Smoothing, then and now
The description of contemporary SSB (Standard Southern British) does not really require centring diphthongs. In this it’s very different from the earlier prestige accent Received Pronunciation. RP’s system of centring diphthongs, though still promulgated in dictionaries and textbooks, was a transitional phenomenon between the rhotic system which pervaded England until the 18th century, and the more monophthongal system we have in the 21st.
The source of the centring diphthongs (diphthongs ending in schwa) was the vocalization of r when not before a vowel, simplistically as follows:
The monophthongization of the resultant centring diphthongs spread first up the back vowels and then up the front vowels. (Splitting the diphthongs into 2-syllable sequences was an alternative fate for uə and iə.)
A dynamic demonstration of the difference between RP’s centring nature and the system of today is provided by an optional speech process known as smoothing. This affects certain long vowels when they occur before a schwa in the next syllable. The 2-syllable sequence may optionally be smoothed into a single (1-syllable) long vowel.
Smoothing has been around for a long time, but contemporary smoothing works differently from the smoothing of old RP. The older smoothing was described by John Wells as follows:
A long vowel or diphthong changes: iː becomes ɪ, uː becomes ʊ, and a diphthong loses its second element, so that aɪ and aʊ become a.
(Longman Pronunciation Dictionary, 3rd ed., p. 173)
Note that in this older smoothing, the following schwa was unaffected, so that the process had the effect of adding even more centring diphthongs to RP, even reinstating historically lost ones:
iː + ə → ɪə
eɪ + ə → eə
aɪ + ə → aə
ɔɪ + ə → ɔə
uː + ə → ʊə
ɑʊ + ə → ɑə
and
əʊ + ə → əə (not actually a diphthong)
So fire faɪə could become faə and power pɑʊə could become pɑə, etc. The prevalence of centring diphthongs in RP was one the features that gave the accent its distinctive flavour – a flavour perceived as old-fashioned today. Here from an old Pathé newsreel is an assortment of old-fashioned centring diphthongs – “they find th[eə] way… t[ʊə]ring the beautiful countryside… natural p[ɑə]… produced ent[aə]ly with the aid of automatic machines”:
The newer smoothing is simpler: treating contemporary FLEECE and GOOSE as diphthongs, we can just say that the process applies to diphthongs. The output of the (still optional) process is a long monophthong:
ɪj + ə → ɪː
ɛj + ə → ɛː
ɑj + ə → ɑː
oj + ə → oː
ʉw + ə → ʉː
aw + ə → aː
əw + ə → əː
Let’s listen to some examples. First, opera singer Simon Keenlyside, smoothing variety from var[ɑjə]ty to var[ɑː]ty, which could rhyme with party:
Here a TV interviewee smooths liabilities from l[ɑjə]bilities to l[ɑː]bilities:
To demonstrate the monophthongal nature of the smoothing, here is the speaker’s smoothed ɑjə edited before –st, which gives a pretty natural-sounding last lɑːst:
(This also supports the analysis of the contemporary PRICE vowel as ɑj, with a first element identical in quality to PALM/START.)
Now let’s look at the MOUTH diphthong aw followed by schwa. Here is chef Sam Clark saying sourdough with an unsmoothed awə in sour:
Nicely demonstrating the optionality of smoothing, he also produces the word in a smoothed version with saː in sour:
The first part of this word sounds like a longish pronunciation of the word sad:
which supports the analysis of the MOUTH diphthong as beginning with the same quality as the TRAP vowel, a.
Next, the FACE diphthong ɛj followed by schwa. Here is UCL’s Prof. Mark Miodownik saying thinner layers with an unsmoothed sequence ɛjə in layers:
Whereas here he produces layer smoothed to lɛː (like lair) in an atomic layer of graphene:
When the GOAT diphthong is immediately followed by schwa, there is no difference between the effects of old RP smoothing and those of the newer kind. Since GOAT begins with schwa, the result either way is a long schwa, ie the NURSE vowel. Here is RP speaker John Wells saying lowered with a smoothed əː vowel:
Words in the NEAR set are often pronounced as sequences of FLEECE + schwa. FLEECE itself is a narrow diphthong ɪj, so that near may be pronounced nɪjə. When this undergoes contemporary smoothing, it produces a long version of the first element, ɪː. The hɪjə pronunciation of here can be heard from Prof Mark Miodownik saying here a carbon fibre composite is being made:
And this is BBC presenter Andrew Plant saying all these Christmas cards here, with the smooth hɪː pronunciation:
(Many younger speakers produce NEAR words generally in the smooth ɪː form.)
