(L&N) Q&A Thread - Quick questions go here [2010-2019]

[Dormouse559]

Oh, I kind of abandoned this conversation. I got sucked down a bit of a rabbit hole reading about the passé simple.

Regarding -i, there are no inherited 3rd conj. verbs with the ending already stressed in Latin (-didī is not stressed in Latin), so the stress shift of -i might be explained as analogical influence of the 4th conjugation (which has lots of verbs like audīvī audīvistī > oḍï oḍïs, fīnīvī fīnīvistī > fini finis, dormīvī dormīvistī > dormi dormis), perhaps indirectly reinforced by the past participle as -ūtum > uṭ is stressed after the stem. And so vīcī > *vincī > venqui, rompī > rompi, fūdī > *fundī > fondi, etc., happened.

There's too much here for me to go through right now, but this caught my eye. I've seen these verb endings explained through analogy with the preterite of dō (dedī, dedistī, dedit). It makes a lot of sense; many of the verbs are derived from dō, and it explains the third person in -ié with regular sound changes (and the first-person singular, as it happens.)

[Sequor]

Dedī itself! That makes sense.

Meanwhile I wonder just how much literature on dialectal variants one studying this stuff seriously would need to go through...

[Zekoslav]

As for dialectal variation, Old Walloon is the real thorn in the eye... but first I'd like to respond to some of your points.

This, this and this paper are a must (has anyone found a full version of the last one?).

1. Vulgar Latin regularisation of perfect endings

The comparison of modern Romance Languages shows that Vulgar Latin really, really hated endingless perfect stems, i.e. those with a simple -ī ending. The only ones which remained were those of relatively common verbs, like dedī (and it's derivatives in -didī), fēcī and vēnī. Other verbs replaced the zero ending with -didī, -uī or -sī, depending on the verb (verbs with participles in -tum and especially -sum favoured -sī, verbs with stems ending in d, especially nd, favoured -didī, and other verbs, especially those of the 2nd conjugation favoured -uī).

So praehendī became praehēnsī, later O. F. pris because it had a past participle praehēnsum, and respondī became respondidī, later O. F. respondi because its stem ended in nd like vēndidī.

Verbs which look like they had no ending actually lost -uī by sound changes: battuere has ttu > tt in all forms, and likewise consuere has su > s in all forms: we have to reconstruct Proto-Romance *battere and *cosere. These two later adopted the weak perfect descended from -didī because it was the regular perfect of the 2nd and 3rd conjugations.

In Italian, when -uī is lost it usually causes gemination of the preceding consonant, so we have habuī > ebbi, rūpī > *rūpuī > ruppi, tenuī > tenni and what's the most interesting, cognōvī > *cognōvuī > conobbi and crēvī > *crēvuī > crebbi. Other languages also indicate this extension of -uī to cognōvī, like Old Occitan conoc and Old French conui.

basically conui can't come directly from cognōvī, it'd be *conuf, but from *cognōvuī > [ˈkonovwi] (V. L. sound changes) > [koˈnuvwe] (Western Romance metaphony) > [koˈnuwwe] (assimilation vw > ww) > [koˈnu] (apocope and simplification of the final consonant cluster, this form is preserved in Walloon) > [koˈnui] (-i probably analogical to fui). *crēvuī > crui is parallel, although the reason why ē became u instead of i is debated (more on that later)

You're (Ser) certainly right that the split in the development of perfects in -uī is due to whether the final consonant of the stem can phonetically disappear before [w] or not. In stems ending in plosives, this left and intervocalic [w], but in stems ending in sonorants this left a cluster of sonorant + [w]. We have reasons to believe this [w] would have become [v] at leas in [rw] and [lw] clusters, so it must be in part analogical: Lat. voluit becomes O. F. volt like Lat. absolvit becomes O. F. assolt, and Lat. voluerunt becomes O. F. voldrent like Lat. absolvere becomes O. F. assoldre. This irregular verb likely preserves the original inherited stem, and the one with u-endings is likely analogical. I personally think it might have been preserved in tremuī and then it spread to verbs ending in sonorants which had a past participle in -ūtum, since otherwise their passé simple would have merged with the present

i.e. I suggest tremuistī > *cremuste and valuistī > *valviste > *valuste because *cremutu = *valutu. The first and third person endings were copied from fui, which had that unusual u-perfect already in Vulgar Latin, to prevent the merger of perfect and present: lat. valet and valuit would both, problematically, give O. F. valt.

The other issue is, is the umlaut a > o (avoir > oi), e > u (devoir > dui) phonetical or analogical? This is where Walloon throws us a wrench: in Walloon, there is no umlaut! latin habuit gives aut, latin dēbuit gives diut, and Latin *cognōvuit gives conut: only the regular Western Romance metaphony is seen. There's not even the change ui > u in the stressed endings: habuistī gives awis, dēbuistī gives dewis! And according to a historical grammar of French I've read, passé simples like valui, valus, valut still exist!

