The unstressed word-final -y vowel in General British pronunciation
(This has been very slightly adapted and corrected from the version which first appeared in 1990 in Studies in the Pronunciation of English (ed.) Susan Ramsaran. London: Routledge pp 159-67)
2. Although Gimson would no doubt not have questioned the phonetic substance of Jones's detailed accounts of the behaviour of these words, he preferred not to deal with them as Jones had. Gimson (1962) simply added after the phonemic transcriptions of such words as had strong forms with /i:/ and /u:/ a re-transcription in square brackets respectively [i] and [u] with no further comment. In Gimson (1975: 49) a minimal comment was supplied 'Often, however, the vowel, though shortened, remains tense'. Windsor Lewis (1976), a review of Gimson (1975), made a clumsy attempt to provoke him to comment on this phonological problem but he was not to be drawn. Part of its interest seems to be that, in traditional "RP", speakers had in miniature what was to become in the second half of the century the problem of the nature and phonological status of what Wells (1982) refers to as the happY vowel.
3. That prodigious storehouse of ideas and information on English pronunciation contains the fullest account so far of the development of the final -y vowel in modern British English. Among its comments (p. 165) are (i) to classify it among 'weak (unstressable) vowels occurring word-finally' whose 'phonemic identification with strong vowels will usually be debatable' and (ii) to say of its quality in his two 'reference accents' that 'Most RP, and conservative varieties of Gen Am, have [ɪ] for it.' This symbol perhaps calls for some clarification. Wells (1982) incorporates as p. xx the chart headed The International Phonetic Alphabet, in the form of a revision produced by Wells himself in 1979, which equates [ɪ] with /ɩ/. However, the rather close placing of [ɩ] on that chart should not be taken as its specific quality. It has not got such a defined quality in the way the cardinal vowels have but is to be understood as defined by the user. Wells (1982: 127) conveys his value for the symbol where the KIT vowel is described as 'an unrounded vocoid [ɪ], centralised from and somewhat closer than cardinal 2'. We shall also use [ɪ] with a half-close centralised value.
4. Wells (1982: 294) remarks that 'a word such as city may have two vowels virtually identical in quality and duration'. He continues with a reference to the only -y variants mentioned by Jones or Gimson as within "RP", ie opener ones, as 'approaching the quality of' DRESS and identifies them as sociologically conspicuous ("Upper-crust RP"). He then observes that it appears that recently RP may be beginning [my italics] to prefer a closer quality, adding that speakers of "adoptive RP no longer seem to regard ['hæpi], [sɪti] etc. as regionalisms which should be avoided in cultivated speech'. This comment one can endorse but at the same time noting that there has never been very much evidence that such pronunciations were noticed as provincialisms – certainly not of [i] as opposed to [i:] – and there have been plenty of elocutionary pundits over the last 200 years willing to issue lists of proscriptions, for example Walker (1791), Sheridan (1789), Ellis (1875).
5. In order to be able to offer some rather more detailed comments on the types of -y value found in General British pronunciation, although admittedly the various types are gradiently rather than sharply divided, it will be convenient to discuss them under the following headings: (1) Traditional (2) Open (3) Russellian (4) Variable (5) Close and (6) Non-enclitic.
6. Speakers with the Traditional treatment of the -y vowel have matching vowel qualities in the two syllables of words like city and fifty and also in ones like cities and pitied (which will thus be homophonous with pitted). If they aim at strong contrast in a sentence like Not fifteen but fifty they are unlikely to give nuclear stress to the final -y but if so it will match the first vowel of the word. They make either no adjustment at all to a closer value before a rhythmically closely following vocoid or, if any, very little. Gimson belonged to this type as can be plainly detected by listening to the recording of Gimson (1975) with its many numbers above 19 of which only A68 tends to sound a little like A-6-D-8.
Certain speakers tend to use slightly opener and/or more central values than the Wells (1982) norm with little or no sociological effect. Hawkins (n.d.) has suggested that 'in RP ... /ɪ/ is gradually developing into a central vowel'. Whether this is an increasing tendency or not, there have been speakers with well-centralised values for a long time. The present writer has an elderly acquaintance (a lady born in 1907, boarding-school education, conspicuously upper-class accent) whose pronunciation of Newbury seems virtually indistinguishable from Newborough.
