Do phonemes exist? (Travis bait)

[Richard W]

In my dialect there are a number of words where historical /stər/ turned into, depending on the speaker, /stʃər/ or /ʃtʃər/, without any schwa elision taking place at all.

That does seem rather like the result of a reversed change @r > r= > @r - Äreo chose his description as schwa reduction well.

[bulbaquil]

But it's not just limited to English. There are analyses of a number of languages that involve extremely complicated synchronic rules to derive surface forms from underlying forms. [...]

The whole motivation for it seems to be to avoid redundancy. But as any computer programmer knows, there's often a trade-off between time and memory usage. Who's to say that the human brain doesn't prefer to store something close to the actual output forms redundantly instead of doing a lot of complicated rule application everytime it decides to speak?

That was exactly my reaction, as a programmer, to much of linguistics when I first studied it. It was reinforced by noticing things like spelling errors by native French speakers, which made it clear that they were missing 'obvious' rules (e.g. "spell all /-e/ infinitives with er"). Because linguists can recover a rule does not mean that the speakers do.

Michael Tomasello is very good on this in syntax; his idea, based on acquisition studies, is that children learn verbs and constructions individually, and only generalize to ideas like 'direct object' far later. Even things like the question transformation are historical, especially as children often hear and master questions first, before statements.

I think there's also the issue that brains aren't computers; optimal efficiency for the one may not be optimal efficiency for the other, and efficiency might not even be the purpose.

[Travis B.]

In my dialect there are a number of words where historical /stər/ turned into, depending on the speaker, /stʃər/ or /ʃtʃər/, without any schwa elision taking place at all.

That does seem rather like the result of a reversed change @r > r= > @r - Äreo chose his description as schwa reduction well.

IMD /ər/ is [ʁ̩ˤ(ː)] except morpheme-initally, where then it is [ə(ː)ʁˤ]. However, IMD /ər/ in many ways does not pattern with onset /r/, in that it does not normally condition the retroflex or postalveolar affrication of /t/ or /d/ before it.

Edit: I added a single word I forgot before.

[zompist]

Particularly, it seems to me that typical purported generative-type phonologies heavily confuse diachronic phenomena for being synchronic phenomena, that is, they presume phenomena to be happening in the process of speech production that actually happened diachronically, with the results of such processes being acquired directly. Likewise, the purported synchronic results of said generative-type phonologies, especially "deep" ones, could be far more plausibly treated in terms of a combination of allomorphy and analogy.

So far so good; that's similar to my own view.

[rotting bones]

I think there's also the issue that brains aren't computers; optimal efficiency for the one may not be optimal efficiency for the other, and efficiency might not even be the purpose.

I agree with the first half, but efficiency is by definition the purpose.

PS. I mean, once you make the terms relative, it can be "efficient" for a system to have redundancies built in.

[Bedelato]

This kind of reminds me of how people seem to tell a lot of stories about small children who speak AmE dialects where /t/ is realized as [tS_h] before /r/ who misspell <tree> as <chree>. I never did this, and I never realized that the [tS_h] wasn't [t] until after taking an interest in linguistics. I suppose you could say that they internalize it as /tSri:/, but I don't think there is any /tSr/ as an onset cluster in English. Or maybe it's that there is simply is no real reason to analyze it as /tri:/ rather than /tSri:/ if you don't know the spelling already.

I don't find it at all unreasonable that some people internalise it as /tʃr/ while others internalise it as /tr/. Similarly with other cases of neutralisation: a friend of mine once said something which seemed to imply he thought of what are normally analysed as /st sp sk/ in English as /sd sb sg/.

I agree with most of what zompist said, but especially:

People without much exposure to other dialects will have a phonemic representation based on their own dialect, not whatever abstraction linguists use for the standard. This should be obvious, but I think we can forget it a little too easily— e.g. Travis comes close to this above when considering what a representation "should be" in NAE.

I always find it a bit bizarre when people seem to want a phonemic (or underlying) representation to apply to all dialects simultaneously.

Yeah. I think of it as /tri/ but still pronounce it [t͡ʃʰɹʷi:].

