Friday 13 November 2009

weak u and ə

Christer Bermheden wrote from Sweden:
I am sometimes still confused when it comes to weak vowels. Today [in LPD] I found ʊ in spatula and u in occupy, although they seem equal and appear in the same environment. I also noted that they are replaced by the same symbol in the transcription of the American-English pronunciation - ə.
Actually, it’s not true that they “appear in the same environment”. A crucial difference is that spatula has a weak final vowel (ə), while occupy has a strong one (). Consequently by my syllabification rules (which you may disagree with) the second syllable in spatula is closed, while that in occupy is open: the l of spatula remains in the default coda position, while the p of occupy is captured into the final syllable by the following strong vowel.
My introspection and observation leads me to think that in RP, or at least for me personally, we typically have a vowel ranging over [ʊ ~ ə] in a closed syllable but over the more firmly back [u ~ ʊ] in an open syllable.
Compare stimulus and stimulate, which I have transcribed ˈstɪm jʊl əs and ˈstɪm ju leɪt respectively (in each case also with -jə- as an alternative possibility).
In American English, yes, -jə- would be normal; but in BrE, for someone of my age group, my reaction to it is that that would sound uncultivated or at least very casual.
You may think that I have made things unnecessarily complicated here. There is obviously a gradual tendency for the British to follow the Americans towards generalizing -jə- in these weak syllables. But we haven’t by any means got there yet.
_ _ _
If you would like to see and hear me chatting about teaching pronunciation, here are two television interviews recorded when I was in Argentina recently. They are in programmes 25 and 26, but you’ll have to wait for 15-16 minutes of other material first each time.

33 comments:

  1. You can use the fast-forward button - have to push it about a hundred thousand times; it jumps in unpredictable gaps.

    ReplyDelete
  2. Seems to go in larger jumps when you do that while it's on pause.

    ReplyDelete
  3. For me spatula is [spætʃələ], and I suspect most Americans have something similar. I wonder why [tj] has become [tʃ] here instead of the more normal American resolution to [t] -- or does that only happen in stressed syllables?

    ReplyDelete
  4. Yod is regularly retained (> tʃ) in weak syllables: rite-ritual, situation, perpetual, etc.

    ReplyDelete
  5. With you 100% today on both this piece to web and the pieces to camera. I hope the 'prónunciation' wasn't as trying for you as it was for me.

    A lot of what a few of us have been discussing on diphthong schmiphthong must have been driven by morphic resonance with those interviews!

    ReplyDelete
  6. Word verification was saying I must certify that I am a non-rhotic monster, and did not believe my attempt to do so. There's a moral in that somewhere.

    ReplyDelete
  7. Isn't the distribution of light and dark (velarized) allophones of /l/ in RP quite a strong argument against a syllabification /spæt.jʊl.ə/ ?

    I, and I believe most RP speakers, have clear (unvelarized) /l/ in spatula, as in all cases or prevocalic /l/. However I have dark (velarized) /l/ where non prevocalic, as in "altar" or prepausal "full".

    I believe that this is true of RP in general, and I think you agree with this in "Accents of English".

    Given this distribution of allophones of /l/, surely it would make more sense for prevocalic /l/ to occur in the onset, rather than the coda, of the syllable. Then there would be a simple rule for the realization of /l/: in the onset it is clear, while in the coda of the syllable it is velarized. So the last syllable of "spatula" would be /lə/

    Of course the opposite argument would hold for GenAm, where I believe all post-vocalic /l/s are dark. For GenAm, a syllabification /spætʃ.əl.ə/ might have more to be said for it.

    ReplyDelete
  8. John, what are the problems you identified on the interviewers's pronunciation?

    ReplyDelete
  9. Well vp I did say the other day 'Most diphthongs are schmiphthongs, actually. Anyone remember who it was that said, "The syllable is not a phonological unit"?' But if we must talk in terms of syllables then JW is right about a vowel ranging over [~ ə] in a closed syllable but over the more firmly back [u ~ ʊ] in an open syllable. Your analysis would seem to imply that we have ʊ in an open syllable, and we can't have THAT. Ugh! (I mean [ʊ], not [ʌx] or anything, but interjections are extra-systematic.) So the l is not the onset rather than the coda. It is functioning as both, and in RP that makes it relatively clear, and in GA doesn't stop it being dark.