Notably, Walloon doesn't do avu, agu > ou like Francien does either, so the difference is probably due to a sound change. Clavus goes clou in Francien and clau in Walloon, fagus goes fou in Francien and fau in Walloon. In other words, there was a second, late change of au > ou, unrelated to that of latin au > O. F. o. This new au was created when intervocalic consonants were lost and was preserved in Walloon but changed to ou in Francien. Now, is the change iu > u equally phonetical or not? If it's analogical it can only come from *cognōvuit > conut and *mōvuit > mut, which is not much. But cognōscere did influence other verbs: in Occitan it's preterite is conoc, and that of parescēre is... parec, completely parallel to conoc! This was the same in Old Catalan, and later conoc assimilated to parec and became conec. Why couldn't the opposite have happened in Old French?

Verbs with perfects in -ui with an original a in the root all had oi phonetically, but this was changed in some of them by analogy. So, jacuit is attested as joth in La Vie de st. Légier, but later becomes jut because the a in the root had disappeared from the present and became e or i.

[Sequor]

Thanks for the long reply, I'll read it more carefully later.


By the way, while writing this other post, it occurred to me that all the correlative pronouns for motion-towards look like ablatives. Does anyone know how that came about? Are these, somehow, maybe remnants of the Indo-European instrumental?

I'm talking about quō 'where to?', aliquō/quōquam/quōpiam 'to some place', eō 'to there', eōdem 'to the same place', quōquō 'no matter where [you go] to', aliō 'to somewhere else'.

EDIT: fixed link

[Zekoslav]

At a first glance it may seem odd for adverbs indicating movement towards come from ablatives, i.e. case indicating movement away from... but in a stepwise manner it can and it did happen. For example, PIE. had an adverbial suffix meaning "away from", *-tos, which gives Sanskrit -taḥ, still meaning "away from" and Latin -tus, meaning "at", as in intus, subtus. In Old French, the meaning could further shift from "at" to "towards", so Lat. intus > O. F. ens "in, into", and of course Lat. subtus > O. F. sous "under" (concerning the other thread, these are good examples of prepositions coming from adverbs).

As for quō etc., it could be from the instrumental, from the ablative or even from the controversial allative. It looks like these spatial adverbs are ancient, with cognates in Hittite and Slavic, so they may have been unanalysable adverbs already in PIE and we may never know which case they derive from. For example in Croatian there's kamo "where to", where ka- corresponds to Latin quō and -mo is some sort of particle.

[KaiTheHomoSapien]

My friend cited the sci-fi terms "subspace" and "hyperspace" as examples of words that seem to be antonyms, but are not (they both refer to surpassing the speed of light). Another quasi-example would be "inflammable" and "flammable" although etymologically these do not contain morphemes that are antonyms, it's just a misunderstanding of the different meanings of "in-"

Any examples of words that seem like they would be antonyms, but are not?

(This is different from a contranym, a word that has two contradictory meanings, like "cleave" which means both to separate something and to stick things together).

[Pabappa]

Sometimes words like this overlap with contronyms. e.g. inflammable is its own antonym, even though the 2nd meaning of "not flammable" is rarely used. Along the same lines there's pregnable and impregnable, both listed in dictionaries as "fertile, capable of becoming pregnant", but the 2nd word also has a separate meaning ... probably the more common meaning .... meaning "solid, impossible to penetrate"/.

[Zekoslav]

Apparently Latin quantitative metre, borrowed from Greek, wasn't something a lay person could spontaneously produce: you needed special training to do it (I remember reading that in Pompeii they found "hexameters" consisting of prose + dum da da dum dum, because lay people couldn't detect the meter when ictus didn't correspond to word accent). Why? What was the difference between Greek and Latin which lead to this? Some ideas...

1. You need mora-timing for quantitative meter: Greek was mora-timed and Latin wasn't.

Old Latin had lots of syncope and vowel reduction, Vulgar Latin had a bit or a lot of syncope depending on the region. There are rules which shorten unstressed vowels, mostly in final or closed syllables. These are not features usually associated with mora-timing, and it's unlikely Classical Latin would have naturally been a mora-time stage in between these two stress-times stages. Latin may have had vowel length, but as long as long vowels weren't exactly twice as long as short vowels, a quantitative metre wouldn't have been obvious. This is quite similar to Serbo-Croatian, where long vowels are about a third longer than short vowels and accented vowels are relatively longer than unaccented vowels (long unaccented vowels may even be shorter than short accented vowels, but still longer than short unaccented vowels, I think the numbers given were short accented 100%, long unaccented 80% and short unaccented 50%).

