Are allophones in big phonological inventories less likely?

[Turama]

I read (or heard) somewhere, that languages with great consonantal phonological inventories are less likely to have consonantal allophones, because small variations of a sound can have an different meaning. I think it was a video about the ubykh language. But maybe I misremember. Nevertheless, are languages with big consonantal phonological inverntories less likely (or slower) to develop allophones, because of the rist of homophones?

For me it is kind of important, because I am making a conlang, which has a big consonantal phonological inventory and many consonantal clusters (right now it has 68 consonants and 15 allophones). And then I want to develop daughter languages, but I do not know whether the merging of consonants is natural, because many homophones could emerge. I know languages like Japanese and Chinese have no great problems with homophones, because of context, but I rather ask anyway. (I don't know whether it is true or not, but I also read that in Japanese, the most homophones come from Chinese, so these homophones do not mainly come from merging consonants).

So what do you mean?

[Pole, the]

So what do you mean?

I don't mean anything, but I think it's somehow true.

[احمکي ارش-ھجن]

I read (or heard) somewhere, that languages with great consonantal phonological inventories are less likely to have consonantal allophones, because small variations of a sound can have an different meaning. I think it was a video about the ubykh language. But maybe I misremember. Nevertheless, are languages with big consonantal phonological inverntories less likely (or slower) to develop allophones, because of the rist of homophones?

For me it is kind of important, because I am making a conlang, which has a big consonantal phonological inventory and many consonantal clusters (right now it has 68 consonants and 15 allophones). And then I want to develop daughter languages, but I do not know whether the merging of consonants is natural, because many homophones could emerge. I know languages like Japanese and Chinese have no great problems with homophones, because of context, but I rather ask anyway. (I don't know whether it is true or not, but I also read that in Japanese, the most homophones come from Chinese, so these homophones do not mainly come from merging consonants).

So what do you mean?

To be honest, I thought large inventories would make allophony become even more likely.

[2+3 clusivity]

At a casual glance, allophone heavy languages appear to fall in an inverse bell-curve in relation to consonant phoneme inventory size:

Compare: on the small end: http://en.wikipedia.org/wiki/Rotokas, vs. on the extremely large end http://en.wikipedia.org/wiki/Voiceless_ ... _fricative (with regard to Adyghe). You know inventory size isn't an obstacle when languages can start throwing in virtually pathological allophones.

I would also expect that in a large inventory, any constraint on consonant allophones patterns with an increase in vowel allophones.

[Nortaneous]

What's the inventory? There's no general rule. Mergers are pretty common in NWC langs; the dialects that get studied just seem to be the ones with the fewest of them. Look up Karacalar Ubykh.

Also, it's totally possible to have allophony or consonant shifts. Some NWC dialects shifted pʼ to fʼ; many developed a long low vowel from ɜ + pharyngeal sequences (note that /ɜ/ and the long low vowel are usually written /a a:/, but the short vowel is higher than the long one); Abkhaz had ʕʷ > ɥ. Also there's loads of homophony in NWC langs because most of the roots are pretty much just a consonant.

You're probably more interested in NEC than NWC; NEC is where the large vowel inventories are. Chechen has >40 consonants and 20 vowels + nasalization.

Mandarin doesn't have a problem with homophones because it's mostly disyllabic now. (cf. ink pɪn vs. stick pɪn)

[Miekko]

You are aware that you'll never perfectly recreate your mouth's setup when uttering the same phoneme twice? That falls under allophony, even though it may be so fine-graded that it's barely measurable.

[Curlyjimsam]

You are aware that you'll never perfectly recreate your mouth's setup when uttering the same phoneme twice? That falls under allophony, even though it may be so fine-graded that it's barely measurable.

This is the most important thing to state in response to this thread, I think. Though in fairness to the OP - this sort of error is an extremely common one amongst conlangers.

But one would presume that in general a language with more phonemes is going to have less variation in the range of articulations permitted for any one phoneme. A language with a stop inventory /p k/ will likely have a much greater range of values for /k/ (including, say, [t d k kʰ g k͡x x]) than one with /p pʰ b t tʰ d k kʰ g q qʰ/.

[Turama]

You are aware that you'll never perfectly recreate your mouth's setup when uttering the same phoneme twice? That falls under allophony, even though it may be so fine-graded that it's barely measurable.

This is the most important thing to state in response to this thread, I think. Though in fairness to the OP - this sort of error is an extremely common one amongst conlangers.

