Rhoticization

[vegfarandi]

Just asking to be clear, "rhoticization" as pertaining to vowels such as ɚ or ɝ is actually retroflexing, correct? Am I correct in assuming that "rhoticization" is a weird English-centric view of the whole thing?

[Nortaneous]

It varies - I've seen "rhotacization" and "r-coloring". Linguistic terminology isn't very standardized.

[Travis B.]

I personally can't tell what people mean when they say "rhoticization", as my own rhotic, and far as I can tell, that of other members of my family is an (initally or post-rounded-vowel labialized) uvular approximant (even though I think on rare occasions it can be a uvular trill), except after coronals (where then it is a messy coarticulated rhotic), and certainly uvularization is not what people mean when they say "rhoticization".

[Vijay]

I think "rhotic" just in general is a very vague term that can refer to a pretty wide variety of things.

[Pabappa]

I think its because only the retroflex variety of rhotic allows distinction between other categories that distinguish vowels. e.g. the languages in India sometimes distinguish multiple rhoticized vowels, whereas the Czech syllabic /r/ is not really a rhoticization of something else, its just a consonant.

https://en.wikipedia.org/wiki/R-colored_vowel implies that at least some languages have rhoticization for the duration of the vowel, meaning that it is just one more way of expanding the language's vowel inventory.

The Chemnitz dialect of German does much the same thing with its pharyngealized vowels, and those arose from vowel + /r/ sequences, ... but perhaps we dont use this term to describe their vowels because we can just say "pharyngealized".

[Vijay]

I think its because only the retroflex variety of rhotic allows distinction between other categories that distinguish vowels. e.g. the languages in India sometimes distinguish multiple rhoticized vowels, whereas the Czech syllabic /r/ is not really a rhoticization of something else, its just a consonant.

wat

[Pabappa]

Sorry if I said something wrong .... but what part of what i said was wrong?

[Vijay]

All of it?

I don't have the faintest idea what you're talking about. What are these "multiple rhoticized vowels" you're talking about? What does it mean for a rhotic to "allow distinction between other categories that distinguish vowels"? What criteria do you use to distinguish "rhoticized vowels" from "not really a rhoticization of something else, it's just a consonant"?

[Whimemsz]

I don't know about the rest of it, but there is at least one Dravidian language with two phonemic degrees of vowel retroflexion, Badaga. (I don't know if there are others, I only know about Badaga because of the UCLA Phonetics Lab page.)

[Nortaneous]

I think its because only the retroflex variety of rhotic allows distinction between other categories that distinguish vowels. e.g. the languages in India sometimes distinguish multiple rhoticized vowels, whereas the Czech syllabic /r/ is not really a rhoticization of something else, its just a consonant.

wat

You can't have vowels coarticulated with an alveolar trill or tap, but you can have vowels that are sort of coarticulated with a retroflex approximant. In some languages, like English, Yurok, and Purepecha, there's only one rhotacized vowel; in others, like Ersu and Serrano, the rhotacized vowel system has gaps relative to the unrhotacized vowel system, but there are multiple rhotacized vowels; and in others, like Qiang and Badaga, every vowel can take rhotacization.

Cf. nasal vowels: in some languages, like Cherokee, Massachusett, and 12th-century Polish, there's only one nasal vowel; in others, like French and Lakota, the nasal vowel system has gaps relative to the oral vowel system, but there are multiple nasal vowels; and in others, like Breton and Maxakali, every vowel can take nasalization.

[Vijay]

In English, it's not necessarily retroflex, and it's controversial whether there's actually a rhoticized vowel or just a syllabic R. EDIT: I've seen the Badaga thing before, but I haven't heard of that in any other Dravidian languages yet so far.

[Nortaneous]

In English, it's not necessarily retroflex, and it's controversial whether there's actually a rhoticized vowel or just a syllabic R.

Probably because many people aren't aware that similar sounds occur in some Chinese languages, like how Africanists tend to have a hard time describing the fricated vowels of Grassfields languages. (Reduction of unstressed /rə/ to /ə˞/, as we see in the word "iron", has exact precedent in some of those languages too, including Mandarin Chinese.)

Some flavors of Shakespearean Original Pronunciation prescribe monophthongal rhotic vowels for SQUARE etc.

