Wikipedia:Reference desk/Archives/Language/2016 July 9

= July 9 =

"ː" as a distinguishing feature in English phonology
My phonology education was very sketchy, and this may explain why I ask the question below.

Vowel length has a short subsection in which it's alleged (without any sources) that "In Australian English, there is contrastive vowel length in closed syllables between long and short /e æ a/ and sometimes /ɪ/" (my emphasis). No mention here of US or British English.

Of course, English isn't limited to the three nations Oz, US, Britain, and US and British English aren't limited to GA and RP; but there are only so many hours in the day, and therefore I'll look primarily at GA and RP.

There's quite a lot of material in Received Pronunciation, but nothing that I notice about any contrast. A table there lists the following phonemes with the lengthening quasi-colon: /iː/, /ɜː/, /uː/, /ɔː/, /ɑː/; but it doesn't list even one of these without the quasi-colon. So there seems to be no phonemic distinction.

International Phonetic Alphabet chart for English dialects has the diaphonemes /ɑː/, /ɔː/, /ɜːr/, /uː/, /juː/, /ɑːr/, /ɔːr/; but it doesn't have any one of these without the quasi-colon. It does have both /iː/ and /i/; but its examples for the former are "see" and "meat", and for the latter "city". Seems to me that the distinction is an automatic result of ±stress. Is the word /ˈsɪtiː/ phonotactically possible in English; and if so, could it contrast with /ˈsɪti/ ("city")? If not, the pair look like mere allophones to me.

In General American, we learn that there are contrasting English diaphonemes, /iː/ and /i/, and that in General American these are realized as [iː] and [i(ː)]. But it's pretty much the same as above: [i] is exemplified by /i/ in unstressed syllables of disyllabic words; [iː] by /i/ in monosyllabic stressed words. Mere allophones again, I think.

What have I misunderstood? -- Hoary (talk) 13:49, 9 July 2016 (UTC)


 * Bad-lad split may address some of your queries. Valiantis (talk) 20:23, 9 July 2016 (UTC)


 * Thank you, Valiantis. It's most interesting. It does suggest that /ː/ is a possibility in English -- but curiously, it does so for /æː/, which isn't mentioned above. It's within an article on the pronunciation of "a", and it does say a lot about what it refers to as /ɑː/, which is mentioned above. However, it has no mention of /ɑ/ in Modern English. If a language lacks /ɑ/, how can it meaningfully have /ɑː/? (Why isn't the one phoneme "/ɑː/" simply written /ɑ/?)


 * The article Australian English phonology does indeed claim both regular and lengthened /e æ a/. The article tells us: "There exist pairs of long and short vowels with overlapping vowel quality giving Australian English phonemic length distinction, which is unusual amongst the various dialects of English", and sources this claim to Robert Mannell, "Impressionistic Studies of Australian English Phonetics". But when I look at the latter (admittedly before my second coffee of the day), I don't see that Mannell says any such thing.


 * Any other phonologists in the house? (Æµ§œš¹?) -- Hoary (talk) 23:23, 9 July 2016 (UTC)


 * I will not go much into details as I myself as well may do not understand well or remember all the theory, but much confusion or misunderstanding comes from the transcription. When Daniel Jones firstly devised the IPA for English in the beginning of the 20th century he actually implied directly the length contrast. This has been called the quantitative analysis. But later the views about the phonology of English (or more precisely of the RP variety) have been revised and the analysis was changed to quantitative-qualitative. American phonologists, though, mostly prefer a plain qualitative analysis, ignoring the length as non-phonemic, as you have mentioned right. There is a short description of that evolution as well as a comparative table I've once made. (I think this section somewhat out of place, because the IPA for English is not a respelling, and its history and overview might deserve its own article.)--Lüboslóv Yęzýkin (talk) 04:14, 10 July 2016 (UTC)


 * Thank you, Любослов. That's really interesting -- and within an article I'd never imagined would exist. This does explain a lot about historical influences on the phonemic representation of English, but it doesn't start to explain why people have clung on to it. According to my reading of a little table in that article, Jones thought that the vowels in "cod" and "cord" differed only in length, as did those in "rid" and "reed". I can't speak for the English of his day, but for the standard US and British English of our day, he's plain wrong. I mean, the mouth is doing things that differ in ways other than mere time. (Even if he got it wrong for the English of his day, I'm reluctant to knock him. He was a pioneer, and pioneers can't be expected to get everything right.) Suppose, though, that Jones had realized that the vowels in (say) "rid" and "reed" were qualitatively different. I think he'd be right to say they're quantitatively different too, and therefore to use a colon for the latter. After all, he was aiming for a phonetic script. However, a century later, while we're using "/   /", we require distinguishing features, no? (Or am I even more confused than I realize?) -- Hoary (talk) 04:39, 10 July 2016 (UTC)
 * I've oversimplified a little bit. And I indeed do not remember all the details as I read Jones' works quite a long time ago, but now I've read them again and found out that he actually knew the difference when he described the vowels verbally. But he still tended not to signify the distinction in his transcription. Either he did it for simplicity or he did not consider quantity as much as important. Anyway, as we know now, this has been reviewed and both the quantity and the quality are considered important and transcribed accordingly. American linguists, however, may and often ignore the length sign, you can see that, for example, in the IPA Handbook.


