Tuesday, 22 March 2011

how do we pronounce train ?

Students of phonetics who are NSs of English are regularly given the task of “doing transcription”, i.e. transcribing a passage of English into phonetic notation. This task may not be as easy as it would seem at first sight. The gifted get it right first time, but the average student is tempted to make frequent mistakes. Many can be attributed either to being dazzled by the familiar orthography (so wrɒŋ, for example, instead of rɒŋ), or to failure to take account of weakening and other features of connected speech (e.g. ˈtuː ɒv ðem hæv ˈfɪnɪʃd instead of ˈtuː əv ðəm əv ˈfɪnɪʃt).

The teacher who sets the transcription task has to exercise a certain discretion when correcting the student’s work. Students are encouraged, after all, to represent their own accent rather than some variety different from their own (though in Britain we normally allow them to choose to transcribe RP if they prefer, even if it is not exactly their own accent). Where do we draw the line between a mistake on the one hand and a permissible local-accent feature on the other?

Consider someone who transcribes train as tʃreɪn, drink as dʒrɪŋk etc (which is by no means unusual).

What I would do faced with this is to draw the student’s attention to the fact that the usual transcription is treɪn, drɪŋk. I would mention the allophonic rule that makes English r fricative in the clusters tr and dr, and also the corresponding allophonic rule that retracts t and d when followed by r. I would then try to get the class to discuss which analysis is right, and why.

Do the initial clusters of train and drink contain any phonetic matter that cannot be attributed to the fricative r that we expect in this context? Is the friction element followed by the r element, or simultaneous with it?

The concept of phonological neutralization is relevant here, though sometimes difficult for practically-oriented students to grasp. The point is that there is no possible contrast in English words between a posited cluster tr and a posited cluster tʃr. This means that the opposition between t and is neutralized in the context _r, at least word-initially, and likewise the d - dʒ opposition. So in a sense it is meaningless to ask which is involved.

It may be relevant to consider the pair century – sentry. Obviously, century is basically ˈsentʃəri and distinct from sentry ˈsentri. However, like other words with this phonetic structure, it is subject to optional compression in the form of the loss of the schwa, leaving ˈsentʃri. Is this still distinct from sentry?

If the answer is no, they are not distinct, it confirms our diagnosis of phonological neutralization. If it is yes, they are distinct (which it tends to be), then we ask whether ther initial affricate of train is like the -tʃr- of compressed century or like the -tr- of sentry. It is like the latter, and we transcribe accordingly.

You will now see why I was not convinced by Joshua Smiles, who wrote to me asking
I wonder whether you might know the name of a phonological change occurring in London speakers of RP (and those who simply watch too much television). What I refer to is the shift of any alveolar plosive preceding a rhotic consonant to a post alveolar affricate. Examples of this would be "tripoli", and "children" which are often rendered /tʃɹɪpəli/, and /tʃɪldʒɹən/ in RP and EE alike.

I don’t believe there is any such phonological change in progress. So naturally I don’t have any name for it.

42 comments:

  1. I'm pretty sure that for me the answer to the century/sentry question is no: a compressed "century" would indeed end up the same as "sentry"; similarly the endings of "wondrous" and "dangerous" sound the same. So I'm still unpersuaded that my intuitive analysis that the initial phonemes of "train" and "drink" are /tʃ/ and /dʒ/ is wrong.

    NB I'm neither an RP nor an EE speaker, and I don't think I "simply watch too much television" either.

    ReplyDelete
  2. They probably see it as insane. 'Why would I transcribe it as /tr/ when no one really says [tr]?' etc. To them the principle of simplicity and recognizability of symbols makes no sense and annihilates the whole purpose of the IPA alphabet. Correcting /ʧr/ back to /tr/ would be the same as correcting /oʊ/ back to /əʊ/ when even in RP it was once /oʊ/ (not to mention /ə/ being wrong and /ɜ/ being probably more suited). And then the flood comes: /ʌ/ vs. /ɐ/, /a/ (which, let's be honest, no one really ever pronounces in RP, though as open as the Mir mine pit, it's still a mixture between /a/ and /e/, thus /æ/)... ... ...

