(Conlangs) Q&A Thread - Quick questions go here

[Omzinesý]

How plausible it would be that a Romlang preserved Latin phonetic stressing after the loss of vowel length?

cantāre has the stress on the penultimate syllable.
=> cantare would have the stress on the antepenultimate syllable

[Creyeditor]

Just to clarify: your idea is to lose vowel length but keep the synchronic latin stress rule leading to many (diachronic) stress shifts, right?
Sounds naturalistic but I would frame it slightly different. You lose vowel length, keep stress positions (so in a way 'lose' the synchronic stress rule) and later regularize stress to the antepenultimate in the cases you mentioned. Does that make sense?

[Omzinesý]

Just to clarify: your idea is to lose vowel length but keep the synchronic latin stress rule leading to many (diachronic) stress shifts, right?

Yes

Sounds naturalistic but I would frame it slightly different. You lose vowel length, keep stress positions (so in a way 'lose' the synchronic stress rule) and later regularize stress to the antepenultimate in the cases you mentioned. Does that make sense?

I did not mean that. Surely anything can happen after we deviate from the real histry of Romance, which is usually "Vulgar Latin".
But I'm wondering if the processes of weight-sensitive stressing somehow could have stayed productive for the loss of vowel length to directly lead to the stress shifts.

Now we get to the argument that Vulgar Latin didn't really lose long vowels but the contrast between short and long vowels.

[Creyeditor]

I know you did not mean that. I was hoping that the two descriptions are extensionally the same even if they are intensionally different.

[Salmoneus]

I think the stress rule has to go, at least temporarily.

The problem is that shortening all long vowels is a massive simultaneous change that effects a vast percentage of the lexicon.

If there's a small change - like loss of nasal codas before fricatives - I can imagine that speakers would immediately say "hang on, why does this short penult have stress? whoops, don't want to look silly by pronouncing it wrong, better move stress to the antepenult like in all other words".

But if half your words suddenly have stressed short penults, then it's hard to imagine everyone simultaneously changing the pronunciations of half their words, in order to maintain conformity with a 'rule' that half the lexicon now breaks.

[I know it's not actually half, but it's a very big percentage].

But of course the rule could later be reinstated, perhaps via shifting all stress to the antepenult and then shifting it back to heavy penults?

------------

The Latin-specific problem I see is that "cantare" would never actually have had a short vowel anyway!

In Proto-Romance/Vulgar Latin, the vowels in stressed open syllables were always long, and unstressed vowels and vowels in stressed closed syllables were always short. We talk about Latin "losing length", as though all the long vowels became short... but that's not what happened. It lost length phonemically, because length came to be determined by stress and syllable structure.

In that sense, the early Romance languages DID maintain the Latin stress rule: if the penult was heavy because of a long vowel or a sonorant coda, it was stressed; otherwise, the antepenult was stressed. It's just that phonemically the direction of causation changed (stressed because long > long because stressed), but the actual position of stress and length of the vowel didn't change at all. [except for short stressed open antepenults lengthening by analogy, and long penults shortening before a coda to avoid overlong syllables (if that hadn't happened already?)]

[interestingly, it seems to have kind of happened twice, because short vowels before nasals before fricative lengthened when the nasal dropped. Normally we'd just say this was due to the effect of nasalisation or handwave it as non-specific "compensatory lengthening". But we could also say that such vowels lengthened in penults to maintain the stress rule despite the loss of the nasal and that this spread by analogy to other vowels... but in this case the lengthening was PRIOR to the quality shift of long vowels. So it's tempting to see this as a process of stress-maintenance that was maintained for hundreds of years, both before and after the quality shift.]

You can't really have "this vowel is no longer phonemically long because it's forced to be long because of its stress" and also "this vowel is no longer stressed because it's short" at the same time. These two ideas go in opposite directions.

[Omzinesý]

Now we get to the argument that Vulgar Latin didn't really lose long vowels but the contrast between short and long vowels.

I think this is quite much what Salmoneus said.

So, the surface form and underlying form/rule cannot change simultaneously.

Either what Greyeditor suggested
1) Surface form changes (phonetic long vowels get short)
2) Stress moves out of the short syllable

Or
1) The underlying form changes, without affecting the surface (I cannot come up with an example at this example which was the problem)
2) The surface form changes.


Do we have to conclude that weight-sensitive stressing rules are very fragile?
(Or maybe not very fragile, just broken by major changes that affect weight.)

[Creyeditor]

I would say that stress positions are very stable and do not usually get 'updated' en masse if there is some other change (except for vowel deletion maybe).

[Salmoneus]

I think I would say that stress "rules" are primarily descriptive. Where a relatively small number of exceptions are created by the diachronics, they can act to regularise those exceptions through analogy. But the more exceptions there are, the less likely they are to be analogised away, and the more a new "rule" will take over, even if that's just "there is no rule, you have to learn it by heart".


I don't think I agree that stress positions aren't updated en masse, because things like "now everything has initial stress" do happen, particularly when stress was irregular before. But it should also be said that when created a new rule purely by analogy, the simpler the rule the more likely it is to catch on. Something like "initial stress" might easily develop as a simple rule, but something like "stress on the penult if heavy but otherwise the antepenult" is probably not going to come out of nowhere.

