Overcorrecting

A forum for all topics related to constructed languages
Post Reply
User avatar
LinguistCat
sinic
sinic
Posts: 246
Joined: 06 May 2017 07:48

Overcorrecting

Post by LinguistCat »

So, lets say a speech community differentiates certain vowels in some positions and neutralizes them in others. Some of these positions are important to grammar. Could a group of L2 learners that eventually have children that speak the language end up over-correcting so the vowels remain unneutralized in these positions that are grammatically relevant?

For example, though the actual vowels involved would be different, if /i/ and /ɪ/ neutralize after alveolar consonants, but the L2 speakers come from a community that has /i/ and /ɪ/ separate in all situations, and verbs have forms that are differentiated by having ~Ci in one form and ~Cɪ in the other (except where C is alveolar), would it be likely the L2 community would use /i/ in the first form and /ɪ/ in the second even after alveolar consonants? Even if not likely or not ALL L2 speakers, would it be possible that SOME would have this? Would they be more likely to neutralize ALL /i/ vs /ɪ/ distinctions?
User avatar
Pabappa
greek
greek
Posts: 483
Joined: 18 Nov 2017 02:41
Contact:

Re: Overcorrecting

Post by Pabappa »

Russian, at least, keeps its unstressed vowels distinct in some grammatical inflections, despite merging them everywhere else. see https://en.wikipedia.org/wiki/Russian_p ... el_mergers and https://en.wikipedia.org/wiki/Vowel_red ... in_Russian ... it's the most ocmplex vowel reduction system I'm ever seen, so full of exceptions and overlapping rules that keep trying to overrule each other, but it's just what you're asking for, at least at the surface level .... don't know about how it got that way.
I'll take the theses, and you can have the thoses.
Salmoneus
MVP
MVP
Posts: 2318
Joined: 19 Sep 2011 19:37

Re: Overcorrecting

Post by Salmoneus »

LinguistCat wrote: 26 Apr 2021 23:04 So, lets say a speech community differentiates certain vowels in some positions and neutralizes them in others. Some of these positions are important to grammar. Could a group of L2 learners that eventually have children that speak the language end up over-correcting so the vowels remain unneutralized in these positions that are grammatically relevant?

For example, though the actual vowels involved would be different, if /i/ and /ɪ/ neutralize after alveolar consonants, but the L2 speakers come from a community that has /i/ and /ɪ/ separate in all situations, and verbs have forms that are differentiated by having ~Ci in one form and ~Cɪ in the other (except where C is alveolar), would it be likely the L2 community would use /i/ in the first form and /ɪ/ in the second even after alveolar consonants? Even if not likely or not ALL L2 speakers, would it be possible that SOME would have this? Would they be more likely to neutralize ALL /i/ vs /ɪ/ distinctions?
What language are these L2 speakers learning?

Let's say you have three languages: A, B, and Z. A distinguished /e/ and /i/, but its daughter, B, does not distinguish them after alveolars. Z, an unrelated language, has both /e/ and /i/ after alveolars.

If speakers of Z learn A, obviously they will have the distinction.

If speakers of Z learn B, they will not have the distinction, because B does not have the distinction to learn.

If speakers of Z learn A, WHILE A is turning into B, then it's feasible that they might keep the more conservative distinction in their dialect, thanks to influence from their mother language. Indeed, L2 speakers are often more linguistically conservative anyway. However, for this to be meaningful the timing and population sizes would have to be exactly right. Because obviously over time the speakers of Z who have learnt A will simply become speakers of A, just like anyone else. If the population of Z-speakers is large enough, they could in theory form a self-sustaining dialect.

But really, we're just stating the obvious here: any subgroup of speakers could form a distinct dialect that would not have to undergo the changes undergone in other dialects - that's how we get divergence into language families.
User avatar
LinguistCat
sinic
sinic
Posts: 246
Joined: 06 May 2017 07:48

Re: Overcorrecting

Post by LinguistCat »

So basically Sal, you assume that if speakers of Z learned B, they would not overcorrect and incorrectly add back the distinction in verb forms or other places where the distinction could be "assumed" from a pattern. I'm just double checking if that's what you meant.
User avatar
cedh
MVP
MVP
Posts: 380
Joined: 07 Sep 2011 22:25
Location: Tübingen, Germany
Contact:

Re: Overcorrecting

Post by cedh »

I agree with Sal that the default situation would be "If speakers of Z learn B, they will not have the distinction, because B does not have the distinction to learn."

However, I think it's plausible that at least some of these native speakers of Z might indeed preserve the distinction even if they only learn the modern form of B, if some or all of the following factors are in play:
  • they mostly learn B in a non-immersive setting (e.g. in a classroom, or through occasional contact), so that they explicitly think about abstract grammatical rules at least sometimes
  • the morphological analogy pattern is very obvious
  • the orthography of B still distinguishes the two sounds in the relevant environment, so that the analogy gets reinforced by spelling pronunciation
  • (there exists a dialect or close sister language to B which retains the distinction, so that native speakers of B are used to hearing and understanding the conservative pronunciation, even if they don't use it themselves)
Of course, this doesn't mean that speakers of Z would necessarily overcorrect there, only that they could indeed do so with a certain level of plausibility.
Salmoneus
MVP
MVP
Posts: 2318
Joined: 19 Sep 2011 19:37

Re: Overcorrecting

Post by Salmoneus »

LinguistCat wrote: 27 Apr 2021 05:43 So basically Sal, you assume that if speakers of Z learned B, they would not overcorrect and incorrectly add back the distinction in verb forms or other places where the distinction could be "assumed" from a pattern. I'm just double checking if that's what you meant.
Well, if speakers of Z learn B, they can't "overcorrect", because they have no reason to think there "should" be a distinction.

To give a concrete example: the English distinction between /u/ and /ju/ is neutralised after certain consonants, such as /l/, in almost all modern dialects (we'll ignore the ones that don't do that for now). So in modern English - i.e. in "B", "loot" and "lute" are homophonous. If a French speaker learns English, will they overcorrect and add a /j/ into "lute"? Probably not, because English just doesn't have a /j/ in that word, and there's no obvious reason why it should.So the French speaker won't have the idea that /ljut/ is 'more correct' than /lut/.

That said, if a French speaker learns English primarily in written form, and realises that ute usually indicates /ju/, it's conceivable that they could add /j/ as a spelling pronunciation, of course. But this relies on them not hearing the word much, and the spelling being transparent and (otherwise) regular to the learner. If the language is not written, or is written irregularly, or in a way that closely matches actual (rather than historical) pronunciation, or if the speaker simply doesn't understand the spelling rules, or if the speaker actually learns the language from contact rather than in a written, scholarly setting, then they would have no cues to lead them to re-insert the /j/.


Basically I agree with cedh.
User avatar
LinguistCat
sinic
sinic
Posts: 246
Joined: 06 May 2017 07:48

Re: Overcorrecting

Post by LinguistCat »

Ok thank you. I was thinking something of a combo of the second and third options Cedh mentioned in his post, possibly with some amount of four. So it sounds like there is some possibility for it to happen, though I haven't decided how much direct contact the learning group would have while learning. They might, at least at first, have a strong accent which would get smaller as they interact directly.

Which is fine for my purposes right now and enough of a divide in the speech communities for what I'm planning. I would have had to find a different starting spot for the split otherwise.
Post Reply