δῖος) Ἀλέξανδρος, Ἑλένης πόσις ἠϋκόμοιο, as our main evidence for Ϝελένη in Homer? Along with the referenced grammarians in the West quote? If we accept that, ἐκ Ἑλένη is not so circular. Still, δ’ seems would seem natural to me there, considering the preceding line and new element now in focus.
I would assume that it is likely that as the oral posture of Greek changed to make the digamma difficult, and effect its drop from everyday speech, it would have remained for a time in more precise speech, old people’s speech, dialects, and special phrases.
An aoidos of the first generations of digamma-droppers, faced with having to fix an unmetrical line when pronounced with his normal pronunciation, would likely find the most natural fix to be to use the digamma, just as he could still hear it from still living rhapsodists. Digamma would remain a fairly natural fit for the language rhythm for some time after it became disused. The next generation of rhapsodists would do the same, and so on. Eventually this would lead to other oral corruptions, of course, but I wonder if a digamma wouldn’t last as long as the oral set phrase itself.
I was thinking exactly this. We should expect to see any particular “fix” technique used in normal oral composition, leading to many unmetrical “fixed-in-performance” lines.
So I’m no longer satisfied by the argument that digamma had dropped out of everyday speech, and therefore that it must have been absent from every line of Homer as performed.
The digamma remained operative in some dialects well into the classical era and beyond. But in all likelihood it disappeared in Ionian well before the time Homeric poems were composed, since it’s observed very inconsistently in them.
Compare
Il. 6.258: ἀλλὰ μέν᾽ ὄφρά κέ τοι μελιηδέα οἶνον ἐνείκω,
Il. 18.545: τοῖσι δ᾽ ἔπειτ᾽ ἐν χερσὶ δέπας μελιηδέος οἴνου
An old accusative formula embedded in the inherited traditional language that observes the digamma, μελιηδέα οἶνον, has been used to fashion a non-digammatic genitive, μελιηδέος οἴνου by the composer(s) of the Iliad who did not pronounce the digamma. There are many other examples where the original initial digamma of the same word is observed in some instances and neglected in others. The poems were composed at a stage of the Ionian dialect where the digamma was no longer present.
It’s not a case of original digammas dropping out over the course of the history of the poems and then “fixed” by later aoidoi or rhapsodes. Older formulas that metrically required initial digammas were in many cases reworked to avoid hiatus or lengthen a preceding consonant by adding particles, but this probably occurred mostly during or before the earliest stage of the composition of the poems.
In this respect, observance of the digamma was just like other archaic features of the epic language, which were modernized (in many cases converted from Aeolic to Ionic) when they could be modernized without doing violence to the meter, but were preserved when they could not. That process is thought to account for the many Aeolic words and forms that pervade the poems.
In Hesiod and in the Homeric Hymns, observance of the digamma is less frequent than in the Homeric epics. The digamma was still observed in these poems (and occasionally even in Pindar), but over time less and less so. That doesn’t mean it was pronounced.
Data can be found at R. Janko, Homer, Hesiod and the Hymns (Cambridge 1982), pp. 46-47.
Responding as I read through:
Paul,
I’m a bit of a hardliner on this. If hexameters cannot be read as hexameters, they are no longer hexameters, and can have no place in Homeric verse.
For some verses—many, both before and especially after loss of digamma—a good deal of prosodic fudging is necessary. This is often hidden by the spelling.
ου τοι επι δεος is no less metrical than εδεισε. The δ would be doubled.
|φιλε κασιγνητε could either lengthen the ι (as e.g. Απολλωνος lengthens the α, or ἠυ- the ε) or double the λ (as e.g. ελλαβε).
|επει would double the π (as e.g. οπποτε).
If |εως wasn’t metathesized it must have been made to fit somehow.
The first foot admits of more laxity/irregularity than the rest of the line (except at caesura), but it too has to be shoe-horned into the form of the dactylic hexameter.
