The Great Vowel Shift & Vocalism in Romance languages

A forum for discussing linguistics or just languages in general.
User avatar
qwed117
mongolian
mongolian
Posts: 3764
Joined: 20 Nov 2014 02:27

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by qwed117 »

Ser wrote:
26 Apr 2020 21:40
qwed117 wrote:
26 Apr 2020 20:09
Notice the alteration between /ɛ/ and /e/! That's I think what you're talking about, and why I was specific in referencing Campidanese, and not Logudorese or Nugorese
I'm afraid that, no, that's still not the allophonic vowel length we were talking about. We were talking about the (allophonic) lengthening of Latin vowels when they are stressed in open syllables, as in, say, Latin pira 'pear' [ˈpɪra] > [ˈpera] > [ˈpe:ra] (and then > early Old French [ˈpeirə]).

The Campidanese example there is different (and irrelevant) as all those [ɛ]s are inside closed syllables in the Latin etyma (cantāssem cantāssēmus cantāssent). Also, Spanish, Portuguese, Italian and Romanian have the same collapse of the system of stress alternations in that particular tense as Campidanese (Sp cantase cansemos cantasen, Rom cântasem cântaserăm cântaseră), so I'm not sure what it is that you read in that database.
From my understanding the Campidanese example experienced a collapse of the system, which is and remains allophonic in most other languages (ie there wasn't any phonemicization of the alteration. For example, :wikip: Spanish. The Latin etymon is not important to the understanding of this since the difference wasn't phonemicized.
Ser wrote:
26 Apr 2020 21:40
]
That Logudorese preterite looks whacky as hell though, to the point that I even wonder if you transcribed it correctly... It looks like it has got the same [-zi-] suffix that appears in a few Old French verbs that descend from the Latin 2nd or 3rd conjugation (facere fēcistī fēcimus > faire fesis fesimes [ˈfairə fəˈzis fəˈziməs], which expanded to e.g. sedēre sēdistī sēdimus > sedeir sesis sesimes [səˈðeir səˈzis səˈziməs]), but with thoroughly whacky stress assignment. Logudorese conserves geminate [ss] (ipsum > isso), so I don't think this preterite has taken forms from the pluperfect subjunctive in the 1PL and 2PL slots. Not to mention this is a 1st conjugation verb we're talking about (domare), which only increases the whackiness (why the -si- suffix if so?), although then again, Latin domāre did have a whacky half-1st half-3rd conjugation (domō/domat domāre domuī domitum), so maybe Logudorese just happens to conserve that. Anyway, what the hell?
Prefix: Romance languages, like all languages are cursed

I wasn't the one who transcribed it, so don't be blaming me. I copied it straight off the Oxford ODRVM. Of the 4 sources I have on Logudorese, I find 3 that give the first person preterit to be -esi (verbix, ODRVM, EB) and 1 that says ei (nativlang). I'd lean to supposing the nativlang preterit to be the mistranscribed version, because all three have distinct features that make me believe that they were all transcribed independently (especially the 3rd person preterit, -erunt in nativlang, -esint in ODRVM, and -eint in verbix).

Second thing to note is that to my knowledge, only Standard Italian and its related languages retain directly inherited geminates. <Ss> is not [ss], but just /s/. <Issos> represents /isos/or /isoz/ depending on the next phone, <usso> /uso/ (from Latin URSUM). In this case the transcription, could be phonologically understood as the pluperfect subjunctive. Obviously, though, the semantic mismatch is very large, enough to rule out the change- especially since, based on nativlang and verbix, Logudorese maintains the pluperfect subjunctive, like Spanish, probably an Aragonese influence.

Anyways, the issue with pushing this onto just domare, is that it's also *all* Sardinian verbs that share the preterit. <Ischire> becomes <ischesi>, <caentare> ("to warm") becomes <caentesi>, <ballare> becomes <ballesi>, so the explanation isn't just related to <domare>'s singular oddity. Instead, the actual answer is extremely cursed.
[url=https://books.google.com/books?id=7Q7H6uE2MMkC&pg=PA891&lpg=PA891#v=onepage&q&f=false wrote:Encyclopedia Britannica[/url]]Next comes the analogical and almost corrupt diffusion of the -si of the ancient strong perfects (such as posi and rosi-) by which cantesi, timesi (cantavi, timui), dolfesi, dolui, are reached. Proof of the use and even the abuse of the strong perfects is afforded, however by the participles and the infinitives of the category to which belong the following examples: tennidu, tenuto; parfidu, parso; balfidu, valso; tennere, balere, &c (Arch. ii. 432-433)
I love the absolute sense of of disgust in the author's voice.
Spoiler:
My minicity is Zyphrazia and Novland
What is made of man will crumble away.

User avatar
Ser
sinic
sinic
Posts: 247
Joined: 30 Jun 2012 06:13
Location: Vancouver, British Columbia / Colombie Britannique, Canada

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by Ser »

qwed117 wrote:
27 Apr 2020 19:12
Ser wrote:
26 Apr 2020 21:40
qwed117 wrote:
26 Apr 2020 20:09
Notice the alteration between /ɛ/ and /e/! That's I think what you're talking about, and why I was specific in referencing Campidanese, and not Logudorese or Nugorese
I'm afraid that, no, that's still not the allophonic vowel length we were talking about. We were talking about the (allophonic) lengthening of Latin vowels when they are stressed in open syllables, as in, say, Latin pira 'pear' [ˈpɪra] > [ˈpera] > [ˈpe:ra] (and then > early Old French [ˈpeirə]).

