Monday, 16 November 2009

bank balance

Castedo Ellerman writes
I am wondering whether you hear /æ/ or /eɪ/ in the sound recordings for US English in your LPD software (3rd edition) for the word “bank”. To my ear I am hearing /eɪ/ in all of them (to varying degrees).
I have a US English accent and I find it natural to say the vowel in bank, sank, tank, or thank as either /æ/ or /eɪ/, so I think I can hear the difference.
For instance, the recording of “bank balance” from the US speaker really sounds like /beɪŋk/ … to me. Does it sound like that or /bæŋk/ to you?

I replied
I hear bæŋk, with a bit of an offglide of the ɪ type such as many Americans use before a following voiced velar nasal. But it's not beɪŋk - compare what you hear at bane.

I believe there is a phonotactic constraint in English against before the cluster ŋk. I suppose that means we could say that the opposition æ-eɪ is neutralized in the context of a following ŋk, which would mean it’s meaningless to ask which of the two we get in this context.
I am sorry I cannot extract the sound files for bank and bank balance from the CD-ROM so as to make them available to you here: but if you have the dictionary you can check them for yourself. Here is the clip in question (thanks to Petr Rösel). I am not sure which of the American speakers that we used recorded this particular sound clip, but he is evidently not one of those who uses a strictly monophthongal æ in all contexts.

I wonder how other Americans would react to Mr Ellerman’s opinion.
_ _ _

Further to my report of the death of Stanley Ellis (blog, 4 Nov), I would refer those interested to Jack Windsor Lewis’s obituary published in the Guardian (you may need to go to the front page, www.guardian.co.uk, then down through Obituaries | Stanley Ellis), as well as to what he has written in his own blog.
Edward Aveyard has drawn my attention to a press report from the Yorkshire Evening Press dated August 1953.

25 comments:

  1. I can't play the sound recordings, as the CD does not work with a Mac. (I think that's a regrettable oversight, but that is by the by.)

    But I can confirm that in most American accents, bæŋk is an accurate transcription. I have heard something closer to beŋk in some American speakers, but in only a few. I assume it is a regional variation.

    As a matter of tangential interest, in some English and Scottish accents, I hear bænk, and in Scotland I have heard the word "thanks" pronounced θaŋks; or perhaps it was θanks - this was some time ago, and it was the more open vowel that impressed itself upon me most.

    ReplyDelete
  2. – I believe there is a phonotactic constraint in English against eɪ before the cluster ŋk. I suppose that means we could say that the opposition æ-eɪ is neutralized in the context of a following ŋk, which would mean it’s meaningless to ask which of the two we get in this context.

    Isn't that opposition neutralized by the ŋ alone? Proper names are of course extra-systematic to a large extent, and therefore permit spelling pronunciations like [leɪŋ] for [læŋ].

    May I commend to all of you the extreme cases of this US eɪ which have oppositions like heɪv~hejəv for have~halve? One informant assured me that in his dialect it was ˈhejəv~ˈhejjəv or even ˈhejjeɪv! My Ulster informants would have none of it, of course. And like Leo on here, they didn’t fall for any dirty tricks!

    This is what I had concocted for them: I have an apple and if I halve it I will have two halves. Not an inch would they give on any possibility of an opposition even at the upper limit of distinctive realization. "It's 'I have an apple and if I half it I will have two halfs'," they averred. "Or at a push 'haves'."

    ReplyDelete
  3. I have often encountered students who insist on transcribing [bæŋk] as [beɪŋk]. In their own speech they tend to nasalize front vowels, and merge PIN and PEN (sorry to manufacture lexical sets). I can't really recall whether these same students merge DRESS with FACE in words like "beggar" and "measure" but it strikes me as consistent.
    Or rather, a student who has even a slight tendency towards this raising of front vowels, particularly in a nasal context, will take in the -æŋk that I speak for their transcription, and after a few internal test drives of the sound within their own ear/mind/mouth they're certain that I'm producing -eɪŋk. Phonotactic constraints seem to have little hold on their perceptions.

    ReplyDelete
  4. Thanks for the link, John and Petr Rösel. I definitely hear bæŋk in "bank" on its own, but I think I hear a little further raising of æ in "bank balance." It goes by so quickly I doubt my accuracy - that's the impression I get at the moment.