The Traditional -y ordinary vowel quality sounds neutral in sociological terms but socially conspicuous varieties are produced by some speakers by drawling and/or adding particular voice-production qualities to it. If drawled it may exactly coincide with one possible realisation of /ɪə/ so that, for example, funny will suggest funnier. It may even possibly be accompanied by a slight schwa off-glide from such speakers but anything of that sort at all marked is probably confined to comic impressionists' characterisations of the "upper-class twit".
More commonplace is the addition of a voice-production quality effect brought about by what is apparently the same kind of pharyngeal contraction as has been noted as characterising some speakers' /æ/ vowels, especially the more old-fashioned closer types (cf Gimson 1980: 108). This may be heard from, for example, the more conspicuously aristocratic type of old Etonian and from some Yorkshire dialect speakers. It was characteristic of Noel Coward's -y, though he belonged not in our Traditional but in our Variable category as can be heard from his distinctly close -y used at house-party in Norfolk on the gramophone recording of an extract from Private Lives.
7. The next two categories are the least common and most old-fashioned types. The Open type is becoming pretty unusual if it ever was very common. It is a type quite familiar from comedy actors, with often the pharyngeal tension mentioned above, portraying Colonel-Blimp-type officers, as Wells mentions.
An example of a once well-known speaker who had a strikingly open -y value was the onetime Foreign Secretary, the late Michael Stewart (born 1906). His variety (possibly more fronted than most) was such that it coincided with some speakers' (monophthongal) versions of the final syllables of essay, survey, etc. He was a public school product but with no particular conspicuous pronunciation features (if his -y is not so counted).
Open -y values also occur in some strongly regional accents, for example in parts of Lancashire and Yorkshire. In such areas one often hears quite markedly close values higher up the social scale.
Daniel Jones and Henry Sweet both had variation between a more and a less open -y. Sweet (1890) even indicated it regularly in transcriptions of connected speech. However, the actual difference between the two values for Jones, though perceptible, was very slight indeed, such that his proper assignment in our categorisations is not to the Open but to the Traditional type.
8. Our third, Russellian, category refers to a type of speaker described in Gimson (1962,1970) as having generally an /ɪ/ which was a 'conservative' form 'much closer than the general ... coming nearer to the quality associated with' /i:/. By 1980 he appears to have decided that the type had become obsolete because he removed the account of it. We have called it after no doubt its most famous exponent, Bertrand Russell (1872-1970) the philosopher. It seems that he had far fewer contacts with other children than is usual when he was growing up so that his speech probably sounded rather archaic to the majority of his contemporaries. (Compare the present Queen's upbringing and occasional archaisms.) Archive recordings show him as capable of close values that are not merely context-prompted but, so it seems, used in various unaccented word-internal syllables as well which would be alien to speakers in our Close category. For example he could say repetition with a quite close second vowel. He was certainly capable of very close -y values. One recording has such a value in twenty feet which may be due to vowel harmony but also an only slightly less close -y in lovely poetess which can't be so explained. This kind of speaker would probably usually make studded and studied homophonous by having a close vowel in both endings. Contrast our Close category below.
9. Speakers in our fourth Variable category exhibit the marked "context-sensitive variation" attributed by Wells (1982: 165) to "Some English northerners, some RP speakers, and some Americans". Such speakers from time to time in varying degree produce -y sounds in pre-vocoid situations which can correspond in quality to their value for /i:/ thus losing on such occasions a contrast available to them when no vocoid follows in close rhythmic relation. For example The khaki is may become indistinguishable from The car-key is, almost invariably so if the second syllable of khaki is boosted by carrying the falling element of a rising-falling tone.
