My experiences teaching myself ancient Greek pitch accent

Greek Learning Greek
1

I thought that it might be useful to perhaps a few people on this site to share my experiences learning ancient Greek pitch accent and to give some humble advice. I’ve been working on this for a few years now (yeah, I’m a slow learner), and I’ve had to do it completely alone as what I did find online was mostly of little use and often ridiculous.

There may be members of this site who have thought about trying pitch accent or have attempted it, or maybe even had some success. If so, I would be interested in learning about their experience.

Mastronarde and Allen both discourage learning pitch accent. I understand why - it would take up a lot of time that you could use for just learning Greek, and you will never really know if you have it right anyway. Plus, there is no one to speak to anyway.

The process of learning pitch accent for me was iterative (as is all language learning). I would think that I had it, and then realize months later that I didn’t have it, and many times I had to go back to the beginning and start from a blank slate.

Along the way I studied Serbo-Croatian and even travelled around Serbia listening as much as I could for pitch accent. It comes and goes in Serbo-Croatian depending on the speaker and the region, so it was a frustrating experience at times.

Go to the livelingua site and grab the free FSI Serbo-Croation course. The introductory material has a lot on the contonations involved, the morae, and a lot of audio samples of the various pitch cases. Serbo-Croatian pitch accent is probably not exactly the same as Greek pitch accent, but at least you will be able to get an idea of the subtle nature of it. (FSI material, although severely dated, is of the highest professional quality.)

Study unit 2 of Mastronarde to become familiar with the contonation, which is the one or two syllables in a word which are subject to a raised pitch and a fall back to normal pitch. Mastronarde does an excellent job explaining the contonation.

(Knowledge of Chinese will be of no use as the syllabic tonal system in Chinese is completely different from Greek pitch accent.)

My goal was to be able to speak in a way that was not exhausting to me and was not startling to the listener. I believe that it should take no more effort to use it than to use the English stress accent, and that it should be rather subtle. Often I will use pitch accent with almost no pitch movement at all. The Czech language is a good example of “pitch accent without pitch.”

I believe that the most important thing about pitch accent is to eliminate as much as possible the stress accent of English (or of modern Greek) and to concentrate on giving each syllable an equal weight and a correct length. Once this has been accomplished, pitch rise and fall can then be applied to the contonation, but in a subtle way. It should not sound like singing. I was after a normal speech that would be used on the street in ancient Greece.

Every Greek word (other than enclitics and proclitics) has a written accent mark. This does not mean that every word that has an accent mark should be pronounced with a noticeable pitch rise and fall. I suspect that in a Greek sentence only the words which are emphasized, that have phrase stress, would have received a noticeable pitch contour. The FSI Serbo-Croatian course goes into that idea.

At this point, I am able to switch between ancient Greek pitch accent and modern Greek stress accent at will. They are completely orthogonal. I have no idea if have got it right, but then nobody else does either. In this, you are on your own.

2

Very impressive katalogon!

Pitch accent is very difficult I think, and I say that as a person whose native tongue is arguably a pitch accent language (Swedish). I try to learn to pronounce Greek words with a pitch accent, but it’s not natural, so it typically ends up with a stress, and a bad one at that.

Something I’ve been looking for a very long time is a resource where native speakers of different languages pronounce the letters - or rather IPA symbols - common to their particular language. I think that’d be a pretty nice way to piece together a full phonetic alphabet not only relevant to ancient Greek but all languages. That’d be better than having one speaker do the sounds for the full alphabet, which is sometimes a bit awkward.

For example, a pronunciation mistake I often find in English youtube videos about ancient Greek is eta (η), being pronounced as “Hey dude” or as some other diphthong, instead of open mid [e] like the french word “tête” (here I quote Vox Graeca by Allen, page 70). Here’s a video for french tête which approximate η and is how I pronounce eta: https://www.youtube.com/watch?v=8H6pfZlPLC0 (this is particularly easy for me, because this is more or less one of the most common sounds in the Swedish language but even so, an accented η is challenging for me to pronounce in the Greek context).

