Latin long vowels

[Travis B.]

Not wishing to sound overly reductionist, am I correct in saying that your major objection to the communis opinio is that you think that the phonetic nature of Classical Latin's "stress-accent" (i.e. the Dreimorengesetz we're all familiar with) was one of pitch, rather than stress? Or are you disputing an earlier stage of strong initial stress as well?

Well, (1) yes and (2) not exactly, rather I doubt if it's been demonstrated that the factor traditionally identified as initial stress ceased to be active before Classical times.

The pitch is the only thing described by the grammarians (with some non-trivial details mentioned), so it's fair to put onus probandi onto those who deny pitch accent; and IMO they fail to carry the load to their chosen destination point.

What do we know besides pitch?

That (vowel/syllable) duration was hardly seriously affected, because quantity-based meter was perceived by ear, so uneducated audience would immediately react to metrical flaws. (And this is already bad for a cumulative accent, i. e. "dynamic stress").

That at least at some point before the Classical times there was a massive vowel reduction, and that word-initial syllables were protected from such by some factor. For the traditional view, this implies the explication that this factor was "stress" as well, and thus could not co-exist with the "stress" on penult/antepenult(/final overlong syllable). IMO, an invalid argument, since we haven't demonstrated yet the involvement in accent of any factor other than pitch; and for example, intensity contours in some modern languages involve initial intensity which proves to be compatible not only with pitch accents but even with true cumulative accent (dynamic stress), e. g. in Russian and IIRC French. In fact, it has never been demonstrated (AFAIK) that this factor didn't work in Classical Latin, since everybody has been mesmerized by the notion of one solid "stress"; and I wouldn't say that data suggesting the activity of this factor (or rather, the type of positional vowel reduction it blocked) in post-Classical language are completely lacking.

Consider simply the case of English, where in addition to a phonemic primary stress, one can have one or more other phonemic secondary stresses, and furthermore initial syllables of non-grammar words with onsets even when unstressed are typically more stressed than other unstressed syllables and syllables tend to fall into a general pattern of unstressed syllables alternating with at least partially stressed syllables, with phonemically unstressed syllables not adjacent to a more stressed syllable actually being given some stress to maintain this pattern.

Also, one idea that has always stricken me as nonsensically farfetched is that Latin-speaking elite might adopt a foreign (Greek) accent which remained alien to people in the street. What I do buy, however, is the hypothesis that educated speakers' pronunciation of foreign loans could be more faithful to the source language and thus allow for more violations of native speech automatisms (which might include the non-initial reduction); I believe this is how one can interpret e. g. the difference between Classical and "Proto-Romance" forms of many Greek loans.

Not necessarily - consider the pronunciation of French loans in more affected speech in many languages, such as English, versus the pronunciations of those same French loans in normal everyday speech in the same languages.

[David McCann]

The pitch is the only thing described by the grammarians (with some non-trivial details mentioned), so it's fair to put onus probandi onto those who deny pitch accent; and IMO they fail to carry the load to their chosen destination point.

What do we know besides pitch? That (vowel/syllable) duration was hardly seriously affected, because quantity-based meter was perceived by ear, so uneducated audience would immediately react to metrical flaws. (And this is already bad for a cumulative accent, i. e. "dynamic stress").

The grammarians just copied the terminology of Greek. These men weren't phoneticians; Rome wasn't like ancient India.

The uneducated did not recognise metrical flaws. If you look at lower-class epitaphs, few of them can get a hexameter right: they do a line of prose with a final da-didy-da-da rhythm. The Greek rules for verse are to get the stresses in a regular pattern. There is no basis for metre other than stress, any more than you can write music without a time-signature. The Romans copied the Greek rules, even though their stresses were independent, and persuaded themselves that the result sounded right. It's not difficult to do; Bridges wrote some English "verse" using Greek metrical schemes and he, and others familiar with Greek, thought they sounded like poetry, although most people don't.