    ReplyDelete
  5. Can *any* free vowel occur before /N/? (Not counting "long" in people with CLOTH=CAUGHT, because it might itself be a case of LOT-CAUGHT neutralizing before /N/, and therefore /QN/ surfacing as /O:N/).

    ReplyDelete
  6. @army1987 /ɔi/ if "boing" /bɔiŋ/ isn't too extra-systematic for you.

    ReplyDelete
  7. Diphthongs and long vowels do no occur before [ŋ], except in a few onomatopoeic words like "boing". Well, that is true for citation forms. However, if one has an alv → velar assimilation then forget the phonotactic constraint.

    It would be interesting to find out if speakers who have an ɪ offglide in "bank" make a difference in the vowels of, say. "painkiller" [peɪŋkɪlə] and "mankiller" [mæŋkɪlə], assuminɡ they could be persuaded to produce versions with the assimilation.

    ReplyDelete
  8. Army,
    – Can *any* free vowel occur before /N/?

    I meant by the way to assert that it can't in BrE (exc for freak examples like the ones Richard and I have given). And I do realize that it's a bit more complicated in AmE because of the long 'long', as Anthony Burgess called it in Earthly Powers (not that I would choose to remember either), but the loss of opposition is the main point that JW introduced here, and of course that is just as you surmise.

    Richard,
    – if "boing" /bɔiŋ/ isn't too extra-systematic for you.

    Well it is, like /leɪŋ/, because it's in solidi, but it is definitely a phonological constraint, not a phonetic one: I'm about to take my RSI to the [drɔɪŋ] room.

    Just seen your point about assimilation, John M, but the constraint holds because of the solidi we've been using.

    ReplyDelete
  9. Wikipedia's article on California English says that /æ/ and /ɪ/ are raised to /e/ (i.e. the FACE vowel in a relatively monophthongal incarnation) and /i/ before /ŋ/ in that dialect, so not only is bank /beŋk/, but king is /kiŋ/ as well. I have met trained phonologists from California who say those transcriptions reflect their intuitions regarding the phonemes, too: that the only difference between pane and pang or between keen and king is the place of articulation of the final nasal.

    ReplyDelete
  10. John Maidment wrote:
    'It would be interesting to find out if speakers who have an ɪ offglide in "bank" make a difference in the vowels of, say. "painkiller" [peɪŋkɪlə] and "mankiller" [mæŋkɪlə] ...'

    I think I am one of these speakers and I am curious how one could NOT make a difference in the vowels of painkiller vs mankiller or pancreas, assuming n->ŋ assimilation.

    Here is a recording of me saying
    "I really need some pancreatic pain-killer.":
    http://www.castedo.com/sounds/pancreatic_painkiller.wav

    Is there anybody here who would transcribe my pronunciation of "pain-killer" in that recording as /pæŋkɪləɹ/? (never mind the əɹ)

    To me /pəɪŋkɪləɹ/ seems more approriate as /pæŋkɪləɹ/ is what I would say reading "pang-killer" (something that kills pangs, not pain).

    Is there anybody who hears/speaks these words the same:
    * pain-killer
    * pang-killer
    when the n-k assimilates into ŋk?

    I have more recordings of me contrasting /æ/ vs. /eɪ/ after bilabial and before velar consonants if anybody is interested.

    ReplyDelete
  11. I blogged about this a while back after discovering that my wife and sons had different intuitions about the vowels in bank and sink. I grew up hearing them as the vowels in bane and seen, and only after taking a phonetics class in college did I listen closely and realize the vowels weren't the same.

    ReplyDelete
  12. How is this bank/baink issue connected to the phenomenon that some people pronounce the older or standard [eɪɫ] as [æɫ] eg in detail? I used to regard this an American thing, but I think I'm hearing it more and more from RP speakers.

    ReplyDelete
  13. Realised with all the RP and notation varieties, the [æ] isn't clear. In this case, I meant the DRESS, not the TRAP vowel.

    ReplyDelete
  14. Just as a follow-up to John's link to the obituary of Stanley Ellis, there's also a lovely letter in the Guardian today written in response by one of Ellis' former students:
    http://www.guardian.co.uk/theguardian/2009/nov/17/top-gear-nick-griffin-barking

    ReplyDelete
  15. For me it's most certainly [bæŋk]. I think the "regional" accent Amy Stoller mentions is most likely the midwestern American accent, with which I became somewhat familiar while I was living in Montana. In this accent æ is raised to eɪ before velars, hence [beɪg] for "bag".