I highly appreciate your pointer to the FSI Serbo-Croatian course, I think that will help me expand my working knowledge of phonetic sounds.

3

Hi, I also tried to learn the pitch accent, especially from Devine & Stephens 1994 on pitch contours. It might be a good idea to start by listening to Stefan Hagel reading the Symposium here, and seeing where your approach differs with his:

https://www.oeaw.ac.at/kal/agp/

Some of your assumptions differ from what I have read in the scholarship (and therefore from Hagel’s reconstruction linked above), e.g. where you say “I suspect that in a Greek sentence only the words which are emphasized, that have phrase stress, would have received a noticeable pitch contour.”

PS I’m not picky about pronunciation methods—any method (reconstructed or otherwise) applied consistently seems fine to me. It may be because I’ve never actually been in a classics classroom or spoken Greek (or Latin) to anyone in the real world (i.e. other than on this forum or over emails), and so for me pronunciation is simply part of my personal appreciation of the language—if I had chosen a different method, it wouldn’t have mattered that much.

Cheers, Chad

4

Very interesting topic katalogon!

I agree that Serbian/Bosnian is a good place to look for ideas about how Greek accent might have sounded. Lithuanian and Latvian would be a couple more to consider. I would suggest visiting sites for voice-over samples. Of course the ideal would be to have a transcript of the text these folks are voicing, with all the accents marked. Probably too much to expect, but a native speaker might be willing to help you out.

I have heard some recordings of “pitch accented” Greek that sound like yodeling! Your idea is better–start with a real, comparable language.

5

brn,
I understand exactly what you are saying about pronouncing η, etc. For me the most challenging is ει, for which the English speaker would have a tendency to produce a slight diphthong at the end. Maybe German speakers would have no problem since I think that ει is pronounced as the first e in Bete. I have had some trouble with η, in particular it took me a long time to get to where I could easily pronounce ἥλιος!

Regarding IPA, I’ve found it to be very useful as a first approximation to a foreign sound. But it only gets you started. You then have to, in the process of iteration, change that first approximation that you made, and many times.

(By the way, once a long time ago I tried to learn Swedish and got frustrated with the accent-1 and accent-2 system. It seemed to me that Swedish was a language that was certainly possible to learn, but to me seemed impossible to describe.)

cb,
I actually have a copy of Devine and Stephens, but havn’t read it seriously through. I also have one of Allen’s earlier books, Accent and Rhythm, 1973, but again I haven’t read it thoroughly. I’ll listen to the Hagel clips. There is a Greek, Stratakis, who also does good recitations using reconstructed plus pitch.

I think that my guess regarding the giving of full pitch contour only to phrase-stressed words may well be incorrect, as you point out. I notice it in Serbo-Croation, but that may not mean much. Early on I studied two Chinese languages, Cantonese and Mandarin. In Cantonese, it seems to me that each syllable must preserve its inherent tone at all points in a sentence; but in Mandarin there is a loss of tonal contour (called neutralization) for many syllables, if not most. So I think it is very dangerous to make general assumptions about pitch accent (or tonal systems) based on observations of one language.

I’m not trying to be a scholar on pitch accent (not that there’s anything wrong with that!). My motivation was simply to find a reasonable way to to help me learn ancient Greek, and to have fun with the challenge of something different in the linguistic world. Wheelock in my 7th edition constantly advises to pronounce Latin words as part of the learning process (audi et pronuntia!). So I talk to myself, or my dog, as a way of learning Greek, of retaining it in my memory. My dog is refusing to answer, so the audi part is not being used.

As you say, it is a personal thing and, outside of academia, it probably has to be that way. Even if you got together a group of people to study Greek, you would have no success at all if you tried to impose a standard pronunciation on the group.

Zembel,
Thanks for the idea about voice-over clips. I’ve read that Lithuanian and Latvian have preserved some pitch accent, but haven’t look into them. Japanese is looked at for ideas, as well.

