derived semivowels

If we define a diphthong as being two vowel qualities in a single syllable, or equivalently as “a complex vowel which changes its quality within a single syllable“ (SID), then we might feel we must recognize rising diphthongs in words such as win u̯ɪn, watch u̯ɒtʃ, yacht i̯ɒt and you i̯uː.

The argument against this is phonological. If we add to our phoneme inventory the rising diphthongs in these words, we shall have also to add those of weave, wet, whack, suave, war, woman, woo, work and yeast, Yiddish, yet, yam, yarn, yawn, York, yearn, i.e. more than the number of simple vowels we have in our inventory; not to mention additional triphthongs that we shall have to recognize in words such as way, woe, wine, wow, weird, yea, yoke, yikes, yowl, yeah. But these nonsyllabic u̯ and i̯ pattern like consonants (being at the margins of syllables), so it is clearly better to recognize just the two semivowels w and j, and to analyse all the polyphthongs just mentioned as /wV, jV/. A semivowel (or ‘glide’, if you prefer) is articulated like a vowel but patterns like a consonant. We no longer attempt to distinguish on the phonetic level between nonsyllabic [i̯, u̯] and [j, w].

You may be familiar with an indelicate limerick (search here, for example) in which Australia and dahlia (BrE for this flower, with eɪ as the stressed vowel) are made to rhyme with failure. Are these good rhymes? Is dahlia ˈdeɪljə an exact rhyme with failure ˈfeɪljə? Well, yes and no. The possible difference between them is not a matter of i̯ as against j, but rather of our awareness of the varisyllabicity in dahlia as against its absence in failure. We know that dahlia can optionally be said with three syllables (by some of us, at least), while failure can only have two.

(I was perhaps being too sweeping the other day when I suggested that this whole matter was a question of BrE vs AmE; but it is striking that Kenyon-Knott and Merriam-Webster do not allow for -eɪl.i.ə in Australia, while LPD and EPD do.) Then what about millennia compared with tenure? Peter Roach’s CPD has these as non-rhymes (mɪˈlen.i.ə and ˈten.jəʳ), just as in LPD I have mɪ ˈlen i‿ə and ˈten jə. Merriam-Webster, too, shows the difference here. So do Kenyon and Knott — though at Virginia K&K give -ˈdʒɪnjə but also add ‘esp. New England’ -ˈdʒɪnɪə. At this word ODP, by the way, gives for BrE only vəˈdʒɪnɪə(r) and for AmE only vərˈdʒɪnjə.)

What about a word like happier? It is clear that it can on occasion be pronounced as a disyllable. So if so pronounced, is ˈhæpjə the correct way to transcribe it? And for various, ˈveərjəs? What about DJ’s valuing ˈvæljwɪŋ? Is genuine truly ˈdʒenjwɪn?

We have seen that the reason why we hesitate to regard these as the underlying (lexical-entry, articulatory-target) representations is our awareness that in each case there is the possibility of a syllabic (= vowel) pronunciation in place of the putative semivowel. There are other possible reasons, too.

• Given that we get noticeable devoicing of j after p in words like pure, why do we not get similar devoicing in happier? (Or perhaps we do?)
• If the sequence -rj- is so awkward in garrulous, virulent, glomerula that we tend to avoid it by dropping the j, why does the same not apply in disyllabic various, barrier, glorious and so on? There is even disyllabic Istria, which must be ˈɪstrjə.
• In some kinds of AmE the sequence -lj- in words such as William, million, failure can get reduced to -jj-. Does this happen in volleying and jollier? If not, why not?
• The ‘semivowel’ solution leads to our treating valuing and genuine as containing sequences of semivowels, -jw-, something otherwise unattested in English and universally unusual.

In supplying a pronunciation entry for any word containing a possible w or j plus vowel, then, we have to ask ourselves: can this semivowel alternatively be pronounced as a syllabic vowel? If yes, then we take it as u, i; if not, then as just j, w. We may not always agree on the answer. As we have seen, British and American lexicographers disagree in the case of Australia. When I transcribed Daniel as ˈdæniəl the other day, I didn’t stop to consider the issue, though if you now ask me I would confirm yes, I can pronounce this name as a trisyllable. But some of those who commented obviously can’t. Anyhow, I also entered it in LPD as ˈdæni‿əl (the possible compression is predictable from context). On the other hand, I entered million in LPD as ˈmɪl jən, only to receive a complaint from one user that I ought to have allowed for a trisyllabic version and entered it as ˈmɪl i‿ən. (So in the current edition I give both.) You can't win them all.

rising diphthongs

Before we start on the promised discussion of iə and related topics, let’s have a bit of history.