As I thought I indicated before but maybe didn’t bother to, I agree with the analogy with French “h aspiré,” the point being that in delivery it is not aspiré. But the analogy goes only so far, since in French the loss of the aspiration is not compensated for [or not until recently; you reported—what was it?—e.g. les Halles or les Hautes-Alpes with the -s sounded, so I wasted years learning not to], whereas in Homer the loss of digamma is. Homer, after all, is in verse, i.e. is metrical, willy-nilly.
It’s true that the tradition does not always smooth out metrical irregularities, but it does so wherever it can. Performers always will have, somehow or other. (As will hearers. We shouldn’t forget them.)
Timothée understands my reasoning, except that I don’t think it was ever unmetrical, only that would have been had there not been prosodic adjustment.
By Q’s time, however, I believe (with West) that the erstwhile digamma left no audible trace.
Loss of digamma results in some awkward hiatuses, but that’s just too bad. It was accepted, and hexameters stayed hexameters.
jeidsath thinks that the focus should be on (δῖος) Ἀλέξανδρος, Ἑλένης πόσις ἠϋκόμοιο. I don’t agree. It’s well known that irregularities are particularly liable to occur at caesura. This is analogous to hiatus, and very insecure evidence (in fact I would say no evidence at all) for Fελενη, otherwise unknown to Homer.
Everyday speech hardly comes into it. Homeric verse was never that. It’s true that with the passage of time the epic language became increasingly distant from anyone’s everyday speech, but it was always distant, distanced, Kunstsprache. The effect of digamma mostly continued to be active in epic, inevitably and perpetually so, even though it was no longer pronounced. Hence verses such εκ Ελενη (assuming Bentley was right to cancel the δ’) and thousands of others, however they were accommodated to the meter. That they were accommodated to the meter I really don’t think is open to question. Otherwise they would not be there.
The argument is not that digamma had dropped out of everyday speech (in aeolic regions it hadn’t) but that it had dropped out of the pre-Homeric tradition. That’s generally accepted, and is supported by the internal evidence of the Homeric poems, where digamma is not always “respected,” i.e. was never there. I won’t revisit the argument.
EDIT. Crossed with Hylander. I have no time to respond to him now. —In fact I see there’s no need to.
Sorry, but I still find this a bit circular. When we talk about metrical laws, we talk about theoretical, idealized descriptions of what is actually happening. Reality is messy. Can we really expect 30000 lines of poetry, however good it may be, to be without a single inconsistency? Bad poetry, as found in some ancient grave inscriptions and some modern children’s books and pop songs (in Finland at least), can be full of unmetrical abominations, every second line or worse. I agree that “a good deal of prosodic fudging is necessary”, but how can we say that the result is always, always strictly metrical? “Homer” wasn’t a scholar anyway, and he was not aiming to please Homeric scholars of future generations. The “rules”, as formulated textbooks ancient and modern, are applied with remarkable consistency, but I submit that in the few remaining cases where they don’t quite work out, the poet composed something that was “good enough” for him and his audience.
We had a slightly similar example of this phenomenon in an earlier thread. Gregory Nagy is an accomplished Homerist, but his performance of A15 χρυσέῳ ἀνὰ σκήπτρῳ καὶ λίσσετο πάντας Ἀχαιούς here is blatantly unmetrical (the second foot has short-short-long; the line scans L L S S L L L L S S L S S L L). Everyone makes mistakes, that’s not my point – my point is that both you and Hylander, accomplished Homerists likewise, refused to accept that the line is unmetrical, even after this was pointed out to you. Why? Because it’s good enough – it doesn’t offend us because it has longs and pairs of short alternating. I’m not saying that this particular case would have been acceptable in antiquity as it is acceptable to us; this is just an analogy.
I’m not sure I understand what you mean by “aspiration is not compensated”.
“H aspiré” is alive and kicking, I don’t want to give you a wrong impression. But there many individual words where “incorrect” usage (according to prescriptionists) is making its entry into the vernacular. I did some googling and found out that there has been a false rumor going around that l’Académie française now accepts “les z’haricots”. I found this thread, where one of the first posters is asking whether he should say “les handicapés” or “les (z)handicapés”.
