Via Laudator Temporis Acti, a quote from Bryan Hainsworth, The Iliad: A Commentary, Volume III: Books 9-12 (Cambridge: Cambridge University Press, 1993; rpt. 2000), pp. 6-7:
A true hapax legomenon seems to present a special problem for those who believe that the techniques of composition used in the Homeric poems are mainly those of oral poetry. The techniques of oral poetry are generic and formular, the hapax legomenon by definition is not. It may not even bear any relation of sound, sense, or form to the formular part of the diction, and it would be gratuitous and implausible to claim that more than a handful make their sole appearances by chance. On the contrary, hapax legomena, being an aspect of the vitality of the Kunstsprache, and of the willingness of ἀοιδοί to experiment with their lexicon, must be accommodated in any satisfactory account of Homeric diction.⁵ Here then the question is how hapax legomena can be deployed in a sentence otherwise made up of formular elements by a composer who relies heavily on such elements. When it is put in that way the problem posed by a hapax legomenon for the singer is not radically different from that posed by an otherwise unused grammatical form of a regular part of his lexicon. The unique grammatical form will indeed bring with it the verbal associations of the regular forms, but since the associated words and phrases would be built around the particular metrical shape of the regular forms they are likely to be as much a hindrance as a help in handling the unusual form.
The scale of the problem presented by true hapax legomena and by many uniquely occurring grammatical forms is quite serious. The printed text of the Iliad is made up of some 111,500 words, i.e. segments of text marked off by verse-ends or spaces, or about 63,000 if particles, pronouns, and prepositions are ignored. Many of these ‘words’ are repeated, but about 11,000, or more than one in six, are found once only. About 2,000 of them according to M. Pope are true hapaxes, lexical items occurring just once in the poem.⁶
⁵ See M.M. Kumpf, Four Indices of the Homeric Hapax Legomena (Hildesheim 1984) for statistics, N.J. Richardson in Bremer, HBOP [Homer: Beyond Oral Poetry] 165-84, for argument, Edwards, vol. v 53-5. Edwards concludes his discussion of hapax legomena with these words: ‘[Homer] was also completely at ease in employing in his verse words which are not only non-formular but which must be considered (on our limited evidence) foreign to the usual epic vocabulary.’ M. Pope, CQ 35 (1985) 1-8, draws attention to new coinages in Homer.
⁶ ‘Word’ is used here as a publisher might speak of a ‘book of 80,000 words’. The composer’s vocabulary or lexicon of course is very much shorter: ἔγχος is one entry in the lexicon but supplies 205 ‘words’ to the text of the Iliad. Statistics are mine. I am indebted to the Revd A.Q. Morton, formerly of the University of Edinburgh, for making available to me computerized word-lists and indices.
An interesting issue I don’t think I had ever thought about.
I am not convinced by this at all, statistically: no matter how “formulaic” a text may be, you’re going to get a fair number of hapax legomena by pure chance (unless it’s a very boring text indeed, and moreover has just a single unvarying theme throughout); and if one means by “hapax legomenon” that the word is not found outside Homer either, the age and dialectal nature of the poetry (Ionic with stray bits of Aeolic and maybe even other dialects) makes it not too surprising that there would be a good number of such cases.
There also seems to be a kind of implicit assumption that the poems, so far as they are based or oral tradition (and in their complete totality as we now have it this is only partly the case), all go back to the same oral bard. That is a romantic idea but unfortunately seems quite implausible. Different griots may very well have have their own particular individual personal linguistic tics even when being as formulaic as all-get-out. Especially as they clearly didn’t even all speak the same dialect, even before you get to personal idiosyncracies. The idea that proto-Homeric bards, coming from different regions and different historical eras, all formed a sort of linguistic monoculture is surely ridiculous.
new coinages in Homer
How would we even know? This is worse than the Shakespeare-as-great-neologist story.
(The Rig-Veda is absolutely full of words not previously documented in Indic: new coinages everywhere you look …)
How would we even know?
I wonder if there are any Homeric hapax legomena that turned up in Linear B texts…
(Might be hard to tell, admittedly, given how hilariously bad the Linear B script was at representing Greek phonology.)
Rather than the first, the Homeric epics surely contain the last occurrences of a great many Greek words that had, perhaps, already retreated into just this archaic/poetic/epic register.
I’ve posted this handout on counting Homers here before, but once again Google refuses to find it (and what I said about it; I think the answer was “at least 10”, but I don’t want to count again). Next time I’ll try a few other search engines.
And not Classical Ionic either! Definite articles are apparently mostly limited to the most obvious latest additions, and on the other extreme there’s ἀνδροτῆτα scanning with a short first syllable – it had one back when it was *[anr̩ˈtaːta]!
I’ve come across the idea that the Aeolic bits were patched in to repair the worst mismatches with the meter that had resulted from such developments.