Sometimes the grammar books are so historically conservative, that they describe a language that doesn’t really exist any longer. For example, if you read a Slovenian grammar (e.g. chapter 8 in The Slavonic Languages, Routledge, 1993), you will find a complex tonemic system described. But if you try to discuss this with a Slovenian from Ljubljana, they will have no idea what you are talking about, as tones are not used (at least consciously) any more except maybe by older villagers in some parts of the country. Similarly for Serbo-Croation, the full tonal system as laid down by Karadžić in the early 1800s is not used so much now, certainly not in Zagreb.

6

As someone with an interest in this topic myself, I would say all of you seem to have worked harder at it and at a greater level of sophistication than I have.

I have listened to some of the recordings on Avery Andrews’ website and would be interested to know if he is included in the “sounds like yodelling” indictment. The one time I tried listening to Stephen Daitz, I certainly thought he was too sing-songy. Rachel Kitzinger reciting Sophocles’ “Electra” was not bad, I thought.

I am bamboozled by the fact some authors recommend us not to try and recreate the pitch accent. Unless you are going to recite in a monotone, the pitch of your voice will naturally rise and fall while speaking. As far as I can see, the question is not whether or not you should incorporate some pitch-accent in your speech but whether or not you should try and make it coincide with the Ancient Greek one. And if not why not? What alternative could possibly be better?

7

Here is how to learn the Greek pitch accents: Get Stephen Daitz’s The Pronunciation and Reading of Ancient Greek, with the recordings. Study carefully his instructions and description of Attic Greek pronunciation. After that, get his recording of the Iliad or the Odyssey. Familiarize yourself with the rules for recitation. Then spend some time listening to the recordings while you read along with the text, paying attention to the pronunciation. Once you’ve done this for a copuple of days, start shadowing. What this means is that you listen to Daitz’s recording, verse by verse, then record yourself, trying to imitate his recording as much as possible. Record yourself and compare it with the original. Keep doing it until you can perfectly mimic Daitz, intonation and everything. After a couple hundred hours of practice, you should be able to start doing it on your own. To motivate yourself, it would help a lot if you also have enough Greek to follow the text’s meaning, but focus entirely on pronuncation, even if you only understand a small part of what you are saying.

This shadowing method is exactly how people who learn how to pronounce foreign accents with near native profiency do it. It is used, for example, in accent reduction methods. It works. Suffice it to say that monolingual English speakers have used it to master Cantonese and Mandarin. I have done it to some degree of success (about 100 hours), but I stopped because I though it more important to focus on reading prose in silence for the time being.

8

Hi David,

I have listened to some of the recordings on Avery Andrews’ website and would be interested to know if he is included in the “sounds like yodelling” indictment. The one time I tried listening to Stephen Daitz, I certainly thought he was too sing-songy. Rachel Kitzinger reciting Sophocles’ “Electra” was not bad, I thought.

I don’t want to judge the results of people’s trying to reproduce Greek accentuation. As I pointed out, there are living languages (Lithuanian, Latvian, some varieties of Bosnian/Croatian/Serbian) with similar systems, and you can find recorded samples of native speech. When I listen, I find that the tonal contours are not easy to make out.

In the same way, a native speaker of English does not easily hear the difference between aspirated and unaspirated stops. Both are there in the phonetics of English pronunciation, but since they are not distinctive it takes some effort to “hear” the difference.

Same goes for double consonants, which also existed in ancient Greek but are absent in English. There’s a funny story about an American at a restaurant in Italy and the difference between pene and penne.

9

I’ve been playing with a free app, VocalPitchMonitor, on my android phone, to see if it could be used to help with pitch accent.

It seems somewhat useful in verifying the proper pitch contours and in correcting mistakes.

Below is an example of its use in correcting a possible mistake with the phrase ἐν Κερκύρᾳ.

The υ is long, and there is the possiblity that an incorrect contour might be applied to the υ in anticipation of the return to base level (the svarita) on the syllable ρᾳ.