Really, 100% accuracy is only possible on the idiolect level. Lower resolution means less precision.

They may still occur in words like history - I have ["hIstri] alternating with ["hIstSrI],

Yeah, I have that distinction too, but in such words (like history) it results from schwa reduction.

And I still have three syllables in ['hɪstɹ̩i].

[Travis B.]

Worry not, I have not forgotten about this thread. So hence I am at least going to answer the original post in parts, with one part following:

But if that's the case, I'd also wonder if the phonetic representations Travis uses are also to some extent a matter of trained abstraction. I mean, why stop at the phonetic level? The 'real' representation could be 'everything captured in a spectrogram'. (NB, this isn't Yet Another Complaint about Travis's phonetic representations. But if he can doubt the reality of phonemes, I can at least wonder about his phones.)

Oh, my phonetic representations most certainly are a trained abstraction. The reason why they are is that they attempt to cover a lot of space in both idiolectal and register variation in a fashion that is at least somewhat succinct while simultaneously not diluting out notable dialectal features with General American influence that is certainly present in varying degrees. Likewise, they seek to be detailed with regard to important phonological features while not simply describing my own idiolect (despite what many allege).

Accomplishing all this requires a great deal of abstraction in what one represents. A good example is how I mark laminal versus ordinary (apical) coronals. A lot of people seem to think that this is just an unnecessary detail that could be omitted. However, an important reason why I mark this is that laminal, aspirated /t/ is actually frequently an affricate in many idiolects; I typically have [t̻͡s̻ʰ] therefor much of the time, but this is not necessarily typical, as [t̠̻͡ɕʰ] is also very common and may actually be more common therefor. However, certainly not everyone affricates it, and most people who do do not necessarily do so consistently or always with the same affricate; I, for one, in more formal or careful speech generally avoid using an affricate and will in more informal speech occasionally use [t̠̻͡ɕʰ] rather than [t̻͡s̻ʰ], and am never consistent about whether I use an affricate or not. Marking all this would add quite a great deal of unnecessary complexity to my transcriptions that can be abstracted away simply by marking laminalness, with the implication that some individuals may pronounce laminal coronals differently in ways other than just laminalness versus apicalness alone.

[bulbaquil]

Travis, half the time I don't even mark aspiration in my phonetic transcriptions, unless it's relevant (like, for example, I might say I pronounce "stand" [stʰæ̃nd] to show that I aspirate it even in this position, but...)

[finlay]

Worry not, I have not forgotten about this thread. So hence I am at least going to answer the original post in parts, with one part following:

But if that's the case, I'd also wonder if the phonetic representations Travis uses are also to some extent a matter of trained abstraction. I mean, why stop at the phonetic level? The 'real' representation could be 'everything captured in a spectrogram'. (NB, this isn't Yet Another Complaint about Travis's phonetic representations. But if he can doubt the reality of phonemes, I can at least wonder about his phones.)

Oh, my phonetic representations most certainly are a trained abstraction. The reason why they are is that they attempt to cover a lot of space in both idiolectal and register variation in a fashion that is at least somewhat succinct while simultaneously not diluting out notable dialectal features with General American influence that is certainly present in varying degrees. Likewise, they seek to be detailed with regard to important phonological features while not simply describing my own idiolect (despite what many allege).

Accomplishing all this requires a great deal of abstraction in what one represents. A good example is how I mark laminal versus ordinary (apical) coronals. A lot of people seem to think that this is just an unnecessary detail that could be omitted. However, an important reason why I mark this is that laminal, aspirated /t/ is actually frequently an affricate in many idiolects; I typically have [t̻͡s̻ʰ] therefor much of the time, but this is not necessarily typical, as [t̠̻͡ɕʰ] is also very common and may actually be more common therefor. However, certainly not everyone affricates it, and most people who do do not necessarily do so consistently or always with the same affricate; I, for one, in more formal or careful speech generally avoid using an affricate and will in more informal speech occasionally use [t̠̻͡ɕʰ] rather than [t̻͡s̻ʰ], and am never consistent about whether I use an affricate or not. Marking all this would add quite a great deal of unnecessary complexity to my transcriptions that can be abstracted away simply by marking laminalness, with the implication that some individuals may pronounce laminal coronals differently in ways other than just laminalness versus apicalness alone.