    Anon, you can NAHT be Sirius!

    ReplyDelete
  10. vp: /l/ allophones determined not by syllabic position, but by the status of the following segment: clear before vowel or /j/, dark otherwise - as you say yourself. Even word-finally you get clear l if the next word starts with a vowel (e.g. fill it, fill up, fill everything).

    ReplyDelete
  11. @JW:

    I understand that, under your theory of syllabification, the allophones of /l/ cannot be predicted from their syllabic position.

    However, if I understand your theory of syllabification correctly, it is justified by its ability to predict allophones of phonemes (for example, aspiration or non-aspiration of voiceless plosives). My suggestion is that a theory of syllabification that also predicts the allophones of /l/ is more powerful than one that does not. Since a theory that makes pre-vocalic (or pre-/j/) /l/ occur in the onset of a syllable does predict the allophones of /l/ successfully, it seems to be preferable on these grounds.

    As for the sandhi effect whereby a word-final /l/ is clear if it is followed closely by a vowel, that seems to be a very close parallel to the linking "R" of RP. Whatever strategy is adopted to syllabify a phrase such as "far off" can be adapted to explain "fill up".

    ReplyDelete
  12. @mallamb:

    Would you care to say why we cannot have /ʊ/ in an open syllable?

    ReplyDelete
  13. Are you really sure that ju in occupy and spatula are Americanisms, that might be spreading in Britain? I suspect our ages are similar but I don't speak RP. I've always said ju here, and affricated spatula. I checked some old dictionaries. An English-Swedish dictionary published in Stockholm in 1948 claims to follow Daniel Jones and EPD, and offers only ju here. I can understand Christer Bermheden's confusion. Equally old EPD and Oxford Pocket also have only ju. So you ju seems to be the conservative prununciation, rather than an import. And uncultivated affrication? Isn't that just another way of saying I don't speak RP?

    ReplyDelete
  14. Sidney: you've got it the wrong way round. I'm suggesting that the spread of /ə/ for /u/ is at least the following of a trend already predominant in AmE. The evidence you adduce supports this.
    I made no comments about the social status of affrication or its absence.

    ReplyDelete
  15. OK, vp.
    "Would you care to say why we cannot have /ʊ/ in an open syllable?"

    I didn't say that. I deliberately used no brackets, having pinched the ʊ out of JW's square brackets, regrettably, as you can see, without holding Ctrl.

    I recently said re blancmange "I note that in these relatively informal communications JW himself dispenses with brackets when the common-sense interpretation of a transcription is obvious." However in this case I had another reason for dropping them: whatever phonematic status you assign to the ʊ, and whether you are in phonology or phonetics, it can never have the indisputably open status for which I have given the interjection [ʊ] for 'ugh'. And as I say interjections are extra-systematic, i.e. they cannot consistently be shown to partake of the double articulation, or indeed of any articulation at all. No doubt you will want to adduce examples of "/ʊ/ in an open syllable", but they would of course beg the question of what a syllable is, which you yourself are disputing here.

    You base your argument against a syllabification /spæt.jʊl.ə/ on the distribution of light and dark (velarized) allophones of /l/ in RP, in the light of which it would make more sense for prevocalic /l/ to occur in the onset, rather than the coda, of the syllable in such a sequence, providing us with a simple rule for the realization of /l/, and with this I have some sympathy.

    But it leaves you with /jʊ/ as a syllable, so in spite of my mantra "the syllable is not a phonological unit" I say if we must talk in terms of syllables then JW is right about a vowel ranging over [ʊ~ ə] (with the ʊ restored) in a closed syllable but over the more firmly back [u ~ ʊ] in an open syllable. And your /jʊ/ would have realizations in the former range, would it not?