The Campidanese example there is different (and irrelevant) as all those [ɛ]s are inside closed syllables in the Latin etyma (cantāssem cantāssēmus cantāssent). Also, Spanish, Portuguese, Italian and Romanian have the same collapse of the system of stress alternations in that particular tense as Campidanese (Sp cantase cansemos cantasen, Rom cântasem cântaserăm cântaseră), so I'm not sure what it is that you read in that database.
From my understanding the Campidanese example experienced a collapse of the system, which is and remains allophonic in most other languages (ie there wasn't any phonemicization of the alteration. For example, :wikip: Spanish. The Latin etymon is not important to the understanding of this since the difference wasn't phonemicized.
That... is still not what Salmoneus and I were talking about. I keep replying to you because it doesn't seem to me you understand the conjectured Late Latin allophonic length we were discussing, as you keep bringing up irrelevant separate topics when you reply to me. And I would like you to understand it.

The Campidanese single /ɛ/ in that pluperfect>imperfect subjunctive tense, opposed to Logudorese /ɛ/ and /e/ in a similar preterite context, is irrelevant because in the Latin etyma all of those Campidanese /ɛ/s are found inside closed syllables (amāssem amāssimus amāssent). The Latin etyma is absolutely important to understand the Late Latin allophonic length Sal and I were discussing because many of the languages it produces effects in (like Old French, Rhaeto-Romance, Dalmatian) have degemination and plenty of cluster simplification.

Sal and I made some comments above about Spanish being completely irrelevant in general to say things about Late Latin allophonic vowel length, because the language had straightforward /e/ > /e/ and /ɛ/ > /je/ in stressed syllables. If anything, Late Latin /ɛ/ might have been turned into [ɛ:] in pre-Spanish no matter if it was inside an open or closed syllable, but maybe not even. Your link to allophony in contemporary Spanish today is stil a further separate topic, especially as the Latin etyma is not relevant for today's Spanish allophones there.

We're really talking about something regarding the pronunciation of spoken Latin at some point in Late Antiquity here.


Let me try to restate the topic at hand as a whole, as clearly as I can possibly do:

It is easily observed that Latin vowels, specifically the Late Latin vowels after the mainstream Romance mergers, evolved differently into Old French and Dalmatian (among other daughter languages) depending on whether they were in an open or closed syllable. A little sample data showing regular sound changes in Old French and Dalmatian (all accusative nouns):

/ɛ/ in open syllable: petram /ˈpɛtra/ > OFr pierre [ˈpjerə], Dal pitra
/ɛ/ in closed syllable: ventum /ˈvɛnto/ > OFr vent [vɛnt], Dal viant
/a/ in open syllable: cīvitātem /tʃiveˈtate/ > OFr citét [tsiˈteθ] (Dal čituot)
/a/ in closed syllable: carnem /ˈkarne/ > OFr char [tʃar] (Dal kuorne)
/ɔ/ in open syllable: novum /ˈnɔvo/ > OFr nuef [nwɛf], Dal nuf
/ɔ/ in closed syllable: porcum /ˈpɔrko/ > OFr porc [pɔrk], Dal puark

The Dalmatian data is interesting, though, because the higher Late Latin /i e o u/ in stressed open syllables (sampled in my 1st post, not here) break into /ai ai au oi/, which suggests that a GVS happened that pulled Late Latin /ɛ ɔ/ in open syllables further up to the outcomes /i u/ (maybe via *[jɛ wɔ] as a middle step). Regarding the diphthongization of Late Latin /e ɛ ɔ/ in stressed closed syllables into /ja ja wa/, the fact that these did not join the /e ɛ ɔ/ > /ai i u/ open-syllable change likely says something about length anyway (and besides, /e ɛ ɔ/ > /ja ja wa/ may be very recent, in line with Romanian /e o/ > /e̯a o̯a/).

And from these observations, noticing that whether the Late Latin etyma had an open or closed syllable had such strong effects, we also conjecture that Late Latin vowels had lengthened allophones in open syllables, at least in the dialects that gave birth to Gallo-Romance and Dalmatian. That is, petram /ˈpɛtra/ was perhaps [pɛ:tra] while ventum /ˈvɛnto/ was [ˈvɛnto], or, piram /ˈpera/ was [ˈpe:ra] while siccum /ˈsek:o/ was [ˈsek:o].

Such length then gets reflected in the outcomes above, as in:

1) novum /ˈnɔwʊm/
2) /ˈnɔvo/ [ˈnɔ:vo]
3) [ˈnwɔvo]
4a) pre-French [ˈnwɛvo] > Old French nuef [ˈnwɛf]
4b) pre-Dalmatian [ˈnwɔf] > Dalmatian [nuf]

1) porcum /ˈpɔrkʊm/
2) /ˈpɔrko/ [ˈpɔrko]
3a) Old French porc [pɔrk]
3b) pre-Dalmatian [ˈpwarko] > Dalmatian [ˈpwark]
hīc sunt linguificēs. hēr bēoþ tungemakeras.

User avatar
qwed117
mongolian
mongolian
Posts: 3764
Joined: 20 Nov 2014 02:27

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by qwed117 »

Spoiler:
Ser wrote:
27 Apr 2020 21:25
qwed117 wrote:
27 Apr 2020 19:12
Ser wrote:
26 Apr 2020 21:40
qwed117 wrote:
26 Apr 2020 20:09
Notice the alteration between /ɛ/ and /e/! That's I think what you're talking about, and why I was specific in referencing Campidanese, and not Logudorese or Nugorese
I'm afraid that, no, that's still not the allophonic vowel length we were talking about. We were talking about the (allophonic) lengthening of Latin vowels when they are stressed in open syllables, as in, say, Latin pira 'pear' [ˈpɪra] > [ˈpera] > [ˈpe:ra] (and then > early Old French [ˈpeirə]).

