Latin's Word Accent: Stress or Pitch? (Discussion Moved from True or False Thread)

B

Bitmap

Guest

You can't call it mispronouncing if you yourself have no idea how to do it correctly, either.

omnīnō falsum. māllem perīre quam tālī modo verba latīna dēfōrmāre.
Reading verse with what is considered to be a Latin prose accent these days is probably just as much of an abomination. (And on top of it, it sounds even worse).
 

Iáson

Cívis Illústris

  • Civis Illustris

You can't call it mispronouncing if you yourself have no idea how to do it correctly, either.
quod scio, nōn dubitātur quīn metrum latīnum sit 'quantitātīvum', nec crēdit ūllus iam grammaticus Rōmānōs aliēnō accentū carmina legere. ergo nōn puto 'male prōnuntiāre' esse verbum ineptum.

Reading verse with what is considered to be a Latin prose accent these days is probably just as much of an abomination.
nescio utrum vīs indicāre prōnuntiātiōnem mōnstruōsam Britannōrum ac Americānōrum, aut prōnuntiātiōnem restitūtam ut (exemplī grātiā) Godmy dīceret. sed sī hanc significās, ā tē magnopere mē dissentīre fatērī debeō...
 
B

Bitmap

Guest

quod scio, nōn dubitātur quīn metrum latīnum sit 'quantitātīvum', nec crēdit ūllus iam grammaticus Rōmānōs aliēnō accentū carmina legere. ergo nōn puto 'male prōnuntiāre' esse verbum ineptum.
That doesn't change the fact that you have no idea how Romans actually read verse.

nescio utrum vīs indicāre prōnuntiātiōnem mōnstruōsam Britannōrum ac Americānōrum, aut prōnuntiātiōnem restitūtam ut (exemplī grātiā) Godmy dīceret. sed sī hanc significās, ā tē magnopere mē dissentīre fatērī debeō...

What I mean is the overrated belief that most people share that a stress-timed accent played much of a role in the Roman pronunciation of Latin. If it did, you would have to explain to me why 'esse videntur' is not an acceptable prose clausula while 'esse dicuntur' (stressed on exactly the same syllables) is.
The natural stress can at best have been extremely mild, and quantity must have been the dominant feature of the language, not only in poetry, but also in every day speech (as prose clausulae that defy all natural stresses serve to show). As a result, any attempt at reading Latin poetry with a stress-timed accent that I've heard so far sounded at the very least just as pathetic and wrong to me as simply stressing the prevalent quantities (which is very likely also not the way Romans did it when reading Latin poetry, but which I at least consider a lot more likely to come closer to the way Romans actually perceived the language upon pronouncing or hearing it).
 

Iáson

Cívis Illústris

  • Civis Illustris

the overrated belief that most people share that a stress-timed accent played much of a role in the Roman pronunciation of Latin. If it did, you would have to explain to me why 'esse videntur' is not an acceptable prose clausula while 'esse dicuntur' (stressed on exactly the same syllables) is.
The natural stress can at best have been extremely mild, and quantity must have been the dominant feature of the language, not only in poetry, but also in every day speech (as prose clausulae that defy all natural stresses serve to show).
Perhaps I have failed to understand you correctly, but this looks like a fallacious argument, from 'stress accent does not affect prose clausula' to 'stress accent was not prominent in speech'. Metre in prose, like in verse, depended on quantities. But this does not imply that stress accent was not present or prominent.
(I mean, on the same argument you could argue that pitch accent in Classical Greek - which also does not affect the quantitative metre - was unimportant in speech.)

On the contrary, it is stress that was maintained into Late Latin, whilst vowel quantity was eventually lost/replaced by quality distinctions. Moreover, syncope is an important phenomenon in the Latin language, and likely occurred more frequently than the spelling reveals: note some of the spelling errors in the documents of C. Novius Eunus (look at p231) for example. Syncope is determined by stress, and thus I'm entirely unconvinced by the argument that stress was unimportant to Latin.