2. You need pitch accent for quantitative meter: Greek had it, Latin didn't.

One would think a stress accent would be more rhythmically prominent than pitch accent and more likely to be a feature of poetic metre. However, there's a bunch of stress accent languages using quantitative metre and at least one pitch accent language using qualitative metre. So probably no.

3. Latin stress dependent on length messed up the metre, Greek accent independent of length didn't.

Under this assumption, Greeks would perceive accent and length separately, leading to the quantitative metre where ictus would be independent of word accent. However, Latin word accent directly depended on length and they didn't perceive accent and length separately, leading to unintelligibly of the metre when ictus and word accent didn't coincide. Again nullified by Sanskrit (though the metre did appear during a time when it had free pitch accent) and Classical Arabic.

In the end it was probably just because, for whatever reason, stress was more perceptible in Latin than length. What do you think, and do you know of any discussion of this topic?

[Salmoneus]

These seem reasonable suggestions. I would say, though:

- stress patterns do change, and quite rapidly - French suddenly switched from a strong stress system that wiped out most of its vowels into a syllable-timed language. Maybe Classical Latin really was going through a phase.

- the fact ordinary people found it difficult to write quantitative metre and often didn't bother doesn't necessarily mean they didn't recognise it. Complex patterns can just be hard to spontaneously produce - many people, myself included, find it hard even to rhyme, although objectively it's easy in English, and very noticeable. Certainly if you asked people to write a sestina, they'd struggle, and maybe try to cheat. Likewise, if you think of some of the more complicated schemes of rhyme and assonance - in older Irish poetry, for example - they would be difficult for the ordinary person to spontaneously generate, and they might well, while writing graffiti, simply write something that only casually called the form to mind, but that doesn't mean that they didn't recognise rhyme or assonance. [you can recognise the elements even if you've never taken the time to work out how they're put together.] For a non-linguistic example: almost everybody recognises the basis harmonic patterns of western music, but many people wouldn't get it right if asked to write a chord progression. I mean, to take a very fixed and formulaic yet famous example, most of us can quickly recognise the harmonic progression of the twelve-bar blues, but many of us wouldn't get it right if asked to write it down - we might just latch on to an element, like the final cadence.

- just as in music, in poetry the ends of the lines are often the most salient. I don't quite understand your description - do you mean that their lines appeared unmetred, except for having a metric pattern at the end of each line? Because that just sounds like they're writing in an easier metre. There are certainly poetic traditions where only the metre of the end of the line counts, and the rest is free.

- in particular, the classical hexameter IS just a line of prose that ends 'dum da da dum dum'. So the graffitists would seem to have understood the basis of the poetic form. The only difference is that the hexameter more strictly limits the number and length of the preceding syllables - but it does so in an extremely charitable manner. There are 16 permissable lines (minus the final dumdadadumdum), ranging in length from eight to twelve syllables. On top of which, some variation (like the odd iamb for a spondee) was iirc permitted now and then. It's easy to imagine an uneducated writer recognising the fixed metric pattern of the end of the line, but not knowing the rules that determined which 16 previous possibilities were permitted and which were not, or even recognising that there WERE rules! In addition, it's not hard to imagine a poet of the people recognising that most of the actual effect of the metre comes from that fixed metric conclusion to each line, and intentionally not bothering with the much more recondite "rules" for the rest of the line. There's also, of course, the possibility of intentional satire.

- one interesting idea might be that the Latin of the literary classes - or at least the literary Latin in which they declaimed and composed - might not have been the same as Vulgar Latin! Specifically: what if Vulgar Latin was indeed a stress-timed language, but Classical Latin (the literary form), under the influence of both an unnaturally ennunciated style of declamation and the poetic theories and role-models of Greek culture, actually was a mora-timed language? This would certainly explain why the non-literary classes might recognise something in classical poetry to try to imitate, yet not be able to do so, due to not understanding the difference in prosody between the two registers?

- finally, it's worth pointing out that a lot of the metres the Romans inherited were much more complicated than the hexameter, and it's easy to imagine them driving less scholarly would-be poets to give up on the whole, seemingly arbitrary business.

[Zekoslav]

Thank you for all your points: it looks like I wrongly assumed a poetic metre suited to a language would be easily improvised,and then concluded Latin poetry was hard because it was influenced by the Greek one. I find it relatively easy to produce rhymed syllabic verse, usually with a trochaic accentual pattern added on top, but that may just be me: the classical hexametre, with the complex rules of what kind of syllables may come before the dum da da dum dum cadence, and complex interplay of word accent and ictus, is certainly something more difficult to conjure!