 * So if I were to try to answer you question I'd say some important thresholds:
 * 1) Both quantity and the quality are important in producing and distincting the English vowels.
 * 2) Both of them may fluctuate according to the environment, stress, temp of speech, accent and so on. It is hard to describe those variations in short. The length is more unstable.
 * 3) In general, and this is most important, the English vowel system, may be described as asymmetric. There are no or few minimal pairs that could be differentiated only by vowel length, but the quality is always accompanied by the quality, they are interrelated. The English vowels cannot be structured in some sort of symmetric grid like, say, it is reconstructed for Classical Latin (Vulgar Latin was another story), or some other languages. However, I would say there are few languages where a change in length is not accompanied by some sort of change in articulation. For example, in German or Hungarian long vowels tend to be tenser, however, if I remember right there is a /ɛ/-/ɛː/ contrast in German (but in some dialects /eː/ and /ɛː/ are merged, hence only the contrast between /ɛ/ and /eː/ is left ). But in Finnish the contrast is only in length.


 * What I said concerns largely RP and GA. Some accents may indeed not observe the quantity-quality interrelation. Scottish English is very remarkable in that respect, where the Scottish Vowel Length Rule is in effect, and in other accents there are splits like "bad-lad" mentioned above.


 * I'm not sure if I've answered your question, but I hope it might be helpful.--Lüboslóv Yęzýkin (talk) 21:28, 10 July 2016 (UTC)


 * Thank you, Lüboslóv, that's most interesting. (If I'd had you as the teacher of my [sole] phonology course, I wouldn't have dozed off during the lectures.) Jones seems to have been a lot more perceptive and skilled than I'd have dared hope. I'd never heard of or even imagined the existence of the Scottish vowel length rule. And even for "more standard" (?) English, where minimal-pair contrasts of length don't exist, I do accept that there are differences of timing. (And I find it easy to imagine that, for example, if a recording of "beat" were altered so that the vowel were shortened but otherwise unchanged, the result could sound like "bit".)


 * Ladefoged's piece about (one Californian speaker of) US English is particularly informative. But it's also particularly odd. Here's a rather randomly chosen quotation from it:


 * When intervocalic and before an unstressed vowel, as in 'city, vicinity', [t] is a voiced flap, resembling [ɾ].


 * (My emphasis.) I expect to see "/t/" instead. Ladefoged's notation is odd, or he's little concerned with phonemes, or I understand even less than I suppose. But if he is little concerned with phonemic distinctions, then of course it's hardly surprising that he adds details that a speaker of English never has to worry about (vowels aside, this would include, say, ±aspiration for [p]).


 * Please don't spend too much time on my question(s), but I do greatly appreciate your explanations. -- Hoary (talk) 02:21, 11 July 2016 (UTC)


 * As an American who took linguistics 201/202 for majors in the 1980's and who already spoke English, French, German and Spanish to lesser or greater degrees, as well as a little of my moribund family Rusyn language, we were tought that all phonemic transliterations differed in quality for American English, while phonetic transcriptions were either diphthongs or differed in length. Hence:


 * iy, ih, ey, eh, æ, ah, ough, o, ooh, yew


 * /i/ /ɪ/ /e/ /ɛ/ /æ/ /a/ /ɔ/ /o/ /ʊ/ /(j)u/


 * /i:/ /ɪ/ /e:/ /ɛ/ /æ/ /a:/ /ɔ/ /o:/ /ʊ/ /(j)u:/


 * Except that, of course, my dialet differs between tense and lax /æ/.


 * μηδείς (talk) 00:24, 14 July 2016 (UTC)


 * Thank you, μηδείς; but unless I'm totally lost, what are in your third line should be enclosed in [ ] rather than / /.


 * Though I wasn't asking about phonetic distinctions, actually these are starting to interest me. I'd uneducatedly guess that it's hard to make simple statements about length; but I suppose that there have been studies of vowel lengths (averages from a lot of tokens of these phones, within more or less "natural" speech by a wide variety of speakers of English that can plausibly be called standard) -- and yet I seldom see any talk about them. Take the article Length (phonetics), for example. This tells us:


 * In certain languages, however, there are pairs of phonemes that are traditionally considered to be long-short pairs even though they differ not only in length, but also in quality, for instance English "long e" which is (as in feet ) vs. "short i" which is  (as in fit ) [...]


 * So, quality aside, they do differ in length. All right -- but however tentatively this must be expressed, and with whatever qualifications: by how much? The article (which NB purports to be about not phonology but phonetics) doesn't even start to say. -- Hoary (talk) 23:18, 15 July 2016 (UTC)