The Campidanese example there is different (and irrelevant) as all those [ɛ]s are inside closed syllables in the Latin etyma (cantāssem cantāssēmus cantāssent). Also, Spanish, Portuguese, Italian and Romanian have the same collapse of the system of stress alternations in that particular tense as Campidanese (Sp cantase cansemos cantasen, Rom cântasem cântaserăm cântaseră), so I'm not sure what it is that you read in that database.
From my understanding the Campidanese example experienced a collapse of the system, which is and remains allophonic in most other languages (ie there wasn't any phonemicization of the alteration. For example, :wikip: Spanish. The Latin etymon is not important to the understanding of this since the difference wasn't phonemicized.
That... is still not what Salmoneus and I were talking about. I keep replying to you because it doesn't seem to me you understand the conjectured Late Latin allophonic length we were discussing, as you keep bringing up irrelevant separate topics when you reply to me. And I would like you to understand it.

The Campidanese single /ɛ/ in that pluperfect>imperfect subjunctive tense, opposed to Logudorese /ɛ/ and /e/ in a similar preterite context, is irrelevant because in the Latin etyma all of those Campidanese /ɛ/s are found inside closed syllables (amāssem amāssimus amāssent). The Latin etyma is absolutely important to understand the Late Latin allophonic length Sal and I were discussing because many of the languages it produces effects in (like Old French, Rhaeto-Romance, Dalmatian) have degemination and plenty of cluster simplification.

Sal and I made some comments above about Spanish being completely irrelevant in general to say things about Late Latin allophonic vowel length, because the language had straightforward /e/ > /e/ and /ɛ/ > /je/ in stressed syllables. If anything, Late Latin /ɛ/ might have been turned into [ɛ:] in pre-Spanish no matter if it was inside an open or closed syllable, but maybe not even. Your link to allophony in contemporary Spanish today is stil a further separate topic, especially as the Latin etyma is not relevant for today's Spanish allophones there.

We're really talking about something regarding the pronunciation of spoken Latin at some point in Late Antiquity here.


Let me try to restate the topic at hand as a whole, as clearly as I can possibly do:

It is easily observed that Latin vowels, specifically the Late Latin vowels after the mainstream Romance mergers, evolved differently into Old French and Dalmatian (among other daughter languages) depending on whether they were in an open or closed syllable. A little sample data showing regular sound changes in Old French and Dalmatian (all accusative nouns):

/ɛ/ in open syllable: petram /ˈpɛtra/ > OFr pierre [ˈpjerə], Dal pitra
/ɛ/ in closed syllable: ventum /ˈvɛnto/ > OFr vent [vɛnt], Dal viant
/a/ in open syllable: cīvitātem /tʃiveˈtate/ > OFr citét [tsiˈteθ] (Dal čituot)
/a/ in closed syllable: carnem /ˈkarne/ > OFr char [tʃar] (Dal kuorne)
/ɔ/ in open syllable: novum /ˈnɔvo/ > OFr nuef [nwɛf], Dal nuf
/ɔ/ in closed syllable: porcum /ˈpɔrko/ > OFr porc [pɔrk], Dal puark

The Dalmatian data is interesting, though, because the higher Late Latin /i e o u/ in stressed open syllables (sampled in my 1st post, not here) break into /ai ai au oi/, which suggests that a GVS happened that pulled Late Latin /ɛ ɔ/ in open syllables further up to the outcomes /i u/ (maybe via *[jɛ wɔ] as a middle step). Regarding the diphthongization of Late Latin /e ɛ ɔ/ in stressed closed syllables into /ja ja wa/, the fact that these did not join the /e ɛ ɔ/ > /ai i u/ open-syllable change likely says something about length anyway (and besides, /e ɛ ɔ/ > /ja ja wa/ may be very recent, in line with Romanian /e o/ > /e̯a o̯a/).

And from these observations, noticing that whether the Late Latin etyma had an open or closed syllable had such strong effects, we also conjecture that Late Latin vowels had lengthened allophones in open syllables, at least in the dialects that gave birth to Gallo-Romance and Dalmatian. That is, petram /ˈpɛtra/ was perhaps [pɛ:tra] while ventum /ˈvɛnto/ was [ˈvɛnto], or, piram /ˈpera/ was [ˈpe:ra] while siccum /ˈsek:o/ was [ˈsek:o].

Such length then gets reflected in the outcomes above, as in:

1) novum /ˈnɔwʊm/
2) /ˈnɔvo/ [ˈnɔ:vo]
3) [ˈnwɔvo]
4a) pre-French [ˈnwɛvo] > Old French nuef [ˈnwɛf]
4b) pre-Dalmatian [ˈnwɔf] > Dalmatian [nuf]

1) porcum /ˈpɔrkʊm/
2) /ˈpɔrko/ [ˈpɔrko]
3a) Old French porc [pɔrk]
3b) pre-Dalmatian [ˈpwarko] > Dalmatian [ˈpwark]
You're completely missing the point. If you have an allophonic contrast among a certain segment, three different things can happen to that contrast.

It can be phonemicized by other environment changes (ie what you mentioned about /pɔrko/ > [ˈpwark], /ˈnwɔvo/ > [ˈnuf]). This is what you're noting happened in Old French, Gallo-Italic and Dalmatian.

It can be completely collapsed, in other words, the allophonic distinction is lost with no phonemicization. This happened in American English, where the [l] and [ɫ] allophones of /l/ have coalesced and become realized as [ɫ] relatively uniformly. This is what happened in Campidanese Sardinian. The long [e:] that appears in stressed open syllables has merged with [ɛ] occurring in unstressed or closed syllables. Sure the example paradigm I chose probably wasn't the best for explaining that, but you can compare the present paradigm of depere, where there's no alteration between the /ɛ/ in the stem. I do want to remind you though that Campidanese does not have phonemic consonant length, unlike the syntactic gemination of Italian.

1SG: ˈdɛpu
2SG: ˈdɛpis
3SG: ˈdɛpit
1PL: dɛˈpeus
2PL: dɛˈpeis
3PL: ˈdɛpint(i)

It can be leveled as environment as the environment changes as happened in Logudorese and Spanish. As gemination was lost (which would ordinarily produce a phonemic distinction, the rule governing the long vowel/short vowel distinction was expanded to all stressed syllables (as Sal mentioned earlier), which is the origin of the [ɛ]/[e] alteration here.