What I meant to say about spontaneous "hexametres" is this: the people writing them understood the dum da da dum dum part, but didn't understand the more complex part preceding it, so what they write doesn't obey classical rules and may as well be prose. From that I deducted that the stress pattern was rather more obvious to them than the length pattern.

Concerning timing, accent and length in Classical Latin, there's an ongoing debate on how much of it is native and how much is influenced by Greek. Latin authors were trying to describe Latin according to the model of Greek which makes things difficult for modern interpreters (they went so far as to declare that acute and circumflex accents existed in Latin, as well as rules describing which one of them can go on which syllable, and then comment that they sound the same in Latin, unlike Greek). I certainly agree that, if popular speech wasn't mora-timed, poetic speech probably would have been in order to make the metre more noticeable, Greek influence or not. As for declaring Classical Latin necessarily stress-timed due to Pre-Classical and Vulgar developments, that was a sort of devil's advocate against the claim that it was necessarily mora-timed and even pitch-accented, which can still be hear from time to time. I agree that these things can change radically and quickly.

[Salmoneus]

Thank you for all your points: it looks like I wrongly assumed a poetic metre suited to a language would be easily improvised,and then concluded Latin poetry was hard because it was influenced by the Greek one.

Oh no, your observation was quite valid and suggestive. It's just that I thought I should point out that the argument may not be as strong as you presented it as being.

I find it relatively easy to produce rhymed syllabic verse, usually with a trochaic accentual pattern added on top, but that may just be me: the classical hexametre, with the complex rules of what kind of syllables may come before the dum da da dum dum cadence, and complex interplay of word accent and ictus, is certainly something more difficult to conjure!

Personally, I find it very easy to produce rhythmic patterns, particularly iambic pentameter; but I find it very hard to spontaneously produce rhymed verse. I know some people also struggle with rhythm, while others can dash off fifty rhyming couplets on any given topic at the flick of a switch (Pope famously explains that this was an innate ability for him:
"WHY did I write? What sin to me unknown
Dipt me in ink,—my parents’ or my own?
As yet a child, nor yet a fool to fame,
I lisped in numbers, for the numbers came."
['the numbers' being metric rhyming verse]

But certainly some structures are more complex, and cannot be generated spontaneously even if the basic rules are understood. Mediaeval Irish literature is a famous example - there's rhyme between lines, rhymes between ends and middles of lines, rhymes between stanzas, alliteration in various patterns within and between lines, repetitions of words or phrases between lines, repetition of sounds from the first to the last word of the poems, and often these rules are based not on the actual sounds in old or middle irish but on what the sounds would have been in primitive Irish (so /d/ and /d/ don't always alliterate, if one is a lenited /t/, but /d/ and /D/ can alliterate...]. The rules eventually became so impossible to follow that the poetry became not only very dull but also largely meaningless (as the same fairly meaningless set phrases had to be inserted, regardless of the syntax, because they were the only phrases that could actually reliably meet the demands of the poetic rules...).

What I meant to say about spontaneous "hexametres" is this: the people writing them understood the dum da da dum dum part, but didn't understand the more complex part preceding it, so what they write doesn't obey classical rules and may as well be prose. From that I deducted that the stress pattern was rather more obvious to them than the length pattern.

But the cadence is also a length pattern.

It's not invalid to suggest that their difficult understanding quantitative rules MIGHT be because they're unfamiliar with quantitative rules. But since they evidently understood simple quantitative rules, like that cadence, it might just be that the complicated quantative rules were hard to grasp because they were complicated, not because they were quantitative.

[Zekoslav]

@ general stuff: No worries! Sometimes I find it hard to get my point across in a written medium and try to be as modest as I can so as not to provoke heated arguments (sometimes maybe too much).

@ Old Irish poetry: I took some Old Irish classes, and the professor just said not to bother with translating the poetic parts, since they're hard even for experts and the meaning of some words has simply been forgotten!

@ Hexameter cadence: This probably boils to the question of word accent and ictus. At the beginning of a verse, poets avoided making word accent coincide with the ictus, while at the end they almost always coincided. This was done by design, to separate the quantitative and the accentual rhythm, which confused the uninitiated both ancient (faulty hexameters of Pompeii) and modern (classicists debating whether to keep the natural word accent or to accent the ictus... Croatian classicists traditionally did the second option). I've read that classical poets thought if word accent coincided with the ictus all the time, the result would be too simple and too regular. So you're right that inherent complexity of the metre must be taken into account. The fact that the uninitiated got it right when word accent and ictus coincided made me think they were looking for accentual cues.

[Clio]

What I meant to say about spontaneous "hexametres" is this: the people writing them understood the dum da da dum dum part, but didn't understand the more complex part preceding it, so what they write doesn't obey classical rules and may as well be prose. From that I deducted that the stress pattern was rather more obvious to them than the length pattern.

