Friday, 18 November 2011

winter and winner

Maybe I’ve just not been keeping my eyes open, but I can’t recall reading any surveys of the prevalence or otherwise of what I would like to call nt-reduction.winterBy this I mean the loss of t from the cluster nt in intervocalic contexts. This makes winter a possible homophone of winner,
painting a possible rhyme of straining and dental a potential rhyme of kennel. As far as I know this is not found in any kind of British speech, and we think of it as an American or Australian characteristic.winner

The possible AmE pronunciation of continental as ˌkɑ̃ːʔn̩ˈẽnl̩ is quite strikingly different from the BrE ˌkɒntɪˈnentl̩.

Several qualifications are needed.
• In the kind of AmE I am referring to, winter may possibly have a nasalized tap, thus ˈwɪɾ̃ɚ, rather than the more deliberate plain nasal of winner ˈwɪnɚ. Trager and Smith (1951) refer to this as a ‘flap-release short nasal’, how accurately I am not sure. In any case, a distinction based solely on ɾ̃ vs. n cannot be very robust. I suspect that in reality for many Americans (and Australians) winter and winner can be, and often are, pronounced identically.
• I have the impression that the incidence of nt-reduction is subject to regional variation in the US. It seems more prevalent in the south and west, less so in the north-east. Is this so? Do Canadians ever do it? It is also probably subject to stylistic variation, with unreduced nt more careful and the reduced variant more casual. Has anyone ever investigated its sociolinguistic characteristics?
• The environments in which nt-reduction operates seem to be the same as those for t-voicing. In particular, it does not happen in the environment of a following stressed vowel, as in intend, contain, nor of a following unstressed but strong vowel as in intake; nor does it apply to ntr clusters, as in country. The t can be lost in centre/center but not in central.
• Some words may be special cases, In particular, I have the impression that ninety in AmE is often ˈnaɪndi rather than the expected ˈnaɪnti or ˈnaɪni. Does the same apply to seventy? Are there other exceptional cases?
• Special cases of a different kind are the handful of words in which a similar reduction is found in BrE, namely in London and some other kinds of popular English. For Brits who do this, the t can be lost from twenty and plenty, and from prevocalic went and want (as in went out, wanted), but not from words such as winter or painting.

This posting was triggered by my hearing an Australian golf commentator on TV referring to ðə ˌsevn̩ˈiːnθ the seventeenth (hole). This violates the constraint barring nt-reduction before a stressed vowel, and I suspect would not be possible in AmE.

Furthermore, I wonder whether Australian English has taken nt-reduction direct from AmE, rather than via some British source? And if so, is it the first instance of such a sound change?


  1. I am American, and I say "sevendy" for "seventy". It must have something to do with the voicing of the previous consonant, because I keep the unvoiced 't' for fifty and sixty, but not twenty, thirty, forty, seventy, eighty, or ninety.

    I was going to suggest that nt->n reduction only occurs after stressed syllables (hence maintaining the t in 'carpenter' and 'seventy') but 'ninety' goes against this. Maybe it's by analogy with thirty, forty, seventy, eighty (but then why reduction in 'twenty'?). Or maybe it only occurs after certain vowels - e.g. it doesn't occur after the 'ai' in ninety. But off the top of my head I can't think of another word with that sequence of sounds...

  2. I speak this kind of AmE, and I am a northeasterner. The distinction between winter and winner for me is flap vs. stop, and seems perfectly stable to me. Same story with dental and kennel: they certainly do not rhyme. However, painting has a true /t/ which appears to belong to the second syllable, though it may or may not be aspirated. This may have to do with the fact that -ing is not reduced (I have the Weak Vowel Merger) whereas -er and -al are.

    Seventy, ninety, eleventy (as in Bilbo's eleventy-first birthday) all have /d/ for me. *Tendy would be the same if it existed, so I think this is a property of the numerical suffix -ty. Seventeen, nineteen keep their aspirated /t/. However, twenty and plenty have only /n/ even in lento speech; pronouncing /t/ would be affected.

