what the ell...?

Our Deputy Musical Director is struggling to find the right words to express what he wants to say to the choir. When we sing the phrase we will sing, he wants us to pronounce the l-sound in will differently from the way we would speak it.

In speech we make our l big and fat, we swallow it a little bit…

And he demonstrates the way he wants us to pronounce will when we sing.

It’s a good thing he does demonstrate it, because he doesn’t have the technical vocabulary to explain what he means. And if he had, the choir members probably wouldn’t have understood it.

If the DMD and the choir members all knew some phonetics, he could have said

When speaking this phrase we use a dark l, or even a vocalized sound. But in singing it I want you to use a clear l.

I have to hand an excellent book entitled A Handbook of Diction for Singers, by David Adams (OUP, 2008). Excellent, that is, up to a point. The author is not a phonetician (and does not claim to be). Here is what he tells us about “the Italian /l/ sound”. Although he is American, and is targeting speakers of AmE, he is attempting to make the same point as our DMD. Let’s analyse this account.

• “…with the tongue rather tense…” I know of no evidence that the two languages differ in the degree of lingual tension for l.

• “…making contact with the palate well behind the teeth…”. That is, English l is alveolar, while Italian l is dental.

• “…with a collapsed pharynx…” I’m not sure what would be involved in the collapse of the pharynx, but I’m sure it doesn’t happen every time we utter English l. Narrowed or constricted pharynx, could be; but not collapsed.

• “…the tongue must be forward…” Now we’re getting to it. Not only is the Italian sound dental, but the body of the tongue has a different configuration.

• “…the vowel shape of the preceding vowel must be behind the tongue…” Sort of true. But how do you put a “vowel shape” “behind the tongue”, rather than somewhere else? Certainly the l of Italian crudele kɾuˈdeːle should have front, e-type resonance, rather than the backer vowel resonance of the l in AmE daily. But it would be wrong to suggest that the l in Italian nulla must have u-resonance. It’s ˈnulla, not ˈnuɫɫa. Likewise, the l in colpo is not o-coloured.

• “…the sound must have resonance as well as forwardness.” But l always has resonance of some kind. The question is, of what kind?

As Kenneth Pike documented in his ground-breaking Phonetics (1943), there is a sense in which the use of vague impressionistic terms, even if they are not well-founded in anatomical and physiological fact, does not matter, as long as the teacher also offers an ostensive definition by directly demonstrating the sound involved. He called these pseudo-facts “imitation labels”.

How would I teach English-speaking students of phonetics to produce a clear l when required?

I’d start by getting them to produce some kind of l involving contact of the tongue tip with the teeth or the alveolar ridge. Keeping the tongue tip in its place there, I’d point out that it is possible to arrange the remainder of the tongue in all sorts of different configurations, which correspond to all the various vowel sounds the students know how to make. I’d get them to produce an ɑ-coloured l, then an ə-coloured, an ɔ-coloured, an ɛ-coloured, an i-coloured, even perhaps an y-coloured or œ-coloured lateral. Try all the possibilities! Make it a game! We call the laterals with back-vowel resonance “dark”, those with front-vowel resonance “clear”.

Then I’d get them to analyse what colour of l they use in various English words. Try let, look, allow, bullet, tell, field, full, all…. Are they all the same? If not, what determines the choice between them? Are the l-sounds different when we compare tell us and tell them?

Then I’d demonstrate some Italian words and get them to imitate them and analyse them in the same way. Does the l in crudele have the same vowel coloration as the l in English daily? (For some British students, it might.) If not, what is the difference? How does Italian alto differ from English alto? How does the Italian pronunciation of coloratura differ from the English pronunciation of the same word? (Not just in the l-sound!) What about fatale faˈtaːle compared with English fatal ˈfeɪtɫ̩? Say molto ˈmolto, bella ˈbɛlla, scialbo ˈʃalbo, all with a clear l.