That’s a good point and well made, Paul. To paraphrase what was said elsewhere on Textkit: can we not allow Homer to be less than perfect?
Paul, first, your scansion of A15 is wrong, and Greg Nagy’s reading of this line is unobjectionable.
χρυσέῳ ἀνὰ σκήπτρῳ καὶ λίσσετο πάντας Ἀχαιούς
This line is a perfectly regular, well-formed line of hexameter. It obeys the rules of Greek poetic prosody in general and the hexameter in particular. It scans:
L S S / L L / L : L / L S S / L S S / L L.
The second syllable is a typical, and regular, example of “synizesis” and “correption”:
-
Synizesis: the spelling έῳ disguises the fact that -σέῳ is pronounced as a single syllable, as εω(ι) nearly always is in epic, as well as other genres.
-
Correption: as is more often the case than not, a long vowel or diphthong at word-end before another vowel is treated as a short/light syllable.
These are basic rules of prosody in Greek poetry, not exceptions.
And -νὰ σκ- is scanned as a long syllable: a short vowel followed by two consonants (actually, simply a closed syllable, with the syllabification -νὰσ-κήπ-).
But I’m not sure you understand mwh’s point. There are some lines that contain syllables that can’t be scanned in a regular fashion. mwh’s point is that in performance these lines would be fudged, mostly by stretching out otherwise short syllables or (less frequently, I think) truncating long syllables.
How do we know this? In many cases, the spelling of our texts demonstrates exactly how the fudging occurred: ελλαβε, οπποτε. In other cases, the spelling doesn’t reflect the fudging, but it must have happened because every line would have to be spoken or sung as a hexameter.
Your examples:
Il. 5.539 φίλε κασίγνητε κόμισαί τέ με δός τέ μοι ἵππους
Il 23.2 ἐπεὶ δὴ νῆάς τε καὶ Ἑλλήσποντον ἵκοντο
The problem in φίλε κασίγνητε is φίλε. As mwh notes this would have been sung or spoken as φιλλε, just like ελλαβε, οπποτε. Similarly, εππει in 23.2.
κασίγνητε isn’t really a problem. With some frequency, short syllables are allowed to occur before the pause at a penthemimeral caesura, just as short syllables occur all the time in the second syllable of the final spondee, before the pause at the end of the verse.
You’re right that some lines don’t scan on the written page if you mark out the longs and the shorts, but in performance they would have been fudged to make them scan, and in many cases the orthography shows us very clearly how this was done. Of course, the orthography of our texts is undoubtedly different from the orthography of the “original” poems, which didn’t reflect long consonants, but our texts show how the fudging was performed in a tradition of Greek verse performance that went back with continuity to the original performances.
mwh’s point, and mine, is not that “Homer” is perfect – (s)he’s not. But the little imperfections that crop up would have been smoothed over in performance, by cheating a little, so that every line would come out as a hexameter.
I know how to scan hexameters. You misunderstood my point completely. I gave the scansion as read by Nagy, not the “correct” one, which is of course what you give. The problem in Nagy’s reading is that he reads -σέῳ, the syllable in correption, as a long, when it should be read short. To compensate, he reads ἀ-νὰ-σκήπ-τρῳ short-short-long-long, i.e. he reads -να short. The result is something the sounds superficially like a hexameter but actually isn’t one - the scansion being what I gave earlier.
I’m not saying that this is exactly the sort fudging that might have occured in antiquity - what I mean is that if you are willing to accept this fudged hexameter that doesn’t scan, something analogous may have happened occasionally in antiquity as well.
Il. 5.539 φίλε κασίγνητε κόμισαί τέ με δός τέ μοι ἵππους
West mentions this line in his Greek Metre, and he too thinks that the first syllable φι is lengthened, as well as -τε in the next word. In cases like these, the Homeric metric licence is just about at its biggest. Also the LS mentions φῑλε. The lengthening apparently allowed only in the vocative, which probably point to a formula or a half.