As a result, any attempt at reading Latin poetry with a stress-timed accent that I've heard so far sounded at the very least just as pathetic and wrong to me as simply stressing the prevalent quantities
Since I reject your earlier argument, I also reject this corollary. However, I do like the logic here: that we shouldn't accept a taste for a certain method of reading based on how poetry is read in our native languages, but on how we think it should be authentically read (based on logical arguments).

which is very likely also not the way Romans did it when reading Latin poetry, but which I at least consider a lot more likely to come closer to the way Romans actually perceived the language upon pronouncing or hearing it
An interesting but (to me) unconvincing argument. If I understand you correctly, what you are saying is that this way of reading is probably not correct, but for you it produces the same (or more similar) effect as it would for a Roman to have the poem read correctly. Perhaps this might be true to some extent, provided that you read Latin poetry in a way that sounds familiar or comprehensible to you based on the manner of reading poetry in other languages. However, I still think that it would in the long run give a more authentic perception of Latin poetry, first to assimilate and acclimatise to the actual way Latin was spoken (according to our evidence), and then appreciate Latin poetry against that standard.

That doesn't change the fact that you have no idea how Romans actually read verse.
What precisely do you mean? Sure, I'll happily admit that I've never visited 1st century Rome and listened to a native speaker read hexameter. But that doesn't mean that one can't make use of ancient evidence and come to balanced and probable conclusions. Whenever I discuss the pronunciation of ancient languages, someone will always come up with the 'it was all long ago and they're all dead and so nobody knows for certain anyway' line. Yes, there's a place for scepticism when studying the linguistics of ancient languages, because we have no audio recordings. But where there is evidence for pronunciation (and Latin is one of the best-evidenced ancient languages in this respect), and where there is no reason to doubt it, why should we?

There is a further point to be made about stress and hexameter. In the vast majority of cases, the stress does fit a regular pattern at the end of the hexameter line, in the last two feet. If stress is that unimportant, why bother to make it correspond to the meter at the end of the line? To me, it seems most likely that the failure of word-stress and metrical ictus to correspond in the earlier part of the line is an important aspect of the metre, brought out by the correspondence in the last third.
 
B

Bitmap

Guest

Perhaps I have failed to understand you correctly, but this looks like a fallacious argument, from 'stress accent does not affect prose clausula' to 'stress accent was not prominent in speech'. Metre in prose, like in verse, depended on quantities. But this does not imply that stress accent was not present or prominent.
There would be no point in having the prose clausulae Cicero used if the quantity had not been the prevalent feature. As I said, if stress is the prominent feature, it makes little sense to avoid the clausula ésse vidétur (–◡◡–x) because it sounds like finishing the sentence like an hexameter, but not the clausula ésse dicúntur (–◡––x) which, if stress had been the prominent feature, would sound like the end of an hexameter as well. (NB I just made up those two phrases here for the sake of the argument... it basically comes down to the fact that the heroic clausula is something that is generally avoided while a catalectic dicreticus is generally welcomed).

(I mean, on the same argument you could argue that pitch accent in Classical Greek - which also does not affect the quantitative metre - was unimportant in speech.)
Doesn't the fact that it was a pitch accent, and not a stress accent, actually underline the argument that stress played a much less important role than people think?

On the contrary, it is stress that was maintained into Late Latin, whilst vowel quantity was eventually lost/replaced by quality distinctions. Moreover, syncope is an important phenomenon in the Latin language, and likely occurred more frequently than the spelling reveals: note some of the spelling errors in the documents of C. Novius Eunus (look at p231) for example. Syncope is determined by stress, and thus I'm entirely unconvinced by the argument that stress was unimportant to Latin.
I have to make a confession here: I read quite a bit about the pronuntiatus restitutus. And don't get me wrong, at least in prose, that's also the kind of pronunciation that I would at least try to get close to when reading prose ... FWIW it's still a better attempt than butchering the language with my native pronunciation (even though a lot of my native pronunciation remains, of course). However, all I ever read about the ante-penultima rule was ... well, what the rule consists of ... I never found any explanation as to what the rule is based on. Do you happen to know the source that demonstrates how this rule was in place in classical Latin? It would greatly enlighten me.
If you don't have any source, either, I would have to expect that you have to reconstruct such a rule from Romance languages. However, that at best tells you that it was in place in later periods of Latin. How would you be able to tell that it was a feature that 'carried on' rather than being a feature the language 'changed into' or at the very least being a feature that became more prominent only in later stages of the language while being rather weakly developed at earlier stages?