1SG: ˈdɔmɛsi
2SG: dɔˈmesti
3SG: ˈdɔmɛsit
1PL: dɔˈmesimus
2PL: dɔˈmeʣis
3PL: ˈdɔmɛsint

Now I want to bring up this post that Sal made.
Salmoneus wrote:
Besides the Langues d'Oïl and Rhaeto-Romance, Dalmatian as shown above also shows different vowel sound changes depending on whether a syllable is open, in a way that's reminiscent of the shift in long vowels that Middle English went through (the higher long vowels break into falling diphthongs, the lower ones get raised to a higher quality, and something funny happens to long "a"). Also, modern Italian shows the same lengthening allophonically, coincidentally or not (not easy to tell due to its huge conservatism...): cadit > cade /ˈkade/ [ˈka:de], cantō > canto /ˈkanto/ [ˈkanto].
Yes, and iirc it also interacts with metaphony in the Italian dialects. But we can't be sure that open syllables were different due to significant lengthening specifically, although it certainly seems the obvious option.
Spanish and Portuguese show no evidence of any ancient vowel lengthening though, zero, nada. And I don't really know about Occitan or Romanian at the moment. Perhaps it only affected languages near the centre...
Iirc, the theory is that the Iberian and Dacian languages generalised the long stressed open syllables to ALL stressed syllables. This "explains" the breaking of stressed mid-vowels, which can then be seen as a general areal change that affected all long mid vowels.
Now perhaps I was a bit unclear with my original wording, I apologize. If you want me to be extra-precise, evidence points to some allophonic vowel length system existing *throughout* Late Latin, a system that was conditioned on open/closed syllables and stress. In some languages that allophonic contrast was phonemicized. In others, that allophonic contrast was expanded to only condition on stress as geminates were lost, thus avoiding the phonemicization of the contrast (Logudorese, DAIM-Romanian). In some languages that allophonic contrast was expanded to condition on stress, resulted in some change and then reformed again with conditioning on open/closed syllables and stress (Spanish, although tbf I'm not sure how parsimonious it would be to assume the system reformed rather than that all /ɛ ɛ:/ diphthongized, and that the current system is just the former with some levelling). The Latin etyma are not important for these forms because the difference was not phonemicized but rather levelled.

I don't want to argue on this anymore, so I'm not going to write any further response.
Spoiler:
My minicity is Zyphrazia and Novland
What is made of man will crumble away.

User avatar
Ser
sinic
sinic
Posts: 247
Joined: 30 Jun 2012 06:13
Location: Vancouver, British Columbia / Colombie Britannique, Canada

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by Ser »

qwed117 wrote:
28 Apr 2020 00:04
I don't want to argue on this anymore, so I'm not going to write any further response.
I find it very unfortunate though. I just read and re-read your post several times (more times than what "several" implies actually), and while I had some things about your posts clarified, I'm also left confused about others.

I think that the larger reason why we are having this weird conversation is because I'm too ignorant about synchronic and diachronic Sardinian, but you assume that I'm much more knowledgeable. So you mention this while assuming that, and I can't fully reconstruct your meaning because I don't know enough Sardinian.

For example, in your post you seem to be saying that Late Latin open-syllable /e/ [e:] and closed-syllable and unstressed /ɛ/ [ɛ] provided the allophones for the single Sardinian phoneme /e/, with [e] in stressed syllables and [ɛ] unstressed ones. Actually, I'm not sure if this is what you're saying, since you don't spell the changes out clearly with phoneme + phones stage by stage, but I'm 70% confident I'm reading you correctly (if I knew more about Sardinian, I wouldn't need to guess because I'd already know the changes, but here we are). I also find this a curious way to think about it, since you could also just say that the new Sardinian phoneme /e/ was realized higher as [e] when stressed and more lowly as [ɛ] when not, without linking them to supposed contributions by unchecked /e/ and checked /ɛ/, but it doesn't really matter.

But then, I don't understand how Campidanese shows a collapsed system whereas Logudorese doesn't. However, in one of my re-readings, I noticed that you refer to the spread of [ɫ] to all positions in American English as a collapse, so maybe you're using the word "collapse" for the collapse of allophones. Which would be a bit odd, but okay. I mean, Spanish is often described as having a simple mid [e̞ ] as the one allophone of /e/, although others describe the slight changes in height depending on the nearby consonants with [e ~ ɛ], and some even make more distinctions than that (I believe I once came across a phonetics textbook from Spain, written in Spanish, at UBC's library that distinguished four or five allophones of /e/).

The thing is, phonetic realizations can be described as minutely as you wish, so normally one doesn't talk much about allophone collapse, especially for vowel quality. If you read about Mandarin phonetics, you'll notice different authors give different allophones for /ə/ and /a/, even when talking about the same sound, e.g. some use [ɥɛ] for /ɥa/, but others use [ɥæ] like the "[æ]" of /an/ [æn], it's just a matter of what range you give to "[ɛ]" and "[æ]".

But then, you give that Campidanese paradigm where -ēmus > [eus], with a close [e], while saying Campidanese had an allophone collapse to [ɛ]. But maybe you're not counting this, because for you it's a separate /eu/ [eu] phoneme. Again, I don't know what model of Campidanese phonology you're using here, because I don't know about scholarship on synchronic Sardinian.


So, if I assume all my interpretations above are correct, I think I'm finally seeing some light about what you meant in your first post when you said that Oxford database says the allophone(?) collapse happened in all of Romance except Campidanese. You meant to say that all of Romance has [e ɛ] as allophones if not outright distinct phonemes for the unrounded front part of the vowel system, except Campidanese which only has [ɛ] for the general Sardinian "/e/" (more accurately "/ɛ/" for Campidanese, ignoring /eu ei/). Although for this we kind of need to disregard that Spanish, Romanian, Sicilian and others may be described with just [e̞], or perhaps way more allophones than [e ~ ɛ], depending on the phonetician's minuteness.