That's an awful lot of waffle you got there. What's the laminal for, anyway? Where does it show up?

[Travis B.]

Worry not, I have not forgotten about this thread. So hence I am at least going to answer the original post in parts, with one part following:

But if that's the case, I'd also wonder if the phonetic representations Travis uses are also to some extent a matter of trained abstraction. I mean, why stop at the phonetic level? The 'real' representation could be 'everything captured in a spectrogram'. (NB, this isn't Yet Another Complaint about Travis's phonetic representations. But if he can doubt the reality of phonemes, I can at least wonder about his phones.)

Oh, my phonetic representations most certainly are a trained abstraction. The reason why they are is that they attempt to cover a lot of space in both idiolectal and register variation in a fashion that is at least somewhat succinct while simultaneously not diluting out notable dialectal features with General American influence that is certainly present in varying degrees. Likewise, they seek to be detailed with regard to important phonological features while not simply describing my own idiolect (despite what many allege).

Accomplishing all this requires a great deal of abstraction in what one represents. A good example is how I mark laminal versus ordinary (apical) coronals. A lot of people seem to think that this is just an unnecessary detail that could be omitted. However, an important reason why I mark this is that laminal, aspirated /t/ is actually frequently an affricate in many idiolects; I typically have [t̻͡s̻ʰ] therefor much of the time, but this is not necessarily typical, as [t̠̻͡ɕʰ] is also very common and may actually be more common therefor. However, certainly not everyone affricates it, and most people who do do not necessarily do so consistently or always with the same affricate; I, for one, in more formal or careful speech generally avoid using an affricate and will in more informal speech occasionally use [t̠̻͡ɕʰ] rather than [t̻͡s̻ʰ], and am never consistent about whether I use an affricate or not. Marking all this would add quite a great deal of unnecessary complexity to my transcriptions that can be abstracted away simply by marking laminalness, with the implication that some individuals may pronounce laminal coronals differently in ways other than just laminalness versus apicalness alone.

That's an awful lot of waffle you got there. What's the laminal for, anyway? Where does it show up?

The laminal (and for postalveolars, alveolopalatal) shows up in two different general kinds of places. The first, and seemingly the most widespread (as I can tell it is often present in General American-like varieties, albeit without the affrication of aspirated /t/) is in coronals before any of /uː/, /ʊ/, /w/, or /ər/. (For instance, I frequently pronounce two and stressed to as [ˈt̻͡s̻ʰʉ̯u] while this one friend of mine, whose idiolect is similar to mine but still somewhat different, typically pronounces the same as [ˈt̠̻͡ɕʰʉ].)

The second, which I can tell is present in my own dialect but I am not sure of its range outside it, is in many consonant clusters involving a sibilant, specifically sibilants before any one of /p/, /t/, /tʃ/, /m/, /n/, /r/, /l/, or /w/ (but not /k/) or after one of /k/, /ɡ/, /r/, or /l/. I should note that the laminal/alveolopalatal assimilates through sequences of adjacent coronals, such that in the case of sibilants before /t/, /tʃ/, /n/, the consonant after the sibilant is also laminal or alveolopalatal. Also note that /st/ clusters tend to assimilate to just [s̻ː] both medially and finally.

[Jipí]

What Travis painstakingly tries to pin down with laminal affricates and a hoard of other diacritics is what other people broadly transcribe as [tʃɹ].

[Travis B.]

What Travis painstakingly tries to pin down with laminal affricates and a hoard of other diacritics is what other people broadly transcribe as [tʃɹ].

Actually, no, the laminal/alveolopalatal affricates are different from that - that is only laminal or alveolopalatal IMD when preceded by /s/ or if the /r/ is followed by one of /uː/ or, hypothetically (in that it applies to words I would make up, but I cannot think of any actual examples), /ʊ/ or /ər/, and then only in idiolects that do not use a (practically inherently apical) retroflex affricate for /t/ or /d/ before /r/ as many more conservative ones IMD do.

