Recovering melodies from ancient Greek documents

Are you reading Homeric Greek? Whether you are a total beginner or an advanced Homerist, here you can meet kindred spirits. Besides Homer, use this board for all things early Greek poetry.
Post Reply
User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Recovering melodies from ancient Greek documents

Post by jeidsath » Tue Jun 07, 2016 9:01 pm

I have been considering a statistical project over the last few days, and I hoped to get some input on it.

My guess is that if an original melody line existed for any document of sufficient length, and it had some effect on words selected (according to their accentuation), it becomes a reasonably straightforward machine learning problem to recover some aspects of the original melody. In fact, if you assume that the likelihood of accent correspondence increases with the magnitude of the tonal difference, you may be able to recover something close to an actual melody line.

The main difficulties would be:

1) Find documents of sufficient length that are composed against a particular melody.
2) Discovering the number of versus before melodic repeat.

The Iliad and Odyssey would be ideal, due to length and a simple meter with no questions about responsion. However, I don't know if they were originally composed against a melody (or more likely a set of melodies), regardless of whether they were performed with or without melody at a later point.

Thoughts about good texts to attempt this against? Has anybody taken steps in this direction in the past?
Joel Eidsath -- jeidsath@gmail.com

Markos
Textkit Zealot
Posts: 2966
Joined: Sun Jun 21, 2009 8:07 pm
Location: Colorado
Contact:

Re: Recovering melodies ancient Greek documents

Post by Markos » Tue Jun 07, 2016 9:39 pm

jeidsath wrote:Thoughts about good texts to attempt this against?
1 Tim 3:16:
θεὸς ἐφανερώθη ἐν σαρκί, ἐδικαιώθη ἐν πνεύματι, ὤφθη ἀγγέλοις, ἐκηρύχθη ἐν ἔθνεσιν, ἐπιστεύθη ἐν κόσμῳ, ἀνελήφθη ἐν δόξῃ.

User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Re: Recovering melodies from ancient Greek documents

Post by jeidsath » Tue Jun 07, 2016 10:10 pm

@Markos It couldn't work for that, I'm afraid. Thousands of lines with a similar melodic pattern are probably necessary. And with Christian verse, it's not possible to lock down syllables to specific points in the meter.
Joel Eidsath -- jeidsath@gmail.com

mwh
Textkit Zealot
Posts: 3366
Joined: Fri Oct 18, 2013 2:34 am

Re: Recovering melodies from ancient Greek documents

Post by mwh » Tue Jun 07, 2016 11:31 pm

A melodic line is often hypothesized for the Homer by analogy with comparable traditions, where it repeats (with many ?accent-influenced variations) verse by verse I think. Martin West did his best with it in a bold article in Journal of Hellenic Studies quite a few years back. Try his Anc.Gk.Music book? The South Slavic epics might be your best bet (though of course there are many others). Albert Lord made recordings which are still available. John Miles Foley, who followed in Parry and Lord’s footsteps in Yugoslavia and died only two or three years ago still in his sixties, did much work in comparative oral traditions (including Mongolian, which could also be useful to you?). You could check out his work and the journal he founded, Oral Tradition.

User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Re: Recovering melodies from ancient Greek documents

Post by jeidsath » Wed Jun 08, 2016 3:16 pm

Thanks for the references. It seems that Ancient Greek Music discusses this on pages 208-9 (and mentions the Slavic epics), but doesn't footnote any paper of West's. A single repeating line of melody would be much easier to recover statistically (if it affected accentuation at all), but I would expect that someone would have already done so by now -- you wouldn't even need a computer for that. I'll see if I can find the JHS paper, and the other references that you mention.
Joel Eidsath -- jeidsath@gmail.com

User avatar
Paul Derouda
Global Moderator
Posts: 2148
Joined: Fri Nov 05, 2010 9:39 pm

Re: Recovering melodies from ancient Greek documents

Post by Paul Derouda » Wed Jun 08, 2016 3:51 pm

The article must be The Singing of Homer and the Modes of Early Greek Music, Journal of Hellenic Studies, vol 101 (1981), pp. 113-129. A rather heavily abridged version of the article is also to be found in Hellenica vol 1.

I have been advertising Stefan Hagel ad nauseam, but I think what he has done is somewhat close to what you have in mind. http://www.oeaw.ac.at/kal/sh/

User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Re: Recovering melodies from ancient Greek documents

Post by jeidsath » Thu Jun 09, 2016 1:57 am

Thank you for the article link. I've made the suggested melody from the article into sheet music. You can hear it here on Noteflight.

Stefan Hagel has taken the same approach as West (he has built a melody from the accents using ancient scales and the Seikilos accent-note dynamic connection), but he reduces note variation approaching the end of lines, citing modern pitch-language singers. His result is more successful than West's, in my opinion.

