Wednesday 18 April 2012

a new weak form?

Petr Rösel asks
Do you still believe there’s no weak form for couldn’t (as you state in LPD)?
To which my answer is yes, I do so believe.

His query arises because he thinks he’s heard people saying kədn̩t for couldn’t.

This is entirely likely. In many English northern- and Midlands-flavoured accents you could reasonably call ə the customary pronunciation of the FOOT vowel (usually merged with STRUT). Even in RP and the south of England, the English FOOT vowel does seem to be getting more central and unrounded, more like ə.

We hear this in all the FOOT words: put, bush, full, butcher, cushion, pudding, bullet, good, wood, cook, look, wool, woman, wolf, bosom…. For the speakers we are thinking of, all these words have ə. And so do could(n’t), would(n’t), and should(n’t).

But the fact that millions of people say (ˈ)ʃəɡə for sugar does not mean that sugar has a weak form, any more than do pudding, cook, or butcher.

What it does mean is that for the handful of FOOT words traditionally described as having a weak form it may be difficult or impossible to detect any difference between the strong form and the weak form. This handful comprises could, should, and would. It means that for these speakers these words have no phonetically distinct weak form: they are always kəd, ʃəd, wəd, no matter whether the phonetic environment calls for a strong form or a weak one.

So if you hear aɪ ˈdənəu ɪf aɪ kəd ˈduː ɪt I don’t know if I could do it it might well be true that the difference between the putative strong and weak forms of could is not detectable.

But if you also hear aɪ ˈdʒəs kədn̩t ˈduː ɪt I just couldn’t do it, you can’t on that basis say that couldn’t, too, has suddenly acquired a weak form (which happens to coincide with its strong form).

We usually speak of weak forms of words only when they are paired with strong forms, the two being phonetically different. Thus we say at has a weak form ət alongside its strong form æt, and we can formulate principles for the use of one or the other.

There is perhaps a parallel in the case of the pronoun it. Most of our pronouns have paired strong and weak forms, thus he hiː — hi, you juː — ju, jə, us ʌs — əs, them ðem — ðəm (and for some speakers also əm ’em). On general grounds you might expect it to undergo weakening in the same way: yet it remains as ɪt. There is no distinct weak form. That is because in RP and similar accents ɪ happens to belong to both the strong and weak vowel systems: KIT coincides phonetically with the vowel in packet, rabbit. So you could say (though I wouldn’t) that it does have a weak form, but that it happens to coincide phonetically with its strong form.

In Australian English, on the other hand, where ɪ does not occur as a weak vowel except in certain restricted phonetic environments, and rabbit is pronounced ˈræbət, it behaves exactly as expected, having ɪt as its strong form and ət as its weak form. Pack it and packet remain homophones.
_ _ _

I’ve got a busy few days coming up. So the next posting will be on 24 April.


  1. There is an unfortunate weak form for "couldn't" in some areas of the country. This is [kʊnt ~ kʊn?]. Petyt (1985, p.182) found this (along with contracted negatives such as [dʊnt] for "doesn't", [ka:t] for "can't", etc.) in West Yorkshire. My impression is that the form extends over a much larger area than that, but it seems that few other linguists have researched contracted negatives in depth.

    Are there related forms in the south?

    1. No, though there are Midlands forms such as [ˈmʊnə] 'mustn't'. Compare the regional forms "dursn't" (= daren't), "divvn't, dinnae" (= don't) etc., from various parts of the UK, not to mention "ain't".

      These are not weak forms (they are accentable!), but contracted negatives.

    2. Thanks for your response, John. I'm always pleased when you take the time to respond to me, as I get to learn more.

      I am wholly self-educated in phonetics and there are times like this when I make a fundamental mistake that reveals my lack of training.

  2. "in RP and similar accents ɪ happens to belong to both the strong and weak vowel systems: KIT coincides phonetically with the vowel in packet, rabbit. So you could say (though I wouldn’t) that it does have a weak form, but that it happens to coincide phonetically with its strong form"

    Can't we say the same about FLEECE and GOOSE and be done with the weak [i] and [u] symbols which cause so much confusion among learners? I know it would be something of a simplification, but there are lots of interesting points of English phonetics/phonology that are not explicitly shown in phonemic transcriptions. Cruttenden's latest revision of Gimson's Pronunciation of English treats [i] as an allophone of /i:/ and the graphics on the cover even give us /prəˈnʌnsiːˈeɪʃn/.

    Most phonemic transcriptions are produced for foreign learners. Including non-phonemic weak vowel symbols goes over their heads and could put them off phonetics altogether.

    Teaching that 'we' has a strong form /wi:/ and a weak form [wi] is most unsatisfactory.

    1. You're right - we could do away with them. In doing that we would revert to the position before 1978 (was it?), when LDOCE first used the weak i and u, followed by the rest of us.