  3. I'm British and have /d/ in "seventy" (but not in "ninety" or any of the others).

  4. "I wonder whether Australian English has taken nt-reduction direct from AmE, rather than via some British source" - it is of course very well possible that the Australians came up with this themselves, independently.

  5. As regards its spreading to British English, I have noticed this amongst younger people in cases involving the word "want":

    1 "Wanted" as [wɒnɪd ~ wɒnəd]

    2 "Want to" as [wɒnuː]

    There is also "twenty" as [tweni], which can be heard from old Brits as well. Perhaps it will spread here. The phoneme /t/ is changing rapidly.

  6. I'm an American from the Midwest and like John Cowan I have /d/ in seventy, ninety and eleventy.

  7. Ed's remark might be valid even for some older Brits :

    Indeed, quite some time ago The Beatles already sang "I wanna hold your hand"...

    Jérôme Poirrier
    Grenoble, France

  8. Ed, Jérôme: that's what I said (the last bullet point)! I even have a Powerpoint screen for one of my lectures setting this out. But I think it's lexically restricted to just these items.

  9. I'm an Australian, and I think it's worth reiterating that Australian English has flapped /t/ in the same positions as AmE. I've always assumed these are related.

    Dropping the /t/ in "seventeenth" (and similar forms) is, I think, optional but restricted to that range of numbers.

    And I'm not aware of any difference between the n and nt; they're exactly identical to my ear. (However, /nt/ triggers [æ:] whereas /n/ does not in underlyingly intervocallic words.)

  10. Oh by my first paragraph I meant to make that comment in relation to your curiosity about its origins. I think you'll find a common origin for both, so if you find t-flapping a British source of Australian English, then it's probably nt-flapping comes from the same source.

  11. Sorry, John. That was quite stupid of me. My memory's obviously getting quite short-term.

  12. Wanna and gonna are special cases, not general phonological reduction of -nt-. As is well known, a syntactic trace (a point from which underlyingly present material has been moved elsewhere in the sentence) between want and to blocks contraction: thus Why do you wanna leave is fine but *Who do you wanna leave is not.

    1. "Why do you wanna leave is fine but *Who do you wanna leave is not."
      I disagree, if one says [huː də jə ˈwɑːnə liːv] (male 49 years old, native of Mattituck on Eastern Long Island, NY)

  13. 10 and 90 have just an n sound.
    30, 40, 80 have a flap t.
    50 & 60 have a t sound.
    70 has a d sound (seventy rhymes with candy)

  14. linguala: and 20?
    The only multiples of ten relevant in the current "nt" discussion are 20, 70, and 90.

  15. To answer one of your questions: Yes, we do it at least in Western Canada.

  16. As a native speaker of American English who grew up in the Chicago suburbs, I have for the numbers in question

    ten: /tɛn/
    twenty: /ˈtwʌni/ (along with "plenty" /ˈplʌni/. "plentiful" is still /ˈplɛntəfəl/ though.)
    thirty: /ˈθɜrdi/
    forty: /ˈfɔrdi/
    fifty: /ˈfɪfti/
    sixty: /ˈsɪksti/
    seventy: /ˈsɛvəndi/
    eighty: /ˈeɪdi/
    ninety: /ˈnaɪndi/

    The assignment of the intervocalic flap to /d/ is somewhat arbitrary, but I feel that I internally keep the phonemic distinction between /t/ and /d/ in the relevant intervocalic flapping position enough in most words that I'm more likely to produce the allophonically shortened vowel that I have before all unvoiced consonants before a tap derived from underlying /t/ than from /d/ in most words, but not in the numbers under consideration.

  17. I'm a speaker of a rhotic, moderately Eastern New England dialect in Massachusetts (with full intervocalic /t/ flapping), and I'm pretty sure that I reduce /nt/ a lot less than most North Americans do. I have it in "twenty" /ˈtwɛni/ and "plenty" /ˈplɛni/ (with "seventy" being /ˈsɛvən(d)i/ and "ninety" being /ˈnaɪndi/), in a few other words like "gentlemen", and after things like "want" and "isn't", but the elision of /t/ in most lexical words is pretty alien to me. I never say "winter", "dental" or "inter-" without a fully articulated [nt].