Regarding syncope: I don't see how stress is required for that to happen. It seems to be a mere matter of convenience to leave out less important syllables. You probably know Greek a lot better than I do, but don't syncopes also appear in Ancient Greek, a language that has a pitch accent rather than a stress accent?

However, I still think that it would in the long run give a more authentic perception of Latin poetry, first to assimilate and acclimatise to the actual way Latin was spoken (according to our evidence), and then appreciate Latin poetry against that standard.
That's fine. You just have a different idea of what the actual way Latin was spoken was :>

What precisely do you mean? Sure, I'll happily admit that I've never visited 1st century Rome and listened to a native speaker read hexameter. But that doesn't mean that one can't make use of ancient evidence and come to balanced and probable conclusions.
That takes me back to what I mentioned above. If you refer me back to the ancient evidence there is, I may cease my argument.

There is a further point to be made about stress and hexameter. In the vast majority of cases, the stress does fit a regular pattern at the end of the hexameter line, in the last two feet. If stress is that unimportant, why bother to make it correspond to the meter at the end of the line? To me, it seems most likely that the failure of word-stress and metrical ictus to correspond in the earlier part of the line is an important aspect of the metre, brought out by the correspondence in the last third.

That is true of Augustean poetry (and even there with the caveat that Horace's hexametres often didn't live up to that standard). It's less true of epic poetry that came earlier, especially Ennius (I tend to think that's why Ovid used to call him arte carens and arte rudis). It's also only true of hexameters (and maybe to a lesser extent of pentameters that by the time of Ovid almost all ended in 2-syllable words, so the 4th and 5th ictus of the pentameter would coincide with the natural stress).
There is little coincidence between natural stress and verse accent iambic verses or in Horace's lyrical metres.
 
E

Etaoin Shrdlu

Guest

As I said, if stress is the prominent feature, it makes little sense to avoid the clausula ésse vidétur (–◡◡–x) because it sounds like finishing the sentence like an hexameter, but not the clausula ésse dicúntur (–◡––x) which, if stress had been the prominent feature, would sound like the end of an hexameter as well.
I don't really follow this. It wouldn't sound like the end of an hexameter, because it couldn't be one – there's a long syllable where the hexameter requires a short one. Stress being the most prominent feature doesn't mean that the force of less prominent features isn't felt. If it weren't, they would hardly be features at all.
 
B

Bitmap

Guest

Stress being the most prominent feature doesn't mean that the force of less prominent features isn't felt. If it weren't, they would hardly be features at all.

Same question as above: What makes you think stress is the most prominent feature? I chose a few examples where the stress coincides with the quantitative elements, I could also give you a lot of other examples where the quantity is more important for the clausula than the actual stress. If I just have a quick look at de amicitia:
8: quod autem Nonis in collegio nostro non adfuisses, valetudinem respondeo causam, non maestitiam fuisse.
17: nihil est enim tam naturae aptum, tam conveniens ad res vel secundas vel adversas.
7: tu non adfuisti, qui diligentissime semper illum diem et illud munus solitus esses obire.
 

Pacifica

grammaticissima

  • Aedilis

Location:
Belgium
The point of dispute seems to have shifted from "Is it right or wrong to put the stresses in unusual places in order to make Latin meter hearable to an accentually minded person?" to "Was stress in Latin more or less important than syllable length?"

Since the discussion started because of me, I feel obliged to specify that I never said or meant to imply that stress was more important than syllable length. On the contrary, I think syllable length is likely to have been more important, otherwise they would probably have based their poetry on stress rather than syllable length. Well, I suppose it's possible to argue that stress was actually more important and that it would have been more natural for the Romans to base their poetry on it* but that, in their blind adoration for Greek literature, they imitated the Greek method even though it didn't really work in Latin; that is, however, probably not very likely.

*According to what I've read, there are those who believe that Saturnian verse, the oldest recorded form of Latin poetry, was accentual, but not everyone agrees. The fragments we've got are so small that the pattern hasn't been able to be determined with certainty; it's possible to read the fragments several ways.
 