The mistake would be to return to base level too early, on the second mora of the υ, with the syllable ρᾳ then being sounded entirely at base level.

For me base level is about A2 or a semi-tone above A2, with an upward excursion of 4 to 6 semi-tones. With phrase-end intonation (not related to lexical pitch), I tend to end a little below base level.

One other thing to note. Syllable onset seems to have a short high-frequency component associated with the beginning consonant (even if the syllable is just a vowel alone, there may be an initial glottal stop) before the vowel takes over, and this is why I usually see a small drop before the vowel.

In the screen capture below, we see ἐν and Κερ at base level A2, and then a severe drop down to a semi-tone below A2. This is on the second mora of κύ. We finish with ρᾳ. The mistake is that I am pronouncing the phrase as if it were ἐν Κερκῦρᾳ.

The starting level of the drop at E3 is so high due to the initial-consonant high-frequency effect.

In the screen shot below, I correct the pronunciation so that the ύ has the rise over the two morae, ending high, and the fall back to base level (the svarita) is now on the ρᾳ. The short squiggle at the end is the subscript iota of the long diphthong.

10

I am glad to see this experiment concerning the pitch accent, as I have tried something similar myself.
It has been said that pitch accents are not used anymore in languages such as Latin, English, Modern Greek, and most of modern languages. My impression, however, is that every language today - and perhaps at any moment of time - has had some musicality of its own.
Below are some samples of sound analysis (the software used is ‘Speech Analyser’ from SIL International); at the bottom of the diagrams the variation of frequency is shown. In terms of ‘musicality’, the three samples are quite similar to one another: frequency rises and drops as syllables are accented).

\

  1. English Reading sample (taken randomly from the Internet)
    “British and American people speak the same language, English, but with some small differencies.”

  1. Modern Greek sample (from the tutorial named 'Φιλογλοσσία)
    “Ψάχνω το άλμπουμ με τις παλιές φωτογραφίες από την Πάτρα…”

  1. Ancient Greek (excerpt from Genesis 1, recorded by myself)
    “Ἐν ἀρχῇ ἐποίησεν ὁ θεὸς τὸν οὐρανὸν καὶ τὴν γῆν. ἡ δὲ γῆ ἦν ἀόρατος καὶ ἀκατασκεύαστος, καὶ σκότος ἐπάνω τῆς ἀβύσσου…”

Remarks:

  • Regarding the pronunciation of ἐν Κερκύρᾳ, the model according to the system that I am trying to keep would, theoretically, look like this:

There a two indicators which helps applying the correct pitch: (1) the acute on -κύ-, which prompts the reader/speaker to prepare to extend the accent effect over the subsequent syllable (or expect a sentence end, followed, of course, by a pause caused by this), and (2) the long final syllable.
If the actual pronunciation of this fully ‘consumes’ the accent effect within the syllable of -κύ-, then automatically the final vowel is forced to shorten itself.

Moderator: I changed the links to the png images.

11

Thanks. The pitch contours produced by your application seem similar to what I’ve seen with mine.

Brief aside since you seem interested in modern Greek:
As far as pitch in modern Greek is concerned, the language is stress-accent, so there is no lexical pitch as in ancient Greek. However, there is the extremely important aspect of suprasegmental pitch contours, which can be so dominant at times that you might mistakenly think that the position of the stress accent has changed in the last word of a yes-no question. If you don’t get this right, you will have a very foreign accent in speaking modern Greek. Unfortunately, the analysis of yes-no question pitch contours in modern Greek is complex. Here is some interesting research.
http://www.lel.ed.ac.uk/~bob/PAPERS/almspecom.pdf
Strangely, FSI seems to get this aspect wrong (stating “Questions of the second category have the highest pitch on the stressed syllable of the last word.” with the second category being yes-no questions). This is absolutely incorrect, as it is quite common for a dramatically raised pitch-fall to occur on the final syllable of the last word regardless of the position of stress accent. What is even worse with FSI is that the speaker seems to verify the wrong approach. I have not been able to explain this and it worries me. Could it be that the yes-no pitch contour has changed over time?