In 1954 Daniel Jones published an interesting article entitled “Falling and Rising Diphthongs in Southern English” in Miscellanea Phonetica ii: 1-12 (issued with Le Maître Phonétique).

The article starts with a general discussion about two types of diphthong, ‘falling’ (with decreasing sonority) and ‘rising’ (with increasing sonority), distinguishing both types from simple sequences of two vowels.

The ‘common’ English diphthongs ei, ou, ai, au, ɔi, he says, (i.e. the FACE, GOAT, PRICE, MOUTH and CHOICE vowels, which we nowadays write eɪ, əʊ, aɪ, aʊ, ɔɪ), are all ‘falling’. ‘Rising’ diphthongs are ‘uncommon’, but as an example of one he adduces the ĕo of Tswana.

He then discusses ‘the vowel elements of words like ruin, bluish’. There may, he claims, be either a succession of two short vowels (ˈru-in, in today’s notation ˈrʊ.ɪn) or a falling diphthong (ruĭn, = rʊɪ̯n); as a third possibility there may be a long vowel plus a short one (ˈruːin, = ˈruːɪn).

In unstressed positions, on the other hand, as in valuing, the possibilities are again a succession of two distinct vowels or a diphthong; but this diphthong is ‘generally a rising one, ŭi’. Furthermore, ‘in many such words there is an alternative pronunciation with wi as well as u-i. Thus valuing may be any of ˈvælju-iŋ, ˈvæljŭiŋ, ˈvæljwiŋ (= today’s ˈvæljʊ.ɪŋ, ˈvæljʊ̯ɪŋ, ˈvæljwɪŋ). Although it may be difficult to distinguish between these possibilities, ‘the distinctions are possible, at least in theory, and are probably felt subjectively by the speaker in slow utterance’.

Applying this approach to words that can have the falling diphthong iə, he distinguishes two classes: those that have alternative pronunciations with i-ə, such as idea, theatre, theory, museum, Ian, and those that do not, such as clear, fierce, nearly, hearing [i.e. distinguishing the varisyllabic first group and the non-varisyllabic second group]. A possible minimal pair for some speakers (though not for most) would be rhea and rear. [Today I would use as an example the more familiar Korea vs career.]

Words with the rising diphthong, such as hideous, easier, luckier, colloquial, theoretical, should be compared with those having a secondarily-stressed, falling diphthong, such as reindeer, Bluebeard, wheatear, realistic. With reindeer (falling diphthong) we can compare windier (rising diphthong or sequence of two separate vowels).

The rising-diphthong words “are sometimes said with two syllables and sometimes with one [, which] is shown by their variable treatment in verse, where the metre sometimes requires two syllables though more often, it would seem, one.” Jones adduces two lines from Hamlet, in one of which the word audience requires disyllabic pronunciation, and in the other trisyllabic.

Have of your audience been most free and bounteous
And call the noblest to the audience,

[I like to quote the British national anthem, in which -ious has to be disyllabic in happy and glorious, and compare it with the hymn Glorious things of thee are spoken, in which it has to be monosyllabic. See blog, 16-17 January 2007.]

Jones then applies a similar analysis to the uə-type sounds in fewer, renewal; tour, poor, skewer; contour, tenure, uranium, neurotic; influence, valuable, statuary, puerility, and again finds in Shakespeare lines in which virtuous must sometimes have two syllables, sometimes three.

He finishes by considering further possible rising diphthongs in words such as narrower, follower, coalesce; shadowy, yellowish, coefficient; forayer; essayist, archaism.

In the eleventh edition of his EPD (1956) Jones introduced two new symbols, for the rising diphthongs in happier (ĭə, corresponding to LPD’s i‿ə) and influence (ŭə, corresponding to LPD’s u‿ə). When Gimson took over as editor, he abandoned them.

In LPD I followed Jones in recognizing the various RP possibilities for these words. So I show museum , for example, as mju ˈziː‿əm, while fierce is just fɪəs. In mju ˈziː‿əm the italicization of the length mark shows that the first vowel may be short rather than long, while the compression mark indicates that between z and m we may have either a sequence of two separate vowels or else a falling diphthong, so that the word as a whole may consist of either three or two syllables.