  18. @ rirelan: I'm from a place that's probably about a 5 hour drive away from you and the stressed vowel in twenty definitely isn't /ɛ/ for me either. I forgot to mention that before (not that it's particularly relevant to today's topic). It's something between /ʌ/ and /ʊ/ for me. Maybe it's /ʌ/, but the preceding /w/ gives it a different quality. I'm not sure about plenty though. I feel like sixty is /ˈsɪksdi/ for me, but I don't think it makes a difference in that environment, does it?

  19. I'd have to agree with Lazar here. I'm from the exact same location, Eastern Massachusetts, about 40 mins from Boston, and I feel the same way about all the words he just said. I was never sure if it was an accent thing or just me, but this seems to support the fact that it's related to where I live.

  20. I have /d/ in thirty, forty but not eighty, perhaps because eight provides the /t/.

  21. This reminds me of a song I heard (in AmE) in which "dentist" rhymed with "hygienist". As "hygienist" is /haɪˈdʒiːnɪst/ for me, this seemed especially weird (as did the song!).

  22. On a similar subject, I've just been listening to Coldplay's What If, where Chris Martin rhymes "maybe you'll get what you wanted" with "maybe you'll stumble upon it".

  23. What about "rt"? I find it difficult to differentiate (when speaking) "mortal" and "moral".

  24. I wonder if one reason it doesn't happen so much in the UK is that there are a lot dialects which already replace that 'nt' sound with a glottal stop.

  25. Replying to John Cowan's comment that "Who do you wanna leave?" is not fine. It doesn't work for me in written English, but in spoken English it's fine, as is "Who are you gonna leave?"

    As for the original topic, t-deletion, I once used inner as a prefix in place of intra-. They do, after all, mean basically the same thing. I got corrected. The issue, pointed out to me, was the inner sounds just like how inter- is sometimes pronounced. Thus the need to use intra- rather than turning inner into a combining form.

  26. I have /d/ in stand-alone "ninety" but for compounds, eg 91, 92, 93, 90th, etc, I say them with the nasalized tap (North Jersey, Gen Y)

  27. Like Sil, I also find it very hard to distinguish between some realizations of "moral" (on the one hand) and "mortal" (on the other hand) in American English.


    Do British speakers feel the same, or is it just a problem of Sil's and mine? (Should I say "mine" or "me" here?)

  28. The crucial difference in AmE is that "moral" has an approximant [ɹ], while "mortal" includes a tap similar to Spanish /r/ [ɾ]. We'd better have a discussion of AmE /rt/: watch this space.

    In BrE "moral" ˈmɒrəl and "mortal" ˈmɔːtl̩ are much more distinct from one another.

  29. Unlike the other two commenters from Massachusetts, I definitely have nt-reduction in such words as "winter" and "inter-". I think following syllabic /l/ ("mental", "dental") disfavors nt-reduction for me, but I wouldn't be surprised if I do it in less-careful speech; and "gentlemen" is exceptional in having a high rate of nt-reduction. I have [d] in "ninety" and "seventy", but "seventy" can also be subject to reduction in fast speech.

    With regard to "Do Canadians ever do it?", nt-reduction is actually a stereotype of sorts for the Toronno accent.

  30. The Tronno accent, surely.

    Would I be right in guessing that considerations of machismo are involved in ˌsevn̩ˈiːnθ not conforming to the constraint barring nt-reduction before a stressed vowel? Or in the no less startling ˌsevn̩ʔˈiːnθ in BrE for not conforming to a similar constraint? And now on the radio I hear kɑːˈʔuːn for 'cartoon'. I suppose anything is possible.