    That makes me wonder: why do you sometimes use /u/ and sometimes /ʊ/? Does an /u/ represent a shorthand for either [ʊ] or [u(ː)], depending on the speaker, or do people really always say [u]? Why is it /pəˌtɪkjuˈlærəti/, but /pəˌtɪkjʊləraɪˈzeɪʃən/ in LPD? Same for 'Europe' and 'European'? I think Peter Roach-edited Cambridge dictionary is clear on it in the preface, he gives this a paragraph (also yours and his syllabification is wildly different).

    ReplyDelete
  3. Furthermore:
    - in my accent compressed "entering" does not develop the [tʃ]-like sound in the same way, i.e. it is not like "sentry" (or indeed "entry").
    - I had a friend at school called Patrick whose name was abbreviated not to "Pat" or "Paddy" but to "Patch".

    ReplyDelete
  4. P. S. Why would it be ˈfɪn ɪʃd as a 'strong' form when LPD has it as ˈfɪn ɪʃt?

    ReplyDelete
  5. "century" and "sentry" are homophones for me (BrE, but not EE or RP). Representing initial tr with /tr/ instead of /tʃr/, seems as counter-intuitive to me as it would be to represent initial shr as /sr/ not /ʃr/, just because initial [sr] doesn't occur in English (excepting some pronunciations of "Sri Lanka").

    ReplyDelete
  6. A quick reply to Silvia's points, not relevant to the topic of today's posting:
    1. In LPD I use the symbols i and u systematically for the weak neutralization of iː-ɪ and uː-ʊ respectively. I also apply the theory of syllabification explained and justified here. This accounts for the difference (which I believe to be real) in pairs such as stimulus-stimulation and the one you quote. I may be wrong, but everything in my analysis is very carefully thought out and, I hope, consistent. (For Peter Roach's syllabification, which is different, see 2.6 in the Cambridge dictionary.)
    2. Beginners tend to think of the -ed ending as always d. Hence this common mistake. I didn't and don't call it a strong form.

    ReplyDelete
  7. Since I saw someone mention ʊ, may I ask: is it possible to say ˈsepʊlkə and seˈpʊlvedə? Or is that entirely incorrect?

    ReplyDelete
  8. If RP is supposed to be non-regional, how can a pronunciations be confined to London speakers of RP? That's like saying that Cornish speakers of RP are rhotic.

    ReplyDelete
  9. For me the compressed form of 'century' is the basic form and is homophonous with 'sentry' (that others might want to transcribe an affricate plus /r/ is new to me). There is the possibility of an 'expanded' form, but in my ideolect this would be a spelling pronunciation.

    I noticed this a while back when I found 'natural' transcribed with an affricate, not my /tr/.

    ReplyDelete
  10. John,
    You say these NS students are supposed to be transcribing a passage of English into phonetic notation, and are encouraged to represent their own accent rather than some variety different from their own. If they recognize their own initial cluster as a post-alveolar affricate (i.e. post-alveolar from start to finish, as in tʃeɪn) followed by an r, and that it is markedly more post-alveolar and affricated than what they hear for the RP output of "the allophonic rule that makes English r fricative in the clusters tr and dr" and "retracts t and d when followed by r", and moreover if they are correct in thinking that in their idiolects all three segments are identifiable (which phenomenon is probably not much less usual in their speech than their transcriptions tʃreɪn, dʒrɪŋk would indicate), is it not reasonable that they should transcribe it accordingly? Unless of course they have opted "to transcribe RP, even if it is not exactly their own accent". They are “doing transcription”, after all, and this is phonetics, not phonology, or they could elect to mark it as an archiphoneme representing the phonological neutralization you mention.

    There is no possible contrast in English words between a posited cluster sr and a posited cluster ʃr either, and if they do observe that there is no alveolar correlate to their tʃr etc, they might even see the parallel with the case of ʃr, and in phonetics at least, no one seems to think "it is meaningless to ask which is involved" there.