Maybe it wouldn’t work any better or worse than Adams’s approach. But at least I’d have the warm glow of believing I had taught my students something true rather than obfuscate the matter with nonsense.

where the cow slips

It is tiresome that the English letter s (single, non-initial) does not always make it clear whether the voiceless sound s is involved or the voiced z. Just think of gas and has, use as a noun and use as a verb, or answer and pansy.

Fortunately there are some rules that hold fairly well. In final clusters, after a voiced consonant we always get z: fibs, adds, begs, slams, hens, tells. The -s is almost always inflectional, but the rule still works when it isn’t: lens lenz.

This is why newsreaders and others tend to mispronounce the name of the city of Homs, now in the news. As the Oxford BBC Guide to Pronunciation reminds us, its is properly hɒms, rather than the hɒmz we often hear. The only way this could be clearly signalled in English spelling would be if we were to write Homss.

In Arabic it’s actually حيمص Ḥimṣ ħimsˁ, with a final voiceless ‘emphatic’ (pharyngealized) alveolar fricative. The city was previously known as Emesa (Greek Ἔμεσα), again with a voiceless fricative.

I notice that in the Japanese Wikipedia it’s ホムス homusu, not ホムズ homuzu. Quite right.

I have known the word cowslip, the name of a wildflower, from an early age, and have always pronounced it ˈkaʊslɪp. As far as I know, so does everyone else. The s here does not trigger pre-fortis clipping, so it is natural to analyse the word as cow plus slip.

Recently I had a sudden thought: is it really from cow’s plus lip? Does the flower or its leaf look like part of a bovine mouth? (There’s another, similar, flower called an oxlip.) That would make the fricative an inflectional z. Are there, or were there, people who say ˈkaʊzlɪp?

But no. The OED tells us that its etymology is

Old English cú-slyppe, apparently < cú cow + slyppe viscous or slimy substance, i.e. ‘cow-slobber’ or ‘cow-dung’ (compare German kuh-scheisse as a plant-name in Grimm)

It’s got nothing to do with a cow’s lip. Where the cow slips, there slip I.

ooh!

Students of phonetics in Britain have to learn to recognize the Cardinal Vowels established by Daniel Jones: at least the primaries (i e ɛ a ɑ ɔ o u) and four of the secondaries, namely y ø œ ɯ. Masters’ students have to learn not only to recognize them but to produce them, too.

Some readers may be surprised to learn that the cardinal vowel that generally proves most difficult for English and Scottish students to produce is primary number 8, u. These students have to learn to make a vowel sound that is considerably backer and rounder than their English GOOSE vowel.

This is also the cardinal vowel that Japanese students find most difficult.

If simple imitation failed, I generally found that the most helpful technique was to start from the English word wall. The BrE vowel in this word is reliably back. More importantly, so is the close and rounded w at the beginning. If you prolong this w instead of immediately gliding away from it, the result may be an acceptable cardinal-style u — properly close, back, and rounded. It may need to be made a little “tighter” (i.e. with a greater degree of tongue raising). Once the learner has produced that satisfactorily, you just need a few fluency and catenation exercises. Then you can compare and contrast English boot with cardinal but and moon with cardinal mun. (NB cardinal vowels have no inherent length. They can be prolonged or not at will.)

Here’s Daniel Jones’s demonstration of cardinal 8, from the recording he made in 1956.

The same difficulty faces the English-speaking learner of German. German long uː is just about cardinal. You can hear some authentic examples here, on Paul Joyce’s German Course site (the URL mentioned for this site in my blog for 10 July 2009 is no longer valid).

(Warning: to my ear the sound clip for this vowel on the Univ. of Iowa site sounds extremely odd and un-German. To hear it, go to Vokale, Monophthonge, hinten, and select /u/.)

Here’s Wikipedia’s sound clip for the word Fuß fuːs.

If I were teaching German uː I would apply the same technique. I’d emphasize the difference in sound between German du duː and English do, German Hut huːt and English hoot. And of course learners of German also have to master the front-back distinction in Brüder — Bruder ˈbryːdɐ — ˈbruːdɐ. (Both tend to get mapped onto English brooder.)