Contemporary smoothing is particularly common immediately before r or dark l. Above we heard Mark Miodownik smoothing layer in layer‿of. Here is Prof. Jim Al-Khalili saying curious and inquiring mind, with smoothing of inqu/ɑjə/ring to inqu/ɑː/ring:
Turning to cases with dark l, I’ll and aisle may be pronounced ɑjəl, where ‘pre-L breaking’ adds schwa between the diphthong and dark l; and this can be smoothed to ɑːl. Prince Harry does exactly this in “he’s going to walk down that aisle with his future wife”:
Similarly, Royal Mail may be pronounced rojəl mɛjəl, but is smoothed here by telecoms honcho Richard Hooper to roːl mɛːl:
I’ll end with GOOSE + schwa. The Oxford Advanced online dictionary’s audio pronunciation of secure is unsmoothed sɪkjʉwə. On the other hand, the audio for security exhibits contemporary smoothing, sɪkjʉːrɪtɪj, with ʉwə smoothed to ʉː before r:
Of course the dictionary sticks to the old-fashioned transcription /ʊə/ in both words, /sɪkjʊə/ and /sɪkjʊərɪti/. This transcription reflects neither of the actual pronunciations. Furthermore, the old account of smoothing as stated by John Wells has nothing to say about why the two words should exhibit this pronunciation difference, since old smoothing didn’t even apply to /ʊə/.
Geoff,
Fascinating post as ever!
Your readers might be interested to read this article of mine as well:
http://alex-ateachersthoughts.blogspot.com/2011/03/pm-address-in-cahro.html
Thanks for the link, Alex. I could have made it clearer in my post that, in contemporary smoothing, [ɑɑ] comes from PRICE + ə, while [aa] comes from MOUTH + ə, e.g. flower pot [flaapɒt]. On a blog post, John Wells suggested that Cowell might possibly be heard as [kɑɑl], but I’d be very surprised to hear this: nowadays [kɑɑl] comes only from Carl/Karl and (by smoothing) from Kyle.
An apparent exception is the common pronunciation of our as [ɑɑ], but that is a lexicalization of old smoothing, like prayer, which I presume was once disyllabic.
Incidentally, I increasingly hear mayor desmoothed as [mɛjə]; I can’t say whether this is American influence, a spelling pronunciation, or a hypercorrection which confirms the productive (but of course optional) status of smoothing in the language today.
Interesting post.
Surely the NORTH vowel must have gone through a similar process before merging with THOUGHT: /ɒr/ -> /ɒə/ -> /ɒː/ -> /ɔː/?
I’m not sure I buy your claim that the NEAR vowel should be seen as containing a hiatus of FLEECE followed by schwa. “Sun” headlines aren’t necessarily the source of linguistic truth 🙂
Thanks for reading and commenting! I described r vocalization “broadly” with five vowel qualities to illustrate, and without differentiating e.g. NORTH v. FORCE. I’m no authority on Middle or Early Modern English, but I believe I’m right about the sequence that the subsequent Modern English monophthongization has followed.
Do you have any actual evidence for not buying NEAR as FLEECE + schwa? Obviously the pronunciation of near as literally [nɪə], with KIT gliding to commA, sounds very old-fashioned today. In a stressed citation form, speakers today will typically produce something that certainly sounds like FLEECE gliding to commA. Of course, in continuous speech with less stress, NEAR words are likely to undergo contemporary smoothing and to be pronounced as [ɪɪ], e.g. n[ɪɪ]sighted or b[ɪɪ]glasses.
P.S. I’m certainly open to puns and language games as a source of evidence, rather than just swallowing what dictionaries (and YouTube tutorials) claim. But as it happens, that Sun headline doesn’t really bear on the issue, since old-fashioned R.P. would have neutralized “beer” and unstressed “be a” too – as [bɪə].
Geoff
Congratulations on your splendid bloggery
JCW seems to have re-invented ‘smoothing’
Sweet used the term repeatedly, first acc to OED in
1888 Sweet Eng. Sounds 22 Smoothing or the levelling of the two elements of a diphthong under a monophthong is the result of absorption.
Many many thanks Jack. (Of course the Wellsian re-boot of ‘smoothing’ is about turning 3 phthongs into 2, not 2 into 1.)
[Please excuse any misapprehensions due to ignorance in what I write. I’m not a professional, just a fascinated reader.]