    ReplyDelete
  11. I say /treɪn/ with the typical nasalization of the preceding vowel of the Canadian priaries. No t-softening present at all though. Crisp, clear /tr-/ and /dr-/ in my idiolect. And I do distinguish "century" /'sɛntʃri/ from "sentry" /'sɛntri/ now that I think about it. I guess that makes me anal. ;o)

    ReplyDelete
  12. mallamb: "There is no possible contrast in English words between a posited cluster sr and a posited cluster ʃr either [...]"

    Whoa! Wait there a sec... I pronounce "Sri Lanka" with /sri/ while "shreek" is /ʃrik/. Am I an outlier?

    ReplyDelete
  13. Erh, that should be "shriek". I'm having spelling difficulties.

    ReplyDelete
  14. I'd use /tr, dr/ in a phonemic transcription and [t͡ɹ̝̥ʰ, d͡ɹ̝] in a phonetic one.

    ReplyDelete
  15. Glen,
    "Sri Lanka" is not exactly an English word, is it? I've checked the online dictionaries and it looks as if a fair number of them give only ʃriː – not surprisingly since even in Sanskrit and Sinhalese it's śrī! LPD however supports you to the extent of putting srɪ first.

    ReplyDelete
  16. I stand with Glen here, except that century has three inelidable (and ineluctable) syllables.

    ReplyDelete
  17. On the basis of my own (originally near-RP, now somewhat influenced by AmE) speech, I would transcribe the clusters as /tʃr/, /dʒr/. The only basis for using /tr/ and /dr would be historical, orthographical, or to accommodate the Glens of the world in a "panlectal" phonology :) I also have no contrast between /*sr/ and /ʃr/.

    ReplyDelete
  18. Goodness, army1987! How does one pronounce that?! [t͡ɹ̝̥ʰ, d͡ɹ̝]! Could you explained your reasoning behind this armada of diacritics?

    In general, I would keep the dychotomy phonetic—phonemic, however, I would homogneize the symbols, which would make the phonetic transcription different in that it would use the diacritics etc., which the phonemic one wouldn't.

    ReplyDelete
  19. ...I was of course using "phonetic" in the traditional, broad, sense, not in contradistinction to "phonemic/phonological". Some phonetic transcriptions are phonemic, some are not.

    ReplyDelete
  20. John,
    So you wouldn’t consider a relatively narrow transcription appropriate for a "phonetic" notation in which NS students, including as you say those for whom RP "is not exactly their own accent", are encouraged to represent their own accent rather than some variety different from their own?

    I know you see phonology as a branch of phonetics, but this doesn't make that view any more comprehensible for me.

    I don't dispute that some phonetic transcriptions are phonemic, some are not, but this is only with respect to the extent to which they are based on some specific phonemic analysis. The more narrowly phonetic they are, the less transparently phonemic, and vice versa.

    ReplyDelete
  21. Thank you, John, for the reply on syllabification. In LPD you point to it in Syllables, but the Guide doesn't have a subentry for it in Pronunciation Noted (CD version).

    ReplyDelete
  22. Are there any examples - on YouTube, for instance - of native speakers of English who use one or the other pronunciation?

    For example, the Queen might pronounce drink as [drɪŋk] whereas David Cameron might pronounce it as [dʒrɪŋk].

    ReplyDelete
  23. ...I was of course using "phonetic" in the traditional, broad, sense, not in contradistinction to "phonemic/phonological". Some phonetic transcriptions are phonemic, some are not.

    Then it's easy: you should have warned them of conventions and given them all the table of phonemes from LPD, Cambridge or one of several others (but not Upton's). So then if they wrote /ʧr/ instead of /tr/, even though it can be [tɹ], [tɾ] or [ʧɹ], it would be considered a 'mistake' thus diminished the final score.

    Do they also learn how to transcribe narrowly?

    For example, the Queen might pronounce drink as [drɪŋk] whereas David Cameron might pronounce it as [dʒrɪŋk].

    You mean, she would've said [dɾɪŋk], no? Does [ɾ] comes only intervocalically or also antevocalically in older RP?