Back to ancient Greek:
I’ve found that long rising tones in ancient Greek, e.g., εί, ώ, ή, present the greatest difficulty in attempting a pitch-accent pronunciation.

My approach is to make sure that I don’t include the first mora of the long vowel/diphthong in the contonation.

Taking the ἐν Κερκύρᾳ example, I break the long υ into two consecutive short υ pieces (morae).

The contonation is then (in bold): ἐν Κερκυύρᾳ

So the first short υ is part of the base-level sequence preceding the contonation.

The pronunciation of the contonation is no longer problematic, just a short υ at raised pitch, and the fall back to base level on ρᾳ (the svarita) occurs naturally.

In order to drill this I can first start by excluding consonants and just concentrating on the vowels in the contonation. The reason for this is that unvoiced consonants (π κ τ φ χ θ σ and rough breathing) obscure the pitch contour since those consonants are not sounded. In our example, it is not a problem to include the ρ because it is a voiced consonant.

In any case, a general approach to drilling is to first exclude consonants, and to concentrate on forming the contonation as one continuous vowel.

Drill, ύᾳ ύᾳ ύᾳ … where the ύ is short. Pronounce this contonation as one continous vowel.

Next, drill, ύρᾳ ύρᾳ ύρᾳ …

Finally, pronounce the entire phrase, ἐν Κερκυύρᾳ, concentrating on not including the first short υ in the contonation.

12

I am indeed interested in Modern Greek, and I would like to be able to speak it but I realize it is quite different from the old language and I prefer to get to some good level of Ancient Greek before I really study its modern version.

My main focus on this matter has been to find a practical approach of what manuals define and describe in relation with pronunciation and accent of Ancient Greek.

I decided to comply with the pronunciation system of Modern Greek to the extent that I do not get in conflict with the accent rules of Ancient Greek. (While studying some foreign languages, I noticed that there are just a few sounds which are really specific to each language, which should provide encouragement for any student to try to learn and imitate those few sounds.) Using a Modern Greek tutorial, I practiced the pronunciation of each letter and combination of letters, plus the sample words given. I am trying to systematically keep the system so defined.

Regarding accents, I apply them as follows:

(1) The acute: rise the pitch on the accented syllable and then let it drop gradually to the initial level on the subsequent one. Graphically, it would look like this:

But what about this accent when placed on an ultima? How to drop the pitch level, since a new word is beginning just after it and there is no subsequent syllable available. I believe it is common sense to drop the pitch to some extent - not completely to the initial level. There is then enough room (given by the inherent pause) to start the new word from the normal pitch level.

(2) The grave accent should modify the intonation in a similar way the acute does, only the maximum of the pitch should move towards the second syllable, dropping completely by the end of it.

(3) The circumflex functions more simply, with the pitch rising and then dropping within the same syllable.

If we accept such explanations, it becomes quite intuitive that the acute/grave accents are very similar to the circumflex, only they cover two syllables instead of just one (as very clearly explained by Mastronarde in his ‘contonation and mora’ principle.

(4) If we try to read aloud Greek text using such guidelines, I think we should easily agree that it sounds very natural, even very close to that which we are used to today in relation with any of the modern languages.
Below is a passage where the variations due to accents are described by means of a line placed above the text, whose shape suggests the way the pitch is raised or lowered, according to the explanations above.

13

Vasile, if you are using unit 2 of Mastronarde, there is one thing that I should point out. The discussion of contonation and mora in section 4 does not go into enough detail, and I think that is why your pitch contour for οὕτως, for example, does not look right to me.

Mastronarde states “In the case of an acute accent, the contonation includes both the syllable on which the accent is written (and on which the pitch rises) and the entire following syllable (on which the pitch falls), if any, whether it counts as long or short.”