I'm still not sure if this is what you meant though. Could you or anyone else confirm I understood you (qwed) correctly?
You're completely missing the point.
Yeah, I think so.

I don't have great hopes that you will reply, but with the above I hope that at least you can appreciate why I've had incredible trouble reading your posts. I don't know what mental models of Sardinian dialects and history you're using here, because of my ignorance of Sardinian, or your use of "collapse", because I'm unfamiliar with the use of that term for allophonic vowel quality (as in pre-Sardinian [e ~ ɛ] > Campidanese [ɛ]). When you talked about "collapse" in your first post, I naturally, but wrongly, assumed you meant the southern Romance phonemic collapse of Late Latin /e ɛ/ > pre-Sardinian /e/ (which was [e ~ ɛ] as you say). And then all the misunderstandings continued happening from that.

I'm not going to deny I've been very amused by how hard I find it to understand your posts in this thread, which only makes me want to try harder to get to the meaning. I've taken it like a challenge. Otherwise I wouldn't have spent the last hour and a half re-reading your post and typing this one. It's so interesting, and I'm not being sarcastic. Partly it's also because I recently had a conversation with someone about how hard it is to understand language pragmatically.

By the way, if you have not-so-hard-to-access recommendations for readings about Sardinian, I'd like to know. I remember trying to find scholarship on it about six years ago, but quite a few publications were in German (which I can't read), or otherwise not available for me in North America even through inter-library loans nor posted on the pirate online libraries. It was really annoying. And I can read Italian. But I think there are some gems to be found among general surveys, paper collections and individual papers here and there, but those are less easy to identify.
In some languages that allophonic contrast was expanded to condition on stress, resulted in some change and then reformed again with conditioning on open/closed syllables and stress (Spanish, although tbf I'm not sure how parsimonious it would be to assume the system reformed rather than that all /ɛ ɛ:/ diphthongized, and that the current system is just the former with some levelling).
I don't know what model you're picturing there in the main sentence for Spanish, but it sounds interesting. I mean, whether a syllable is open or closed generally doesn't matter for the development of Latin phonemes > Spanish phonemes, but here you say it was part of the conditioning, and I'm not sure why. (Maybe you have the allophones on Wikipedia in mind, but those modern [e ~ ɛ] are so far removed from Latin etyma, just as you have also said before, that they can't count in the conditioning, so I think I'm failing to understand your meaning here.)

The alternative analysis you mention is basically the normal explanation, namely that Late Latin /ɛ/ [ɛ] (no allophonic vowel length) underwent straight diphtongization to [je] when stressed. Alternatively, stress is used as the only condition, so that /ɛ/ [ɛ ~ ɛ:] becomes /ɛ/ with long [ɛ:] whenever it's stressed, and then maybe [e̯ɛ] > [jɛ], then Old Spanish [je]: dextera > dextra > [ˈdɛstɾa] > [ˈdɛːstɾa] > ([ˈde̯ɛstɾa]) > [ˈdjɛstɾa] > OSp diestra [ˈdjestra].
hīc sunt linguificēs. hēr bēoþ tungemakeras.

User avatar
Zekoslav
sinic
sinic
Posts: 340
Joined: 07 Oct 2017 16:54

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by Zekoslav »

Should all of this maybe moved to the Romance tidbits thread? This is now quite off-topic and any further reply would be as well.
Languages:
:hrv: [:D], :bih: :srb: [;)], :eng: [:D], :fra: [:|], :lat: [:(], :deu: [:'(]

A linguistics enthusiast who occasionally frequents the CBB.

- Guide to Slavic accentuation

User avatar
Aszev
admin
admin
Posts: 1381
Joined: 11 May 2010 05:46
Location: ꜱᴇ

Re: Apparently Old Prussian underwent the Great Vowel Shift

Post by Aszev »

Zekoslav wrote:
28 Apr 2020 09:39
Should all of this maybe moved to the Romance tidbits thread? This is now quite off-topic and any further reply would be as well.
Since the posts on the different topics were so intertwined, I resolved it like this instead! :)

Salmoneus
MVP
MVP
Posts: 1917
Joined: 19 Sep 2011 19:37

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Salmoneus »

Maybe we should set things out in black and white, for clarity?

Here's how I think Proto-Romance worked (considering only short *e, and ignoring cases where there is a following /j/, /l/ or a nasal, which often worked out differently):

Code: Select all

Latin	||	Early Proto-Romance
	||	Stressed Closed	|	Stressed Open	|	Unstressed
e	||		ɛ		|		ɛ:		|		ɛ (merged with /e/)

At this point, length is allophonic, as it is predictable from context. The loss of some unstressed vowels immediately brought about the breaking of this system:

Code: Select all

Latin	||	Early Proto-Romance	||	Late Proto-Romance
	||					||	Stressed Closed	|	Stressed Open	|	Unstressed
   e	||			ɛ		||		ɛ		|		N/A		|		ɛ
   e	||			ɛ:		||		ɛ:		|		ɛ:		|	      N/A	
		
Length is now distinguished only in stressed closed syllables. We can therefore think of this as there being effectively two types of stress: lengthening stress and non-lengthening stress (particularly since some of these stressed vowels would still be in transparent derivational and inflectional relationships with the unstressed vowels, and some would be in alternation between originally 'open' and 'closed' forms). This was a bit of a mess, particularly when you consider that there's also e and e: in the system, and that all four vowels reduce to one unstressed vowel...