[Travis B.]

I should also note that whether laminal aspirated /t/ is affricated in clusters may differ significantly from whether laminal aspirated /t/ is affricated before a vowel. For instance, marked affrication thereof in /tw/ clusters seems to be far less frequent than affrication thereof directly before a vowel; I for instance either do not affricate laminal aspirated /t/ in /tw/ or only slightly affricate it, and this seems to be the case with most people back in southeastern Wisconsin. (However, I have heard people who do reliably turn aspirated /tw/ clusters into [t̠̻͡ɕʰw], which I honestly find very annoying for some reason... maybe it is just that lawyer commercial back in Wisconsin where the person speaking pronounces twenty that way over, and over, and over...)

[TaylorS]

This kind of reminds me of how people seem to tell a lot of stories about small children who speak AmE dialects where /t/ is realized as [tS_h] before /r/ who misspell <tree> as <chree>. I never did this, and I never realized that the [tS_h] wasn't [t] until after taking an interest in linguistics. I suppose you could say that they internalize it as /tSri:/, but I don't think there is any /tSr/ as an onset cluster in English. Or maybe it's that there is simply is no real reason to analyze it as /tri:/ rather than /tSri:/ if you don't know the spelling already.

In American English this seems to be simply a completed conditional sound change, /t/ and /d/ have shifted to /tS/ and /dZ/ before /r/, since the later are already phonemes in English.

[chris_notts]

This kind of reminds me of how people seem to tell a lot of stories about small children who speak AmE dialects where /t/ is realized as [tS_h] before /r/ who misspell <tree> as <chree>. I never did this, and I never realized that the [tS_h] wasn't [t] until after taking an interest in linguistics. I suppose you could say that they internalize it as /tSri:/, but I don't think there is any /tSr/ as an onset cluster in English. Or maybe it's that there is simply is no real reason to analyze it as /tri:/ rather than /tSri:/ if you don't know the spelling already.

In American English this seems to be simply a completed conditional sound change, /t/ and /d/ have shifted to /tS/ and /dZ/ after /r/, since the later are already phonemes in English.

before r, surely?

[TaylorS]

before r, surely?

Fixed!

[Travis B.]

This kind of reminds me of how people seem to tell a lot of stories about small children who speak AmE dialects where /t/ is realized as [tS_h] before /r/ who misspell <tree> as <chree>. I never did this, and I never realized that the [tS_h] wasn't [t] until after taking an interest in linguistics. I suppose you could say that they internalize it as /tSri:/, but I don't think there is any /tSr/ as an onset cluster in English. Or maybe it's that there is simply is no real reason to analyze it as /tri:/ rather than /tSri:/ if you don't know the spelling already.

In American English this seems to be simply a completed conditional sound change, /t/ and /d/ have shifted to /tS/ and /dZ/ before /r/, since the later are already phonemes in English.

In my own dialect, this is not the case. While some younger individuals such as myself do have this completed change, there are still younger people who lack this merger and rather (medially) contrast /tr/ and /dr/ with /tʃr/ and /dʒr/, with the former pair being [t͡ʂɹ͡ɰˤ] (1) and [d͡ʐɹ͡ɰˤ] (2) and the latter pair being [t̠͡ʃɹ̠͡ɰˤ] (3) and [d̠͡ʒɹ̠͡ɰˤ] (4) rather than merging them as the latter pair.

(1) [ts`)r\M\)_?\], (2) [dz`)r\M\)_?\], (3) [t_-S)r\_-M\)_?\], (4) [d_-Z)r\_-M\)_?\]

[Travis B.]

I figured I would cover more of the original post, so I will cover the following:

People without much exposure to other dialects will have a phonemic representation based on their own dialect, not whatever abstraction linguists use for the standard. This should be obvious, but I think we can forget it a little too easily— e.g. Travis comes close to this above when considering what a representation "should be" in NAE.