Iáson

Cívis Illústris

  • Civis Illustris

if the quantity had not been the prevalent feature. ... if stress is the prominent feature,
Stress being the most prominent feature doesn't mean that the force of less prominent features isn't felt.
What makes you think stress is the most prominent feature?
I don't think it's really a question of which of syllable quantity or stress is 'more prominent' or 'more important' (I've never heard of such a distinction made in a linguistic context...). I think that they're both important, just in different contexts: quantity is what metre is primarily based on, but stress is important to linguistic changes like syncope. Thus there is ample evidence (as I see it) that both were 'prominent' in Latin speech, and I don't think that the fact that clausula are based on quantity undermines this.

Doesn't the fact that it was a pitch accent, and not a stress accent, actually underline the argument that stress played a much less important role than people think?
No, because Greek is a different language.

Regarding syncope: I don't see how stress is required for that to happen. It seems to be a mere matter of convenience to leave out less important syllables. You probably know Greek a lot better than I do, but don't syncopes also appear in Ancient Greek, a language that has a pitch accent rather than a stress accent?
No. Certainly not on the scale of Latin in Classical Greek. There are a few examples (ἦλθον, ἔσται, οἶμαι) but they're comparatively much rarer. It's not impossible for syncope to occur in languages which lack a stress accent, though.

all I ever read about the ante-penultima rule was ... well, what the rule consists of ... I never found any explanation as to what the rule is based on. Do you happen to know the source that demonstrates how this rule was in place in classical Latin? It would greatly enlighten me.
If you don't have any source, either, I would have to expect that you have to reconstruct such a rule from Romance languages. However, that at best tells you that it was in place in later periods of Latin. How would you be able to tell that it was a feature that 'carried on' rather than being a feature the language 'changed into' or at the very least being a feature that became more prominent only in later stages of the language while being rather weakly developed at earlier stages?
That takes me back to what I mentioned above. If you refer me back to the ancient evidence there is, I may cease my argument.
Okay.
Having spent half an hour trying to find the relevant Latin grammarians online, I had less luck than I'd hoped for, as they are mostly poorly digitised; but there are accounts in Capella pp65-8 ed. Eyssenhardt, Diomedes Grammaticus, cf. also Consentius (pp88-90). Of course, all of these are relatively late, though they probably preserve material from earlier grammarians who only survive in fragments: eg. Servius de Accentibus mentions that Varro discussed the accent. Probably this book would be very useful and comprehensive on the matter, but I don't have access to it at the moment. Oh, another early account is in Quintilian Inst. Or. 30-1:

Apud nōs vērō brevissima ratiō: namque in omnī vōce acūta intrā numerum trium syllabārum continētur, sīve eae sunt in verbō sōlae sīve ultimae, et in iīs aut proxima extrēmae aut ab eā tertia. Trium porrō dē quibus loquor media longa aut acūta aut flexa erit, eōdem locō brevis utique gravem habēbit sonum ideōque positam ante sē, id est ab ultimā tertiam, acuet. Est autem in omnī vōce utique acūta, sed numquam plūs ūnā nec umquam ultima, ideōque in disyllabīs prior. Praetereā numquam in eādem flexa et acūta, quia īn flexā est acūta; itaque neutra clūdet vōcem Latīnam. Ea vērō quae sunt syllabae ūnīus erunt acūta aut flexa, nē sit aliqua vōx sine acūtā.

The grammarians borrow the Greek terminology, which is why they talk about acute and circumflex accents: by this period these likely meant much the same thing, since the Classical Greek pitch accent had probably changed to the stress accent found in modern Greek.

In addition to the statements of grammarians and authors, there is some evidence from poetry and linguistic features. The coincidence of accent and metrical ictus in the last two feet of hexameters works if we follow the rule. Sihler, for what it's worth (Sihler is terrible at citations and this does contradict your observation above), remarks 'There is in fact a pretty good alignment of accent and ictus overall in Plautus and Terence, and in the surviving hexameters of early poets like Ennius the alignment is considerably closer than in later poets. Indeed, later taste faulted early poetry on exactly that score, regarding it as inartistic and pedestrian by comparison to the sophisticated dissonances of accent and ictus in the poetry of a more refined age.' Iambic shortening in Plautus is affected by the position of the accent (according to the rule). In Seneca, Horace, and Phaedrus' iambic trimeters, in the third foot the long syllable of the iambus is the accented syllable of a word (or if it is resolved into two shorts, the first is the accented syllable) according to the rule. Syncope only affects unstressed vowels.

