IPA vowel symbols for British English in dictionaries

This is a very slightly amended version of the article published and copyright by Cambridge University Press which appeared in the

Journal of the International Phonetic Association

Volume 33 Number 2 in December 2003

Abstract

The set of vowel symbols introduced by Gimson in 1977 into the Daniel Jones English Pronouncing Dictionary for the representation of British pronunciation became so widely used that for two decades an unprecedented harmony reigned and an ever wider non-specialist audience became accustomed to accept that pronunciations should be presented to them via the International Phonetic Alphabet in this form. This unity was disrupted by the publication by one branch of the Oxford University Press of important reference books in which one in four of the Gimson twenty vowel-phoneme symbols were changed. The grounds for welcoming or deprecating each of these changes are discussed.

1. Introduction

The publication in 1977 of A. C. Gimson's revision of the Daniel Jones English Pronouncing Dictionary (EPD14) was a watershed in  respect of the representation of British pronunciation in dictionaries. Gimson's most marked departure from Jones's notation in the EPD of the previous sixty years lay in his use, as in his Introduction to the Pronunciation of English of 1962, of the IPA non-cardinal vowel symbols ɪ, ʊ, ɜ and the non-roman secondary cardinal ɒ in order to effect more explicit or, as he expressed it, "more realistic" (EPD14 p.xiii) representations of vocalic contrasts. The old EPD notation had already contained the non-cardinal items æ and ə as well as the non-roman cardinals ɛ, ɔ and ʌ. Gimson always candidly acknowledged that his action was directly contrary to what Jones himself advocated.

In his EFL textbook An Outline of English Phonetics (1956:344/5) Jones described how at one time he had experimented with narrow transcriptions for EFL classes and been disappointed by the lack of any perceptible benefit from their use. Accordingly he had devised at the end of the twenties what he came to call his "simplified" transcription which made use of only two non-roman letters ʌ and ə. This became his declared preference for a set of symbols for EFL and general use. Though he encouraged others to use it he failed to convert to it either his EPD or any other work than The Phoneme (1950). It was used in various books and periodicals up to the 1970s (see Jones 1956:348), but after then mainly only in one or two minor dictionaries influenced by David Abercrombie. It had been the initially projected notation for the Longman Dictionary of Contemporary English (LDCE).

2. The move to IPA in Oxford dictionaries

The great Oxford English Dictionary originally contained a highly complex pre-IPA notation (OED1:xxxiv) devised in 1882 by Murray. When that work appeared in a new 20-volume edition (OED2) in 1989, it embodied EPD14 notation, except for ɛ instead of e and ɛə in place of eə. These departures from EPD14, were deemed "necessary" on the grounds of harmonising their treatment with representations of French and German. The further comment that, while employing /iː/ and /uː/ eg at the third syllables of delineate and perpetuate rather than Gimson's /ɪ/ and /ʊ/, "The IPA [sic] transcriptions convey the misleading implication that such vowels have ... greater duration" seemed to ignore the neat originally LDCE device of the dropping of the length mark to give /i/ and /u/.

In advance of the OED2 transfer, various lesser items in the Oxford series of dictionaries had begun to turn over to IPA in (unchanged) EPD14 notation starting with the Oxford Dictionary of Current English in 1985. This revision of what had previously been issued as the Pocket Oxford Dictionary is not to be confused with A. S. Hornby's Oxford Advanced Learner's Dictionary of Current English (ALD) which had been published from 1948 by an independent arm of the Oxford University Press and had always employed IPA segmental symbols. The ALD's rival, the LDCE, when it finally appeared in 1978, followed virtually exactly the notation Gimson had introduced into the EPD in the previous year.

There was, however, that notable device just mentioned by which vowels like the final one of the word happy were represented by neither /iː/ nor /ɪ/ but by /i/ and many word-medial occurrences of the /uː/ phoneme were converted to representation with /u/. This innovation will be referred to as EPD14b. It was something that J. C. Wells judiciously embraced when he compiled his Longman Pronunciation Dictionary (LPD) of 1990. By that year the very widely used Concise Oxford Dictionary (COD) had turned over to IPA symbols, at that time in unchanged EPD14 (not in OED2 or EPD14b).