So, Iberia and Romania (and apparently Wallonia?) as peripheral areas perhaps less comfortable with the prevailing system and more at liberty to change (having less contact with other Romance speakers and more contact with speakers of other languages), both independently regularised this system, eliminating the phonemic distinction by analogising the rule that all (as opposed to only most) stressed vowels are long:

Code: Select all

Latin	||	Early Proto-Romance	||	"Peripheral"
	||					||	Stressed Closed	|	Stressed Open	|	Unstressed
   e	||			ɛ		||		ɛ:		|		N/A		|		ɛ
   e	||			ɛ:		||		ɛ:		|		ɛ:		|	      N/A	
		
In some places, there is then diphthongisation of many long vowels. This may have originated in metaphonic diphthongisation of long vowels depending on the quality of the following vowel, which then spread to all long vowels. This diphthongisation occured in central Italy, in a band up to French, apparently - although in gallo-italic dialects it's often obscured by later monophthongisation - and then back down into northwestern iberia. It also happened in some romansh dialects and in eastern romance, but apparently it's complicated, and much of it can be explained as later changes.

Then many languages merged some of their short vowels, and then most languages lost phonemic length altogether (it remains in Friulian).

User avatar
qwed117
mongolian
mongolian
Posts: 3764
Joined: 20 Nov 2014 02:27

Re: GVS & Romance vocalism

Post by qwed117 »

Ser wrote:
28 Apr 2020 02:43
qwed117 wrote:
28 Apr 2020 00:04
I don't want to argue on this anymore, so I'm not going to write any further response.
I find it very unfortunate though. I just read and re-read your post several times (more times than what "several" implies actually), and while I had some things about your posts clarified, I'm also left confused about others.

I think that the larger reason why we are having this weird conversation is because I'm too ignorant about synchronic and diachronic Sardinian, but you assume that I'm much more knowledgeable. So you mention this while assuming that, and I can't fully reconstruct your meaning because I don't know enough Sardinian.

For example, in your post you seem to be saying that Late Latin open-syllable /e/ [e:] and closed-syllable and unstressed /ɛ/ [ɛ] provided the allophones for the single Sardinian phoneme /e/, with [e] in stressed syllables and [ɛ] unstressed ones. Actually, I'm not sure if this is what you're saying, since you don't spell the changes out clearly with phoneme + phones stage by stage, but I'm 70% confident I'm reading you correctly (if I knew more about Sardinian, I wouldn't need to guess because I'd already know the changes, but here we are). I also find this a curious way to think about it, since you could also just say that the new Sardinian phoneme /e/ was realized higher as [e] when stressed and more lowly as [ɛ] when not, without linking them to supposed contributions by unchecked /e/ and checked /ɛ/, but it doesn't really matter.

But then, I don't understand how Campidanese shows a collapsed system whereas Logudorese doesn't. However, in one of my re-readings, I noticed that you refer to the spread of [ɫ] to all positions in American English as a collapse, so maybe you're using the word "collapse" for the collapse of allophones. Which would be a bit odd, but okay. I mean, Spanish is often described as having a simple mid [e̞ ] as the one allophone of /e/, although others describe the slight changes in height depending on the nearby consonants with [e ~ ɛ], and some even make more distinctions than that (I believe I once came across a phonetics textbook from Spain, written in Spanish, at UBC's library that distinguished four or five allophones of /e/).

The thing is, phonetic realizations can be described as minutely as you wish, so normally one doesn't talk much about allophone collapse, especially for vowel quality. If you read about Mandarin phonetics, you'll notice different authors give different allophones for /ə/ and /a/, even when talking about the same sound, e.g. some use [ɥɛ] for /ɥa/, but others use [ɥæ] like the "[æ]" of /an/ [æn], it's just a matter of what range you give to "[ɛ]" and "[æ]".

But then, you give that Campidanese paradigm where -ēmus > [eus], with a close [e], while saying Campidanese had an allophone collapse to [ɛ]. But maybe you're not counting this, because for you it's a separate /eu/ [eu] phoneme. Again, I don't know what model of Campidanese phonology you're using here, because I don't know about scholarship on synchronic Sardinian.


So, if I assume all my interpretations above are correct, I think I'm finally seeing some light about what you meant in your first post when you said that Oxford database says the allophone(?) collapse happened in all of Romance except Campidanese. You meant to say that all of Romance has [e ɛ] as allophones if not outright distinct phonemes for the unrounded front part of the vowel system, except Campidanese which only has [ɛ] for the general Sardinian "/e/" (more accurately "/ɛ/" for Campidanese, ignoring /eu ei/). Although for this we kind of need to disregard that Spanish, Romanian, Sicilian and others may be described with just [e̞], or perhaps way more allophones than [e ~ ɛ], depending on the phonetician's minuteness.

I'm still not sure if this is what you meant though. Could you or anyone else confirm I understood you (qwed) correctly?
You're completely missing the point.
Yeah, I think so.

I don't have great hopes that you will reply, but with the above I hope that at least you can appreciate why I've had incredible trouble reading your posts. I don't know what mental models of Sardinian dialects and history you're using here, because of my ignorance of Sardinian, or your use of "collapse", because I'm unfamiliar with the use of that term for allophonic vowel quality (as in pre-Sardinian [e ~ ɛ] > Campidanese [ɛ]). When you talked about "collapse" in your first post, I naturally, but wrongly, assumed you meant the southern Romance phonemic collapse of Late Latin /e ɛ/ > pre-Sardinian /e/ (which was [e ~ ɛ] as you say). And then all the misunderstandings continued happening from that.

I'm not going to deny I've been very amused by how hard I find it to understand your posts in this thread, which only makes me want to try harder to get to the meaning. I've taken it like a challenge. Otherwise I wouldn't have spent the last hour and a half re-reading your post and typing this one. It's so interesting, and I'm not being sarcastic. Partly it's also because I recently had a conversation with someone about how hard it is to understand language pragmatically.
First, I'd like to say, I think you understand what I'm misusing terminology to say, at least when you're asking about clarification. The reason I got irritated with that was the comment about "mistranscribing Logudorese", primarily because I was just copypasting from the Oxford site. It wasn't my fault. With that in mind, I may have taken your terminological criticism, definitely meant out of good faith, as being a personal attack, rather than just expressing confusion.