Can we not at least have a fake pseudo-Classical English plural of “hapax” that’s more pretentious than “hapaxes”?
Hapacia.
Hypyxi
a fake pseudo-Classical English plural of “hapax”
You mean, a halp-axed one?
fake pseudo-Classical English plural
Been there, done that.
I’ve posted this handout on counting Homers here before
October 14, 2023?
1 in 6 is a very low frequency. Should be about 1 in 2. Or at least this is what auntie Wiki says.
Maybe because the Iliad is so formulaic.
Speaking of that, I decline to believe that the 14-billion-word iWeb corpus of English contains anywhere near 7 billion hapaxes.
Wikipedia says, “According to classical scholar Clyde Pharr, “the Iliad has 1097 hapax legomena, while the Odyssey has 868″.[19] Others have defined the term differently, however, and count as few as 303 in the Iliad and 191 in the Odyssey.[20]”
On the original topic, if I understand the quoted material correctly, the author says hapaxes pose a problem for the composer of the line. I don’t get that. Even if the singers frequently relied on formulas, they must have been capable of composing non-formulaic metrical lines. The history of popular music shows that people who have never been taught what a dactyl and a spondee are can compose words to fit a tune. Even for a formulaic line, all you have to do is find a word that fits into the metrical slot of the word it replaces. Or if someone puts in a word that doesn’t fit well metrically, some later singer who thinks it’s an appropriate word may be able to regularize the line. I can see that words that were rare when they were put in the poem, or not used all at other times and places when they were sung, pose a problem for other singers’ memory, but I don’t see the problem for composition. Am I missing something?
And has anyone suggested that ἀοιδός is the origin of “AOI” in The Song of Roland?
I decline to believe that the 14-billion-word iWeb corpus of English contains anywhere near 7 billion hapaxes
Probably not. Obviously the rule can reasonably be applied only to the cases where the corpus size is less than the number of words (however defined) in a language. English has a great capacity for words, but probably not more than a million. Wiki cites Brown corpus that has 50 thousand words and 50% hapaxes and Moby Dick with a bit more than 200 thousand words and 44% of hapaxes.
P.S. About a million words in English. By thinking a bit more about it, maybe it’s an upper estimate for the number of “head words”. Let’s say with various forms, ten million tops.
Yes, but that’s not what I was thinking of; I thought I had commented it a bit more.
I decline to believe that the 14-billion-word iWeb corpus of English contains anywhere near 7 billion hapaxes
It has 14 billion tokens, but far fewer* types. If hapaxes make up 50% of the types, they are a tiny fraction of the tokens.
*A cursory Google did not find any numerical estimate.
Thanks, mollymooly. So we don’t know what the fraction is for the Iliad. (And “tokens” and “types” are words that Hainsworth could have used, though he may have purposely avoided them as too technical or something.) Anyway, I’m getting a better idea of how the claim of 40–60% hapaxes could be true for a big corpus.
@DE. Thank you for ‘ griots’. Had to look it up. A fine hapax legomenon. Or happenstance.
Griot at LH.
As it happens this past Saturday night I saw a musical ensemble whose style was supposedly praised by the Village Voice as “a futuristic, psychedelic, neo-griot frenzy.” They are in fact a bunch of (majority-white) American dudes who like and adapt various West African musical styles associated with Mali and Guinea and the Ivory Coast etc. But it’s still rather purple prose.
I’m suspicious — googling the phrase, or even just “neo-griot frenzy,” gets only Toubab Krewe sites with the attribution, and I’m pretty sure the entire run of the Voice has been digitized.
“Neo-griot frenzy” looks like something out of a UK-style cryptic crossword clue.
Neo-griot frenzy, but drug-free, for an old-fashioned parent (7)
You’re saying I can’t trust Toubab Krewe to be accurate about the attribution of their blurbs?
Sorry, should be
Old-fashioned parent loses a bagel in neo-griot frenzy (7)
(The original version might have appeared in the Guardian, though.)
Note neo-griot has as anagram: E[ddyshaw] rooting.
Spooky!
So we don’t know what the fraction is for the Iliad.
I am going to believe (provisionally) this
Reddit poster
According to Perseus the Iliad has 8112 unique words and 1357 hapax legomena.
The Odyssey has 7291 unique words and 1198 hapax legomena.
The two combined have 9954 unique words and 1541 hapax.
FWIW, if you actually use the search function on the VV’s own website you will find a 8/21/2013 thing with Richard Gehr’s byline saying inter alia “A Krewe show usually opens with a percussive introduction; gradually picks up velocity as more surf, jazz, dub, and hip-hop ideas are stirred into the mix; and ultimately climaxes in a psychedelic neo-griot frenzy.” https://www.villagevoice.com/toubab-krewe-2/
“Doctor Who’s ‘Stones of Blood’ lost among neo-griots.”
Thanks, D.O. Still a much lower fraction than the Wikiparticle says is typical.