    ReplyDelete
  16. I grew up in the suburbs of Washington DC USA in the 80s and "bag" and "bagel" contrast as /bæg/ and /beɪgl/ to my ear. Here is my natural pronunciation of
    "First I bagel bags of dough into bagels and then I baggle the bagels into bags."
    http://www.castedo.com/sounds/bags_of_bagels.wav
    (I've added the fictitious verbs "to bagel" and "to baggle")

    To my ear hearing myself say "bang" and "bank" contrast in the same way as "bag" and "bagel", although i can try to say "bank" as though it is "bang" followed by a "k" but my natural inclination is to say "bank" similar to how I would say "pain-killer" with n to ŋ assimilation.

    I'm still curious to hear from anybody who does not speak like me how they say "pain-killer" vs. "pang-killer" and "mane-killer" vs. "man-killer" with all n's before k assimilated to ŋ's. Or perhaps such speakers never assimilate n's into ŋ's when following a /eɪ/ dipthong?

    ReplyDelete
  17. Lippers (I now see Lippo was wrong: too many ps. I do hope you saw my reply to your last post on diphthong, schmiphthong yesterday, about this lambdacism being a shared affliction!)

    – How is this bank/baink issue connected to the phenomenon that some people pronounce the older or standard [eɪɫ] as [æɫ] eg in detail? I used to regard this an American thing, but I think I'm hearing it more and more from RP speakers.

    It seems to be a diametrically opposed phenomenon, doesn’t it? With either the DRESS or the TRAP vowel, but especially the latter, it strikes me as being another of the mockneyfications you were talking about, deriving from [æo].

    But what this detail does seem connected to is what I was saying about 'pale' on wholly holy. It's closer to an [eː] monophthong that to DRESS, but Daniel Jones drew attention to this allophone of [eɪ] before ɫ
    quite early on.

    ReplyDelete
  18. I think my "pale" is more like [peəɫ], with [eə] an allophone of FACE.

    I've occasionally misheard other people's DRESS as FACE in this environment, e.g. I once heard "Wells" as "Wales".

    ReplyDelete
  19. It seems to be a diametrically opposed phenomenon

    Or the same, if one says the vowel/diphthong depends on how "palatal" or "velar" the following consonant is.

    But frankly, I'm not sure this [æ] occurs in front of [ɫ] only, or if people say ret[æl]er just as well as ret[æɫ]. (Or ret[ɛl]er, ret[ɛɫ].)

    ReplyDelete
  20. OK, the same in reverse.

    But I don't think people do say any of those things. Except perhaps with compensatory lengthening: ret[æ:l]er. But [ˈriːtæolə] sounds likely, and I think we would get the same sort of spectrum as we had for wholly~holly, depending on whether people have the Janus effect that John M and I were pushing as an explanatory model on the 'weak u and ə' thread.

    So quite consistently in my case, the diphthong comes back in [ˈriːteɪlə].

    ReplyDelete
  21. I meant the *phoneme* /N/, so the allophone of /n/ doesn't count. As for onomatopeias, I'd be tempted to dismiss them as extra-systematic, but "oink" has an inflected form "oinking" with 53k ghits, so I think I shouldn't.

    ReplyDelete
  22. Of course you did, army. You put it in solidi. I didn't bother with them when I first said above "Isn't that opposition neutralized by the ŋ alone?" because I have observed that people don't use brackets on here when they can be deduced, as they can from the context "Isn't that opposition neutralized". But I sympathize if you are annoyed that your explicit solidi were misunderstood, as you can see from my lecture here to a poster who seemed to think neutralization could be refuted by purely phonetic considerations.
    You'd think I'd made it clear enough on this thread too, as I said above:
    "… it is definitely a phonological constraint, not a phonetic one: I'm about to take my RSI to the [drɔɪŋ] room.

    Just seen your point about assimilation, John M, but the constraint holds because of the solidi we've been using."

    Good idea to google the onomatopoeias!

    78,400 for boings
    81,200 for boinged
    94,100 for boinging

    But in the nature of things we should always have suspected accidental gaps for diphthongs. It is indeed the free vowels for which it's virtually impossible to imagine any phonologically distinctive function before /ŋ/, whatever the dialect.

    ReplyDelete