The detail that is missing is that a long vowel or diphthong is divided into two morae, and in the case of an acute only the second mora of the syllable having the acute has a raised pitch. So what Mastronarde says is correct, it is only that he leaves out some detail as to exact pronunciation. It is usually considered that in Greek the raised pitch occurs on a specific mora, not the entire syllable of a long vowel or diphthong.

This is what I was trying to emphasize in my post using the ἐν Κερκύρᾳ example: that the raised pitch only occurs on the second mora of the long υ.

So for οὕτως the contonation should start on the second mora of the long diphthong ου and encompass the entire following syllable τως, as you have shown.

SImilarly for πολλοί, the rise in pitch should occur on the second mora of the diphthong οι.

For long vowels or diphthongs having a circumflex, the rise in pitch occurs on the first mora, and drops on the second mora of the same syllable, which you have shown correctly in your οὗτος and λαλεῖ.

Regarding your question about the grave accent on the last syllable. There is no full contonation, as you point out, since in Greek you cannot use the following word as part of the contonation of the grave. Looking through the book by Devine and Stephens, they say on page 182 that the grave may be taken as a “lowered High tone.” W. Sydney Allen in his books also discusses this. What I usually do is to very slightly raise the pitch on a grave, and leave it that. Similarly for pre-pausal acute, I just slightly raise the pitch at the end of the word. Another option would be to just ignore the grave and stay at base level, since the rise in pitch associated with the grave is not thought to have been nearly as much as the rise in pitch associated with an acute and so I consider it to be a second-order thing.

So the full contonation (excluding the grave case) is always a raised pitch on one mora followed by a descent to base level on the next one or two morae. This applies to both the circumflex and acute cases.

For pre-pausal acute, if the last syllable is closed, for example, ἀγαθός, the ς at the end seems to nicely terminate/absorb the slightly raised pitch of the ο. For something like τιμή, you should be careful not to get carried away and produce a long rising tone on the η that might resemble a rising Chinese tone.

14

This indicates to me that some artificial division into syllables would occur, i.e. Κερ-κυ-ύ-ρᾳ. Would there be a good reason to do so?

Imagine a sentence where several words in a row bear the grave accent: the result of ignoring the grave and staying at the base level sounds quite strange to me: it would be like turning from open speech to talking to oneself (or some private whispering) and then back to open speech and so on… For instance:

εἶπεν δὲ πρὸς αὐτοὺς τὴν παραβολὴν ταύτην (Lk 15:3)
ἦρεν τὴν χεῖρα αὐτοῦ τὴν δεξιὰν εἰς τὸν οὐρανὸν καὶ ὤμοσεν… (Ap 10:5,6)

My understanding is that word accenting is necessary so as to obtain a contrast, fragmenting thus the uttering. If all syllables (or several of them in a row) are pronounced with the same tone and intensity, the message delivered cannot be received correctly in real time.

15

I’m puzzled. Why on earth would anyone even try to read the gospel of Luke with classical pitch accents? Let alone the Apocalypse! It would make no sense at all to the authors or their contemporaries.

16

Pitch accent was probably well on they way out by the end of the fourth century, and by the time the New Testament was written, only stress accent remained. People continued to speak ancient Greek successfully as a highly inflected language. Trying to replicate how 5th century Athenians spoke is an interesting exercise, and gives us some insight into the historical development of the language, but for our purposes is a bit like tilting at windmills.

17

Barry elaborates my post. But I think this is more than a bit unfair to katalogon’s and others’ well-informed and illuminating posts, which concern earlier Greek. I must confess that I find it impossible to eliminate my English stress accent (though I’m ok at speaking Italian or mandarin Chinese, or was), and distrust all attempts to reproduce an authentic oral rendering of any ancient Greek. (I’ve referred before to production of a factitious simulacrum of ancient Greek, which I abhor.). Besides, even if we could hear native ancient speakers we’d be hearing with our ears not theirs, so it would be a radically different experience.

So for myself I don’t think the game is worth the candle. But kudos to those who are more accomplished.