Bill, your analysis of A 15 is slightly misleading. You speak of synizesis + correption, but analyse it as
χρυσε’ ἀνὰ σκήπτρῳ…
i.e., without synizesis, rather -ῳ elided.
PS. We have spent some Homeric nights with Paul, and—take it from me—he has scanned excellently.
1.15 doesn’t need to be fudged at all. It’s a perfectly well-formed line of hexameter. Synizesis is simply a matter of spelling, and correption is a regular feature of Greek prosody. It’s sometimes referred to as “epic correption,” but it occurs in other genres, too.
I listened to Nagy, and didn’t hear anything wrong with his articulation. He scans the line the way I scan it.
But now I’m not sure what your (Paul’s) point is. No one is claiming that “Homer” is perfect, but the imperfections would have been smoothed out in performance so that each line would be read or sung as a hexameter, and the spelling of our texts shows how this was done in some but not all cases.
Bill, your analysis of A 15 is slightly misleading. You speak of synizesis + correption, but analyse it as
χρυσε’ ἀνὰ σκήπτρῳ…
i.e., without synizesis, rather -ῳ elided.
I don’t analyze it that way at all, and I don’t see how you drew that conclusion from what I wrote. σεωι is pronounced as a single syllable, with εωι treated as a triphthong (synizesis), and the entire syllable is truncated (corrreption). Greg Nagy does this in his reading, to my ear. (Admittedly, I’m somewhat hard of hearing and I didn’t have my hearing aids on when I listened to him.)
Sorry, you’re right. All these different levels of analyses managed to confuse me. My apologies.
But in Nagy’s reading, the syllable/vowel after χρυ- is definitely long, as Paul says. His scanning doesn’t sound bad at all, but is it historically quite correct?
I don’t hear it as long – he pronounces it as a triphthong, but he truncates it properly.
And, by the way, I think the Spartan meathook was for helots.
I’m sure Nagy intended to scan it exactly as Hylander says, but he is so intent on demonstrating the synizesis that he forgets the correption, so to speak. Anyway, his intended audience, native speakers of English, can’t make the difference, which I think is a thing demonstrated, the distinction short vs. long vowels not being a feature of your phonetics. He also gets away with it because he unconsciously makes -να- scan short to compensate.
What Nagy says:
Khruu-syoo-ya-na-skeep-trooy
What he intended to say:
Khruu-syo-ya-nas-keep-trooy
I only meant that if a good poet uttered a verse that doesn’t quite scan every 1000 lines or so (and I don’t mean A15, which is a perfectly regular hexameter verse), would that really have bothered his audience so much? If Nagy gets away with an unmetrical line, how do we know Homer didn’t, when his audience weren’t trained philologists but people who wanted a good show? Most likely φιλε κασιγνητε was pronounced with the first iota long, if not by all rhapsodes, at least some of them, but I wouldn’t be so uncompromising as to completely exclude occasional unmetrical performances, like Nagy’s here. Something that isn’t a hexameter can still sound close enough.
I’m getting a hearing aid as well, btw. Can’t hear mosquitoes any more, which is a blessing, but I think I still recognise a long vowel when I hear one…
I find myself hoping lately that a crying child can’t cause hearing loss. Given family history though, it will probably go eventually.
E.B. Clapp presents some statistics for hiatus in Homer, for Τ-Ω (I’m haven’t checked these numbers myself). He claims about 2000 instances.[1]
- ~400 from a lost consonant, mostly digamma.
- 7 cases of synizesis, where two words are almost fused, like ἦ οὐχ
- 66 cases of a vowel already being elided off of the first word
- 1122 cases of epic correption (he calls it semi-elision)
- 41 cases of the first word ending in ι or υ
- 175 cases of hiatus at the masculine caesura
- 100 cases of a pause in sense which we mark by punctuation
- 45 otherwise unexplained
The two categories that would have to be fudged in performance are (1) & (8). If Homer’s fudge for cases of (1) was something other than audible digamma, it’s surprising that category (8) does not represent a much larger proportion of lines. That is, to repeat Paul’s question, why didn’t he use this fudging technique in composition as well? If his fudge was an audible digamma, then the problem is solved, of course.