    So John,
    In your blog for 27 September 2011 "Details", you discuss the realization ˈdɪiʔɛ̈oz in your running coach's strong working-class south London accent, pointing out that in AmE t-voicing is blocked if the following vowel is unreduced and there is no word boundary, so that you get it in 'later' ˈleɪɾə(r) and 'my late uncle' maɪ ˈleɪɾ ˈʌŋkl̩, but not in 'latex' ˈleɪteks, or of course 'detail' ˈdiːteɪl. Never mind before primary stress as in 'cartoon'! I'm consumed with curiosity as to whether your coach can also do t-glottalling in 'cartoon'.

    On that occasion I pointed out that in LPD3 the syllabification sometimes does the trick. In the entry for 'latex'
    latex ˈleɪt eks latex|es ɪz əz
    the syllabification explains the clipping of the vowel of the first syllable (?unnecessarily, especially for AmE) but not the blocking of the t-voicing in the second.

    But the LPD3 syllabification ˈdiː teɪ ə l for 'detail' explains the American non-t-voicing but not the coach's glottal stop, for which you would need to recognize the syllabification ˈdiːt.eɪᵊl. But kɑːt. ˈuːn strikes me as the last straw.

    When Paul Kerswill asked about 'aks' for 'ask' re "cockney then and now", 2 July 2010
    I said the funny thing is that so many words with the same (or similar, e,g, 'wasp') synchronic structure have diachronically been so subject to this metathesis, and 'aks' is merely a celebrated survivor.

    I now wonder whether that's the whole story: I find I have no glottalization anywhere in 'ask' or 'asking', but the usual variance in mɑːʔks vs ˈmɑːksɪst. If glottalization is so macho I'm sure ˈmɑːksɪst has to be ˈmɑːʔksɪst, but 'ask' and 'asking' can hardly be glottalized without metathesis to ɑːʔks and ˈɑːʔksɪŋ. So ɑːʔks may have been reborn out of pure machismo!

  31. mallamb: I rather suspect he could do t-glottalling in "seventeenth", "cartoon" and "guitar". But not in "attack" or "detest". Mysterious.

  32. Ellen K.: I realize now that my example is a poor one, because it is semantically ambiguous. Who do you wanna leave? in the sense 'Who do you want to go away from?', i.e. Who do you wanna leave (t) (where (t) designates the trace, and refers to the same thing as who) is unexceptionable: it calls for an answer like I wanna leave George.

    My apologies if this explanation is over-terse.

    But *Who do you wanna leave? in the sense 'Who do you want to go away from you?', where the answer would be I want George to leave, is impossible. The underlying form of this question is Who do you want (t) to leave, and the (t) blocks wanna-contraction there.

  33. I live in Toronto and grew up in Western Canada. The 'nt' reduction certainly occurs in this country, but I believe it is far less widespread here than in the US. On American television, for example, words like "gentle" with the reduction always strike me; I'd never say that without the aspirated [t], even in the most rapid and casual conversation. It seems particularly odd before a syllabified consonant like [l̩].

    I'm also not sure that the clichéd pronunciation of "Tronno" qualifies as the same phenomenon, since John explains that it doesn't occur before an unstressed but strong vowel. "Tonto" and "pronto" with the reduction are not anything I've ever heard in Canada. I think "Tronno" is therefore one-off of lazy delivery.

    On a similar topic, I was watching a documentary about the Coca-Cola company, where every US businessman interviewed pronounced the word "bottler" as two syllables: [bat ̚lɚ], instead of the (near) three-syllable [bɑɾᵊlɚ] I would have expected. The latter is the only way I would imagine the word would be pronounced in this country.

    How about "gentler", then?

  34. nedecky: My impression was that most Torontonians have a schwa in the last syllable of "Toronto" most of the time.

  35. Lukas: most Torontonians say [tɹɑnoʊ] or [təɹɑnoʊ]. When people make fun of this pronunciation, yes, you get something like [ʧɹãnə].

  36. …same story with "Santa". Seems like this is another reduction that is prevalent in the States, but less so here.