    2. The problem is that in some accents (e.g. old-fashioned RP) the weak vowel HAPPY coincides phonetically with the strong vowel KIT, not FLEECE (making studied a homophone of studded, for example). Also, while a strong vowel and a weak vowel might have the same quality, they still e.g. affect syllabification in different ways (flapped t in humanity but not in manatee, for example). I'm not a fan of using the same symbol for a weak and a strong vowel, and I'd even use ᵻ and ᵿ (small caps I and U with stroke) for the rosEs and beautifUl vowels, as OED3 does.

      (BTW, it seems to be that lack-of-HAPPY-tensing is one of the few conservative accent features which are more common in rock singing than in ordinary speech; is it just my impression or has someone else noticed that?)

    3. One solution would be for pronouncing dictionaries to simply change the status of the i and u symbols. Instead of representing neutralisation (which must be a tricky concept for most casual LPD users), they can be described as part of the dictionary's abbreviatory conventions and represent /ɪ ʊ/ for more conservative speakers and /iː uː/ for more 'cutting-edge' speakers. And it would follow from this that [i] and [u], like the other abbreviatory conventions, should not be used in phonemic transcriptions.

    4. Paul

      Are you implying that pronouncing dictionaries should use strictly phonemic transcription? My feeling is that the number of distinctions to be represented is a pragmatic question, based exclusively on what will help and what will confuse the foreign speaker. In some cases it may be too great a burden for the foreigner to learn rules of allophone that are instinctive to native speakers.

    5. No implication intended.
      Had my mind on the weak vowels, not the bigger picture.

    6. Preserved elsewhere there's an exchange between teardrop and Paul Carley on this theme. Were the two posts deleted? Deliberately? Should they be reinstated?

    7. Wikipedia's system of diaphonemes includes ɨ in cases such as pockit, roses, etc.

      I don't think this dictinction from ɪ is necessary, but the Wikipedia community thought otherwise. I noticed it in the IPA for my home town.

    8. Where you say the "Wikipedia community", I'd hazard a guess that in most cases the transcription is generally left as chosen by whoever gets there first, unlikely to be changed by others unless egregiously wrong.

      Compare: Bassett (I've linked to the current version just in case someone changes it just to prove me wrong!)

    9. I think the point is that Wikipedia has a standard transcription key, which includes ɨ as Ed says. Of course some entries do not follow the key; I'd conclude that the one on Ossett does and the one on Wootton Bassett does not, as I can't see any reason for treating them differently.

    10. David,
      I don't know what happened to my comments. When I entered the first one, the engine placed it into the main comment stream rather than into this branch. Then I removed it and I wanted to add another comment here, but it ended up as a sub-comment below my previously deleted comment in the main stream, and then I gave up, I thought you'd find out that my comment was meant to be a reply to this discussion of yours, and you might reply to it. But a couple of minutes later when I checked back to see if there were any replies, my comments were gone. I suspect the engine may have considered them spam for some reason. If so, maybe John can restore them.

    11. teardrop

      It may be connected with the removal of the dustbin icon after an author's post — to be replaced by blue underlined 'Delete'. However, the change doesn't affect the stream of messages sent to subscribers, so here's your text:

      Yes, too much information would be a burden, but where would one draw the line between allophones to be shown or not shown?

      If dictionaries were to use phonetic rather than phonemic transcriptions, they'd have to indicate other important allophonic features as well, at least aspiration of stops, dark l's, diphthongal FLEECE and GOOSE, word final /ə/ as [ɐ] or [ʌ]. This might be too much information -- more descriptive than a phonemic one, though. Besides they'd have to pick a single BrE accent, but which one? RP? Definitely not, as most EFL students will hardly hear or meet an RP speaker.
      And which AmE accent would be given? One with a cot-caught merger? Not necessarily, as it's not a majority feature yet.

      A phonemic transcription, however, could cover all (OK, most) accents provided the reader knows which phoneme represents which sound in the target accent, or their own foreign accent.

    12. teardrop

      the removal of the dustbin icon after an author's post — to be replaced by blue underlined 'Delete'

      I should have added that 'Delete' is now positioned where 'Reply' used to be. or rather that it appears to be there. I suspect that one or both of the links is doing the wrong job. It can't be human error. It should be impossible for you to delete Paul Carley's post, or for Paul to delete yours.

      Here's the text of Paul's reply:


      "most EFL students will hardly hear or meet an RP speaker"

      I disagree. I will even say that the opposite is likely to be true - most EFL students have heard an RP speaker. Speakers of modern mainstream RP and accents very similar to it are massively overrepresented in the UK media. That's the south east bias that we live with here. I'm not saying it's fair (I'm not from the south east - far from it!), but that's the way it is.

      As for meeting an RP speaker, that's probably about as likely as meeting a speaker of any other particular accent you care to name. What are the odds of meeting a speaker of Swansea English outside of Swansea? Small. About as small as meeting an RP speaker, I dare say.