Inspired by Jones’s pair reindeer — windier, another phonetician (I think it was Bjørn Stålharne Andrésen, but I can’t lay my hands on the reference, so this is from memory) performed a listening experiment in which he got speakers to imagine that as well as reindeer and roedeer we also have a kind of deer called a windeer; he asked them to pronounce in suitable carrier sentences the words windeer (kind of deer, with its falling diphthong in the second syllable) and windier (more windy, with its putative rising diphthong), and then played the results to listeners who were asked to decide which of the two words had been said. They proved unable to do this with better than random success. So the distinction between NEAR (my ɪə) and happY plus schwa (my i‿ə) may indeed be ‘felt subjectively by the speaker in slow utterance’, but the hearer cannot reliably detect it.

(to be continued on Monday)

angelic names

My partner’s name is Gabriel. Like other English-speaking bearers of this name, he pronounces it ˈɡeɪbriəl. However not everyone he meets appears to be familiar with this name, which is not as frequently encountered nowadays as perhaps it once was. People who do not know him quite often read his written name as ˌɡæbriˈel, which is properly the pronunciation that belongs with the female version of the name, Gabrielle. Some use a compromise pronunciation ˈɡæbriel.

I don’t know how long the female version has been around, presumably first in French and now also in English (not to mention Italian Gabriella and German Gabriele).

Although angels are supposed to be genderless, the archangels Michael and Gabriel are treated in English grammar as masculine (taking ‘he’, not ‘she’ or ‘it’ as their anaphoric pronoun), and as personal names are exclusively masculine.

Nevertheless, Michael has now acquired female equivalents — Michelle and the rare Michaela, and Gabriel has likewise acquired Gabrielle. As for other archangelic names, I’ve never come across a female form of Raphael or Uriel.

Interestingly, in standard spoken French the masculine form Gabriel and the feminine form Gabrielle are homophonous, both ɡabʁiɛl; though the feminine form has a final phantom ə that can surface in singing or in regional (southern) speech.

But in English the female form is regularly stressed on the final syllable, which gets a strong vowel, and is thus distinct (usually!) from the male form, which has initial stress and a reduced vowel in the last syllable.

Compare Daniel, the prophet cast into the lions’ den. As a man’s name in English he is ˈdæniəl, and again there is now a female form ˌdæniˈel, spelt in English as Danielle, though the French form is actually Danièle (again homophonous in French with the male form, give or take a schwa).

I don’t know enough about French to know why Gabriel forms the feminine by doubling the l while Daniel does it by adding a grave accent (but compare appeler — j’appelle as against geler — je gèle). Nor do I know enough about the history of English to know why Gabriel ends up with eɪ but Daniel with æ from what was presumably the same vowel in Latin/Greek/Hebrew (for the quantity of a in these phonetic contexts, compare Abraham and germanium).

an archiepiscopal mnemonic

Many of you will be aware that the Church of England is currently in the throes of choosing a new Archbishop of Canterbury. One of the candidates is the present Archbishop of York, Dr John Sentamu.

The religious correspondent on the Sky News morning programme yesterday referred to him as Bishop senˈtɑːmuː. But, as those who consult LPD or the Oxford BBC Guide to Pronunciation will know, his own preferred pronunciation is ˈsentəmuː. (Those who consult Wikipedia, on the other hand, will find there the implausible ˈsɛntɑːmuː. Perhaps one of you will now correct it.)

The Archbishop has given us an easy way to remember the correct pronunciation of his name. He asks us to imagine three cows standing in a row. Each cow moos. On the left we have a left moo, on the right we have a right moo, and in the centre we have a centre moo. And he’s like the centre moo, ˈsentəmuː.

(Sorry this doesn’t work for AmE or even for the Scots.)

more syllable-based allophony

Jacob (Monday’s blog) isn’t ready to give up yet.

In LPD, for the word sequel the given phonemic transcription is ˈsiːk wəl, which indicates that the vowel /iː/ should be fortis-clipped by k.
However, the pronunciation heard on the CD which comes with the book is clearly [ˈsiː]+[kwəl], which shows no clipping at all, as the k does not belong to the first syllable, being the onset of the second syllable.

Could you tell me how to resolve the discrepancy?

No, I can’t, beyond reiterating that speakers are not consistent in whether or not they reflect these boundaries in their pronunciation, and that there are considerable differences between different speakers and different accents. All I know is that when I say this word myself, I believe that I normally do have fortis clipping of the iː. I have no idea why the actor who recorded the word in the studio on the occasion in question appears to have pronounced it as if it were a compound such as sea quest. And if I had shown sequel in the dictionary as ˈsiː kwəl, as you imply I ought to have done, you can bet your bottom dollar that the actor would have chosen to say ˈsiːk wəl, as I do, and you would still be complaining.