[Vijay]

In English, it's not necessarily retroflex, and it's controversial whether there's actually a rhoticized vowel or just a syllabic R.

Probably because many people aren't aware that similar sounds occur in some Chinese languages

I think those are controversial, too.

Also, even in Badaga, the contrast hasn't been that strong for almost a hundred years now; currently, speakers only show retroflexion for a few vowels. I suspect that's part of the reason why I don't actually hear one most of the time in those recordings.

[Nortaneous]

In English, it's not necessarily retroflex, and it's controversial whether there's actually a rhoticized vowel or just a syllabic R.

Probably because many people aren't aware that similar sounds occur in some Chinese languages

I think those are controversial, too.

Sometimes you get debate over whether to classify certain sounds as rhotacized vowels or Vr sequences, but IME it's very common for rhotacized vowels to be included in a phoneme inventory with no comment.

[Travis B.]

I prefer to analyze them as Vr sequences with r-coloring of the vowel in English, at least here, due to there being words where formerly impossible Vr sequences have become possible through elision of consonants in words such as every and other, such as DRESS or STRUT followed by /r/ distinct from NURSE, even though the dialect here is Mary-merry-marry merged and fir-fur-fern merged.

[Nortaneous]

I analyze them as vowel + semivowel sequences, except for NURSE, which is a unit and occupies the position of a high central vowel.

The problem is that it'd be nice to be able to analyze lettER as schwa + r, but then you have V alternating with VC in e.g. "pervert". On the other hand, you also have V alternating with VC in stress-shifted noun/verb pairs like /ˈdijˌfekt/ - /dəˈfekt/ and /ˈdajˌdʒest/ - /dɨˈdʒest/.

[missals]

I favor the rhotacized vowel interpretation for my American variety, i.e. to consider them a unitary tense vowel or diphthong, because vowel contrasts are vastly reduced before /r/ and there is no clear phonetic or phonological motivation to consider any of them directly equivalent to a non-rhotic vowel.

E.g. before most consonants you have the full gamut of the 10 main vowels /ɪ i ʊ u ɛ eɪ ʌ oʊ æ ɑ/ (no /ɔ/, I'm cot-caught merged) plus the three other dipthongs /aɪ ɔɪ aʊ/. But aside from the stressed rhotacized schwa /ɚ/ itself, there are only four contrastive pre-rhotic vowels - /ir er ar or/. It is not at all clear that /er/ is "really" /ɛr/, considering it is a merger of at least three historical front vowels (Mary-merry-marry). Moreover, /r/ cannot occupy the syllable coda following a diphthong - e.g. fire is two syllables, [fʌɪ.ɚ] - so there really are no other monosyllabic vowel-/r/ sequences besides the four I listed, even on a phonetic level.

[Nortaneous]

I favor the rhotacized vowel interpretation for my American variety, i.e. to consider them a unitary tense vowel or diphthong, because vowel contrasts are vastly reduced before /r/ and there is no clear phonetic or phonological motivation to consider any of them directly equivalent to a non-rhotic vowel.

E.g. before most consonants you have the full gamut of the 10 main vowels /ɪ i ʊ u ɛ eɪ ʌ oʊ æ ɑ/ (no /ɔ/, I'm cot-caught merged) plus the three other dipthongs /aɪ ɔɪ aʊ/. But aside from the stressed rhotacized schwa /ɚ/ itself, there are only four contrastive pre-rhotic vowels - /ir er ar or/. It is not at all clear that /er/ is "really" /ɛr/, considering it is a merger of at least three historical front vowels (Mary-merry-marry). Moreover, /r/ cannot occupy the syllable coda following a diphthong - e.g. fire is two syllables, [fʌɪ.ɚ] - so there really are no other monosyllabic vowel-/r/ sequences besides the four I listed, even on a phonetic level.

I analyze Mid-Atlantic AmE as having recently simplified its vowel system by resolving some vowels into VC sequences.