One of Joyce’s examples, Stuhl ʃtuːl, is particularly interesting. For many English people the vowel in this word, because of the following dark lateral, is not all that different from that of their English stool: the initial ʃ is no problem, but the final clear lʲ is strikingly different from the usual English ɫ used in this position.

My picture shows an eagle owl, German der Uhu ˈuːhu. Its name is onomatopoeic. It hoots in a cardinal way.

university challenged

Thanks to Stephen Bullon for drawing my attention to the BBC TV programme University Challenge, which this week had a question requiring the identification of Shakespearean plays from IPA transcriptions of quotations from them. (If you are in the UK you can still, for the next day or two, access this episode on the iPlayer. From about 8:00 in.) It started off well enough, except perhaps that Jeremy Paxman, the questionmaster, characterized this starter question test as “rendered into phonetic English”. (As opposed, we might ask, to what other kind of English? Better, “transcribed into phonetic symbols”.) A member of the Balliol team correctly recognized “Beware the Ides of March” as coming from Julius Caesar.

But the follow-up “picture questions” were rather shocking. What do you make of this? OK, it’s “I know a bank where the wild thyme blows…”. But what kind of accent is this? Is it some kind of American or Irish English, with ɑ rather than ɒ for the LOT vowel in oxlips and nodding? Well, no, because it’s non-rhotic. Why is it so oddly transcribed? Why is the TRAP vowel rendered as æ in bank and and but as a in over-canopied? You can make a case for writing the PRICE diphthong as aj, but not for writing it indifferently as aj, ai and ɑj in just four lines. It gets worse. OK, here line 1 is still non-rhotic, but then we go rhotic. No problem. But what on earth is waːnd itsɛlf? Turns out it’s meant to be wound itself. Strangely, the last transcript offered was fine (though we might quibble about inconsistent stress marking, not to mention prevocalic tə alongside preconsonantal tu).

How could it happen that the BBC, in this prestigious programme aimed at a highly literate demographic, could make such a mess of simple phonetic transcription? The second and third screens were worse than average beginners, doing their very first transcription exercise, would come up with.

Yet the BBC, in its Pronunciation Unit, employs three highly qualified phoneticians, any one of whom could in ten minutes have supplied accurate transcriptions for the programme.

I think that using ignorant amateurs to set questions for a university-level quiz on national television is nothing short of scandalous.

[Since writing this, I have just noticed that John Maidment has said much the same thing in his blog, too.]

desiccated

Do you ever use desiccated coconut, when making a cake for example? If so, how do you pronounce the verb to desiccate?

Perhaps the more usual question about this word is how to spell it. The only pronunciation I have ever heard, as far as I remember, is ˈdesɪkeɪt. This stressing and vowel pattern leads people to think it should be spelt dessicate. It’s only those few of us with a knowledge of Latin who immediately see it as containing the adjective siccus ‘dry’, which explains its unexpected spelling.

But let’s return to the question of its pronunciation. Surprisingly, the OED tells us that until 1864 deˈsiccate was the only stressing given in dictionaries. The OED itself (in the second edition, 1989) still gives priority to the pronunciation dɪˈsɪkeɪt. I wonder if anybody alive actually says that, or indeed if there was anyone who still said it twenty years ago.

For the historical change in stress the OED refers us to contemplate. Shakespeare apparently stressed that word as we do today (ˈkɒntəmpleɪt).

When this is known, then to divide the times:
So many hours must I tend my flock;
So many hours must I take my rest;
So many hours must I contemplate;
So many hours must I sport myself;
So many days my ewes have been with young;
So many weeks ere the poor fools will wean:
So many years ere I shall shear the fleece:
So minutes, hours, days, months, and years,
Pass'd over to the end they were created,
Would bring white hairs unto a quiet grave.
Ah, what a life were this! how sweet! how lovely!
—Henry VI, Part III: II, 5

Nevertheless, the OED tells us,

the orthoepists generally have conˈtemplate down to third quarter of 19th cent.; since that time ˈcontemplate has more and more prevailed, and conˈtemplate begins to have a flavour of age.