    13. David,
      thanks a lot for getting them back.

      if you call what we hear in the media modern RP, then teaching this modern RP is fine. But it's quite different from the traditional RP that most grammar and textbooks still describe and prescribe, and whose phonetic symbols most dictionaries use to this day. For TradRP [æ, ʌ, ɔː] are the right symbols to use, but in ModRP they might be better transcribed with [ɛ/a, ɐ, oː], respectively. Or the TradRP diphthongs of [aɪ, aʊ, eɪ] are quite often pronounced as something like [ɑɪ, æʊ, ɛɪ] in the media, and one must not fail to notice the more remarkable diphtongs in FLEECE and GOOSE -- just to name a few vocalic differences. But a dune-June merger is also quite frequent, still a TradRP-based book or dictionary says nothing about it.

      By RP I meant TradRP in my previous comment, which might explain why I said what I said. I don't think this accent should be the model used in dictionaries anymore.

    14. teardrop

      On symbols: Personally, I'm not at all bothered by the disparity between the range of sounds our phonemic symbols stand for and their values as phonetic symbols. Only phoneticians will appreciate the difference and they should be well used to the limitations of phonemic transcription.

      On textbooks and dictionaries: Cruttenden's latest revision of Gimson describes all the features you mention and LPD3 has dune/June. Non-specialist works are simplified and far from cutting-edge, that's normal in all fields.

      I agree that an out-of-date description of RP shouldn't be the model used in dictionaries.

  3. Historically it is already a weak form; the old strong form was hit, and in dialects (like traditional Ozarks) where hit remains the usual form, we get regular strong/weak alternation: Hit's entirely new to me: I've never heard of it..

  4. And, of course, there's the old (but probably not only old) /t/, as in "'tis".

  5. I'm afraid I still don't understand what we gain in maintaining that there is no weakform for couldn't, wouldn't, shouldn't. John's explanation hinges on the fact that there is an ongoing change from the FOOT vowel to /ə/, as in sugar to use one of his examples, with the result a) that we hesitate to call /ʃəɡə/ a weakform and b) that with /kədn̩t/ we aren't able to tell whether it is a strongform or a weakform.

    If all this is correct, then why does LPD3 accept /kəd/ as a weakform albeit an "occasional" one only? Wouldn't the same arguments apply here?

    And - is it too heretic to widen the concept of weakforms and allow words other than those 40 odd ones to have weakforms?

    1. If I understood that correctly, the point was only to refute the proof from the very occurrence of kədn̩t, which could simply be the speaker's default FOOT vowel.

    2. Kraut

      Our usual use of the term 'weak form' should really be understood to mean 'weak form for EFL purposes'. We usually only list those weak forms whose occurrence can be neatly described and recommended for use by learners in 'careful colloquial' speech (if that's the right phrase).

      I think there are more than the usual EFL bunch mentioned in Gimson's Pronunciation of English and then there's Obendorfer's Weak Forms in Present-Day English (1998, Oslo: Novus Press).

  6. On reflection, I believe I have in my repertoire what might qualify as a weak form of couldn't.

    My best effort at transcribing it is kʰdn̩t — i.e.
    • no oral vowel for the could element
    • sonorant nasal release after d

    I can use it before an emphatic possibly — thought not, I think, before a lexical verb form.

    I don't think I can reduce can't in the same way.

    1. I should have added

      • no oral release between the t of couldn't and the p of possibly

    2. This comment has been removed by the author.

    3. @David Crosby: I know that it takes more than one swallow to make a summer, but then there's hope!

  7. Army, why are you against the same symbol? I think that to this day I don't really get it why is it that we use ɜː instead of əː. If you look at a vowel chart, the place of articulation is almost the same, if not the same, or perhaps sometimes ɜː is even a bit more close than ə.

    On the other hand, I am against ɒ becoming anything else than that.

    I have to say that I don't get ɚ and ɝː either, which to me more often than not seem like a syllabic ɻ, or whatever would be the correct symbol for the American ‹r›.

    1. By symbol I meant ‘the sequence of characters used to transcribe a phoneme’, not necessarily a single character, so I wouldn't object to use ə for COMMA and əː for NURSE.

  8. In other news, I have finally found the cure for comments on Phonetic Blog being in Verdana. I've installed the StyleBot add-on, right-clicked > Stylebot - Style Element and then changed the typeface.

  9. Haloo pak^^

    Kami dari SENTANAPOKER ingin menawarkan pak^^

    Untuk saat ini kami menerima Deposit Melalui Pulsa ya pak.

    *untuk minimal deposit 10ribu
    *untuk minimal Withdraw 25ribu

    *untuk deposit pulsa kami menerima provider

    untuk bonus yang kami miliki kami memiliki
    *bonus cashback 0,5%
    *bunus refferal 20%
    *bonus gebiar bulanan (N-max,samsung Note 10+,Iphone xr 64G,camera go pro 7hero,Apple airpods 2 ,dan freechips)

    Daftar Langsung Di:


    Kontak Kami;

    WA : +855 9647 76509
    Line : SentanaPoker
    Wechat : SentanaPokerLivechat Sentanapoker

    Proses deposit dan withdraw tercepat bisa anda rasakan jika bermain di Sentanapoker. So… ? tunggu apa lagi ? Mari bergabung dengan kami. Pelayanan CS yang ramah dan Proffesional dan pastinya sangat aman juga bisa anda dapatkan di Sentanapoker.


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