18

I believe my comments regarding the grave accent would be more meaningful if examples were selected from earlier Greek, rather than koine. May I propose as follows:

τεκτονικὸν μὲν γὰρ ἢ χαλκευτικὸν ἢ γεωργικὸν ἢ ἀνθρώπων ἀρχικὸν (Xenophon – Memorabilia 1.1.7)
ὑποθέμενοι τῷ λόγῳ, θερμὸν ἢ ψυχρὸν ἢ ὑγρὸν ἢ ξηρὸν ἢ ἄλλο (Hippocrates - Ancient Medicine Hp.VM.1)

19

This indicates to me that some artificial division into syllables would occur, i.e. Κερ-κυ-ύ-ρᾳ. Would there be a good reason to do so?

Vasile, there is a good reason to divide long vowels or diphthongs into two morae. But morae are smaller than syllables, so the artificial division that you refer to is not into syllables.

The concept of the mora seems to have been formally developed by linguists of the Prague School in the early 1930s. One of the main linguists involved in this was Roman Jakobson.

Have a look at lecture V of this collection of Jakobson lectures; you will find a lot on the mora.
https://monoskop.org/images/8/85/Jakobson_Roman_Six_Lectures_on_Sound_and_Meaning.pdf


The mora is difficult to define, but very useful in some languages.

From a paper by Peter Auer, “Some ways to count morae”, 1989

He gives a phonemic working definition of the mora:
“The mora is the smallest prosodic unit of language. It constitutes the lowest level of prosody in a hierarchy that contains on top of it, in ascending order, the syllable, the phonological word, the foot, the rhythmic group, the intonational phrase, and even larger textual units.”

He gives this as one of the useful applications of the mora:
“In languages with a so-called musical accent, such as Classical Greek, Japanese, or Lithuanian, a two-mora syllable may be accented on the first or the second mora”

So in ancient Greek, a long vowel or diphthong is considered to have two morae, while a short vowel is considered to have one mora.


I’m using the book “Accent and Rhythm, Prosodic Features of Latin and Greek: a Study in Theory and Reconstruction” by W. Sidney Allen, 1973, for the following discussion of the accent of ancient Greek in terms of morae (page 235).

“In the case of the vowels marked ‘circumflex’, the morae are accented high+low, the low having automatically its falling variant.”

What is meant by “falling variant” is the glide from high pitch back to base level that occurs in the second mora - the one following the mora with the high pitch. This is often referred to as the svarita, from Vedic Sanskrit grammar.

“In the case of the long vowels marked ‘acute’, the morae are accented low+high (whatever may have been the precise variant of the ‘low’ in this environment, e.g. ? rising).”

So it is not known exactly what is happening with the first mora (the ‘low’ one). My approach is to keep the first mora at base level (or at least I start into the first mora at base level), and then there is a transition from low to high which occurs at about the midpoint of the long vowel or diphthong. I try to make that transition sound smooth, but I cannot really say much other than that. The main thing is that I avoid starting the syllable at high pitch - I want only the second mora to have high pitch.

Note that the svarita occurs in this case on the one or two morae of the following syllable, after the mora with high pitch. The svarita in the acute case always covers the entirety of the following syllable (i.e., it doesn’t matter what the length of the vowel in the following syllable is).

Allen gives this evidence from Greek as support for the above analysis:

“Historical support for these analyses may be seen in the fact that, for example, an original πάϊς contracts to give παῖς (with circumflex) whereas δαΐς contracts to give δᾴς (with acute).

20

Aren’t morae what Greek writers from Aristoxenus on (already Pythagoras perhaps?) called χρόνοι? and isn’t that a matter of syllables rather than vowels? Long (aka heavy) syllables and diphthongs have two, short syllables one. No doubt acoustic length, as distinct from conceptual, will have been less precise, unless musically regulated; perceptual length too I imagine. And wouldn’t pitch (absolute and relative) and pitch contour have varied not only according to accent and local environment but also changes over time and place and other factors?