There are problems with putting speakers into this category because the rhythmical integration of the -y with the following vocoid comes under the heading of the speaker's prosodic choice. The criterion for including speakers in the category is that they do produce close values at times. By contrast with Wells's "some" I suggest that there are very large numbers of speakers in this category. Among the radio and television newsreaders I have observed especially between 1963 and 1978 they were the majority, with as many in the older age groups as the younger, including, for example, John Snagge, Alvar Lidell, John Freeman, Ludovic Kennedy, etc. One could mention many well-known public figures in this group, some of them Victorians. The earliest speaker I have noted is Ellen Terry, born in 1847, who was recorded in 1911. She had a very clearly close value in the last final -y of The quality of mercy is not strained even with the intervention of an intonation unit boundary as also in ... happy / if I ... etc. Max Beerbohm (born 1872) is on record with close values in gallery in, poetry about, etc. At the present day the Queen could be heard with them, for example in her Christmas broadcasts in 1967 (unity in) and 1974 (uniquely effective), etc. Many actors have them, for example Lord Olivier in the Henry V (1942) film at the dead with charity enclosed.
Speakers in this, probably the largest GB (General British) category, can have close values triggered off not merely by an immediately adjacent phoneme but by vowel harmony (the name Healey is often heard with a distinctly closer final vowel than the same speaker would use in, for example, Harley) and even by certain prosodic features. Some speakers more frequently make assimilations across intonation unit boundaries than others. The presence of a complex tone seems to be able to conduce to this when it falls on the pre-boundary -y word. Most curious of all such phenomena is the tendency for a Drop (high-to-mid) simple tone to conduce to a close -y value with nothing else following. It is not perhaps surprising when for instance Ivy is heard on a Drop tone as a separate rhythm unit before Watkinson on a falling tone (from a speaker whose -y is not normally close, in this case Mr John Timpson) for the -y to be so close that it cannot be certain that one has not heard I. V. Watkinson except from the context. But it is curious that a speaker may call out on a Drop simple tone a name like Johnny or Mary with the -y a distinctly close type. The only explanation one can suggest is that the speaker is experiencing a sort of accentual double-take, perceiving the second syllable as quasi-accented because the pitch pattern exactly corresponds to a succession of down-stepping level accentual tones. Speakers in this category almost always avoid accenting -y in a sentence like I said fifty not fifteen. The occasional isolated uses of close -y values we have mentioned don't warrant categorising such speakers along with our next group.
10. Speakers of our fifth Close type employ -y sounds of varying degrees of closeness with their weakest forms not necessarily very noticeably different from the traditional [ɪ] but mostly producing a clearly closer form and one that is not employed only before a following vocoid. The phonetic quality will probably, most usually, be more retracted than, but about as close as, their accented /i:/ but not usually diphthongised. They may or may not make homophones of pitied and pitted, taxis and taxes, Rosie's and roses, elc. Not fifteen but fifty will not be ordinarily accented on the final-y but if it were it would probably not have strong diphthongisation.
Wells (1982: 299) says this type "would traditionally not have been considered to fall within RP" but "Recently, however, such forms have begun to be heard from speakers who on all other grounds would be considered as speaking RP." He might well have added that this Close type is not merely free from social sanction but is on the contrary often heard from younger speakers who display other phonetic characteristics which are conspicuously fashionable, notably among the products of some girls' boarding schools. This is somewhat reminiscent of the way a lowered /æ/, perhaps especially if pharyngealised, became a notable feature of the speech of 1950s debutantes. It may not be too fanciful to attribute something of the popularity of each of these phenomena to the creation of an attractively youthful impression because both types of articulation could probably be said at one time to be habits that the juvenile grew out of. Cf Gimson (1980: §7.l2(3)) on /æ/. The young female's variety of this -y is just possibly slightly closer on average than that of others of the Close type. Although, juvenile voices, being high-pitched, will naturally tend to have closer-sounding vowels. The Queen at 14 could be heard with a distinctly closer value, for example in victory, than she was to use subsequently in non-linking contexts.
Wells (1973) (a
review of the present writer's Concise
Pronouncing Dictionary)
appeared to be less guarded: "In the absence of properly sampled
surveys, of course, one can only hazard
guesses at the number of speakers using particular pronunciations, but
I wonder how Windsor Lewis justifies the representation [in a work
specifically excluding socially conspicuous usages] of the vowel in happy ... etc. exclusively with
/ɪ/. I would wager that a clear majority of both American and British
speakers use /i/."