The approach that I mentioned at the start of this thread would be the complete reverse of what they are doing. My guess is rather than Homer basing melody entirely off of accent, he sometimes chose words to fit a preconceived melody. This is what is happening is Seikilos, I believe. That haunting line was not the result of a random conjugation of pitch accents.

If some small percentage of accents match a pre-conceived melody, it should be possible to go in the other direction, and recover the melody through a Bayesian inference algorithm.

Image
Joel Eidsath -- jeidsath@gmail.com

User avatar
Paul Derouda
Global Moderator
Posts: 2148
Joined: Fri Nov 05, 2010 9:39 pm

Re: Recovering melodies from ancient Greek documents

Post by Paul Derouda » Thu Jun 09, 2016 8:02 pm

Thanks for the music! Is it possible to change the instrument to a string instrument?

User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Re: Recovering melodies from ancient Greek documents

Post by jeidsath » Thu Jun 09, 2016 8:22 pm

I've set it back to guitar. I had used clarinet before since it's like an αὐλός and the guitar does not sound good. There is no harp option, I'm afraid.
Joel Eidsath -- jeidsath@gmail.com

User avatar
Paul Derouda
Global Moderator
Posts: 2148
Joined: Fri Nov 05, 2010 9:39 pm

Re: Recovering melodies from ancient Greek documents

Post by Paul Derouda » Fri Jun 10, 2016 7:47 pm

Thanks! I still find it hard to conceive what it might sound like if someone was singing, though...

User avatar
Scribo
Global Moderator
Posts: 886
Joined: Fri Feb 22, 2008 2:28 pm
Location: Between Ilias and Odysseia (ok sometimes Athens).

Re: Recovering melodies from ancient Greek documents

Post by Scribo » Fri Jun 10, 2016 8:58 pm

I'll be back at home in a few days, I'll do a quick run through (of West's schema) on my kithara, though I have some disagreements with it and the kithara has much better projection than the phorminx.
(Occasionally) Working on the following tutorials:

(P)Aristotle, Theophrastus and Peripatetic Greek
Intro Greek Poetry
Latin Historical Prose

Hylander
Textkit Zealot
Posts: 1938
Joined: Mon Aug 17, 2015 1:16 pm

Re: Recovering melodies from ancient Greek documents

Post by Hylander » Tue Jun 14, 2016 12:50 pm

Without having read West's article (and recognizing his pre-eminence and my ignorance in the field of Greek music), I would have to question whether his effort at reconstruction bears any resemblance to what an accurate transcription of an aoidos singing or chanting the Iliad would have looked like. I suspect a singer would have been much freer, giving less regular time-values to the individual syllables. The reconstruction seems to proceed like a march, only in 7/8 time! Or rather 3+2+2/8.

While the rigid bar lines reproduce the metrical schema of the hexameter foot by foot, they don't capture the internal dynamics of the hexameter, which aren't based on metrical feet, but rather on cola articulated by significant word-breaks, most importantly caesuras. These don't seem to be reflected at all in the reconstruction.

And would the singer have actually improvised a melody based on the accentual contours of each verse? I imagine something more like a chant on a more or less fixed pitch, with slight variations of pitch giving effect to the Greek pitch accent, and perhaps some melodic inflection at the beginning and end of the verse, like psalm tones (although that would have gotten in the way of enjambment).

Timothée
Textkit Enthusiast
Posts: 564
Joined: Fri Oct 09, 2015 4:34 pm

Re: Recovering melodies from ancient Greek documents

Post by Timothée » Tue Jun 14, 2016 6:04 pm

Hylander wrote:While the rigid bar lines reproduce the metrical schema of the hexameter foot by foot, they don't capture the internal dynamics of the hexameter, which aren't based on metrical feet, but rather on cola articulated by significant word-breaks, most importantly caesuras. These don't seem to be reflected at all in the reconstruction.
What strikes me here is that West obviously knew the significance caesura in dactylic hexametre perfectly well; it is lucidly opened in his Greek Metre, for instance. Did he, therefore, have some particular reason for his deductions? I always feel terribly uncomfortable criticising W., but it has to be done when it has to be done. (His Helicon, for instance, contains, it must be said, much Quatsch.)

mwh
Textkit Zealot
Posts: 3366
Joined: Fri Oct 18, 2013 2:34 am

Re: Recovering melodies from ancient Greek documents

Post by mwh » Tue Jun 14, 2016 9:30 pm

I think it might be as well to read his article before criticizing him.

Hylander
Textkit Zealot
Posts: 1938
Joined: Mon Aug 17, 2015 1:16 pm

Re: Recovering melodies from ancient Greek documents

Post by Hylander » Tue Jun 14, 2016 9:39 pm

West does recognize that the note values are approximate.