I do I think, though, I owe an apology. I've been completely misunderstanding the Oxford source. I would like to reiterate, first, that from my understanding, Sardinian *definitely* has the stress/vowel length system that Italian has- that is the lengthening of vowels in a stressed open syllables (which would have been the system in Late Latin, which later was phonemicized, or otherwise changed).

But that's not what the Oxford source shows. The Oxford source is referring to metaphony, which Campidanese acts weirdly with regards to, because final vowels raised and Logudorese doesn't. That would explain why Logudorese has the alteration between [ɛ] and ['e], which is that unstressed vowels must not experience metaphonic alterations. That doesn't discount what I said earlier about the stress/vowel length system. It just means that I understand why everything that I said made no sense. Because it was wrong. [:P]. Re the Campidanese paradigm though, I think it's a mistranscription- because Sardinian shifted all the 2nd declension verbs into the 3rd declension, at least to my knowledge, meaning that stress should be rhizotonic in depeus. Adding to that mess, all the other verbs in comparable forms show short [ɛi] and [ɛu], even in forms where I would expect a metaphonic [ei] [eu] cf. fai-faˈdɛus (from facio). So I think there's clear problems in the Oxford database. But as stated earlier, I'm pretty certain that Sardinian retained/obtained the vowel-lengthening system that Italian uses, which is what we care about.
Ser wrote:
28 Apr 2020 02:43
By the way, if you have not-so-hard-to-access recommendations for readings about Sardinians, I'd like to know. I remember trying to find scholarship about six years ago, but quite a few publications were in German (which I can't read), or otherwise not available for me in North America even through inter-library loans nor posted on the pirate online libraries. It was really annoying. And I can read Italian. But I think there are some gems to be found among general surveys, paper collections and individual papers here and there, but those are less easy to identify.
In some languages that allophonic contrast was expanded to condition on stress, resulted in some change and then reformed again with conditioning on open/closed syllables and stress (Spanish, although tbf I'm not sure how parsimonious it would be to assume the system reformed rather than that all /ɛ ɛ:/ diphthongized, and that the current system is just the former with some levelling).
I don't know what model you're picturing there in the main sentence for Spanish, but it sounds interesting. I mean, whether a syllable is open or closed generally doesn't matter for the development of Latin phonemes > Spanish phonemes, but here you say it was part of the conditioning, and I'm not sure why. (Maybe you have the allophones on Wikipedia in mind, but those modern [e ~ ɛ] are so far removed from Latin etyma, just as you have also said before, that they can't count in the conditioning, so I think I'm failing to understand your meaning here.)
It's hard to find good stuff on Sardinian. I've found a couple of things in Sardinian about Sardinian which is *incredibly* useful as a monolingual Sardinian speaker trying to learn Sardinian /s. A handful of short Italian works on LCS (which is not a spoken language, with documentation about as comprehensive as the average stub page on Cebuano Wikipedia) but I can't read Italian,and LCS is kinda not important to understanding Logudorese, Nugorese or Campidanese (the actual spoken dialects).

Now I guess the problem when we're talking about models is that I'm perceiving it as a system rather than a set of changes. Take for example Spanish fin from fīnem. In Latin, that would be [fi:nem] (Latin vowels were lengthened behind m). By the Late Latin of Iberia, that would be ['finẽ]. With stress on the first syllable, the initial vowel would be lengthened, and then when the final [ẽ] is lost, so we would expect [fi:n] (ie, fin with an open allophone, if the system was retained). But that would only hold if you imagined the process happening once and done. If you think of the process as a continuous system that can operate over a period of time, one that would include both the time of [fine] and [fin], making ['fine]>['fi:ne] -> ['fi:n] -> ['fin], with the closed-syllable open-allophone that we see [fi̞n]. It might be better to think of my belief as "we see three very very similar vowel systems in the West Mediterranean among Romance languages; maybe it's a common inheritance from Late Latin that's just been edited in different ways?"

Now, I'm tired, and I spent a day procrastinating on conlanging (and studying but who cares about that?) to write this. [:P] ah c'est la vie.

Salmoneus wrote:
28 Apr 2020 14:17
Maybe we should set things out in black and white, for clarity?

Here's how I think Proto-Romance worked (considering only short *e, and ignoring cases where there is a following /j/, /l/ or a nasal, which often worked out differently):

Code: Select all

Latin	||	Early Proto-Romance
	||	Stressed Closed	|	Stressed Open	|	Unstressed
e	||		ɛ		|		ɛ:		|		ɛ (merged with /e/)

At this point, length is allophonic, as it is predictable from context. The loss of some unstressed vowels immediately brought about the breaking of this system:

Code: Select all

Latin	||	Early Proto-Romance	||	Late Proto-Romance
	||					||	Stressed Closed	|	Stressed Open	|	Unstressed
   e	||			ɛ		||		ɛ		|		N/A		|		ɛ
   e	||			ɛ:		||		ɛ:		|		ɛ:		|	      N/A	
		
Length is now distinguished only in stressed closed syllables. We can therefore think of this as there being effectively two types of stress: lengthening stress and non-lengthening stress (particularly since some of these stressed vowels would still be in transparent derivational and inflectional relationships with the unstressed vowels, and some would be in alternation between originally 'open' and 'closed' forms). This was a bit of a mess, particularly when you consider that there's also e and e: in the system, and that all four vowels reduce to one unstressed vowel...