Assuming the spontaneous poet splits his final "shave-and-a-haircut" (as defined only by the stress pattern) over two words, and ignoring whether the final syllable is long or short, there are twenty-four quantitative patterns he can produce that satisfy his stress pattern. Of these twenty-four patterns, three make good meter (_ υ υ | _ x, _ υ | υ _ x, and _ | υ υ _ υ); another four produce two final spondees (υ _ | _ _ x, _ _ | _ _ x, υ | _ _ _ x, and _ | _ _ _ x) with either no bucolic diaeresis (the former two) or bucolic diaeresis but no stress accent in the penultimate foot (the latter two).

I'm not at all familiar with the inscriptional evidence, but unless there are many lines of apparent verse with sequences like υ υ υ | υ x or _ υ _ | _ x, the odds seem to suggest that these spontaneous poets were rather aware of the quantitative pattern _ υ υ _ x. That doesn't quite contradict what you've said here and in your subsequent posts, but it might raise the question of how much more obvious the stress pattern was than the length pattern. I think this is roughly what Salmoneus meant when he said the spontaneous poets

evidently understood simple quantitative rules, like that cadence

and hopefully adds to the discussion about the rule they understood.

[Zekoslav]

Thank you for this information (and it would be nice if you joined this discussion). I've never studied Latin metre like you seem to have, but I'm rather interested in Classical Latin pronunciation and how the language may have sounded, spoken or recited.

This is quite a complex topic and whenever one tries to go further than "there was phonemic vowel length" one opens a can of worms. I was trying to see what poetry and/or failure to write poetry might tell us about it.

It looks like, for whatever reason, spontaneous poets had a hard time producing quantitative verse, and tended to produce accentual verse (a book about Vulgar Latin gives as an example a poem attributed to Caesar's soldiers, written in accentual verse). I listed some possible factors which depend on the language, and Salmoneous added some possible factors which depend on the metre itself.

Failure to produce quantitative verse has often been claimed as evidence that Vulgar Latin had no phonemic vowel length, since you need phonemic vowel length to produce quantitative verse. I think this is weak evidence at best and that the true reason for the failure is more nuanced: my native language, Croatian, has phonemic vowel length, some dialects in both accented and unaccented syllables, some only in accented syllables - and no quantitative verse has ever been written in it, only syllabic/accentual verse (in fact, when the Iliad and the Odyssey were translated, the translator choose to make the hexameters accentual rather than quantitative even though he could have made them quantitative). So I thought timing may be the culprit: mora-timing makes quantitative metre obvious, syllable and stress timing makes it less obvious. Greek and Sanskrit were almost certainly mora-timed*, Croatian almost certainly isn't... maybe Latin wasn't mora-timed, so writing quantitative verse required special training and reading it required special diction. But then, it may just be that the metre was relatively complex and hard to produce regardless of the language...

*A look at their phonological histories shows an abundance of vowel contractions and compensatory lengthenings which always preserve mora count, and a complete lack of vowel reduction and syncope.

After this discussion I'm inclined to think that the culprit for the difficulty of producing quantitative metre was the complex interplay between the word accent and the ictus. What do you think?

[Clio]

You probably know about as much about Latin meter as I do; I've just read a lot of stichic verse.

I think you're right that absence of quantitative verse isn't strong evidence for absence of phonemic vowel length. Creating consistently incongruous ictus and accent probably wasn't easy for anyone (except maybe Ovid); even early poets like Ennius wrote lines in which ictus and accent agree more frequently than they do in their later counterparts' lines. By the time that unaccented ictus was de rigueur, stylistically "correct" hexameters were probably very difficult to compose.

In contrast, it's very easy to produce a sequence with the shape _ υ υ | _ x such that it has accent on the ictus. Vergil and other skilled poets do even better than chance at creating these sequences (which makes lines like Aeneid I.105 ... aquae mons all the more effective), but Latin accent rules were very much in their favor. The converse, as I tried to show in my last post, is not necessarily true. That is to say, starting with the long syllables will normally give you the correct word accents--but starting with the word accents will rarely give you the correct long syllables. To me, this suggests that our spontaneous poets had enough knowledge of quantity to at least produce it where they could follow a common (from 99% of hexameters in Vergil, Catullus, Ovid et al.) and distinctive (with the alternation of long and short syllables probably being exaggerated by the word accent falling on long vowels) shape.

I think given all of this it's not unreasonable to say that

the culprit for the difficulty of producing quantitative [hexa]metre [lines before the bucolic diaeresis] was the complex interplay between the word accent and the ictus.

Out of curiosity, do the vulgar poems you've been looking at exhibit any other phenomena common to Latin quantitative meter (e.g., penthemimeral caesura--maybe defined somewhat more forgivingly as word-end after a long syllable, or after the seventh syllable of the line)?