That's all I can find right now, but I think it is sufficient to demonstrate that the rule was not created based on the Romance languages, but is a simplification of what we find in the Latin grammarians, and corresponds to the available linguistic evidence.
 
B

Bitmap

Guest

Thank you for putting in all the work! At least I know now where that antepenultima rule comes from. I expected it to be Quintilian somehow, but I never found the source. Btw. where did you get the Quintilian text from which you quoted? I can't seem to find it online.

It still doesn't convince me of a Latin stress accent, though. Let me go through this one by one:

I don't think it's really a question of which of syllable quantity or stress is 'more prominent' or 'more important' (I've never heard of such a distinction made in a linguistic context...). I think that they're both important, just in different contexts: quantity is what metre is primarily based on, but stress is important to linguistic changes like syncope. Thus there is ample evidence (as I see it) that both were 'prominent' in Latin speech, and I don't think that the fact that clausula are based on quantity undermines this.

(...)

No. Certainly not on the scale of Latin in Classical Greek. There are a few examples (ἦλθον, ἔσται, οἶμαι) but they're comparatively much rarer. It's not impossible for syncope to occur in languages which lack a stress accent, though.
Well, which one is it? If syncopes can occur in languages that have no stress accent, then you can't claim they are proof of a language having a stress accent.

No, because Greek is a different language.
Well, ok ... you brought up that example.

Okay.
Having spent half an hour trying to find the relevant Latin grammarians online, I had less luck than I'd hoped for, as they are mostly poorly digitised; but there are accounts in Capella pp65-8 ed. Eyssenhardt, Diomedes Grammaticus, cf. also Consentius (pp88-90). Of course, all of these are relatively late, though they probably preserve material from earlier grammarians who only survive in fragments: eg. Servius de Accentibus mentions that Varro discussed the accent. Probably this book would be very useful and comprehensive on the matter, but I don't have access to it at the moment. Oh, another early account is in Quintilian Inst. Or. 30-1:

Apud nōs vērō brevissima ratiō: namque in omnī vōce acūta intrā numerum trium syllabārum continētur, sīve eae sunt in verbō sōlae sīve ultimae, et in iīs aut proxima extrēmae aut ab eā tertia. Trium porrō dē quibus loquor media longa aut acūta aut flexa erit, eōdem locō brevis utique gravem habēbit sonum ideōque positam ante sē, id est ab ultimā tertiam, acuet. Est autem in omnī vōce utique acūta, sed numquam plūs ūnā nec umquam ultima, ideōque in disyllabīs prior. Praetereā numquam in eādem flexa et acūta, quia īn flexā est acūta; itaque neutra clūdet vōcem Latīnam. Ea vērō quae sunt syllabae ūnīus erunt acūta aut flexa, nē sit aliqua vōx sine acūtā.
Thanks again for all the work ... but if I take Quintilian (and Capella, which I admittedly only had a brief look at) at face value, it sounds to me like he's pretty much describing a pitch accent for the Latin language that falls on either the penultimate syllable or the one before that. For him to be talking about a stress-accent, you would have to come up with a number of additional hypotheses; mainly that he was blindly following a Greek model and that he had no idea that Greek had a pitch accent once – and even if all of them are true, it would still seem strang why he is making a difference between 'media longa', 'acuta' and '[circum-]flexa' when the description of a stress accent would have no need of such a distinction — and also why he is making the disctinction by using aut rather than vel when, as you say, the terms pretty much meant the same thing.