However the above does seem to indicate that we aren’t seeing too much metrical flattening in transmission, at least in regards to hiatus. Otherwise (1) and some of the other categories would have gone to zero as they were “fixed.” So I think that Paul’s other claim is unlikely, at least as far as hiatus goes.
[1] https://books.google.com/books?id=9s1fAAAAMAAJ&lpg=PA430&ots=S1mGZH8Xug&pg=PA430#v=onepage&q&f=false
Joel, there was no need to “fudge” for hiatus in performance. Hiatus doesn’t break up the rhythm of the hexameter. The singer or speaker could move on to the next word and not let himself be bothered by the hiatus. There was no need for “audible digammas.” Where the loss of a consonant left a short open syllable in a slot where a long syllable was needed, he could stretch the short syllable or insert an elided particle.
For the most part, “Homer” didn’t compose by stringing together individual words–he composed with groups of words that came pre-packaged as part of the traditional language, and these packages of words generally conformed to the metrical patterns of the hexameter. Some of the packages were very old and, in “Homer’s” day, included hiatuses resulting from lost consonants or other metrical irregularities. Sometimes he would stick in a particle to correct the metrical irregularities, sometimes he didn’t let the hiatus bother him and moved on, sometimes he would fudge, drawing a short syllable where a long was needed or shortening a long syllable where a short was needed (ανδροτητα και ηβην). The poems are full of these fudges, some of which are reflected in the orthography.
“Homer” wasn’t working from a written set of rules. The language in which he composed conformed more or less (mostly more) to the rhythm of the hexameter. The “rules” are simply a codification of the observed metrical patterns of his verse. But when he needed to, when the packages of words no longer fit the rhythm of the hexameter, “Homer” could fudge to keep the rhythm of the hexameter intact.
The singer or speaker could move on to the next word and not let himself be bothered by the hiatus.
I would expect category (8) to be orders of magnitude larger if this didn’t bother the singer or his audience. And again, if he tolerated hiatus, so long as they came from traditional material, I would expect (8) to be orders of magnitude larger. But from the above, it looks like non-metrical phrases didn’t survive in the oral tradition very long at all, except in the case of digamma.
I take Paul’s point about Homer’s verses not being all well formed (far from it!), but the fact that they’re all “good enough” for poet and audience, as he well puts it, is what counts. For me that means that they’re acceptable as hexameters. Fudging (by performer and audience alike) makes them fit the meter, so they’re metrical. Any Homeric line can be scanned in accordance with the metrical scheme. We can observe what the limits of metricality are when those limits are stretched to breaking point. I think the difference between us was merely definitional, or mostly so. Perhaps we should have been talking about rhythm rather than meter. Aristoxenus talks of language as το ρυθμιζομενον; the poet’s material is plastic.
As to the frequency of hiatus consequent on loss of digamma. Well yes I think we have to say that by Homer’s time the rules have changed. Once the digamma’s gone and no longer prevents hiatus, then hiatus that previously would have been barely tolerable at best is now accepted, somewhat reluctantly perhaps but nonetheless accepted, as part and parcel of Homeric versification; with digamma’s disappearance the norms have shifted.
For
χρυσέῳ ἀνὰ σκήπτρῳ καὶ λίσσετο πάντας Ἀχαιούς
with less severe compression of the pronunciation, I am tempted to suggest an older form:
χρυσέῳ ἂν σκήπτρῳ καὶ λίσσετο πάντας Ἀχαιούς
“an” for “ana” also occurs in ἄμπνυε.
Some Homeric lines were likely composed so long ago and passed on by generations of oral poets, that many phonetic and linguistic changes had meanwhile happened.
Likewise ἀνδροτητα, which intrudes a cretic into a common Homeric formula, was likely long ago pronounced ἀνρ̥τητα (ἀ-νρ̥-τη-τα) or ἀνρ̥τᾱτα, with a vocalic R.