[Salmoneus]

Failure to produce quantitative verse has often been claimed as evidence that Vulgar Latin had no phonemic vowel length, since you need phonemic vowel length to produce quantitative verse.

I would have put it the other way: the fact that Vulgar Latin may be presumed to have had no phonemic vowel length* would seem at a blow to explain the loss of quantitative verse.

I would quibble with two things here, though, in the interests of debate. First, you don't actually need phonemic vowel length to produce quantitative verse: you just need a distinction between light and heavy syllables. In most languages with quantitative verse, CVC syllables, or at least some of them (eg those ending in sonorants) are counted as long; so it's conceivable that even a language without long vowels could have quantitative verse, providing that it had a robust division into closed and open syllables.

Second, I don't think it's that clear that there was a 'failure to produce quantitative verse' at all. This seems to fall into the rhetorician's trap of assuming that classical Greek is always the aim of poetry. Latin-language poets who wrote syllabic or accentual verse may not have been trying to write quantitative verse - and hence not failing. And if there was no failure, it's hard to draw conclusions.

It's important to remember that verse forms can change for seemingly no reason. Mediaeval Irish, for example, switched from accentual to syllabic verse (due to contact with the poetry of other languages). In this regard, we might need to bare in mind that many writers of 'failed' Latin poetry may have been non-native, or at least multilingual, and that their expectations of poetry may have had as much to do with their own native poetic systems and languages as with those of Latin.



*All descendents of Vulgar Latin lost vowel length. And the whole of Western Romance at the least (don't know about Romanian) shared the development of an alternative system of vowel length (created in stressed short syllables), either before they broke apart or shortly after. The parsimonious solution is surely simply that the Classical Latin length system had already been lost in Vulgar Latin, or at the very least had become severely eroded.

I think this is weak evidence at best and that the true reason for the failure is more nuanced: my native language, Croatian, has phonemic vowel length, some dialects in both accented and unaccented syllables, some only in accented syllables - and no quantitative verse has ever been written in it, only syllabic/accentual verse (in fact, when the Iliad and the Odyssey were translated, the translator choose to make the hexameters accentual rather than quantitative even though he could have made them quantitative).

That's interesting. It may indeed be to do with timing. But also, while quantitative differences between syllables enable quantitative metric poetry, they don't necessarily impose it, even leaving timing aside, just as the possibility of rhyme doesn't impose rhyme.

I think in each poetic tradition there's a question of how hard poetry is meant to be; and each language different techniques are easier or harder. Techniques are chosen to match the required degree of difficulty. In terms of epics, for example, the great poetic form in English is blank verse: blank verse (largely) abandons rhyme, because although rhyme is possible in English, it's very difficult to sustain regular rhyming in English without repeating yourself; but it retains accentual metre, because accentual metre, particularly iambics, is relatively easy in English.

So even though Greek and Croatian both have the possibility to create regular quantitative poetry, it my be that the more subtle details of Croatian make quantitative poetry too difficult (or too easy!) for its poets.

Or, of course, it could be purely cultural. Irish adopted syllabic verse primarily through the influence of Mediaeval Latin church hymns, which obviously had great cultural authority. Maybe the models on which Croatian poetry was based just happened not to be quantitative?

*A look at their phonological histories shows an abundance of vowel contractions and compensatory lengthenings which always preserve mora count, and a complete lack of vowel reduction and syncope.

On the other hand, you can't always trust sound changes. English has had several sound changes that appear predicated on mora-timing, but it is not mora timed.

Off the top of my head, the three obvious ones are:
- before Old English, high vowels were lost in the second syllable if the first syllable was long, but not if it was short. They were lost in the third syllable if the second was long, OR if the first two were BOTH short, but not if the first was long and the second was short. In other words, if we take long second syllables as metrically short (and in any case rather rarer to begin with), this rule appears to result in high vowels being 'weak' and eventually dropping if they were in the third mora of a word. In other words, this rule acts to shepherd word boundaries to correspond with the boundaries of bimoraic feet.

- between Old English and Middle English, trisyllabic laxing reduced a long vowel in a stressed syllable to short if two more syllables followed. This appears to shepherd words toward having trimoraic feet. Interestingly, it originally operated only on long syllables in closed syllables (followed by two more syllables) - i.e. on metrically overlong, trimoraic syllables.

- similarly, open vowel lengthening is a moraic process, in that it likewise aims at creating trimoraic words. If the first syllable is long, the vowel is left alone, but if the first syllable is short and followed by an unstressed syllable, the vowel lengthens.


Looking at these rules, we might think English had always been mora-timed!