If monophthong + vocalised r gave a centring diphthong, then surely diphthong + vocalised r gave a centring triphthong? Your examples of smoothing (eg fire, sour) are all of that nature, and that’s more special than the general diphthong + schwa. For example, cue-er and shoer (one who cues or shoes) behave differently from cure and sure.
That said, your account of those triphthongs being smoothed to monophthongs feels right to me as someone brought up in West London. But in that case there’s a gap in your vowel inventory for the lengthened version of the TRAP vowel, the smoothing of sour. (It seems to me that it is definitely not the PALM vowel: eg power is distinct from pa.)
So, first point, there’s a missing vowel.
Second point, perhaps it was already there.
It has long seemed to my own introspection that in some words, ostensibly with the TRAP vowel, it is actually lengthened to the way you describe your smoothing of sour. It seems rare. The only examples I can think of are bad, mad, sad and man, so on the face of it could be just a variability or even my own idiosyncrasy. But it can make a difference. I would make madder (more mad) distinct from madder (the plant), madly from Madley (in Herefordshire), and manly from Manley.
So to summarize:
* There’s a gap in your list of long tense linking-r vowels for a lengthened version of TRAP.
* It may well be needed as the destination of sour.
* It has already been present in isolated examples.
* It makes a more systematic correspondence between long tense linking-r vowels and short lax vowels. (And, by the way, is there really any difference in SSB between STRUT and PALM other than length?)
ps Just discovered a Wikipedia discussion of the bad-lad split.
Very interesting things about this new smoothing.
Still, I think it’s simply not correct (other than politically maybe, which I don’t imply) to claim the SQUARE diphthong is a matter of the past. Obviously, one could label every living instance to be either obsolete or PLE, but that wouldn’t be useful, I think. (To see if I’m imagining this, I searched YouTube for “leicester square”.) Also, given that there are still RP speakers around who have a diphthong in FORCE vowels, the 19th century sounds a bit early for its general death.
Thanks for reading and commenting! I didn’t exactly say that centring SQUARE is a matter of the past – I pointed out that [eə] is well preserved in less prestigious accents, but otherwise is perceived as very old-fashioned. (I’ve made the same point about STRUT and the allophony of commA in my latest post.)
I just checked the audio for square on Google Translate, Oxford Advanced Learners, Macmillan and Cambridge (I also made spectrograms). The first three are monophthongs; Cambridge glides slightly, but goes nowhere near schwa. I don’t think these dictionaries have trendy, politically correct agendas; if they did, why would they all give skweə as the transcription, which they do?
As a practical standard for learners, I actually do consider that centring SQUARE is – or should be – a matter of the past; I think it’s silly to teach it to the general (and generally young) learner of British English today. In audio terms, reference sources seem to agree with me.
Interestingly, my biggest problem was with your comment on the CURE vowel. Are you familiar with Bente Hannisdal’s 2006 “Variability and change in Received Pronunciation: A study of six phonological variables in the speech of television newsreaders”?
Hi, I re-read your old blog entries whenever I can and always find something I had missed the first time. Thanks for demystifying English transcription system! I’d always been mislead by the great gap between the IPA vowel chart and the actual phonetic quality in contemporary English. Now I’ve got a question for you, and I’d really appreciate it if you could answer it.
I’ve never been able to hear the final schwa of centring diphthongs when followed by a fully pronounced ‘r’, as in ‘hearing’, ‘Europe’, or the old-fashioned diphthongal variant of ‘hairy’. I’ve been demonstrated these diphthongs in this context by my (young, native, Southern British) English teachers in careful speech, but I’ve never heard it in natural speech.
I know your views on SQUARE, NEAR and CURE, and how each of these are disappearing due to their instability. But I suspect that where a native speaker hears a diphthong, I can’t hear one. Can you point to some audio clips where (even in dated English) these schwas are actually fully pronounced? Or are dictionary transcriptions also mythical in this respect? Is there a phonological rule that states that the schwa is lost in the presence of a following /r/? Or that the schwa and the /r/ merge into one segment? Or does the schwa become ultra-short? I’ve searched Cruttenden’s revision of Gimson’s Pronunciation of English, Wells’ Accents of English, Roach’s English Phonetics and Phonology, and more, but I found no such phonological rule or even a comment on this, so I suspect it’s lack of ear-training on my part.
Thanks a million!! (even if you don’t get round to replying)
Hernán