Code: Select all

i ɚ u | ij     | ir     | uw
e ʌ o | ej  oj | er  or | ʌw
æ   a |     aj |     ar | æw

Code: Select all

KIT   NURSE FOOT    | FLEECE        | NEAR         | GOOSE
DRESS STRUT THOUGHT | FACE   CHOICE | SQUARE NORTH | GOAT
TRAP         FATHER |        PRICE  |        START | MOUTH

Other factors:
- Canadian raising adds phonetic [ʌj]
- A lot of Mid-Atlantic AmE has the TRAP-BATH split; variants that don't have it have æ-breaking before /m n/, which is arguably marginally phonemic (e.g. panko [ˈpeə̯ŋkəw] vs. bank [bæi̯ŋk]) - /æ/ > [æi̯] before voiced velars is regular. The alternative is /pænkəw/ vs. /bæŋk/ - which is almost certainly where this contrast comes from (aside from yeah, which could probably be captured by æ > eə̯ / _#, unpredictable [eə̯] only occurs before <nk> in unfamiliar words; the only other case I can think of is Bernanke), but which requires positing nasal place assimilation to operate both diachronically and synchronically.
- Probably to parallel the development of /eə̯/, /o/ breaks in some environment or other (probably before coronals?) to [ɔə̯]; for some speakers, there's also /uə̯/ in the single word on.
- Some people have [oˤ], especially in the word both (sometimes misspelled *bolth), but sometimes in words that contain /l/, like social.

One advantage of this is that it explains the realizations of the vowels:
- FLEECE and GOOSE are clearly diphthongized [ɪi̯ ʏu̯]
- Neutralization of all vowel features except height before coda /w/ can explain centralization of GOOSE and GOAT. Absence of centralization in MOUTH can probably be explained by the Southern tendency to weaken /j w/ after low vowels -- /æw/ can be realized as [æə̯], or even monophthongal [æ] before /n/. This tendency isn't as strong above the Mason-Dixon line (someone from New Jersey was under the impression that I merged "howl" and "Hal"), but maybe the phonetic resistance to centralization carried over.
- The only vowels that can occur before /j r/ are /a e o i/, so the raising and backing of the onsets of CHOICE and NORTH can be explained by pressure toward maximal differentiation. In the Vr subsystem (which, unlike the Vj subsystem, doesn't contain a phonetic low-mid back vowel), this is followed by raising and backing of the onset of START to [ɔ].

Another advantage is that it explains l-breaking: 0 > ə / V{j r}_l. Coda cluster disallowance seems theoretically much neater than positing the development of a restriction on /l/ occurring after certain (diphthongal) vowels. OTOH, you still need the latter for German.

(Some speakers have this for uw_l also, and... OK, I haven't formally studied this and don't know of any formal studies on it, but one time I wrote up a wordlist and asked my mother how many syllables each word had, and she had a very hard time with orthographic ow(e)l. This is the southern end of Mid-Atlantic, though -- I'm not sure if similar problems obtain in New Jersey. My impression is that there are some words that clearly have one syllable (howl), some words that clearly have two but admit of a monosyllabic pronunciation (towel - [tæwəl] ~ [tæl], maybe lengthened in the second example; cf. the optional realizations [fa(:)r] for fire, [fa(:)l] for file, [tæ(:)r] for tower), and some words that I'm just not sure about (dowel, trowel, owl).)

A third advantage is that a resyllabification rule, where semivowels are preferentially syllabified into the coda (cf. Wobzi Khroskyabs), explains both CHOICE in "lawyer" and the hurry-furry merger.

One problem is that əw uw > oˤ uˤ / _l is an ugly rule. But uw > uˤ / _l isn't universal, and /oˤ/ is maybe marginally phonemic for some speakers. And anyway, you can have semivowel+vowel medials being affected by initials in some Sino-Tibetan languages, and this is just the reverse of that.

[Travis B.]

Does anyone else have optional breaking of NEAR? I notice that in the dialect here near is either a monosyllable with a near-close vowel, namely [nɪ(ː)ʁ], or a disyllable with a close vowel, namely [ˈniːʁ̩(ː)], depending on whether it is emphasized or not.

[missals]

No, but I have ambiguous breaking of /il/. Nort's description is fairly apt for my variety, and I have breaking of /l/ after all diphthongs (including /eɪl/, but not /oʊl/, because it's not actually /oʊ/ + /l/, it's an aberrant "pure" /ol/) - I couldn't tell you for sure whether "seal" is one or two syllables.