[although the give-away may be that all three rules underlying seem to assume that unstressed vowels have no metric length distinction. This explains both high vowel loss occurs after short-long sequences (the unstressed second syllable's phonemic length doesn't make it metrically long) and why laxing and lengthening don't care about the phonemic length in the following, unstressed syllables. This may be a symptom of English having on some level been, we might say 'syllabo-moraic', rather than actually moraic]


Intriguingly, of course, open-syllable stressed vowel lengthening also operated in Romance. And, interestingly, the results were different, as shown by breaking. French initially tolerated this quantitative rule, but Spanish and Occitan both rejected it: the latter by immediately shortening the vowels, avoiding any sort of breaking, and the former making all the stressed vowels long (leading to the breaking of the mid-low vowels occuring everywhere, rather than only in long syllables as in French). But then French caught up to the syllabic verse party: by shifting its falling diphthongs to rising diphthongs, it helped remove quantitativity - likewise by losing many coda consonants, and by removing the length distinction that in some cases arose as a result). The Romance languages seem to have not only transitioned toward syllable-timing, but done everything in their power to protect it! [other than rhaeto-romance, which, perhaps due to the influence of accentual Germanic, has iirc maintained the long vowels in some languages]

[Zekoslav]

Out of curiosity, do the vulgar poems you've been looking at exhibit any other phenomena common to Latin quantitative meter (e.g., penthemimeral caesura--maybe defined somewhat more forgivingly as word-end after a long syllable, or after the seventh syllable of the line)?

Trouble is, the sources where I got the information about this don't give any concrete examples, they just say such things have been found.

I would have put it the other way: the fact that Vulgar Latin may be presumed to have had no phonemic vowel length* would seem at a blow to explain the loss of quantitative verse.

I would quibble with two things here, though, in the interests of debate. First, you don't actually need phonemic vowel length to produce quantitative verse: you just need a distinction between light and heavy syllables. In most languages with quantitative verse, CVC syllables, or at least some of them (eg those ending in sonorants) are counted as long; so it's conceivable that even a language without long vowels could have quantitative verse, providing that it had a robust division into closed and open syllables.

Second, I don't think it's that clear that there was a 'failure to produce quantitative verse' at all. This seems to fall into the rhetorician's trap of assuming that classical Greek is always the aim of poetry. Latin-language poets who wrote syllabic or accentual verse may not have been trying to write quantitative verse - and hence not failing. And if there was no failure, it's hard to draw conclusions.

It's important to remember that verse forms can change for seemingly no reason. Mediaeval Irish, for example, switched from accentual to syllabic verse (due to contact with the poetry of other languages). In this regard, we might need to bare in mind that many writers of 'failed' Latin poetry may have been non-native, or at least multilingual, and that their expectations of poetry may have had as much to do with their own native poetic systems and languages as with those of Latin.



*All descendents of Vulgar Latin lost vowel length. And the whole of Western Romance at the least (don't know about Romanian) shared the development of an alternative system of vowel length (created in stressed short syllables), either before they broke apart or shortly after. The parsimonious solution is surely simply that the Classical Latin length system had already been lost in Vulgar Latin, or at the very least had become severely eroded.

I indeed recognised that distinguishing open and closed syllables may be enough to write quantitative verse, but since that wasn't the typical kind of quantitative verse I decided to ignore it.

There's actually pretty solid evidence that the Italian/French allophonic vowel length system existed already during the Roman Empire. During the 300's a grammarian whose name I forgot reminded people to be careful not to lengthen short stressed vowels, and during the 400's st. Augustine complained that grammarians say in cano the "a" is short and the "o" is long, while everyone pronounces the "a" as long and the "o" as short. When his contemporary grammarians try to describe the difference between short and long "e" and short and long "o", they seem to be describing the qualitative difference of their romance descendants, i.e. short = open and long = close. So it looks like during the late empire Latin, even Classical Latin, was pronounced with more-or-less Italian vowel quality and length (in that context, Augustine's comment that Africans don't distinguish short and long "o" is understood that they had a Sardinian-like vowel system where the two merged when phonemic vowel length was lost).

That's interesting. It may indeed be to do with timing. But also, while quantitative differences between syllables enable quantitative metric poetry, they don't necessarily impose it, even leaving timing aside, just as the possibility of rhyme doesn't impose rhyme.

I think in each poetic tradition there's a question of how hard poetry is meant to be; and each language different techniques are easier or harder. Techniques are chosen to match the required degree of difficulty. In terms of epics, for example, the great poetic form in English is blank verse: blank verse (largely) abandons rhyme, because although rhyme is possible in English, it's very difficult to sustain regular rhyming in English without repeating yourself; but it retains accentual metre, because accentual metre, particularly iambics, is relatively easy in English.