Jacob continues

Further, as there are a large number of words for which the phonemic syllables (based on a number of syllabification principles) do not align with the phonetic syllables (an example is Sundridge ˈsʌndr ɪdʒ phonemically, but [ˈsʌn] +[drɪdʒ] phonetically), it seems that from an ESF perspective, a phonetic transcription which stipulates the phonetic syllables would be of great help to foreign students. It would be a godsend if such a dictionary were made available.

Actually, Sundridge, the name of a village in Kent, is a particularly interesting case. All three possible syllabifications ˈsʌndr ɪdʒ, ˈsʌn drɪdʒ, ˈsʌnd rɪdʒ are phonotactically well-formed (if you accept my argument in favour of recognizing syllable-final (n)tr, (n)dr, as in ent’r a plea, und’r a cloud). The etymological one is the first: the name comes from OE sundor ‘separate’, cognate with the stem of modern asunder, plus an element ersc ‘ploughed field’, which is also to be found in the name Winnersh (a place near Reading). Popular etymology, though, might seem to favour the the second, as if it were a compound of sun, or the third, as if it were ‘ridge of the Sund’. My choice was of course the first, just as in sundry, which following my general principles I syllabify as ˈsʌndr i.

I repeat that speakers (and accents) differ widely in the extent to which they make these boundaries audible in their speech and in what articulatory means they employ to do so.

I’m afraid, Jacob, that as things are you have to choose between my LPD, where I at least try to supply a syllabification that predicts the likely boundary-adjacent allophones in accents like mine as accurately as I know how, and Peter Roach’s EPD/CPD, which divides syllables entirely on phonotactic grounds, making no claim about boundary-adjacent allophones. If you think there’s a gap in the market for a third approach, do feel free to try and fill it. Peter and I have done our best.

Machynlleth

Machynlleth in mid-Wales is quite a small town, not often featured in the national news. But yesterday morning as I finished my breakfast the television presenters on the BBC1 Breakfast show handed over to their correspondent in what they called məˈkʌnlɪθ. The correspondent in question, one Rhun ap Iorwerth, duly told us about the missing five-year-old in the town (at the time of writing she’s still not been found), the town which he, clearly being a speaker of Welsh, pronounced maˈχənɬɛθ.

Note the unreduced vowels in the first and last syllables and the schwa in the middle, stressed, syllable. In Welsh, in this respect strikingly unlike the Germanic languages, ə is often stressed, but is restricted to non-final syllables (clitics such as the definite article y(r) count as non-final).

In passing, I might comment that I have never previously come across the forename Rhun (riːn, or north Welsh r̥ɨːn). I see from Wikipedia that it was the name of a sixth-century king of Gwynedd.

I was in my mid-forties when I sat my Welsh A-levels after studying in evening classes. I remember that in the English to Welsh translation paper the passage set began “The sign on the station platform read ‘Machynlleth’”. I dutifully recast my Welsh version so that Machynlleth was the first word of the sentence rather than the last, as is required by Welsh syntax.

There was recently a brief discussion on the web (I’ve forgotten just where) on the subject of digraphs. As Wikipedia explains,

In some language orthographies, like that of Croatian (lj, nj, dž), traditional Spanish (ch, ll, rr) or Czech (ch), digraphs are considered individual letters, meaning that they have their own place in the alphabet, in the standard orthography, and cannot be separated into their constituent graphemes; e.g. when sorting, abbreviating or hyphenating. In others, like English, this is not the case.

Someone pointed out, correctly, that in Welsh CH, LL, NG, and RH are treated as digraphs [= as single letters]. I pointed out that in the case of NG this can lead to problems. NG is treated as a digraph [ = a single letter] for collation, and ordered between G and H, if pronounced [ŋ], but not if pronounced [ŋɡ], in which case it is treated as N plus G. So angau (death, [ˈaŋaɨ]) comes before ail (second); but dangos (show, [ˈdaŋɡos]) comes after damwain (accident).

the wardrobe in the bedroom

The start of the month today marks exactly fifty years since I first entered gainful academic employment. On 1 October 1962 I became Assistant Lecturer in Phonetics at UCL, where I remained on the staff until my retirement in 2006.

