Wednesday, 23 February 2011

mysteries of existence

Like other teachers of elementary phonetics to NS students, when teaching I would regularly set exercises in “doing transcription”. I would give the students a passage of English in ordinary spelling. Their task would be to convert it into phonetic transcription.

This is a valuable exercise for NSs as much as for NNSs. It familiarizes them with the phonetic symbols. It makes them conscious of the difference between spelling and pronunciation. It alerts them to the characteristics of connected speech as opposed to individual words in isolation.

Good students sail through this task. The weaker ones often find it remarkably difficult. I would frequently have to point out that write does not actually begin with a w-sound, nor looked end with a d-sound (still less a syllabic ). The first vowel in particular is not normally pronounced ɑː, and the second vowel in information is not ɔː. Orthography has a distressingly dazzling effect on the phonetically unsophisticated.

But setting and marking (AmE ‘grading’) transcription can also be valuable for the teacher. Early in my career I noticed some students transcribing exist as ɪkˈzɪst (and similarly with example, exhausted, exams etc). Since the usual pronunciation is ɪɡˈzɪst I would mark this wrong. (Initial e-or ə- rather than ɪ- is OK, of course, though you have to check that that’s what they genuinely say, not just spelling-driven.)

However some students protested: they really do pronounce the word as ɪkˈzɪst. I checked it out, and they appeared to be correct. I came to realize that some speakers in the southeast of England, at least, have an unexpected dissimilation of voicing in these words. Their kz in exist seems to be different both from the gz of eggs or big zits and from the ks of exceed.

Being now a fortis plosive, it is also a candidate for glottal reinforcement: ɪkʔˈzɪst.

Convinced now of their reality, I decided to include these variants in LPD. But no other dictionary seems to recognize their ɪkˈzɪstəns.


  1. I'm very suprised at this, and I've never noticied it myself. It seems to go against the principles of phonetic change too, in that /kz/ takes more articulatory effort than /ks/ or /gz/.

  2. I can confirm that a lot of my students transcribed thus and that I too first took these transcriptions to be errors, but was convinced otherwise eventually.

    John: did you mean
    for the ɡlottally re-inforced version, or should it have been

  3. Marijana Krleža23 February 2011 12:44

    Does that mean that there is a new edition of LPD coming?

  4. JM: Impressionistically, the glottal closure masks (renders inaudible) the velar release. Not sure how best to transcribe that.

    Marijana: No. These variants have been in LPD since the first edition (1990).

  5. I would think this an extremely difficult assignment to grade for anyone not as familiar with the details of NS accents as you are, for precisely this sort of reason. To be stringently fair it would be necessary to validate what each student wrote against what the student says, a time-consuming business.

    Apropos of which, I think people whose schooling is divided between two or more anglophone countries, especially if one of them is the U.S., must have an extraordinarily hard time. Having been taught, or having taught myself, to write competent American English, I would find it very difficult to try to write British English correctly: orthography, syntax, lexicon, and idiom would be tripping me up at every turn. (Per contra, to understand AmE completely you must know the full implications of "Johnny went to the bathroom in his pants.")

    It seems to me that writing BrE would be scarcely easier than trying to write in a language I didn't know, grammar and dictionary in hand. What do those of you who teach professionally do with papers written in a foreign variety of English?

  6. I've definitely heard the kz pronunciation often enough not to consider it too remarkable. To me, the voicing of the /z/ component of 'x' is the salient feature of these words.

    I wonder if the words with 'voiced x' actually passed through a stage where they represented /kz/ before they became /gz/ through voicing assimilation. In French, regressive voicing assimilation is so strong that an underlying /kz/ would be pronounced [gz] anyway.

  7. Jongseong, I confess that until I read your suggestion about 'voiced x' I hadn’t realized what a puzzle it is. But any relatively recent regressive voicing assimilation such as you hypothesize seems so unlikely in English that I'm still inclined to think that /kz/ is the innovation. The k does always seem to be glottalized, whereas /gz/ usually seems to have some degree of release. It could be just another example of galloping glottalization in BrE. And it's glottalized throughout, which is why I don't think either ʔk or kʔ is really satisfactory. I think it would probably be best to represent them as co-articulated: k͜͜ʔ, for example, if you can see that. As a Korean I don't suppose you'd be happy with kʼ. Another indication of its recency is that it certainly seems to be age-related. And it must have happened quite rapidly. I find it hard to imagine that the two Johns who marked it wrong were unaware of it even then. I'm almost their age and it seemed to have overtaken a lot of my generation even then.

