How does r-coloring impact phonological analysis?

Question

Edit: I realized I asked this very confusingly. I think what I really should have said was, are there any phonemic implications to r-coloring? Or thinking about it slightly differently, is there a phonological counterpart to r-coloring, or is it purely a phonetic feature?

I suspect it's the latter, I just wanted to see if I was missing something about, say, how the terminology is applied.

Original post: I'm curious as to when a purely phonological examination might be concerned with highlighting r-coloring as a phenomenon in itself in opposition to a vowel + rhotic consonant sequence. (I'm interested particularly with English analyses.)

Insofar as English goes, I haven't yet seen an instance where one might care that phonemically-speaking an /ə/ followed by a rhotic consonant in the same syllable has its pronunciation impacted. (Though I'm certainly no expert!)

Other questions on SE either address this from a purely phonetic angle, e.g., this question, while others blur phonetics and phonemics. Answers given for the questions here and here imply it wouldn't be of any real concern phonemically (at least in the phonemics of American English) without explicitly saying so.

So, are there times in English when a phonemicist might need to draw attention to when an [r] (etc.) affects the quality of a vowel? Or is this always a purely phonetic distinction?

Answer 1

I'm not sure I entirely understand the question, but there are a lot of reasons why in a phonological analysis of English, r-colored vowels might be treated as something other than a vowel + rhotic consonant sequence.

In many varieties of American English, when we try to use that kind of analysis, we end up with an unusual set of allowed vowel + /r/ sequences. For example:

• the NURSE vowel, often transcribed [ɚ]: if we phonemicize it as /ər/, then it's generally going to be the only context where /ə/ shows up in stressed syllables (a few speakers say they have stressed /ə/ in other words, such as just or because; see Peter Shor's post here on English Language and Usage SE]: /ə/ in a stressed syllable?).

  The NURSE vowel could instead be transcribed /ʌr/, but that doesn't actually accord with my own internal sense of what the identity of the vowel is. If anything, to me NURSE sounds closer to /ʊr/ than /ʌr/. (It may be relevant that, like many NAE speakers, I merge PURE (historically /jʊə̯r/) into NURSE.)

• There are a large number of vowel mergers that are frequent in this context. Some can perhaps be generalized as "tense-lax neutralization". The merry-marry-Mary merger may be the most well known, where /ær/ = /ɛr/ = /eə̯r/: another widespread merger is /ɪr/ = /ɪə̯r/, as in Sirius = serius. Rounded vowels also show a number of possible mergers. In certain cot-caught merged accents, analyzing the NORTH/FORCE vowel as a sequence requires either treating it as /or/ with "tense" /o/ (the same phoneme as the GOAT vowel), or as /ɔr/ with a vowel /ɔ/ which does not exist as an independent phoneme in any other context.

• As a result, in some North American accents of English the inventory of vowel + r sequences arguably looks similar in phonological structure to diphthongs such as /aɪ/ /oɪ/ /aʊ/ or /eɪ/ /oʊ/: in my own accent, I could analyze r-colored vowels as a monophthong NURSE /ɚ/ plus four ɚ-final diphthongs START /ɑɚ/, SQUARE /eɚ/, NORTH/FORCE /oɚ/, NEAR /iɚ/.

Answer 2

The question has a contradictory presupposition:

one might care that phonemically-speaking an /ə/ followed by a rhotic consonant in the same syllable has its pronunciation impacted

The contradiction arises from limiting this to "phonemically-speaking", but then asks about having "pronunciation impacted". Pronunciation is phonetics, phonemes are phonology. Phonemes are eventually pronounced, but that doesn't happen in the phonology. So you have to re-conceptualize the question. Let me also point out that "r-coloring" is not phonological terminology, is is a crypto-phonetic description. In phonology, we have vowels and a coda consonant /r/ (or /ɹ/ or however you want to describe it), and then we have rules of pronunciation. When you are talking about coda r phenomena, you can't talk about "English", you have to talk about a specific type of English – more fine-grained that "US English" and "UK English".

There are some purely phonological issue that relate to /r/, but they are more broadly applicable than just "after /ə/" (/a, ɔ, ɛ/ etc can also appear before coda /r/). The only thing that can be said about /əɹ/ and only /əɹ/ is that in most rhotic US dialects, it is pronounced [ɹ̩] i.e. as a syllabic rhotic without a separate [ə] transition. But that is about phonetics, i.e. how a phonemic sequence is physically pronounced. There is no phonology to that.

In phonology, r-coloring is "having /r/ in the coda", or similar prosodic position (there is controversy over the syllable position of ˈVrV). The set of vowels that can appear in that position is reduced. In my (PNW American) dialect, you get [ɪ ɛ ɔ a] and [ʊ] in sociolinguistically complex circumstances (when talking to people who speak a different dialect, and you have to say "moor", "poor", "tour"). The fact of getting [ɔ] before [ɹ] is of some interest, because in that dialect, the vowel [ɔ] only exists in pre-glide position (core, coy, coat), hence "caught" and "cot" are homophones.

Additionally, you should treat syllabic [ɹ̩] as /ər/, e.g. [bʌɾɹ̩] "butter", [bɹ̩] "burr". That treatment explains alternations like [ˈtɪɹəni ~ tɹ̩ˈɹænɪkl̩] "tyranny~tyrannical", that is, vowels reduction before /r/ yields syllabic r.

/r/ is treated more like a glide w.r.t. syllable-related consonant allophony. For example, /t/ is glottalized or even becomes [ʔ] in the coda after [j w ɹ], as in bit, beat, pout, put, putt, bite and bart but not belt (and in the case of "can't" it depends on whether you nasalize the vowel and get rid of /n/), and not after any obstruent such as "bust, pushed". This does not distinguish /r/ from the glides /j,w/.

The upshot of this is that the rhotic behaves phonologically like a glide. The one thing that is just about coda /r/ is unrounding. In this dialect, /r/ is rounded in the onset – [ɹʷæbɪt] "rabbit" – and in the coda after a round vowel ([kɔɹʷ], vs [bɪɹ, bɛɹ, baɹ]). The simplest account of this is that the phoneme is intrinsically rounded, and unrounds after an unrounded vowel. The other rounded glide, [w], does not unround in [bæw] "bow", though the only non-round vowel that can appear before [w] is [æ] which can't appear before [ɹ] – in this dialect.