    ReplyDelete
  24. @Sylvia: he's made it clear that "students are encouraged, after all, to represent their own accent...". Prescribing a particular list of phonemes based on RP wouldn't really be consistent with that.

    Anyway, prescribing a list of phonemes wouldn't stop people writing /tʃr/: both /tʃ/ and /r/ are surely going to be in the list.

    ReplyDelete
  25. Actually I've just realized that I do produce phonetic [tr] after /s/. So for me

    "strain" is [stɻeɪn]

    while

    "train" is [t͡ʃʰɻeɪn] or [t͡ʃɻ̥ɻeɪn]*

    * These are both trying to represent the same utterance. I want to show that the ɻ is voiceless initially and subsequently voiced. Anyone know the best IPA way to do this?

    ReplyDelete
  26. Mainly between vowels, the second of which is stressed, and after interdentals, I think.

    Anyway, the question of phonetic vs phonemic isn't really new on this blog. It makes sense to have a narrower transcription sometimes (eg. when one wishes ˌhapʰi ˈbɜːθˌdeɪ and another ˌhɛb̥ɪ̈ ˈbɛðdɛː, both of which could still be narrower, of course), but mostly we don't need it because we know what's implied.

    ReplyDelete
  27. vp,
    «Actually I've just realized that I do produce phonetic [tr] after /s/. So for me "strain" is [stɻeɪn]»

    Well, for a lot of these people it's [ʃʧɻeɪn], to use the disliked combo of t and ʃ. As clear as day. This is pretty good evidence for the credibility of their transcription [tʃreɪn], is it not?

    ReplyDelete
  28. mallamb: "'Sri Lanka' is not exactly an English word, is it?"

    If you want to take your quibbling to such absurd extremes, *nothing* is.

    ReplyDelete
  29. Not a true native speaker, but for me a compressed 'century' and 'sentry' are all but identical. Maybe the former is a tiny bit more labialized, but hardly enough to make an auditory distinction. For me the pairs /tr/ and /tʃr/, /dr/ and /dʒr/ are virtually neutralized.

    On the other hand, reading the comments above, I surprised myself pronouncing 'children' in my head without any 'affrication' (r actually does not turn into a fricative, let alone neutralize the preceding /d/ with /dʒ/). It's driving me crazy trying to explain this. I consistently apply the affrication to 'heraldry' or 'cauldron', and even to compressed forms like 'dormitory' or 'thunderous', and when I pronounce them the way I would pronounce 'children' it sounds horribly unnatural for me. Maybe I have this single word stored with an exceptional pronunciation. Maybe I unknowingly picked something up from rhotic speakers who have it stored as /ˈtʃɪldərn/ or some other combination that doesn't produce the affrication.

    ReplyDelete
  30. @ John Cowan: Century has to have 3 syllables for me too. It must be an American thing.

    ReplyDelete
  31. Regarding century and sentry... It reminds me of how the dictionaries can be confusing. For example, one gives ˈsteɪʃ^ənəri, the other /ˈsteɪ.ʃ^ən.^ər.i/, where both are unpronounceable. As written, taken literally, without certain implications. The choice of sounds not to be pronounced is also so random, the prescriptivity arbitrary... Then you have the recorded actor's voice and it doesn't match the written.

    ReplyDelete
  32. Phil: "@ John Cowan: Century has to have 3 syllables for me too. It must be an American thing."

    Believe whatever you like but I'm Canadian (not American) and consistently speak it with only two syllables. Here's a sample of even a Brit clearly using two syllables in "century":
    17th Century Life.

    ReplyDelete
  33. Jongseong: "On the other hand, reading the comments above, I surprised myself pronouncing 'children' in my head without any 'affrication' (r actually does not turn into a fricative, let alone neutralize the preceding /d/ with /dʒ/). It's driving me crazy trying to explain this."

    Maybe it's dissimilation because of the initial affricate already in "ch"?