  8. John Cowan,
    > What do those of you who teach professionally do with papers written in a foreign variety of English?<

    What I did, sadly late in life, was become far more permissive and ready to believe that what seemed like foreignisms to me might not do so to other NSs, and even to believe in the Englishes of NNSs. For so long I was ignorant of the extent of my own ignorance of the things of which you speak in anything but BrE and a smattering of AmE like "Johnny went to the bathroom in his pants", which I think I did appreciate the hilarity of from my earliest youth. I didn't watch Westerns because I really did have great difficulty understanding them. I felt I had good evidence for the belief that contemporaries who were fans only thought they were understanding them, but no doubt all the time they were acquiring competence that I wasn't. We never thought of Americans as foreigners, but I later came to realize that some of the "foreignisms" that I thought non-native were actually Americanisms. And there were, and still are, an awful lot of shockers like me, academics or not.

  9. Mallamb: Ah, I don't have a sense of how this pronunciation correlates with age. If it is age-related then it does seem more likely that /kz/ is the innovation. If this is indeed related to glottalization in BrE, does it affect phonetically similar combinations that are unrelated to latinate 'x' like 'eczema' (in case one voices the 'z') or 'zigzag'?

  10. Jongseong, Aha indeed. I should have thought of 'eczema'. I have in fact thought of it before, reflecting on the fact that AmE has /ɪgˈziːmə/, and thinking it might be because of the different stress. With initial stress I thought anything but /ˈeksɪmə/ unimaginable, but earlier hardback Oxford dictionaries have nothing but transcriptional equivalents of /ˈekzɪmə/, which given their venerability I suppose can hardly have been a spelling pronunciation, though I would suspect it of being one in contemporary English. Both the old and new versions of the online OED still only have that, so they do recognize the existence of /kz/, even though as John says they don’t recognize the existence of ɪkˈzɪstəns etc., strangely only giving Northernisms like ɛgˈzɪstəns and ɛgˈzæm.

    I see LPD reassuringly gives UK ˈeksɪmə and US ɪgˈziːmə first, with ˈeksəmə as an alternative for both, but UK ˈekzɪmə and US ˈegzəmə are right there too.

    I think 'zigzag' is immune for all sorts of reasons: phonetic symbolism, 2-phonotagm structure/secondary stress etc.

  11. As for "eczema": don't forget BrE ˈeksmə, as heard here (skip to 1:29).

  12. And surprisingly, you might think, ˈeksmə may be heard in the UK sound file in LPD, tho it's not acknowledged in the transcriptions.

    ˈegzmə, anybody? Somehow I don't think so.

  13. John Cowan said: "Johnny went to the bathroom in his pants." I never realized how funny that must sound to foreigners until now. Of course, we're really just trying to avoid saying, "Johnny pissed his pants" when we say that. We only say that when we're trying to be extremely polite and/or if young kids are in the room. But you're American so you already knew this. Maybe everyone else here did too. I don't know.

    @ Mallamb: I always hear /ˈeksəmə/ here in America, so that's what I say. I've never heard /ɪgˈziːmə/. Not that anyone asked. It's just that I've come here before and I've seen pronunciations that people say are American pronunciations, but I haven't heard some of them and I don't want people to get the wrong idea of how Americans speak. I'm not sure why this bothers me. But then again, I don't have this condition so maybe I haven't heard other people, e.g., dermatologists, pronounce the word as much as those who do suffer from it.

  14. It may be that I can regale you beyond your recent revelation of how funny "Johnny went to the bathroom in his pants" must sound to foreigners. To more than a few of us foreigners, "pants" are underpants, and therefore the surface interpretation of the expression seems less implausible: he paraded to the room with the bath in it (and quite likely no purpose-built pissing facility in it at all) in his underpants, rather than the more decorous attire that trousers (AmE pants) would have been.

  15. 'Of course, we're really just trying to avoid saying, "Johnny pissed his pants" when we say that.'

    Really? I took it to mean 'Johnny cacked his kecks' - that is, solid material was involved.

  16. My posting was about voicing dissimilation! Your comments seem to be drifting off-subject even more than usual.
    Any moment now someone will be nipping outside to grab a quick fag.