Then again, in Latin it did just come out of nowhere one day for no apparent reason and in a weird way, so, who knows.

But I'd also say: bear in mind that Latin hadn't even eliminated its own stress irregularities - there were a whole bunch of words that didn't obey the normal rules and weren't regularised. So expecting immediate regularisation of a whole heap of new irregulars may be unrealistic.

[Creyeditor]

I don't think I agree that stress positions aren't updated en masse, because things like "now everything has initial stress" do happen, particularly when stress was irregular before. But it should also be said that when created a new rule purely by analogy, the simpler the rule the more likely it is to catch on. Something like "initial stress" might easily develop as a simple rule, but something like "stress on the penult if heavy but otherwise the antepenult" is probably not going to come out of nowhere.

Just to clarify, what I was trying to say was that stress positions do get updated en masse, as part of diachronic changes, i.e. if a new synchronic stress rule emerges. I think they rarely (or never) get updated en masse due to other changes AND in order to keep a synchronic stress rule active. Vowel deletions might be an exception but I am not too sure of this.

This means a "new rule: stress initial syllable">"update stress position of half your vocabulary" is an expected scenario, but "lose vowel length"> "update stress positions of half your vocabulary" is not an expected scenario.

[Salmoneus]

Yes, I think I'd agree with that.

EDIT: I'd also add, a lot of stress shifts aren't really about creating new stresses, but about reprioritising existing stresses - primary stress shifting to a syllable that has some stress already. I suspect "move to initial stress" is common because often initial syllables already carry some stress, particularly in a language that otherwise has weak and unpredictable stress.

[LinguoFranco]

How weird would it be for a language with split-ergativity to encode grammatical aspect in the verb, but uses the ergativity split for some other criteria than aspect? Like the split is based on animacy, for instance?

[Omzinesý]

How weird would it be for a language with split-ergativity to encode grammatical aspect in the verb, but uses the ergativity split for some other criteria than aspect? Like the split is based on animacy, for instance?

Not weird at all.
Aspect is just one possible split of a split alignment. I don't see why coding aspect somewhere would change that.

[Arayaz]

How weird would it be for a language with split-ergativity to encode grammatical aspect in the verb, but uses the ergativity split for some other criteria than aspect? Like the split is based on animacy, for instance?

An animacy-based split is I think more common than an aspect-based one.

[Omzinesý]

But I'd also say: bear in mind that Latin hadn't even eliminated its own stress irregularities - there were a whole bunch of words that didn't obey the normal rules and weren't regularised. So expecting immediate regularisation of a whole heap of new irregulars may be unrealistic.

I didn't know that. Can you give some examples?

[LinguoFranco]

How weird would it be for a language with split-ergativity to encode grammatical aspect in the verb, but uses the ergativity split for some other criteria than aspect? Like the split is based on animacy, for instance?

An animacy-based split is I think more common than an aspect-based one.

Is that so? I was under the impression that it was the other way around. Most of the ergative languages I am familiar with tend to split based on either tense or aspect, like Hindi, Burushaski, Kashmiri, Kurdish, Mayan, and Basque (don't quote me on any of those.)

[Arayaz]

How weird would it be for a language with split-ergativity to encode grammatical aspect in the verb, but uses the ergativity split for some other criteria than aspect? Like the split is based on animacy, for instance?

An animacy-based split is I think more common than an aspect-based one.

Is that so? I was under the impression that it was the other way around. Most of the ergative languages I am familiar with tend to split based on either tense or aspect, like Hindi, Burushaski, Kashmiri, Kurdish, Mayan, and Basque (don't quote me on any of those.)

Arayaz Neverburn's Guide to Developing Split Ergativity
Disclaimer: I make a lot of very bold claims from a position of what looks like robust knowledge, but where I do not have as much authority as I seem to present myself as having.

Split ergativity usually comes about in one of three ways:

1. Development of new syntactic constructions for tense, aspect, or some other verbal element wherein the two arguments are not syntactically the subject and object.
2. Animacy hierarchy splits.
3. Use of the ergative for certain intransitive verbs with more agentivity on the experiencer's part, and/or lexicalization of the antipassive in some contexts.

The first of these produces tense- or aspect-based splits.
An example of how this comes about would be the development of a new future tense that is literally X goes, and Y is Ved. In this case, both X and Y are in the absolutive, so no marking at all occurs; this could be considered split ergativity.
Engála (by David and Jessie Peterson) has a negation-based split. The negative was literally X falls while Ving Y. Again, both are absolutive.

The second of these produces animacy-based splits.
I said that these were the most common, but I don't actually have any numbers to justify that; it's just a hunch.
Basically, what happens with these is that with arguments higher on the animacy hierarchy (where exactly you draw the line is up to you) use the ergative for intransitive sentences. Thus the ergative becomes a nominative for them, since it's also used when they're the agents of transitive verbs, and thus exactly matches the role of a nominative.
If you start out with absolutive-ergative-verb word order, which can come about from lexicalized passives, this can lead to OSV word order.