When I said "should be" here, I meant precisely the opposite: I was speaking of what the surface forms in my own dialect "should be" if one tries to generate them from the conventionally-posited phonemic analysis of General American-like North American English varieties, and these do not necessarily match the actual surface forms. So hence either the analysis, for my own dialect, is incorrect, or the underlying forms themselves are different than those conventionally posited.

But beyond this, I have come to the conclusion that trying to go from the conventionally-posited phonemic analysis of GA-like NAE varieties to the everyday speech I am familiar with is not very plausible on a synchronic level. That is, it requires multiple layers of non-obvious changes which, if evaluated synchronically, should be likely to have other outcomes when one considers idiolectal variation. This is particularly the case with vowel length and nasalization; if these were not set in the underlying forms already, there are many changes that occur that should allow these to vary considerably depending on how one orders rules, in ways that one does not actually see; likewise, the rules that do allow these to come out as they do yet show certain surface forms are in cases very obscure, such that one should see people learn them differently or not learn them at all, but they do not.

My favorite examples for this are the words width and breadth in my own dialect, synchronically [wɪːθ] and [brɜːθ], contrasting with with [wɪθ] and breath [brɜθ]. In a conventional analysis, these should then be /wɪdθ/ and /brɛdθ/, with with and breath being /wɪθ/ and /brɛθ/, with allophonic vowel length being applied and then the rule coda dθ > θ being applied.

The problem is that this rule only applies to these two words, as far as I know. I for one doubt the plausibility of a phonological system where individual words or very small sets of words such as these have rules all to themselves, and were such rules the case, I would expect them to be quickly simplified away. Hence I would expect there to be a likelihood to lose the rule dθ > θ applied after allophonic vowel length, and hence have width and breadth be merged with with and breath as [wɪθ] and [brɜθ], analyzed as /wɪθ/ and /brɛθ/, or to be reanalyzed as /wɪð/ and /brɛð/, assuming allophonic vowel length is still active for these words.

However, all this would not be a problem if width and breadth were learned by children as indeed being /wɪːθ/ and /brɛːθ/ (and with and breath being /wɪ̆θ/ and /brɛ̆θ/, already being assigned underlyingly short vowels). There would then be no need for a still-synchronically-active essentially one-off rule coda dθ > θ, as this essentially would be merely a limited-scope diachronic change which said children would be completely unaware of, at least until they started learning to read and saw them spelled as width and breadth. (This honestly seems to actually be the case for me; as I kid I always wondered what those "d"s were doing there in those two words, and they always sounded quite weird to me actually pronounced out, sounding like very obvious spelling pronunciations to me.)

MInd you this is just one example being used to illustrate a point that it seems like, in practice, things of this sort make far more sense if one posits much "shallower" underlying forms that those used in traditionally analyses, and hence moves much of what is traditionally posited as being part of synchronic phonological processes into the underlying forms themselves (which in turn would little resemble traditionally-analyzed underlying forms), with the supposed effects of these processes at morpheme boundaries in turn being treated as a consequence of allomorphy. Note that I still am not saying that said underlying forms would be equivalent to phonetic realization as, yes, there are clear signs that individuals do have at least some limited level of phonemic awareness and that allophony does apply to it.

[Tropylium⁺]

What's the laminal for, anyway? Where does it show up?

The laminal (and for postalveolars, alveolopalatal) shows up in two different general kinds of places. The first, and seemingly the most widespread (as I can tell it is often present in General American-like varieties, albeit without the affrication of aspirated /t/) is in coronals before any of /uː/, /ʊ/, /w/, or /ər/.

…But not before /j/, /iː/, /ɪ/ etc?

[Travis B.]

What's the laminal for, anyway? Where does it show up?

The laminal (and for postalveolars, alveolopalatal) shows up in two different general kinds of places. The first, and seemingly the most widespread (as I can tell it is often present in General American-like varieties, albeit without the affrication of aspirated /t/) is in coronals before any of /uː/, /ʊ/, /w/, or /ər/.

…But not before /j/, /iː/, /ɪ/ etc?

Nope except that /t/ is often affricated to plain old palatoalveolar [t͡ʃ] before /j/, as is typical in much of North American English.