So, Iberia and Romania (and apparently Wallonia?) as peripheral areas perhaps less comfortable with the prevailing system and more at liberty to change (having less contact with other Romance speakers and more contact with speakers of other languages), both independently regularised this system, eliminating the phonemic distinction by analogising the rule that all (as opposed to only most) stressed vowels are long:

Code: Select all

Latin	||	Early Proto-Romance	||	"Peripheral"
	||					||	Stressed Closed	|	Stressed Open	|	Unstressed
   e	||			ɛ		||		ɛ:		|		N/A		|		ɛ
   e	||			ɛ:		||		ɛ:		|		ɛ:		|	      N/A	
		
In some places, there is then diphthongisation of many long vowels. This may have originated in metaphonic diphthongisation of long vowels depending on the quality of the following vowel, which then spread to all long vowels. This diphthongisation occured in central Italy, in a band up to French, apparently - although in gallo-italic dialects it's often obscured by later monophthongisation - and then back down into northwestern iberia. It also happened in some romansh dialects and in eastern romance, but apparently it's complicated, and much of it can be explained as later changes.

Then many languages merged some of their short vowels, and then most languages lost phonemic length altogether (it remains in Friulian).
I'll agree with the general framework of this, but I wonder how it would approach diphthongizing metaphonic changes like that in Southern Italy, which don't appear to be related to open/closed syllables (I presume generalizing the length a la Spanish?) f. pɛre -> pjeri; lɛddʒe -> ljeddʒi
Spoiler:
My minicity is Zyphrazia and Novland
What is made of man will crumble away.

User avatar
Zekoslav
sinic
sinic
Posts: 340
Joined: 07 Oct 2017 16:54

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Zekoslav »

Concerning the ideas about Proto-Romance vowel length, I find the speculation about what would happen to allophonically lengthened stressed vowels once loss of unstressed vowels enters the play quite bold! Yes, logically, length distinction would become phonemic in closed syllables, but unless the loss of unstressed vowels was very extensive (as in Gallo-Romance, for example), long vowels in closed syllables would be much rarer than short ones and minimal pairs even rarer. I can see that as a part of the argument: the length distinction was marginal, a puzzle to language learners, and was quickly done away with. You say peripheral Romance just made all stressed vowels long, but what did central Romance do? Shorten long vowels in closed syllables so that the rule open syllable > long vowel - closed syllable > short vowel was respected once again, which is what I always assumed that it happened? Do nothing and keep the marginally phonemic long vowels? Is there even a way to tell? French had definitely shortened them by the time [eː], [oː] [aː] became [ei̯], [ou̯] and [e]. On the other hand, the situation with [ɛː] and [ɔː] is quite chaotic: in former proparoxytones, we sometimes see diphthongization, and sometimes not; sometimes the syncope happens after lenition, and sometimes before it (effectively preventing it from happening). I don't know what that can tell us.
Languages:
:hrv: [:D], :bih: :srb: [;)], :eng: [:D], :fra: [:|], :lat: [:(], :deu: [:'(]

A linguistics enthusiast who occasionally frequents the CBB.

- Guide to Slavic accentuation

Porphyrogenitos
sinic
sinic
Posts: 254
Joined: 21 Jul 2012 08:01
Location: Buffalo, NY

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Porphyrogenitos »

For more Romance vowel fun - according to Loporcaro in The Dialects of Italy (Maiden and Parry), various dialects in Puglia undergo diverse vowel changes, including various process of diphthongization in [stressed [final] and [open penultimate]] syllables. Very broadly, various dialects are at various points along the following chain shifts (starting from the ordinary Western seven-vowel system); some shifts in some dialects do have more specific conditioning environments:

*a → æ → ɛ or various diphthongizations

The following have an implicational pattern, with 2 not generally happening without 1 present, etc

1.
*e → ɛi̯ → ai̯ → ɔi̯
*o → ɔu̯ → au̯

2.
*i → ɪi̯ → ɛi̯ →əi̯ / øi̯ → oi̯
*u → ʊu̯ → ɔu̯ → əu̯ → ɛu̯ / iu̯

3.
*ɛ → ɛi̯ / ei̯
*ɔ → ɔu̯ / ou̯

User avatar
Dormouse559
moderator
moderator
Posts: 2644
Joined: 10 Nov 2012 20:52
Location: California

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Dormouse559 »

Porphyrogenitos wrote:
12 May 2020 03:42
2.
*i → ɪi̯ → ɛi̯ →əi̯ / øi̯ → oi̯
*u → ʊu̯ → ɔu̯ → əu̯ → ɛu̯ / iu̯
This is interesting. My focus is on a different area, but some dialects of Arpitan have somehow changed /i/ into /ə~ø/, and I've been at a loss to explain it.

Porphyrogenitos
sinic
sinic
Posts: 254
Joined: 21 Jul 2012 08:01
Location: Buffalo, NY

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Porphyrogenitos »

Dormouse559 wrote:
14 May 2020 18:25
This is interesting. My focus is on a different area, but some dialects of Arpitan have somehow changed /i/ into /ə~ø/, and I've been at a loss to explain it.
I know that Piedmontese and possibly some other Gallo-Italic varieties have Western Romance /i e/ > /ə/ in closed syllables except before nasals, e.g. sëcca 'dry', pëss 'fish', and chërdo 'moment, interval' (semantic shift from Latin credo 'creed'). Perhaps that could be related in some way?

Salmoneus
MVP
MVP
Posts: 1917
Joined: 19 Sep 2011 19:37

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Salmoneus »

Doesn't that happen in some Rhaeto-Romance, too? I don't know, I can't remember anymore...

Ælfwine
greek
greek
Posts: 872
Joined: 21 Sep 2015 01:28
Location: New Jersey

Re: The Great Vowel Shift & Vocalism in Romance languages

Post by Ælfwine »

Also, doesn't Romanian centralize vowels in certain environments?
My Blog
Current Projects:
Crimean Gothic — A Gothic language spoken in Crimea (duh)
Pelsodian — A Romance language spoken around Lake Balaton
Jezik Panoski — A Slavic language spoken in the same area
An unnamed Semitic language spoken in the Caucus.

Post Reply