Tuesday, 31 August 2010

linking semivowels?

Lourdes Achard (blog, 27 Aug) also wrote
I am a Phonetics teacher and am teaching linking /w/ and /j/ at the moment. Have you got any text to practise this kind of simplification?
I found this request puzzling, and replied
I wonder what your students' first language is. Since native speakers don't normally use such linking semivowels, this is not a topic I would teach! And I don't understand how it could be seen as a "simplification". Perhaps I have misunderstood your question.

Peter Roach’s widely used textbook English Phonetics and Phonology (CUP, 4th ed. 2009) quite rightly makes no mention of this topic. Looking further afield, I wonder if Lourdes was influenced by Cruttenden’s formulation in his current (7th) edition of Gimson’s Pronunciation of English (Hodder Education, 2008).
(Click to enlarge this for better legibility.)

Cruttenden is being rather naughty here in his phonetic notation. The IPA symbols [ʲ, ʷ] are properly no more than diacritics, indicating palatalization and labialization respectively. He, though, is obviously using them to denote very short, transitional, non-phonemic glides.

I suppose he is right in saying that these not-quite-segments may sometimes be “heard”, since experience shows that some naïve transcribers are convinced that they exist. Personally, I take the line that they are figments of our imagination: the supposed “[ʲ]” in my arms merely represents the point of maximum upward excursus of the tongue body as it moves from [a] through [ɪ] towards [ɑ]. How could one possibly detect the presence vs. absence of this entity on a spectrogram?

If it meant anything, it would presumably mean that some small part of the is to be regarded as falling in the arms syllable rather than the my syllable. But we do not say *maɪ jɑːmz, as evidenced by minimal pairs such as Cruttenden’s I earn vs. I yearn: there is no phonological (phonemic, linguistic) j present in my arms or I earn.

And there’s no w in two evils (see below).

In this respect English differs from, for example, Serbian, which obligatorily inserts a phoneme j between i and a following vowel, as reflected in the spelling Srbija (not *Srbia): the syllables are sr . bi . ja.

Whatever we conclude from this discussion, I can certainly see no case for wasting any time in the EFL classroom on teaching such a dubious topic.


  1. I would blame it on Headway (L. & J. Soars, New Headway Advanced. Student's Book. OUP 2003; p. 36).

  2. And on Redston & Cunningham's "Face2Face" (Cambridge, pre-intermediate student's book, pp.31-32); and on Hancock's "English Pronunciation in Use" (Cambridge 2003; pp.86-87).

  3. The image is absolutely hilarious.

  4. ... and on Underhill's "Sound Foundations: Learning and Teaching Pronunciation" (Macmillan 1994; p.67); and on Thornbury & Watkin's "The CELTA course" (Cambridge 2007; Trainee Book, p.156, exercise E1, example g)...

  5. ... and on Baker's "Ship or Sheep: An Intermediate Pronunciation Course" (Cambridge 2006, 3rd edition; p.142, p.146)...

  6. Note in particular Adrian Underhill's description of "intrusive /w/ and /j/" in his "Sound Foundations" (Macmillan, 1994) I mentioned above:

    "These are also used to link certain vowel-vowel combinations at word junctions. The intrusive sound may not be distinctly heard especially where the overall vowel sequence is fairly relaxed. Nevertheless you will find it noticeable in all sorts of recorded material, and it has great value as a learning device when helping learners towards a smooth linking of words in continuous speech."

    "Noticeable in ALL sorts of recorded material"?

    And does it really have such great value as a learning device?

  7. Martin Hewings, too, discusses the use of "linking semivowels" in his "Pronunciation Practice Activities: A Resource Book for Teaching English Pronunciation" (Cambridge 2004; pp.82-84).

    This topic really seems to be discussed pretty much everywhere, but how useful is it for the EFL learner?

  8. See my blog for Jan 8th 2010 and the comments thereon.


    I agree entirely that the idea of j and w linking should be consigned to the bin.

  9. The phrase my ears made me think of the phrase donkey's years 'a very long time (ago)', which the OED2 explains thus: "with punning allusion to the length of a donkey's ears and to the vulgar pronunciation of ears as years". A phonetically knowledgeable person from Sussex once told me that this was a feature of his childhood pronunciation (around 1940), but that in fact he pronounced years as ears, and that donkey's years was a hypercorrection by outsiders.

    This leads me to the pronunciation of donkey. The OED2 gives only /dɒnkɪ/, with their usual non-happY-tensed rendition, but says "the original pronunciation apparently rimed with monkey (whence the spelling)" and suggests that it began as a hypocoristic form of Duncan. I have always used this STRUT pronunciation; m-w.com, NID3, and AHD4 all give LOT=PALM as the primary pronunciation, with STRUT and THOUGHT as alternatives.