In addition to the statements of grammarians and authors, there is some evidence from poetry and linguistic features. The coincidence of accent and metrical ictus in the last two feet of hexameters works if we follow the rule. Sihler, for what it's worth (Sihler is terrible at citations and this does contradict your observation above), remarks 'There is in fact a pretty good alignment of accent and ictus overall in Plautus and Terence,
That's not a convincing argument. If you claim that a language had a stress-timed accent with the antepenultima-rules postulated for Latin, you are bound to run into coincidental aligment between verse accent and what you might call 'natural accent' in any metre. You would have to be able to account for the exception that occured on more than a regular base, e.g. for Plautus and Terence having a substantial number of lines ending in two-syllable words ... and you can't. [NB. you would also have to account for the fact why that language didn't just start producing stress-timed poetry at least at *some* point]

and in the surviving hexameters of early poets like Ennius the alignment is considerably closer than in later poets. Indeed, later taste faulted early poetry on exactly that score
He's completely wrong with respect to hexameters. Augustean poetry did a much better job at having the verse accent coincide with the pitch accent than Ennius did. In Ennius's times, it was still acceptable to end an hexameter with a monosyllabic word:
unus homo nobis cunctando restituit rem (only 2/6, 33%, coincide with what is believed to be the natural accent)
[...] saxo cere comminuit brum (as far as I know, the first half of that hexameter is lost ... but the second half coincides at 0%)

Iambic shortening in Plautus is affected by the position of the accent (according to the rule).

Iambic shortening (which is not just a feature of Plautus's poetry, but of Latin in general) simply means that in iambic words the second syllable gets shortened. As with the syncope phenomenon, there is no evidence that stress is the cause for that ... and it probably wasn't considering Latin wasn't a stress-timed language.
 
 

Godmy

Sīmia Illūstris

  • Censor

Location:
Bohemia
I don't know anymore on this than Iáson does, but I can only add that, as far as I remember, Allen (Vox Latina) says (in the chapter on accent) that there are now more than enough proofs that the Roman terminology like "accent" (=accompanying singing) or "circumflex", "acute" etc. is a Greek borrowing, forcing the Greek prosody (including the word prosody itself) - the Greek three accents onto Latin which most probably had only one and since even the word "accent" is likely forced too, it may have nothing to do with singing (pitch) either [not sure if those are the exact words really], then the best guess is non-singing -> stress. (and I don't know if he backs it up by the same linguistic reasons than Iáson stated or not, I bid you to read the book, I'm merely quickly answering... )

But an intereting discussion, sadly I have nothing else to add. But I think Iáson wrote quite a good article on it (although I can't speak towards the poetry that you discuss).
 
 

Godmy

Sīmia Illūstris

  • Censor

Location:
Bohemia
He might have also mentioned that the archaic Latin accent position having been different and the vowel changes happening as a result of that (ad-cantus -> accentus, ad-gradior - aggredior) was an evidence that Latin had at that time a stress based accent, which however doesn't mean it still had it in the classical/later times... But I'm not really sure if I'm not making this up, making up false memories, I'm just not going to open the book at the moment :) I hope someone will.
 
B

Bitmap

Guest

He might have also mentioned that the archaic Latin accent position having been different and the vowel changes happening as a result of that (ad-cantus -> accentus, ad-gradior - aggredior) was an evidence that Latin had at that time a stress based accent, which however doesn't mean it still had it in the classical/later times... But I'm not really sure if I'm not making this up, making up false memories, I'm just not going to open the book at the moment :) I hope someone will.

Yes, archaic Latin must have had some kind of stress-accent. It must actually have been a very strong accent that always affected the first syllable ... which is why you have vowels weakened in composita:
rapio, but corripio
raptus, but correptus
(etc.)
 

Pacifica

grammaticissima

  • Aedilis

Location:
Belgium
I do wonder what all that talk by Latin grammarians of acute, grave and circumflex accents is about. At first sight, it would indeed seem to imply a pitch accent like in Greek. How do we know it isn't the case? And if it isn't, what did they mean by acute, grave and circumflex accents? — Well, to be fair, as far as I know we aren't even sure what they meant in Greek.
 
 

Dantius

Homo Sapiens

  • Civis Illustris

Location:
in orbe lacteo
How does a pitch accent work? Or rather, how does a language without stress accent work in general? How can you not stress certain syllables of words?
 

Pacifica

grammaticissima

  • Aedilis

Location:
Belgium
How does a pitch accent work?
Frankly, I don't know. Well, in theory I think it's something like a rising of the voice, saying some syllables louder... but when I try to do that it sounds pretty similar to a stress.
How can you not stress certain syllables of words?
I don't know... Though I speak that way every day in French. Well, almost. Just try to keep your voice even over all syllables.
 