The third of these produces lexical splits.
This is essentially the same as fluid-S verbs. The ergative can be used in intransitive clauses when the subject is perceived as having more agentivity (e.g. in I swerved versus I died).
As for lexicalization of the antipassive ─ the antipassive just takes the ergative argument and makes it the subject while leaving the absolutive as an oblique argument. If it becomes required for some verbs, it would have a similar effect.

I may post this in the Teach & Share forum if nobody debunks it.

[Salmoneus]

But I'd also say: bear in mind that Latin hadn't even eliminated its own stress irregularities - there were a whole bunch of words that didn't obey the normal rules and weren't regularised. So expecting immediate regularisation of a whole heap of new irregulars may be unrealistic.

I didn't know that. Can you give some examples?

AIUI, the big categories are:

- the suffix -que moves stress to the penult, regardless of syllable weight

- Grammarians claimed many adverbs to have final stress for no reason; I wonder if this is really just that they had very little stress at all (as in earlier forms of the language/PIE), and they interpreted this as final stress due to some form of prosodic, intonational contour. Or maybe it's due to analogy (see next point).

- many prefixes are ignored - forms of "inhabeo" have the same stress as the corresponding forms of "habeo", etc. This results in some adverbs with final stress, like "adhuc". Compounds with "facere" ignore the entire first element of the compound

- where vowels have been lost or contracted, stress allocation often ignores this. This particular affects compounds with the irregularly shortened imperative "duc" (rather than "duce") and in post-classical times "dic" for "dice", words suffixed with the inquisitive -ne (so satis + ne > satín, tanto: + ne > tantón), and forms of nouns in -ius in which the disyllabic sequence -ii have been contracted to monosyllabic -i:, but still act as disyllabic for stress allocation purposes. This also affects words (often names) formed in -a:s, an old contracted form of -atis - so Horace's patron Maecenas and the mediaeval theologian Aquinas "should" both have final stress; interestingly, Wiktionary gives them normal stress, even though it notes that the -a:s suffix they contain imposes final stress. Likewise the adjectives "nostras" (of our country" and "vestras" (of your country). This also applies to forms of the perfect written (and presumably pronounced) without the -v- and with vowel loss (-i for -ivi, -at for -avit, etc), which still assign stress as though they had remained uncontracted.

[Omzinesý]

But I'd also say: bear in mind that Latin hadn't even eliminated its own stress irregularities - there were a whole bunch of words that didn't obey the normal rules and weren't regularised. So expecting immediate regularisation of a whole heap of new irregulars may be unrealistic.

I didn't know that. Can you give some examples?

AIUI, the big categories are:

- the suffix -que moves stress to the penult, regardless of syllable weight

- Grammarians claimed many adverbs to have final stress for no reason; I wonder if this is really just that they had very little stress at all (as in earlier forms of the language/PIE), and they interpreted this as final stress due to some form of prosodic, intonational contour. Or maybe it's due to analogy (see next point).

- many prefixes are ignored - forms of "inhabeo" have the same stress as the corresponding forms of "habeo", etc. This results in some adverbs with final stress, like "adhuc". Compounds with "facere" ignore the entire first element of the compound

- where vowels have been lost or contracted, stress allocation often ignores this. This particular affects compounds with the irregularly shortened imperative "duc" (rather than "duce") and in post-classical times "dic" for "dice", words suffixed with the inquisitive -ne (so satis + ne > satín, tanto: + ne > tantón), and forms of nouns in -ius in which the disyllabic sequence -ii have been contracted to monosyllabic -i:, but still act as disyllabic for stress allocation purposes. This also affects words (often names) formed in -a:s, an old contracted form of -atis - so Horace's patron Maecenas and the mediaeval theologian Aquinas "should" both have final stress; interestingly, Wiktionary gives them normal stress, even though it notes that the -a:s suffix they contain imposes final stress. Likewise the adjectives "nostras" (of our country" and "vestras" (of your country). This also applies to forms of the perfect written (and presumably pronounced) without the -v- and with vowel loss (-i for -ivi, -at for -avit, etc), which still assign stress as though they had remained uncontracted.

I've to remember them.

[Ahzoh]

I intend my language to have (optional) polypersonal agreement. How naturalistic would it be for only the affixes agreeing with the subject to distinguish gender/animacy while the object affixes only mark person and number?

The result would be thus:

3fs>3s: -tu-kka
3ms>3s: -ti-kka
3ns>3s: -ta-kka
3fs>3p: -tu-kkan
3ms>3p: -ti-kkan
3ns>3p: -ta-kkan

3fp>3s: -tū-ya
3mp>3s: -tī-ya
3np>3s: -tā-ya
3fp>3p: -tū-yan
3mp>3p: -tī-yan
3np>3p: -tā-yan

[Creyeditor]

Does the subject agreement distinguish person and number? If yes, your system sounds very naturalistic. There is a strong tendency in the languages of the world for object agreement to encode less distinctions if subject and object agreement differ in the number of distinction they make.