Hylander
Textkit Zealot
Posts: 1938
Joined: Mon Aug 17, 2015 1:16 pm

Re: Recovering melodies from ancient Greek documents

Post by Hylander » Sat Jun 18, 2016 3:01 pm

I've read the article. West acknowledges that the note values he assigns are approximate. In that respect, the bar lines West puts in his musical example are misleading, and in Greek Music he makes exactly the point that ancient Greek metrical feet aren't analogous to modern bars. I'm still a little uncomfortable with some of his speculative leaps--although he recognizes that what he writes is speculative. One thing that troubles me is the instrumental flourishes between verses--wouldn't they disrupt enjambment, especially dramatic enjambment in places like Il. 1.52? West writes that the South Slavic singers recorded by Parry insert these flourishes between verses even when the syntax carries over to the next line. But I'm not sure that the practices of Homeric aoidoi can necessarily be inferred from those of 20th century South Slavic singers--that's the sort of argument that I find troubling when Nagy and his adherents do it (e.g., "Homer" couldn't possibly have been active when writing became available because the South Slavic singers lose their ability to compose orally when exposed to writing"). Another question I have: would the pace of singing (as opposed to some sort of recitation) be too slow to retain an audience's interest over a large chunk of hexameters? But that's just some random thoughts.

mwh
Textkit Zealot
Posts: 3366
Joined: Fri Oct 18, 2013 2:34 am

Re: Recovering melodies from ancient Greek documents

Post by mwh » Sat Jun 18, 2016 7:37 pm

I share these reservations and queries. As to enjambement, however, it’s already disrupted by the metrical independence of each individual verse. Whether or not there was “pause,” let alone instrumental twiddles, the fact that there’s no inhibition against having even enjambed verses end in a light syllable shows that prosodic continuity is broken. E.g. 1.1-2, 3-4, etc etc. Or could it be argued that in such cases, unlike 51f., the thought and the syntax of the previous line is at least potentially complete, so that the next line is a sort of disconnected add-on? I think I once considered this and found it didn’t work, but it might be worth having a new look? In Sappho for instance each line of a stanza runs on to the next even where word-end is invariable (this would appear to make each stanza a single verse, contra West), so perhaps it’s feasible that a minority of lines in Homer did the same? But how about (just from a random glance at the text) 1.111f., 113f., where admittedly the enjambement is less extreme than at 52? Pindar and Bacchylides will sometimes enjamb across stanza boundaries rather than end-stopping, deferring sentence-end to the beginning of the next stanza, e.g. Bacch.3.29 …Κροισον ο χρυσαορος ||| φυλαξ’ Απολλων., 51 … ανα ματρι χειρας ||| εβαλλον·. This at least looks parallel to Il.1.52 …βελος εχεπευκες εφιεις | βαλλ·, and I think it’s reasonably safe to assume there was at least a break of some kind. Comparative evidence goes well beyond 20th-cent. guslars. West’s music book is perhaps a better guide than his earlier bravado article. His second-generation pupil D’Angour (he whose self-advertising memoir of West was just linked by Timothée in the MLW obit thread) has done some excellent work here. These are just some even more random thoughts.

User avatar
Peitho
Textkit Neophyte
Posts: 16
Joined: Fri Jun 03, 2016 12:36 am

Re: Recovering melodies from ancient Greek documents

Post by Peitho » Fri Jun 24, 2016 12:13 am

Stephen G. Daitz has a good 2-hour audio guide to reading Greek (“The Pronunciation and Reading of Ancient Greek: A Practical Guide,” cassette tape with accompanying booklet) that touches on a bunch of the concepts of Hagel and West in an easy-to-digest sort of way. It might be useful as a point of reference. The trouble is that I have no idea how to get it these days. I got it through a torrent myself.

ailuros
Textkit Neophyte
Posts: 91
Joined: Fri Sep 16, 2011 8:40 pm

Re: Recovering melodies from ancient Greek documents

Post by ailuros » Fri Jun 24, 2016 11:21 am

took about 3 seconds to find on amazon us. please don't encourage others to rob mr. daitz's estate.

User avatar
jeidsath
Administrator
Posts: 3342
Joined: Mon Dec 30, 2013 2:42 pm
Location: Γαλεήπολις, Οὐισκόνσιν

Re: Recovering melodies from ancient Greek documents

Post by jeidsath » Sat Jun 25, 2016 2:28 pm

I edited the above to remove the mention of the torrent name. (I'm not sure, however, that profit from this particular Amazon transaction goes to the Daitz estate so much as the Jeff Bezos estate. These tapes must be long out of print.)
Joel Eidsath -- jeidsath@gmail.com

ailuros
Textkit Neophyte
Posts: 91
Joined: Fri Sep 16, 2011 8:40 pm

Re: Recovering melodies from ancient Greek documents

Post by ailuros » Sat Jun 25, 2016 6:04 pm

that's a good point. the re-sellers do, though, help estates out by buying collections at sale. but, yeah, amazon is probably ruthless with these merchants.

Post Reply