    Apparently the STRUT pronunciation is or was thought of as rustic. Scout in To Kill A Mockingbird (1960, but set in 1936) refers to it with other marks of Southern rurality: "liking fiddling, eating syrupy biscuits [AmE, remember!] for lunch, being a holy-roller, singing Sweetly Sings the Donkey and pronouncing it dunkey, all of which the state paid teachers to discourage."

    Google can't find any decent video of the song; this version is grossly overdone, but the best of a bad lot.

  10. Years for ears - I remember reading an article about that somewhere; instancing one of the Mitford sisters.

    In linking, there are still differences, eg with this pair of words, th[ɘ] years vs th[ɪ(ʲ)] ears.

  11. Well, there are dialects of English that merge "ear" and "year", and often also "here". There's certainly a tendency in the English of South Wales to make all of them something like /jɜː/, resulting in "the year" and "the ear" being homophones. I tend to be relatively close to RP, but I do have this merger when I'm not being emphatic (which returns the vowel to NEAR), and rather amusingly I have been picked up on complimenting someone on her "lovely [jɜː]rings" before. I don't know how widespread this is (it certainly stretches from Newport to Neath and inland as far as Merthyr), but I think we can safely say that the Mitford sisters would be appalled, and it's almost certainly not something for an EFL class!

  12. Lourdes raises a valid and important issue.

    While I wouldn't necessarily notate it in Cruttenden's way for myself, I would absolutely discuss with an EFL learner the concept of gliding from one vowel to another, and the audible phenomenon of a "shadow" or "phantom" [j] or [w] (as the case might be) "forming itself."

    And I would also at some point personally demonstrate to an EFL learner the difference in sound between allowing a glide to form itself in the process of transition from one vowel or diphthong to another, and intruding a full [j] or [w]. (As an example, it may help to imagine the difference between "I am what I am" and Popeye's "I yam what I yam.") It would not bother me in the slightest if an EFL client wished to use a [w] or [j] (or a diacritical version) in their own notation, provided this did not lead them to overdo the sound. Nobody is going to grade their written homework in life! The people they speak to will only want to understand them.

    I can assure you that addressing this issue in some way does indeed profit many EFL learners in coming to grips with finer points of pronunciation and the flow of connected speech; for most such learners, orthodoxy in IPA transcription is of far less value than is developing strategies, however unorthodox they may seem, that will bring about the desired result of greater ease in making oneself understood by native speakers.

    As for donkey, it rhymes with monkey in my idiolect, both words belonging to the STRUT set, and in my case pronounced with a close to mid-central [ʌ] in the first syllable.

  13. Edith Skinner's text is commonly used in many theatre programs and teaches this idea and this specific notation as well. For actors - even native English speakers - this concept is often used to avoid strong glottal attacks that can cause vocal issues. It's yet another way that Skinner (and those before her) "adapted" the IPA for their own uses.

  14. I own a little book with the sort of material Lourdes Archard is seeking. It's one of a set of books called Elements of Punctuation by Colin Mortimer with drawings by various cartoonists. This one is call Link-up and the drawings are by Daria Glen. I'll make a link to some sample material when I find time. Meanwhile, here's a flavour...

    Dialogue 37 practices only linking w:

    A Do‿I have to do‿every question?
    B You‿ought to try.
    A How much time do‿I have?
    B We give you‿about two‿hours.
    A Two‿hours?
    B Those who‿are quick can go‿early.
    A And those who can't do‿it?
    B They can go‿early too,‿I suppose.
    A Good.

    Like the other books in the series, Link-up had cassette tape that provided a lively model for interactive fluency.

    Other dialogues practiced other junctures of consonants before initial vowels.

    So, like the Schenker stuff I described earlier this year, the materials exaggerated the lack of juncture — to counteract the exaggerated juncture sounds that students make.

  15. Linking w definitely exists, at least in the north of England. "Saw it" can be /sɔːw it/

    See here for an example http://sounds.bl.uk/File.aspx?item=021T-C0900X08553X-0100A0.pdf
    "I never saw a [sO:w @] marriage license for my dad"

  16. Here's a link to a sample double page of Link-up. I used to have a copy of the cassette; if I find it, I'll post a sample.