Back then one of the most discussed minimal pairs of English was nitrate — night rate. Everyone agreed that they were distinct, despite consisting of the same phonemes in the same order. In the then dominant American ‘structuralist’ approach, the difference between the two was ascribed to ‘juncture’, or more exactly close vs. open juncture, the latter symbolized /+/. So for Trager and Smith and their followers nitrate would be analysed /náytrèyt/, but night rate as /náyt+rêyt/. (There was also the more dubious Nye trait, /náy+trêyt/.)

In LPD I indicate these same differences by the use of spacing. So I transcribe nitrate as ˈnaɪtr eɪt, while for night rate (if that were a headword) I would write ˈnaɪt reɪt, and for Nye trait ˈnaɪ treɪt. You can interpret these spaces as indicating the boundaries of morphemes or (as I chose to) of syllables.

The reason we can “hear” these junctures/boundaries is that the choice of allophones is sensitive to their presence/absence. Take another famous pair (one of Gimson’s favourites), great ape vs. grey tape. The t in great ape ˌɡreɪt ˈeɪp is a typical of t in final position: it has little or no aspiration, it causes pre-fortis clipping of the preceding eɪ; it is susceptible in BrE to becoming glottal, and in AmE to becoming voiced (‘flapped’). None of this applies to the t in grey tape (or, if you’re American, gray tape) ˌɡreɪ ˈteɪp, where the t is a typical initial one, being aspirated, not susceptible to glottalling or voicing, and not having any clipping effect on the preceding vowel.

If t and r or d and r are contiguous, i.e. have no intervening juncture, then in English they are pronounced together as a postalveolar affricate, as in train, drain, mattress, Audrey, entry, laundry. Compare what happens when they are separated, as in that rain, good reign, what result, saw drifts, ten trips, dawn drips. For more on all this, see my article setting out the syllabification principles I applied in LPD.

There are one or two exceptional cases where a putative or etymological morpheme boundary gets treated, by some speakers at least, as non-existent. I know that I do this in the word wardrobe. Although I know that etymologically it is a place for warding (keeping) robes (clothes), I pronounce its dr as an affricate, as in Audrey, not separated as in board room. Doubtless this is because I think of the word as a single item, not a compound. Personally, I do the same with beetroot and bedroom, though I am aware that some other speakers pronounce one or both of these with a boundary. I imagine that wardrobe, bedroom and beetroot are words that I knew well before I learned to read and write, and certainly well before I became aware of their etymologically compound status. (Note for Americans: in BrE a wardrobe is an everyday piece of bedroom furniture. You would probably use a 'closet' instead.)

This is what explains my different treatment in LPD of bedroom and headroom. My main prons are ˈbedr uːm, ˈhed ruːm. (Let’s ignore the irrelevant question of the vowel in -room — some people have ʊ rather than uː.) Although I ignore the boundary in the first, I think I usually preserve it in the second: headroom is a word I would not have learned before the age of nine or ten or so, and its compound nature as head plus room is fairly transparent. (Note for Americans: headroom is the BrE for ‘vertical clearance’.)

Furthermore, even when there is a boundary between t or d and r, people are not consistent in always reflecting it in their pronunciation. If I say there is no good reason to think that, I can still sometimes create an affricate out of the last consonant in good and the first in reason, even though there is an undoubted word/morpheme/syllable boundary between them. Similarly with the plosive and liquid in what rubbish!.

You may think that all this is rather good news for EFL learners. We can safely encourage them to treat all cases of tr and dr identically, namely as affricates.

But that’s to ignore people like my correspondent Jacob Chu, who has been listening carefully to the sound files that come with LPD and is dismayed by what he finds in two words we have been discussing.

The main pronunciation listed in LPD for bedroom is ˈbedr uːm. On the other hand, the main pronunciation listed for headroom is ˈhed ruːm and the pronunciation ˈhedr uːm is visibly absent, but the recording shows clearly, for British English, ˈhedr uːm. Please check the recordings from LPD. My query is, how should the discrepancy be resolved?

Beyond telling him to get a life (an idiom he might not be familiar with), what can I do but hold my hands up and congratulate him on his diligence and on the accuracy of his observations?

OK, I agree: on this occasion the actor who recorded ‘headroom’ in the studio happened to pronounce it as an exact rhyme of (my version of) ‘bedroom’. That’s life.

Of course, if I’d been in the studio monitoring the recordings (which I wasn’t, though I was for most or all of those entries in LPD that are not also in LDOCE, and also for some that are), I’d have jumped on it and got it re-recorded. Possibly. Or possibly not.