The OED continues (in the second edition, 1989, still)

This is the common tendency with all verbs in -ate. Of these, the antepenult stress is historical in all words in which the penult represents a short Latin syllable, as acˈcelerate, ˈanimate, ˈfascinate, ˈmachinate, ˈmilitate, or one prosodically short or long, as in ˈcelebrate, ˈconsecrate, ˈemigrate; regularly also when the penult has a vowel long in Latin, as ˈalienate, ˈaspirate, conˈcatenate, ˈdenudate, eˈlaborate, ˈindurate, ˈpersonate, ˈruinate (Latin aliēno, aspīro, etc.). But where the penult has two or three consonants giving positional length, the stress has historically been on the penult, and its shift to the antepenult is recent or still in progress, as in acervate, adumbrate, alternate, compensate, concentrate, condensate, confiscate, conquassate, constellate, demonstrate, decussate, desiccate, enervate, exacerbate, exculpate, illustrate, inculcate, objurgate, etc., all familiar with penult stress to middle-aged men. The influence of the noun of action in -ation is a factor in the change; thus the analogy of ˌconseˈcration, ˈconsecrate, etc., suggests ˌdemonˈstration, ˈdemonstrate. But there being no remonstration in use, reˈmonstrate, supported by reˈmonstrance, keeps the earlier stress.

Except that nowadays remonstration is in use, and the stress pattern ˈremonstrate has accordingly become usual (in BrE at least). The OED acknowledges this, commenting in its third edition (2009) as follows.

N.E.D. (1906) gives the pronunciation as (rĭmǫ•nstreⁱt) /rɪˈmɒnstreɪt/, but in O.E.D. (ed. 2, 1989) this is marked as being ‘older’ and a pronunciation with first-syllable stress is given as the dominant one. Editions of D. Jones Eng. Pronouncing Dict. up to and including 1963 record the pronunciation with stress on the second syllable as the dominant one, and this is still indicated as an alternative pronunciation even in British English in subsequent editions.

True. But nowadays we’re fed up with football managers who ˈremənstreɪt with the referee. Aren’t we?

ˈtrʌbo

EFL teachers in Argentina receive excellent detailed training in English phonetics, and quite rightly want to apply this knowledge in the classroom.

From time to time I receive queries from classroom teachers there which reflect the fact that like all teachers they want a clear defined set of facts to teach, whereas we academics who train them and who write about the language tend to be conscious of the chaotic nature of the real world, in describing which any generalization has to be qualified by uncertainties and indeterminacies.

María Inés Orge asks about the usage of the schwa symbol in words like trouble and people.

The first option given in the phonetic dictionary of each word is with schwa while the the second is not. They are considered “optionals”. “Optional”,according to the dictionary, means “something you do not have to do or use, but you can choose to if you want to”.

She wants to know, then,

During a dictation of phonetics, can the option ə be considered as a mistake? Is it possible to take the sound between b and l or p and l completely out? Can I really choose the option or not?

I told her

Both pronunciations are possible. But on any given occasion the schwa is either there or not there.

It is perhaps clearest in cases like garden. If there is no schwa between the d and the n (the usual pronunciation) then the tongue remains in contact with the alveolar ridge as we move from d to n, and the only change is the movement of the soft palate, which comes down to allow the air to explode through the nose. If, on the other hand, the tongue tip leaves the alveolar ridge at the end of d and then returns to the alveolar ridge for the n, then there is a schwa between the two consonants.

In marking dictation, it is for you to decide your policy. I would not penalize presence/absence of schwa between a fricative or an affricate and n or l (as in listen, heaven, kitchen; oval, puzzle, satchel), but might penalize it after a plosive (as in happen, garden, organ; apple, middle, eagle), where the difference is perceptually more salient.