    ReplyDelete
  34. Okay this "century" thing is making me get obsessive. LOL! I found an American clip that distinctly exhibits the 3-syllable word: Andrew Bacevich: Farewell, the American Century. Interesting. Sampling a Canadian however, I'm having a hard time telling how many syllables the quick-tongued yappermouth is uttering in National Launch of The Canadian Century. Frankly I wouldn't doubt a few Canadians or Brits uttering three syllables too and I'm skeptical that it's "an American thing" necessarily but I'm still sticking to my 2-syllable pronunciation, hehe.

    ReplyDelete
  35. I'm sure that I would transcribe this as tʃreɪn and can't even conceive of not doing so.

    My reasoning is maybe a little foggy, but I feel certain that to my ear "choo choo train" alliterate and "choo choo taint" do not (I think my 3-year-old daughter actually first pointed this alliteration out to me).

    Also, and I realize this kind of data is highly suspect, when I prepare to say "train" and "taint" or "drain" and "Dane", my tongue moves into different positions for each, with my mouth being the same for "chain","train","drain" and "Jane" and then the same (but different) for "taint" and "Dane."

    ReplyDelete
  36. Here in NZ you hear "hundy" as slang for a $100 note (or $100 generally) but I've heard Australians call it a "hunjy".

    ReplyDelete
  37. David Marjanović24 March 2011 at 23:54

    If RP is supposed to be non-regional, how can a pronunciations be confined to London speakers of RP?

    I suppose RP is underspecified with regards to phonetic detail, just like Standard German is. For instance, people from places where Low German dialects are or were spoken (in many cases haven't been spoken for several generations) make their alveolar consonants apical like in English or Dutch, all others use laminal ones like in Rest-European, and neither option seems to be spreading as far as I can tell.

    Maybe I have this single word stored with an exceptional pronunciation.

    That used to happen to me a lot, and probably still does more often than I notice. But of course dissimilation is likely, too.

    Also, and I realize this kind of data is highly suspect

    Why? The English /tʃ/ and /dʒ/ are postalveolar from start to finish, while /t/ and /d/ are alveolar.

    ReplyDelete
  38. Hi everyone. I'm asking my question here because I think I might not get laughed out of the forum (as I would if I posted this question on Facebook). I know hardly anything about phonetics (I read the IPA symbol charts, but then usually forget the contents), but I think that it's valuable to be able to use phonics in reading. My pre-Kindergarten daughter has recently become interested in "the sounds that letters make," and she and I have been having a blast playing with single letters and now, dipthongs. Until we got to "dr" and "tr." I'm a western-derived American, and I pronounce these combinations as "jr" and "chr." I *can* pronounce "tr" as "tr," but it takes mental effort. I don't think I can do the same for "dr;" my "dragon" comes out as "d-uh-rag-un." Without a "j," my "drive" becomes "derive."

    My question is, how big a deal is it to teach my daughter that the dipthong "tr/dr" makes a "chr/jr" sound, as opposed to making a conscientious effort to change our pronunciation? Or I could perhaps explain to her that many people pronounce these letters as "tr" and "dr," but we pronounce them our way. Do you have any thoughts on this? I do think my pronunciations are fairly standard American English so I don't think my instructing her this way would cause her too many problems. But I wanted to get experts' opinions as well. Thanks very much.

    ReplyDelete
    Replies
    1. I think your pronunciation is more standard than you think. I'm British, and I would say my pronunciation of these combinations of consonants is about the same as your 'jr' and 'chr'.

      Incidentally, the term 'diphthong' is usually restricted to vowels, but the Oxford English Dictionary does recognize the concept of a 'consonantal diphthong'.

      --
      Steve

      Delete
    2. If this is your biggest worry about your pre-Kindergarten daughter's language skills, then I'd say you have nothing to worry about :)

      Delete
  39. I use a labiodental approximant pronouncing the initial consonantal cluster of train as i.e. [t₍ʋ̥-]

    ReplyDelete
  40. At least in most American English varieties, the word initial [tr] cluster becomes [tʃr]. I am puzzled that the author maintains that train is pronounced with [tr] instead of [tʃr], when I produce and perceive so much evidence to the contrary.

    ReplyDelete