    JW does grant us permission to disagree with his syllabification rules, but here they unambiguously predict those ranges. Your proposed rule gives an appealing intuitive solution to the problem of the allophony of /l/, which JW accounts for just as well by the status of the following segment irrespective of syllable or even phonotagm structure. But the syllabification your rule requires gums up the prediction of vowel allophony here.

    I was merely trying to ungum it for you by offering a not particularly outlandish Saussurean solution: that the /l/ is implosive with respect to the /ʊ/ and explosive with respect to the /ə/. Perhaps now that you are talking of a strategy adopted to syllabify a phrase such as "far off" being adapted to explain "fill up", you are more open to my ungumming strategy being adopted here to deal with this bit of gumming together rather than up.

    ReplyDelete
  16. Is there any verb or noun ending with that thing? If so, and if they are *really* allophones, one could expect an ending of [-u ~ -U] to turn into [-Uz ~ -@z] or [-Ud ~ -@d] when it is inflected for plural/3rd person/past.

    ReplyDelete
  17. Sorry John, I was overinterpreting your:

    "In American English, yes, -jə- would be normal; but in BrE, for someone of my age group, my reaction to it is that that would sound uncultivated or at least very casual."

    ReplyDelete
  18. David Marjanović15 November 2009 at 17:57

    Is there any verb or noun ending with that thing?

    Has snafu ever been used as a verb, the way fubar is so regularly?

    ReplyDelete
  19. No. The most plausible candidate is to value, occasionally with final weak -u though usually with strong -u: (both versions in LPD).

    ReplyDelete
  20. I am one of those who has an [ɒʊ] allophone of /əʊ/ before [ɫ]. However, I still have [əʊ] in Polish, even though the /l/ is clear because it is prevocalic.

    Sorry if that upsets anyone. ;-)

    ReplyDelete
  21. Ooops. I meant of course [ɒʊ] in Polish. Sorry!

    ReplyDelete
  22. - verb: issue?

    - John M, while that didn't upset me, I did wonder for a split second what Polish dialect with [əʊ] that might be.

    ReplyDelete
  23. LOL, so did I.

    John M, how does that work for wholly holy? It suggests you would not have JW's opposition between holey (and possibly wholly) and holy. The discussion on that entry of his had a new lease of life. If you get a chance to have another look at it I would be interested to know where you fit in.

    John W, could you elucidate why the most plausible candidate is to value, rather than Lipman's issue or say tissue or cashew? For me the funny thing about that lot is that they can have a geminated [ʃ].

    ReplyDelete
  24. At any rate they can lose the [j] one way or another, but so can spatula. And value can have a geminated [j]. Mostly in AmE, of course.

    ReplyDelete
  25. mallamb:

    in "holey" "holy" and "wholly" I have [ɒʊ] and a clear [l]. They are homophones for me. However....wait for it...in "slowly, hollowly" etc. I have [əʊ].

    It seems that my [ɒʊ] allophone occurs before any /l/ in the same morpheme. Or maybe I have a "Janus segment". It looks like a dark /l/ on the left but like a clear one on the right. Wouldn't that be sumpin' ta chew on?

    ReplyDelete
  26. But John, that's exactly what I was proposing earlier on this thread when I was trying to get vp's argument for a different syllabification than JW to fit the Procrustean bed: "the l is not the onset rather than the coda. It is functioning as both, and in RP that makes it relatively clear, and in GA doesn't stop it being dark"! He apparently thought that post too cryptic, so I explained it in more detail in my subsequent post, claiming a Saussurean pedigree for such a "Janus segment".

    So I didn't have to wait for it! I would have expected you to have [əʊ] in "slowly, hollowly" etc. even if you had [ɒʊ] and a clear [l] in all three of "holey" "holy" and "wholly", because, as you say, of the morpheme boundary. No doubt you were not amused by my treatment of those three on that thread, but all that happens is that some of these morpheme boundaries get lost for some of us, but not for others. I give other counterexamples besides holey and wholly, and point out that there's a whole spectrum of variance. I would on the other hand not have expected 'holy' to be anywhere on it, but of course there is a perfectly good etymological reason why the morpheme boundary should be between the l and the y. Thank you for reminding me of it. Can't say I think many people are still keeping that etymological flag flying, but I note JW is very careful to say of hole-y and wholly, "In both the /l/ is treated as morpheme-final. But in holy it isn’t." (My italics.) Well in your 'holy' it is!