  17. The four American dictionaries I usually look at, namely NID3, RHD2, AHD4, and, all list the pronunciations /ˈeksəmə/, /ˈegzəmə/, and /ɪgˈzimə/; NID3 also lists /egˈzimə/. I myself say /ˈeksəmə/. Presumably the dictionary makers have the weak vowel merger (and so do I).

    Graham Asher: It can mean either or both. Furthermore, there is an implicature that the subject is a child rather than a (presumably impaired) adult.

  18. The above comment is by me; I've borrowed my daughter's computer.

  19. I have ˈɛksmə for "eczema".

    @mallamb: the OED online entry for "eczema" is still a second edition entry, so maybe the third edition will have more variants (hopefully including ˈɛksmə). (And why shouldn't the OED give "Northernisms"?)

    I've never noticed these /kz/ pronunciations; while I guess that's because I'd have to be listening quite carefully to notice, how common are they?

  20. I see that, for exhort, the OED gives the odd pronunciation ɛgzˈhɔːt. I would have thought that if you are pronouncing the h, there would be no voicing of the x: ɛksˈhɔːt.

  21. My daughter has had eczema (luckily only a mild case), so I've heard the word spoken by quite a few medical staff here in Northern California.
    I've heard /ˈɛksmə/ and /ˈɛksəmə/, but never anything so exotic as /ɪgˈziːmə/.

    I myself (raised in England) have /ˈɛksmə/: this is definitely a word I learned through spoken language rather than spelling.

  22. JHJ,
    I have to keep the confessional ball rolling: so many of us are acknowledging that we have not been as observant as we would like to think. I had never consciously heard ˈeksmə before I heard the UK sound file in LPD, and as I said it's not acknowledged in the transcriptions there. I have often noticed how the readers for these sound files seem to be totally innocent of the phonetics and norms, and that was no doubt John's intention. But until I read Leo's post about BrE ˈeksmə (which vp now confirms for AmE as well) and was completely convinced by the clip he linked to in it, I thought the LPD sound file might be just a slip in performance, or blip in recording. I had to check it again before I posted to tell him that that was ˈeksmə too, and to trawl for ˈegzmə, which it seemed to me might still be outlandish enough to remain unattested. Still no takers for that.

    By 'the OED online entry for "eczema" is still a second edition entry', do you mean both the old and new versions of the online OED that I referred to? The old one just says "Second edition, 1989", and the new one says "Second edition, 1989; online version November 2010", so I supposed that may be a new version, if not a new edition. Does it just mean that was when the new version of the software was put online? It does have a link "Previous version" to a page headed "From the second edition (1989):" which also gives ˈɛkzɪmə, and that does look just like a prettified version of the old online version.

    I don't for a moment think the OED shouldn’t give "Northernisms", just as LPD does, though it marks them with § as far as I can see, but don’t you think it's odd that "Northernisms" are all the OED gives in the exposition of these particular existential mysteries?

    You ask of these /kz/ pronunciations "how common are they?" I'm afraid that question runs a very real risk that someone will answer "As muck"!

    OED's ɛgzˈhɔːt is completely loopo, isn't it? Don't suggest ɛksˈhɔːt. It'll only encourage them. LPD often recognizes credible spelling pronunciations like eksˈhɔːt, but reassuringly has only pronunciations with z and absolutely no h.

  23. The OED entry for eczema appears to be unchanged from the first edition (this volume edited by Henry Bradley, 1891), apart from the conversion of the original phonetic transcription (e·kzĭ) to the IPA. It's Bradley also who's responsible for the alleged ‘northernism’ ɛgˈzɪst for ɪgˈzɪst (e rather than ė in the original).

  24. I'm pretty sure that when it says "Second edition, 1989; online version November 2010" that's just a second edition entry reformatted for the new website, and not any sort of revised entry. (The revised entries say "Third edition".)

    My understanding is that second edition pronunciation information is usually essentially the same as in the first edition but retranscribed in IPA (a different IPA-based system from Upton's system used in the third edition, but one which does lose some of the distinctions marked in the first edition). So I wouldn't expect second edition entries to be particularly up to date pronunciation-wise.

  25. Thanks, both of you. Steve, did you see the protracted discussion between Lipman and me about these 'ex-' entries? It seems very relevant to this thread: - c9020905802489021827