(The answer, by the way, is that the
lexicographer limiting himself to the simplest possible form of
transcription places himself over such matters in the position of the
non-wife-beater who is obliged to reply yes or no to 'Have you stopped
beating her?') It is of relevance to our topic to notice that most
users of English as a foreign language manage perfectly well with -y
but have great trouble with /ɪ/ unless we except those who, like
especially the Germans, place themselves in our Non-Enclitic category. At any rate
the Close value is nowadays
commonly heard from media newsreaders and presenters and seems
particularly widespread among those born after 1960 or so.
11. Our final category, the Non-Enclitic, seems generally to correlate with the Wells (1982: passim) notation [i:]. The Close/Non-Enclitic distinction (Wells uses [i] for the former) is no doubt the one referred to in Jones (1950) by specifying a "long and close" -y at §77 as opposed to the close "but short" value of §78. Surely what is being identified is not only or even essentially length but strength, that is, primarily degree of rhythmic separation of the final -y from the rest of its word. In terms of rhythmic structure it may be useful to go further than Wells's division into weak (unstressed) and strong (stressed) to three categories; (i) strong, (ii) rhythmically independent and (iii) rhythmically dependent, that is, enclitic. The speaker who gives -y a non-enclitic value may make it as strong as any syllable ever is that doesn't carry a nuclear stress. He is even likely to feel quite at home giving contrastive nuclear stress to the final -y- of Not fifteen but fifty which is not so of any of our previous types. Large numbers of Australian and New Zealand speakers fall into the Non-Enclitic category. So do many Americans and some speakers in parts of Britain. It is the least likely usage to be heard from General British speakers.
It shouldn't be imagined that Non-Enclitic category speakers use the heaviest values of -y all the time but they will normally have identical values for khaki and car-key, commentary and `common tree, shortie and `short E, property and `proper tea, normally and `Norma Leigh, trusty and trustee (savings bank). They will exactly identify the final vowels of mimicry and pedigree, happily and jubilee, policy and pharisee, and (if American) effigy and refugee. Pike (1945: 78) showed that his distinction of these last two (both front-stressed) placed his type of General American in our Close group.
Although this type was apparently never heard from national radio and television newsreaders when they were very few in number before the sixties, since then such usages have become fairly familiar, only drawing attention in exceptional circumstances as when the quality on a Drop tone produces the two-Alts double take mentioned above as, for example, when one BBC radio newsreader (Mr Don Durbridge) seemed to be referring to the Four Tees Oilfield, which was homophonous on that tone with what he was actually saying, viz. Forties.
12. Wells (1982: 257/8) is no doubt right in suggesting that from round about 1950 there has occurred a markedly 'increasing tendency ... to use a closer quality' of -y. However, the suggestion there that 'between the seventeenth century and 1950' -y was 'regularly analysed by phoneticians as [ɪ] and implicitly assigned to the /ɪ/ phoneme' calls for some comment. Knowing how much ebb and flow there is in phonetic change it would be unjustifiable to interpret paucity of comment as positive evidence. At one end of that period the evidence presented in Dobson (1957) at §§338 and 350 shows a very mixed pattern which was hardly likely to have become regularised by the eighteenth century. On the contrary we find the two most successful pronouncing dictionaries of the century Sheridan (1789) and Walker (1791) (severely critical though the latter could be of the former) in perfect harmony on recognising the matching of the two vowels not in city but in easy. The former (1789: x) said that "the sound of Y perceived in the last syllable of lovely is only the short sound of e in beer". And the latter (§181) distinguishes values of the letter y which include "its short sound, heard in system, syntax etc" and "the unaccented sound ... always ... like ... e ... thus vanity, pleurisy etc, if sound alone were consulted, might be written vanitee, pleurisee etc." Both of these authors wrote at length giving advice to those who wished to eliminate Cockneyisms, provincialisms or foreign accent but with no suggestion of problems with -y sounds.