Iáson

Cívis Illústris

  • Civis Illustris

where did you get the Quintilian text from which you quoted? I can't seem to find it online.
Look here.
Well, which one is it? If syncopes can occur in languages that have no stress accent, then you can't claim they are proof of a language having a stress accent.
This is strictly-speaking true. But I was trying to be nuanced. Syncope is strongly associated with stress-accent languages, to the extent that Adams (Social Variation and the Latin Language, p90) straight out says 'Syncope is the loss of short unaccented vowels in a language with a stress accent.' The fact that there are roughly three common examples in Greek, compared to hundreds of examples in Latin, is a good indication that Latin has a stress-accent.

Thanks again for all the work ... but if I take Quintilian (and Capella, which I admittedly only had a brief look at) at face value, it sounds to me like he's pretty much describing a pitch accent for the Latin language that falls on either the penultimate syllable or the one before that. For him to be talking about a stress-accent, you would have to come up with a number of additional hypotheses; mainly that he was blindly following a Greek model and that he had no idea that Greek had a pitch accent once – and even if all of them are true, it would still seem strang why he is making a difference between 'media longa', 'acuta' and '[circum-]flexa' when the description of a stress accent would have no need of such a distinction — and also why he is making the disctinction by using aut rather than vel when, as you say, the terms pretty much meant the same thing.
I do wonder what all that talk by Latin grammarians of acute, grave and circumflex accents is about. At first sight, it would indeed seem to imply a pitch accent like in Greek. How do we know it isn't the case?
I don't know anymore on this than Iáson does, but I can only add that, as far as I remember, Allen (Vox Latina) says (in the chapter on accent) that there are now more than enough proofs that the Roman terminology like "accent" (=accompanying singing) or "circumflex", "acute" etc. is a Greek borrowing, forcing the Greek prosody (including the word prosody itself) - the Greek three accents onto Latin which most probably had only one and since even the word "accent" is likely forced too, it may have nothing to do with singing (pitch) either [not sure if those are the exact words really], then the best guess is non-singing -> stress.
The words used by the grammarians do indeed imply a pitch accent, and suggest a distinction between acute and circumflex.

Most linguists nowadays, as far as I know, think that this is a misunderstanding based on applying Greek terminology to Latin. The terms the grammarians use are direct translations from the Greek: acūta (ὀξεῖα), accentus (προσῳδία), (circum)flexa (περισπωμένη), gravis (βαρεῖα). This is pretty much what happened in the rest of Latin grammatical terminology, and is paralleled elsewhere when the terminology used to describe one language is transferred to another. cf. pp282-8 of the Latin Grammarians book, which for me at least shows up in the preview.

As is frequently brought up nowadays, it is very rare for an accent to be wholly pitch-based or wholly stress-based; usually there is an element of both involved - although we can still meaningfully call eg. English a stress accent language. So it's probable that the Latin accent included some element of pitch, which eased the adoption of the Greek terminology, whilst still being a stress accent.

By the time Varro or Quintilian was writing, it is quite probable (from Greek linguistics) that the pitch accent had been mainly lost in Greek, replaced by a stress accent. So if they were to read the account of a Greek grammarian and listen to Greek being spoken by native speakers, then when they came to apply the same terms to describe Latin, they would naturally use them to describe the Latin stress accent much as they heard them used to describe the Greek stress accent.

I guess I can't think of an absolutely fool-proof argument that the Greek pitch accent didn't survive a little longer and the Latin accent wasn't primarily a pitch accent (except for the argument about syncope, and perhaps the fact that there are no pitch accents in any Romance languages). But I'm not sure it helps your argument very much (depending on quite what you're arguing at this stage), once you've accepted that there was an accent of some form on the penult or antepenult according to the rule.

Good luck pronouncing Latin with a pitch accent...

And if it isn't, what did they mean by acute, grave and circumflex accents? — Well, to be fair, as far as I know we aren't even sure what they meant in Greek.
In Classical Greek, the acute accent signifies a higher pitch on a short vowel, or a higher pitch on the last mora of a long vowel or diphthong. A circumflex accent signifies a higher pitch on the first mora of a long vowel or diphthong. After the point of higher pitch, there was likely a fall in pitch. A grave accent is probably either a slightly higher pitch (but not quite on the level of an acute) or just a sign that the pitch continued to rise. It's not unlike the 'downstep' in Tokyo Japanese. In modern Greek, the accent is one of stress, but it took until the 20th century for the acute-circumflex-grave writing system to be replaced by just acute accents.