On the other hand you could decide not to penalize this at all, since the two possibilities (i) schwa plus sonorant and (ii) syllabic sonorant are phonologically equivalent. Barring marginal cases, there are no pairs of words distinguished only by this difference.

The “marginal cases” I was thinking of would be, for example, BrE ˈpætən, ˈbɪtən (pattern, bittern) vs. ˈpætn̩, ˈbɪtn̩ (Patton, bitten), which a few non-rhotic speakers may have as minimal pairs, although they are normally homophonous for me as ˈpætn̩, ˈbɪtn̩. Compare also modern as a rhyme (or not) for trodden.

I could have continued by mentioning the likelihood, these days, of l-vocalization in trouble, people and other words shown in the dictionary as having “^əl”. If we represent the output of vocalization conventionally as o, that gives ˈtrʌbo, ˈpiːpo. In a transcription exercise (orthography to phonetics) I would be delighted to see these forms (particularly if phrase-final, or if the next word begins with a consonant sound). In a dictation exercise, however, I would not consider them correct if I had actually uttered l̩ (i.e. ˈtrʌbɫ etc). In the general scheme of things, though, this would count as a very minor error. People who fail phonetic dictation do so because of multiple gross errors, not because of subtleties such as worry María.

Remember, though, that optional symbols in the dictionary should not be shown as optional in these practical exercises. They should either be there or not be there. Their inclusion in the dictionary is an abbreviatory convention. In real-life performance nothing is optional. You either do it or you don’t.

an indefinite oddity

In my Phonology of English course I used to give my students a homework task as follows. I am sure the readers of this blog would have no difficulty in answering this question. We all know that the choice between preconsonantal a and prevocalic an is generally determined by pronunciation, not by spelling.

• We say and write an uncle, an urgent message ən ˈʌŋkl̩, ən ˈɜːdʒənt ˈmesɪdʒ but a unit, a Eurocrat ə ˈjuːnɪt, ə ˈjʊərəkræt, because unit and Eurocrat, despite being spelt with an initial vowel letter, begin with the sound j, and in English the two semivowels count as consonant sounds.
• We say and write a house, a heavy weight ə ˈhaʊs, ə ˈhevi ˈweɪt but an hour, an honourable man ən ˈaʊə, ən ˈɒnərəbl̩ ˈmæn, because hour and honourable, despite being spelt with an initial consonant letter, are pronounced with an initial vowel sound (the h being ‘silent’).

There’s a little problem that crops up in words beginning with the letter h in an unstressed syllable: words such as historic and hotel. Nowadays we normally say ə hɪˈstɒrɪk ɪˈvent, ə həʊˈtel. Do we write, correspondingly, a historic event, a hotel? Yes, most of us do. But some people prefer an historic event, an hotel (which to me could suggest a certain literary prissiness or conscious archaizing). Anyhow, both are accepted in writing. The Concise Oxford Dictionary puts it thus:
The h used to be dropped in these words precisely because it was in an unstressed syllable, and in polite Victorian-era English the sound h was apparently restricted to the position before a stressed vowel. A few people, not vulgar h-droppers, perhaps still today pronounce historic(al), historian with no h, at least when preceded by the indefinite article.

So far so good. But over the weekend I was reading the legal judgment in the case before the Court of Appeal concerning the Christian owners of a bed-and-breakfast who had refused a room to a gay couple. The learned justices had frequent occasion in this judgment to use the words homosexual and heterosexual preceded by the indefinite article. And in their written report they chose the form an.

This does seem strange. The initial syllables in these words do not bear the main word stress, but they do bear a lexical secondary stress, which (i) protects the vowel from reduction and (ii) is often realized as a rhythmic beat. I have never previously seen a published (non-dialectal) text that uses an in this position.

I wonder if the judge(s) in question pronounced the phrases with the n when delivering the judgment. I think it unlikely that they would have dropped the initial h.

Interestingly, the legal regulations that the judges quote refer to “a hotel”.