When we come to the hundred years or so before 1950 it is perfectly true that there is nothing to reveal the existence of anything but our Traditional and Open types in the descriptions of Ellis, Bell (spelling-hypnotised on unstressed syllables anyway), Sweet, Soames, Murray, Wright, Wyld or even Jones insofar as 'receiving' them was concerned. One supposes it is safer to consider Jones as holding to a rigid definition of "RP" rather than failing to observe the closer values, but they were to be heard from people he knew well that most of us would regard as General British speakers. Dr Julian Pring, a member of his own staff known to him from pre-war days, actually transcribed his speech under Jones's own editorship in an article (Pring 1947) which showed him to belong to our Variable category by displaying -y before vocoids (other than /w/) as /i/ and elsewhere as /ɩ/. Murray's suggestion, in his fussily learned transcription fortunately encumbering the OED no longer since 1989, that people had a small category of words like Psyche in which they used a non-long /i:/ sound consistently was almost certain fantasy.
13. Wells observes that Wright (1905) made 'absolutely no mention' of close -y in any of the English dialects. Virtually the same can be said about the Orton (1962) coverage of the Northern Counties: the fieldworkers simply generalised one symbol for all -y tokens because their interest was focused on other matters. In a similar way, one can hear in recordings of phonetic readers, for example, MacCarthy (1956: 51 country, aren't, they), Lee (1963: 109 only in), Arnold and Tooley (1972: 67 Sorry. It's ... ), examples of -y values so close that /i:/ would have been more appropriate than the normalised versions shown in their texts. O'Connor (1971) and O'Connor (1973) show a commendable realism. In the writer's own WL (1977) there were 216 uninflected polysyllables with unstressed final -y of which about 15% exhibited close values. They were quite well distributed amongst the four main speakers though two of them did have markedly higher proportions than the others. Three-quarters of the close values were pre-vocoid: the other quarter seemed to be mostly triggered by prosodies including rather wide (falling-)rising tones or Drops. Each of the speakers was thus classifiable in our Variable category. Only eleven inflected -y words occurred. Each main speaker had at least one. All of them were heard with [ɪ]. This last distributional pattern is reflected in all our categories except Close and Non-enclitic, only the latter of which regularly has a close vowel in the inflected forms.
The point to remember about speakers who seem to vary haphazardly between close and open qualities of -y is that, and this is, it seems, not very widely understood, vowel-quality variation is itself part of the prosodic/paralinguistic element of speech. How else would the same speaker in the same sentence have successively open and close values in (spend our ... ) money on an ugly old (bedpan) (WL 1977 No. 12, line 2).
A curious fact that has emerged in observing the inflected forms was that a large number of very well-educated speakers (including Ian Fleming and Sir John Gielgud) not in our Close or Non-enclitic categories have clearly /i:/ in the word Indies. This pronunciation has presumably come about through equating it with a non-existent Greek-style form *Indes.
This is of course no worse treatment, except for being based on a false spelling, than we all accord to Andes or British speakers to Andes, Antilles, Averroës, Cervantes, Los Angeles etc all of which are spelling pronunciations reflecting a time when the classics were familiar if Spanish was not.
This may have seemed a lot of fuss about a trivial sound, but -y does, as Wells (1982: 165) points out, have considerable indexical and accent-diagnostic interest and is of fairly frequent occurrence – to judge from WL (1977) about 4% of all the words in everyday conversation end with it so that we utter one about every 8 seconds on average. (Of course we don't necessarily utter a -y sound for every -y word we use. Some suffer elision, for example, twenty four hours said briskly should sound natural as /twent fɔːr aʊəz/ and many use them is often heard indistinguishable from men use them.
14. Wells (1982: §3.4.3) says 'Where and when the [i] pronunciation arose is not certain. It has probably been in provincial and vulgar speech for centuries.' It is suggested that the documentary and archival evidence adduced above indicates that it has never entirely gone away from educated speech either. The account presented of present usage has tried to add some detail to the Wells (1982) picture which it essentially confirms suggesting only that the proportions of speakers with context-sensitive -y variation is probably rather higher than Wells seems to allow and is in all probability the major group in present-day General British pronunciation.