    Sumptuous pabulum indeed!

    ReplyDelete
  27. "Of course the opposite argument would hold for GenAm, where I believe all post-vocalic /l/s are dark. For GenAm, a syllabification /spætʃ.əl.ə/ might have more to be said for it."

    With /l/ between vowels, I'd say that many, and I think most, Americans have a Dark L following the first vowel that rolls over to a Clear release into the second vowel. This may not be a very good technical description, but it's the best way I can express it. The way I pronounce spatula, or really, or ballast, sounds very different from the way it would with the truly Dark L one hears in most Scottish accents, and those few American accents that have Dark L all the time, such as Valley Girl/Surfer Dude.

    One of the ways in which English actors consistently get American accents wrong is in following the instructions of their dialect coaches (who should know better) that all Americans have Dark L in all environments. Drives me mad.

    ReplyDelete
  28. "Of course the opposite argument would hold for GenAm, where I believe all post-vocalic /l/s are dark. For GenAm, a syllabification /spætʃ.əl.ə/ might have more to be said for it."

    Forgot to say that I agree with the above syllabification. It reflects what most Americans think is going on, and is reasonably close to what they really say. It's not how I'd syllabify a word in English writing, but phonetically? Absolutely.

    ReplyDelete
  29. So, assuming people use the same vowel in "issues" or "issued" as in "issue", there's such a thing as weak /-uz/ (or /-ud/) which might, in principle, contrast with weak /-Uz/ or /-Ud/. Or I am speculating waaay too wildly. (I was trying to determine whether there's an analogue of "studded"-"studied" with back vowels.)

    ReplyDelete
  30. To vp: If I understand correctly you'd analyse "far off" as /fA:.rQf/; but there are speakers who distinguish "yaw rise" from "your eyes", typically using a labialized r in the former but not in the latter. So I'd still treat "far off" as /fA:r.Qf/, and "feel it" as /fi:l.It/. Allophones can be determined by surrounding sounds even across syllable boundaries ("it sucks" v. "it all").

    Also, syllabifying "volume" as /vQl.jUm/ is consistent with the fact that the /j/ is also pronounced by those who would drop it in "lute". On the other hand, the "wholly"-"holy" thing might be one argument to syllabify /l/ with the syllable at the right in absence of morpheme boundaries.

    Does anybody consistently distinguish "feel eye" or "fee lie" (please find a more plausible example if you can!)? (As a non-native speaker, I can't tell.) If not, then /.l/ ~ /l./ might be a distinction without a distinction in some cases.

    ReplyDelete
  31. @army1987:

    I agree with your suggestion wholeheartedly.

    I still have the wholly/holy split in my own speech, and the syllabilifications that are intuitively correct to me are "wholl-y" and "ho-ly". I have a velarized/dark "l" in "wholly" and a clear one in "holy".

    It seems likely that the origin of the "wholly"/"holy" split is backing of [əʊ] influenced by the following velarized [l].

    ReplyDelete
  32. I always have the feeling that it would be better off for a Spanish/Greek person to learn Northern/Scottish/Irish accent, or an accent which that is closer to their own phonemic system - Cut = /kUt/ or /kut/, book = /buk/, third = /t(h)3:d/...
    Similarly with Chinese people, the Cockney accent would be more suitable for them, Well = /weU/, cut = /k@?/, look =/lU?/.

    Ok, it might not be the "standard", whatever that might mean, but it would presumably be easier for EFL learners to achieve "A" native accent, so why not? If the standard is so important then perhaps every single non-RP native speaker should go and learn the standard. There is no reason why an EFL learners shouldn't speak like a Yorkshireman or a Scouse.

    ReplyDelete

Note: only a member of this blog may post a comment.