  17. The notion of linking j/w has been enshrined in British-based EFL teacher training since the 1970s, at least, when I started out as an EFL teacher and began to develop a particular interest in pronunciation. The main sources of background knowledge and practice activities that were drawn to my attention at that time were J D O'Connor: Better English Pronunciation (CUP 1967)and Colin Mortimer's four booklets (1976?), subsequently repackaged as Elements of Pronunciation (CUP 1984).

    O'Connor uses the convention of a small superscript j/w (as well as small superscript r for linking r in RP). He says the glide between the two words in 'the idea' is "very like a j but a very gentle one." He distinguishes between 'my ears' and 'my years' by saying that in the latter the j is "longer and stronger".

    Later, Joanne Kenworthy: Teaching English Pronunciation (Longman 1987) was widely circulated and influential. She uses the term 'sound insertion' for linking j/w, as a subcategory of 'linkage'.

    This is by way of information, not an attempt at justification. This is the tradition that Headway and innumerable other coursebooks of its ilk - and pronunciation practice materials, including - mea culpa - mine, derive from.

  18. I think Mortimer had it about right. His five books [You missed one, Jonathan] together tackled the problem of rhythmic fluency. They weren't perfect, but I don't know of anything better.

    Stress Time trained students to allot the same time to single-sylllable and multi-syllabie stressed 'feet'. The other four trained them to say words quickly enough to fit them into a foot. Clusters addressed the problem of articulating consonant clusters without slowing down. Contractions and Weak Forms alerted students to choose forms that did not slow them down. Link-up addressed the possible bars to fluency at word juncture — specifically at the juncture between a word-initial vowel and a preceding consonant or diphthong. Some students (notoriously Italian students) have a strong tendency to 'grunt' when a consonant is word-final. Others are tempted to insert minute pauses or glottal onset to the vowels. Neither is a grave pronunciation error, but both may disturb the fluent rhythm of delivery.

    The main problem with the books is that not enough students were far enough advanced to constitute a market. I hardly ever encountered students who were advanced enough to tackle these fluency problems and yet deprived of opportunities to listen to and interact with native speakers operating at natural speed. The only time I ever used them was with extremely advanced German students destined to be English teachers — as part of a course in the difficult art of reading aloud.

  19. @ James D: I think I've heard something similar from South Africans. I'm not sure how that's related to this topic, but oh well.

  20. Is it worth teaching linking /j, w/? Well, one benefit is to avoid an overabundance of "linking" glottal stops in such situations.

    I assume you would say the same thing about linking /r/. Just as 'two evils' and 'two weevils' are not the same, so Mark Anthony did not say "Friends, Romans, countrymen, lend me your rears."

  21. The supposed “[ʲ]” in my arms merely represents the point of maximum upward excursus of the tongue body as it moves from [a] through [ɪ] towards [ɑ]. How could one possibly detect the presence vs. absence of this entity on a spectrogram?

    Well, my understanding is that, in vacuo, there is no inherent phonetic difference between [j] and [i] (or indeed between [j] and [ɪ]). The difference lies only in syllabicity, with [j] being non-syllabic and [i], [ɪ] syllabic. And since syllabicity lies at the phonological rather than the phonetic level, it would be in principle impossible to detect a [j], as opposed to an [i], on an isolated spectrogram. The same applies, mutatis mutandis, to [u] vs. [w], or to [ɹ] vs. [ɚ].

    However, if I examine my own speech, I observe that my [j] is always slightly more constricted than the immediately adjacent vowel(s), whatever they may be. Thus my initial /j/ in "the yacht" is probably slightly more open than my normal FLEECE vowel; whereas my initial /j/ in "my yeast" is clearly closer than FLEECE. So, in the context of my speech as a whole, there probably would be spectrographic evidence to say that there is no /j/ in "my arms".

  22. @Ed:

    See here for an example http://sounds.bl.uk/File.aspx?item=021T-C0900X08553X-0100A0.pdf
    "I never saw a [sO:w @] marriage license for my dad"

    Are you sure that the superscript "w" in the original PDF does not just denote a vowel with extra lip-rounding? That's what it means according to Wikipedia's SAMPA chart.

  23. In this respect English differs from, for example, Serbian, which obligatorily inserts a phoneme j between i and a following vowel, as reflected in the spelling Srbija (not *Srbia): the syllables are sr . bi . ja.

    It's also very interesting to note that, when English words are transliterated into Indian languages, semivowels tend to be used where other varieties of English would have a hiatus or diphthong. For example, the Hindi Wikipedia page on Queen Victoria renders her name विक्टोरिया ("vi-kṭo-ri-yā"), while Tony Blair's page renders his surname as ब्लेयर ("ble-ya-r")

    Both the Indian grammatical tradition, and the Indian scripts themselves, strongly encourage a CV syllable structure.