[Basilius]

The pitch is the only thing described by the grammarians (with some non-trivial details mentioned), so it's fair to put onus probandi onto those who deny pitch accent; and IMO they fail to carry the load to their chosen destination point.

What do we know besides pitch? That (vowel/syllable) duration was hardly seriously affected, because quantity-based meter was perceived by ear, so uneducated audience would immediately react to metrical flaws. (And this is already bad for a cumulative accent, i. e. "dynamic stress").

The grammarians just copied the terminology of Greek. These men weren't phoneticians; [...]

Yeah, right. Also, they were just stupid, they just lied, just wanted to mock future linguists, etc.

However, somewhere around the beginning of 5th century they suddenly become phoneticians and begin to stress the dynamic component of the accent (accentus in ea syllaba est, quae plus sonat &like).

Also, can I use your time machine, please? I want to be able to speak of these matters as a live witness, too ;)

OK, jokes aside...

I can easily accept the idea that Roman grammarians noticed only the pitch component of their accent because of their Greco-centric linguistic/philological background.

However, besides stating that accent is essentially elevated pitch, the grammarians go further and speak of pitch contours (acute vs. circumflex), pointing to rather peculiar details of their distribution.

The usual way to explain this away is to refer to the fact that the distribution of accents as described by the grammarians is, in certain points, vaguely similar to what one finds in Ancient Greek.

But I don't find this convincing.

Firstly, because of that very vagueness. If Romans were copypasting their descriptions from Greek grammars, it's not clear to me why they failed to do that neatly, adding instead details not found in Greek, or making firm rules out of what was only a tendency in Greek.

Secondly, and more importantly, I don't understand the very idea of adding such detail. OK, one is describing only the pitch component of one's native accent because the only available science-like paradigm of the time offers no other notion of accent. But why wouldn't one just stop at this point, instead of fabricating pitch contours and their distributions where they don't exist in relality? How could contemporary readers, being native speakers themselves, fail to notice this forgery? How did a very practical book like a grammar become an authoritative source whose quite expensive copies were ordered again and again over centuries - if in fact it was full of such fantasies? Ultimately, why do we, when discussing other subjects, make reference to such grammars at all, if we think they're so bad and we are so much better informed?

(Probably, I should have adduced a couple passages from the actual ancient grammars here, for my rhetorics is certainly not as convincing; I think I'll do that when I have time, and if there is interest.)

The uneducated did not recognise metrical flaws.

I'm afraid the natives witnessed otherwise.

If you look at lower-class epitaphs, few of them can get a hexameter right: they do a line of prose with a final da-didy-da-da rhythm.

This is interesting; I'm not familiar with this type of material or its analysis, and I'd be grateful for any pointers.

However, I don't think this argument is a strong one.

First, bad verse and bad poets can appear (and even prevail) in every tradition, which doesn't mean that people in general cannot hear metrical flaws.

Second (and this may be a more adequate treatment of the facts you're referring to), those inscriptions (?) could represent a looser metrical canon (especially if, as you seem to say, the metrical pattern in question prevailed in a certain genre). I am no specialist in the history of versification, but I have heard of theories saying that this type of meter (with only a few last feet in a line being metrically regular) was the starting point for several versification traditions, including the best explored ones of those involving quantity-based meters. Preservation of an archaic looser canon in some genres would be natural, I guess.

The Greek rules for verse are to get the stresses in a regular pattern. There is no basis for metre other than stress, any more than you can write music without a time-signature. The Romans copied the Greek rules, even though their stresses were independent, and persuaded themselves that the result sounded right.

This is a very good illustration of why I'm trying to avoid the use of the term "stress". Actually, I can hardly make any sense of your words as quoted above. Partly, because I'm more familiar with the usage where "stress" is essentially a word-level phenomenon (while both Greek and Latin verse is based on syllable quantities which are doubtlessly syllable-level prosodies, irrespective of phonetic detail). Also, for me "a time-signature" is something referring to duration in the first place, so I wonder if stress is essentially duration for you. I guess it's all sheer misunderstanding on my part, but you need a less confusing terminology to make your point clear to me.