That's not a convincing argument. If you claim that a language had a stress-timed accent
considering Latin wasn't a stress-timed language.
I'm not claiming that Latin was a stress-timed language, I'm claiming that it had a stress accent. I would guess that Latin was probably syllable-timed or mora-timed (cf. Czech).

That's not a convincing argument. If you claim that a language had a stress-timed accent with the antepenultima-rules postulated for Latin, you are bound to run into coincidental aligment between verse accent and what you might call 'natural accent' in any metre.
So you think that the correlation in the last part of the hexameter in Augustan poetry is a chance coincidence? If so, why is it more true for later poets than in Ennius, and why specifically in the last two feet of the line?

You would have to be able to account for the exception that occured on more than a regular base, e.g. for Plautus and Terence having a substantial number of lines ending in two-syllable words ... and you can't.
I don't understand what you mean here. Perhaps restate it in different words?

NB. you would also have to account for the fact why that language didn't just start producing stress-timed poetry at least at *some* point
Why should it, when the influence of Greek culture was so strong? We know very little about 'native' Latin poetry; as Pacifica said, it was suggested at one time that Saturnians were based on accent. Considering that the stress accent used to be on the first syllable in the prehistoric period, the Classical Latin accent must be 'relatively' recent, though, making this less plausible.

He's completely wrong with respect to hexameters. Augustean poetry did a much better job at having the verse accent coincide with the pitch accent than Ennius did. In Ennius's times, it was still acceptable to end an hexameter with a monosyllabic word:
unus homo nobis cunctando restituit rem (only 2/6, 33%, coincide with what is believed to be the natural accent)
[...] saxo cere comminuit brum (as far as I know, the first half of that hexameter is lost ... but the second half coincides at 0%)
In the last two feet, yes. But I think Sihler's idea is that Ennius' poetry has a greater tendency to overall agreement between ictus and stress accent, whereas Augustan poetry intentionally avoids coincidence of ictus and stress accent in the first four feet, whilst encouraging it in the last two.

As I said, Sihler is not reliable, and doesn't cite sources. I suspect this might come from the statistical study of Sturtevant (1919). Even Ennius (p379) exhibits some tendency towards agreement in the last two feet.

(which is not just a feature of Plautus's poetry, but of Latin in general)
Quite, but it is considerably more productive in Plautus (it affects new words and phrases rather than specific words, as far as I remember).
Iambic shortening (which is not just a feature of Plautus's poetry, but of Latin in general) simply means that in iambic words the second syllable gets shortened. As with the syncope phenomenon, there is no evidence that stress is the cause for that ... and it probably wasn't considering Latin wasn't a stress-timed language.
The role of stress in iambic shortening is complicated. There's a very short article here which has so much linguistic terminology in it that you'll undoubtably dismiss it without consideration (to be fair, I don't fully understand it, so it may go either way). But there is some relationship: stressed syllables (according to the rule) are not shortened.
 

Iáson

Cívis Illústris

  • Civis Illustris

How does a pitch accent work? Or rather, how does a language without stress accent work in general? How can you not stress certain syllables of words?
Frankly, I don't know. Well, in theory I think it's something like a rising of the voice, saying some syllables louder... but when I try to do that it sounds pretty similar to a stress.
I don't know... Though I speak that way every day in French. Well, almost. Just try to keep your voice even over all syllables.
There is a distinction between stress-timed vs. syllable-timed and stress accent vs. pitch accent. French is syllable-timed but has a stress accent. English is stress-timed with a stress accent. If you want to hear what a pitch accent sounds like, Japanese is a good place to go (eg. this video).
 

Pacifica

grammaticissima

  • Aedilis

Location:
Belgium
French is syllable-timed but has a stress accent.
Well, yeah, we do have stress (hence why I said "almost") but to a rather limited extent. A dispassionate French sentence will sound roughly evenly unstressed throughout, with only the last syllable before every pause receiving a slight stress. We aren't constantly hammering like in English, so to speak.

I said a bit more about it here.
 
Top