  24. Hence, with additional h-dropping, the joke about the task to form a sentence with "green", "pink" and "yellow", which the Indian student answered with: The telephone goes reen-green-green. I pink it up and say "yellow?"

    Not entirely PC, but phonetically interesting.

  25. Well, my understanding is that, in vacuo, there is no inherent phonetic difference between [j] and [i] (or indeed between [j] and [ɪ]).
    Well, the ending of French soleil and that of English day sound quite different. I'd say the former is [j] and the latter is [ɪ]

  26. Just to verify: do you mean [ɪ̯] by that? Otherwise, I'd be surprised to hear a vocalic [ɪ] in RP or GenAm.

    Provided I understood you correctly, I think you're right. An actual [j], or [i̯], for that matter, in day sounds Irish or Scottish to me. (Never mind the equivalent [e:] &c. for the moment.)

  27. @army1987:

    Well, the ending of French soleil and that of English day sound quite different. I'd say the former is [j] and the latter is [ɪ]

    Sure -- but you have no justification for saying that French soleil has [j] rather than [i], except via a general phonological account of French. If French were an unknown language and you had heard just that word, it would be impossible to posit [j] over [i].

  28. Are you talking about syllabicity here? Or an approximant vs a fricative? Or the place of articulation? Or a mix?

  29. In Serbian it's:

    Srbija (with j)
    Marija (with j)
    Mia (whitout j)
    Mario (without j)
    avion (without j)

  30. @Lipman: Yes, I meant [ɪ̯] for English day
    @vp: I've heard that French contrasts /i/ with /j/ (though I can't remember the minimal pairs used to show that), but that wasn't my point. (I wrote "sound quite different" and used brackets.) My point was that in French soleil the last segment is significantly higher than typical for vowels, so using [i̯] wouldn't look right to me (unless you take the upper edge of the vowel quadrilateral to be where friction begins). (To my Italian ear, that j sounds more akin to ʎ than to i.) And if there was a language using such a high vowel as the nucleus of a syllable, I would even write [j̩] (j with syllabicity undermark, in case the font screws it up).

  31. My initial reaction was the same as John’s, to dismiss the idea of treating such a topic in the EFL classroom as waste of time, but it was a pity that Lourdes didnt make it clear just what was worrying her. I think John’s commenters have revealed the likely chief explanation namely that various textbooks have raised the matter making her feel she shd be doing something about the (alleged) problem. My guess is that she was being led to worry unnecessarily. She teaches in Uruguay. I’ve spent what amounts to best part of a year teaching various native speakers of Spanish and never remember any problems of this sort from them. Tho it’s not a very surprising situation. One easily can think of material offered as practice on English features of kinds that of themselves presented no difficulties at all for the students who were being drilled in them.

    However, I’ve certainly he’rd occasional speakers who, no dou’t from anxiety to be enunciating with maximal clarity, have produced the effect of what might be called linking yods etc. Amy’s ref to Popeye was very pertinent. John’s remark “we do not say *maɪ jɑːmz” is of course completely true about ordinary NS conversation but it immediately reminded me of a dreadful singer who ruins for me an otherwise tolerable (80-year-old) jazz performance in which that’s exactly what I feel I hear him repeating when I play it.

    I’d be surprised if it was Alan Cruttenden’s paragraph that John quotes that prompted Lourdes’ worries. By the way, so far from considering Alan’s use of superscript symbols for “very short, transitional, non-phonemic glides” as “nau’ty”, I’m inclined to congratulate him on his inventiveness. I’ve long been a staunch supporter of the IP Alphabet but dont think it’s comparable to Mosaic tablets. If audible unnatural links are really being made by her students it isnt too surprising that Lourdes shd request advice on how to deal with them. At any rate by now she has received in these comments plenty of information of the kind she seemed to be seeking. I’m pleased to endorse the recommendations of the materials by my well remembered sometime student Colin Mortimer.

  32. @army1987.

    My claim is that the inherent value of the sound denoted in the IPA by [j] is no closer than denoted by [i].

    If you look at the IPA handbook, an approximant is defined as a sound "in which the airflow is not turbulent and no frication is audible" (p. 8), while cardinal [i] is described thus: "any further narrowing in the palatal region would cause the airflow to become turbulent, resulting in a fricative". (p. 10).

    It seems as though the IPA handbook is describing a phonetically identical sound. Of course, in a particular langauge, it can be stipulated as a convention that /j/ is, or can be, closer than /i/.