So even though Greek and Croatian both have the possibility to create regular quantitative poetry, it my be that the more subtle details of Croatian make quantitative poetry too difficult (or too easy!) for its poets.

Or, of course, it could be purely cultural. Irish adopted syllabic verse primarily through the influence of Mediaeval Latin church hymns, which obviously had great cultural authority. Maybe the models on which Croatian poetry was based just happened not to be quantitative?

Croatian poetry is indeed based on medieval Romance poetry. Our traditional verses are basically those found in medieval French poetry: octosyllables, decasyllables and dodecasyllables, usually rhymed and with a trochaic accentual pattern added on top. There was a period during the development of Proto-Slavic where vowel length was a redundant allophonic feature of some vowels (rather like the situation in Modern English) and as you commented concerning Vulgar Latin, this probably signed a death blow to any hypothetical quantitative verse inherited from PIE.

There is, though, a quantitative dressing on our syllabic/accentual verse: the penultimate syllable in dodecasyllabic verse is almost always long, and recited as extra-long. This serves to accentuate the cadence and is rather like the last foot of Latin hexameter.

On the other hand, you can't always trust sound changes. English has had several sound changes that appear predicated on mora-timing, but it is not mora timed.

Off the top of my head, the three obvious ones are:
- before Old English, high vowels were lost in the second syllable if the first syllable was long, but not if it was short. They were lost in the third syllable if the second was long, OR if the first two were BOTH short, but not if the first was long and the second was short. In other words, if we take long second syllables as metrically short (and in any case rather rarer to begin with), this rule appears to result in high vowels being 'weak' and eventually dropping if they were in the third mora of a word. In other words, this rule acts to shepherd word boundaries to correspond with the boundaries of bimoraic feet.

- between Old English and Middle English, trisyllabic laxing reduced a long vowel in a stressed syllable to short if two more syllables followed. This appears to shepherd words toward having trimoraic feet. Interestingly, it originally operated only on long syllables in closed syllables (followed by two more syllables) - i.e. on metrically overlong, trimoraic syllables.

- similarly, open vowel lengthening is a moraic process, in that it likewise aims at creating trimoraic words. If the first syllable is long, the vowel is left alone, but if the first syllable is short and followed by an unstressed syllable, the vowel lengthens.


Looking at these rules, we might think English had always been mora-timed!

[although the give-away may be that all three rules underlying seem to assume that unstressed vowels have no metric length distinction. This explains both high vowel loss occurs after short-long sequences (the unstressed second syllable's phonemic length doesn't make it metrically long) and why laxing and lengthening don't care about the phonemic length in the following, unstressed syllables. This may be a symptom of English having on some level been, we might say 'syllabo-moraic', rather than actually moraic]

There's a paper about Lithuanian somewhere which makes it clear that phonological rules dealing with morae can and do exist in languages that aren't mora-timed, so you're completely right about this. A very similar trisyllabic shortening rule operated in the history of Croatian as well... but it targeted only one of the three pitch accents, and short high vowels counted as half a mora while short low vowels counted as one mora! So it depended not simply on mora count, but on other phonetic features.

The English (actually West Germanic, IIRC) syncope rules are often understood in the context of stress patterns depending on mora count: that is, you divide a word into moraic trochees: long vowel/two short vowels = one trochee. You count trochees from left to right, and the relevant vowels are preserved if they belong to one of the trochees and lost of they don't (when they'd extend it into a dactyl).

short short, ˈx x > kept

short short short, ˈx x x > lost

long short, ˈ(x x) x > lost

long short short ˈ(x x) ˌx x > kept

I'm not sure how one would analyse short long short sequences. Maybe the first short vowel would be "extrametrical" as the terminology goes.

[brblues]

It might not be the right corner to ask, but I think the right people are likely to read it here, so I want to just go ahead and ask - has anybody here tried their hand at Hungarian and does have any good resources? I'd like a book, and possibly some good online resources just to add in the mix and have multiple sources. Thanks in advance!

[Clio]

Out of curiosity, do the vulgar poems you've been looking at exhibit any other phenomena common to Latin quantitative meter (e.g., penthemimeral caesura--maybe defined somewhat more forgivingly as word-end after a long syllable, or after the seventh syllable of the line)?

Trouble is, the sources where I got the information about this don't give any concrete examples, they just say such things have been found.

That is troubling. What are these sources, and would you mind quoting them?

[KaiTheHomoSapien]

So what's the deal with Rashida Tlaib's surname? It's pronounced /tə'liːb/ as if the "l" and "a" were transposed in the spelling, similar to the pronunciation of "Favre".

Knowing nothing about Arabic names or pronunciation, I have to ask: is this pronunciation an Anglicization or does it reflect the original pronunciation in Arabic? Was the spelling altered or is it an accurate transliteration of the Arabic name?