But one would presume that in general a language with more phonemes is going to have less variation in the range of articulations permitted for any one phoneme. A language with a stop inventory /p k/ will likely have a much greater range of values for /k/ (including, say, [t d k kʰ g k͡x x]) than one with /p pʰ b t tʰ d k kʰ g q qʰ/.

So do you mean a lang with a quite big inventory (for example my stop inventory consist of 16 stops) kind of has lesser sound changes in form of merging of phonemes and/or slower sound change development?

And do speakers bother if the allophones (or sound changes) create many homophones? Of course a language like japanese has no big problems with homophones, but (if I read right) the most homophones came from the chinese language and not from allophony and/or merging. Or do speaker rather avoid a large homophony if they can?

You are aware that you'll never perfectly recreate your mouth's setup when uttering the same phoneme twice? That falls under allophony, even though it may be so fine-graded that it's barely measurable.

Yeah, I am kind of aware of it, even if I never really thought about that. But I mean allophones, which sound different to us. It's like the pitch, even if a small difference is measurable, they are nevertheless the same tone, because we can't distinguish/hear this small difference. In the case of phonemes of course what is undifferent to one, can be different to another, but I think there is a certain limit to it.

What's the inventory? There's no general rule. Mergers are pretty common in NWC langs; the dialects that get studied just seem to be the ones with the fewest of them. Look up Karacalar Ubykh.

Also, it's totally possible to have allophony or consonant shifts. Some NWC dialects shifted pʼ to fʼ; many developed a long low vowel from ɜ + pharyngeal sequences (note that /ɜ/ and the long low vowel are usually written /a a:/, but the short vowel is higher than the long one); Abkhaz had ʕʷ > ɥ. Also there's loads of homophony in NWC langs because most of the roots are pretty much just a consonant.

You're probably more interested in NEC than NWC; NEC is where the large vowel inventories are. Chechen has >40 consonants and 20 vowels + nasalization.

Mandarin doesn't have a problem with homophones because it's mostly disyllabic now. (cf. ink pɪn vs. stick pɪn)

Yet, my inventory is quite regular. It has a plain, breathy, labialized and ejective version of almost every consonant. These are: p t k q ts tʃ tθ kx qχ(uvular) s ʃ θ x χ(uvular) r and l. But there is only the plain n m j w. As I do not want to write down every phoneme, I just write down the "series". The breathy voice makes the consonant voiced.

Actually, I am interested in NWC, because there are my roots and because I learn Kabardian right now. I want kind of create a conlang which is more or less related to or has features of NWC languages.

[...]You know inventory size isn't an obstacle when languages can start throwing in virtually pathological allophones.[...]

Yeah, but then the problem would be that I wouldn't be able to pronounce all phonemes ^^. And maybe I didn't express me good enough, but I mean allophones which merge with phonemes.

[finlay]

You are aware that you'll never perfectly recreate your mouth's setup when uttering the same phoneme twice? That falls under allophony, even though it may be so fine-graded that it's barely measurable.

This is the most important thing to state in response to this thread, I think. Though in fairness to the OP - this sort of error is an extremely common one amongst conlangers.

But one would presume that in general a language with more phonemes is going to have less variation in the range of articulations permitted for any one phoneme. A language with a stop inventory /p k/ will likely have a much greater range of values for /k/ (including, say, [t d k kʰ g k͡x x]) than one with /p pʰ b t tʰ d k kʰ g q qʰ/.

So do you mean a lang with a quite big inventory (for example my stop inventory consist of 16 stops) kind of has lesser sound changes in form of merging of phonemes and/or slower sound change development?

And do speakers bother if the allophones (or sound changes) create many homophones? Of course a language like japanese has no big problems with homophones, but (if I read right) the most homophones came from the chinese language and not from allophony and/or merging. Or do speaker rather avoid a large homophony if they can?

Allophones and homophones are different things, caused by different things. Homophones tend to be caused by mergers, which are tangentially related to allophones – in the case of Japanese, there were many Chinese loanwords which were merged together because some sounds were changed and tones went missing. One obvious one that they have actively changed recently due to a bit of a conspicuous merger was "science" (科学) and "chemistry" (化学), both pronounced kagaku – however chemistry is often pronounced with the alternative reading bakegaku to be clear. An advantage of kanji when reading is that homophones like these aren't a problem, but it means that in some cases, written Japanese can be more precise than the spoken language.

An allophone is a form of sound change, but one that explicitly doesn't make a merger and doesn't change the phonemic status of a word, or something like that.