[Bedelato]

I figured I would cover more of the original post, so I will cover the following:

MInd you this is just one example being used to illustrate a point that it seems like, in practice, things of this sort make far more sense if one posits much "shallower" underlying forms that those used in traditionally analyses, and hence moves much of what is traditionally posited as being part of synchronic phonological processes into the underlying forms themselves (which in turn would little resemble traditionally-analyzed underlying forms), with the supposed effects of these processes at morpheme boundaries in turn being treated as a consequence of allomorphy. Note that I still am not saying that said underlying forms would be equivalent to phonetic realization as, yes, there are clear signs that individuals do have at least some limited level of phonemic awareness and that allophony does apply to it.

I kinda get what you're trying to say here, and actually I kinda agree with it too.
The minimal pairs test (and other methods of identifying phonemic contrasts) works for identifying reasonable underlying forms.

But then people try to go further than that and posit that e.g. English "has" is underlyingly "have"+"-s" even though speakers are more likely to memorize the two forms seperately. Sometimes they go too far and the supposed "underlying" forms become little more than pointless abstractions.

When you try to get more abstract than phonemes the whole thing starts to break down.

So there's my paraphrase of it. How close am I?

[Travis B.]

I figured I would cover more of the original post, so I will cover the following:

MInd you this is just one example being used to illustrate a point that it seems like, in practice, things of this sort make far more sense if one posits much "shallower" underlying forms that those used in traditionally analyses, and hence moves much of what is traditionally posited as being part of synchronic phonological processes into the underlying forms themselves (which in turn would little resemble traditionally-analyzed underlying forms), with the supposed effects of these processes at morpheme boundaries in turn being treated as a consequence of allomorphy. Note that I still am not saying that said underlying forms would be equivalent to phonetic realization as, yes, there are clear signs that individuals do have at least some limited level of phonemic awareness and that allophony does apply to it.

I kinda get what you're trying to say here, and actually I kinda agree with it too.
The minimal pairs test (and other methods of identifying phonemic contrasts) works for identifying reasonable underlying forms.

But then people try to go further than that and posit that e.g. English "has" is underlyingly "have"+"-s" even though speakers are more likely to memorize the two forms seperately. Sometimes they go too far and the supposed "underlying" forms become little more than pointless abstractions.

When you try to get more abstract than phonemes the whole thing starts to break down.

So there's my paraphrase of it. How close am I?

The logic you get out of it is largely what I mean, but you are applying it to conventional phonemes as opposed to "deep" phonemic models, whereas I was speaking of some kind of "shallow" phonemic model as opposed to even conventional phonemes; i.e. that while people likely still have some sort of "underlying forms" that they process phonology in terms of, it likely does not actually look like conventional phonemes in all cases, and that many of the cases where deeper phenomena associated with conventional phonemes are seemingly predictive in a productive fashion can be explained away as allomorphy and analogy at work.

[Bedelato]

So what sort of processes would you incorporate into your "shallow" forms and what ones would you keep as rules?

[Bedelato]

Okay, I just thought of something else.

Some people have suggested that phonemes are really a projection of alphabetic writing, and there may be something to this. I think it's striking how seemingly poorly a phonemic representation works for Mandarin, for instance: there's little value in assigning phonemes to each vowel. What works far better is the division the Chinese have been using for centuries: initial consonant vs. final. You virtually have to learn each final (e.g. -e, -uo, -ian, -üe) as a unit.

One phoneme doesn't necessarily equal one segment. Think of things like N-phthongs and affricates, which are composed of multiple segments but treated as a unit.

It's not too hard to extend this idea to more complex situations; for example English /ju/ could be considered a vowel phoneme (it's actually a diphthong with an onglide instead of an offglide), similar to how /tʃ/ is not quite the same as /t/+/ʃ/.

[zompist]

One phoneme doesn't necessarily equal one segment. Think of things like N-phthongs and affricates, which are composed of multiple segments but treated as a unit.

Yes, they are sometimes. Now is that a unit of language, or a unit of analysis? How do you know?