OED2 contained several pages in its introductory matter headed THE TRANSLATION OF THE PHONETIC SYSTEM in which its authors explained how they set about converting the original Murray transcriptions into the International Phonetic Alphabet the introduction of which into the new edition "was regarded by many whom the project team consulted as among the highest priorities" (OED2:xviii). In their original electronic version they retained the Murray transcriptions alongside the quasi EPD14 ones but they kept only the latter in printed versions and even in the CD ROM, a policy that it may be hoped will be reconsidered in the future because it involved the loss to the user of much historically valuable material.

[PS The OED3 which began online publication in 2000 has included various quotations of Murray et al 'translated' into IPA notations.]

3. The background to the EPD14 vowel-symbol choices

The symbol set which Gimson finally decided upon for EPD14 had been originally devised by Jones in 1916 and appeared in a succession of slightly differing versions in Le Maître Phonétique from 1923 onwards. Gimson's EPD14 transcription included IPA length marks. These Gimson freely acknowledged to be redundant in terms of identifying phonemes within his rich vowel set. His preference for this strategy was based on the desire to preserve the traditional appearance and to facilitate maximum legibility. The EPD14 choice of symbols, having been adopted for LDCE by the influential publishing house of Longman, within a few years became almost universally employed in EFL and general use.

There is another point that may be considered here. Although I respect the elegant simplicity of the symbol set employed in several of the earlier EFL books by Peter MacCarthy, when confronted with such items in his texts as eg ford as /food/ and bought represented as /boot/, I found the coincidences with the traditional spellings of totally unrelated words quite disturbing. In fact I am inclined to believe that a similar reaction was an important element in the feelings of those who reacted unfavourably to the notation I used in my own CPD (Concise Pronouncing Dictionary 1972, 1979) in which eg beat appeared as /bit/ and boot as /but/. It is noteworthy that Sweet went out of his way to avoid using \oo\ in his Broad Romic transcriptions which, as Jones commented in the 1956 Outline Appendix A p. 346 footnote, “would not have been out of keeping with the rest of his system”.
I have in fact come to the opinion that, at any rate for non-specialist use, it could well be adopted as a principle of symbol choice that

ANY PHONEMIC SYMBOLISATION OF A WORD THAT COINCIDES WITH THE TRADITIONAL ORTHOGRAPHY OF A DIFFERENT WORD SHOULD BE AVOIDED.

Another recommendation I feel to be well worthwhile is that

ANY PHONEMIC SYMBOLISATION, WHETHER OR NOT SURROUNDED WITH SLASHES etc, SHOULD PREFERABLY BE INSTANTLY RECOGNISABLE AS SUCH.

EPD14 largely accords with these principles.

At any rate, however one perceives its principles of vowel symbol choice, the general acceptance of EPD14 was very much a fortunate development because it is undoubtedly a benefit to all not to have to be making constant mental readjustments (however slight these may seem to be to the expert) as one encounters a succession of texts containing different phonemic notations.

4. A schismatic move

This harmony in transcription between very large numbers of English dictionaries and textbooks from a variety of publishing houses, which was no doubt more complete than had ever been the case before 1977, was in the mid 1990s disrupted by the introduction of notably altered representations of five of the twenty EPD14 vowel-phoneme symbols by one division of the Oxford University Press. Kretzschmar (1994:83-84) revealed this move to have been decided upon at the advice of Clive Upton, the distinguished dialectologist now of the Leeds University School of English. This altered EPD14 transcription seems first to have appeared in 1993 in The New Shorter Oxford English Dictionary (NSOED) and then in 1996 in the Concise Oxford Dictionary (COD). The resulting set of symbols exhibits the following five major substitutions: (i) e replaced by ɛ (ii) æ replaced by a (iii) ɜː replaced by əː (iv) aɪ replaced by ʌɪ and (v) eə replaced by ɛː.

It must be emphasised that we are talking only of general-purpose and non-specialist applications: in analytical, comparative and dialectological phonetic studies, the recommendations of the IPA should of course be followed rigorously because otherwise communication will simply break down. The problem is that, in works not directed at specialist phonetic scholars, any changes of notation that can not be seen as essential positive improvements must surely be considered highly undesirable in that they damage an existing very beneficial consensus. To be justified, such alterations must be seen to spring not from the personal transcriptional preferences of individual phoneticians but from a wide agreement that the existing notations of the items in question have come to be regarded as gross misrepresentations of current usage.

It seems not to be a matter of dispute that the kind of British English pronunciation they are being used to represent is the variety (or set of varieties) which is least attributable to any specific region of England. This has been identified in the past couple of generations chiefly by the 1926 Jonesian term "Received Pronunciation" (cf eg EPD15 p.v) for which various alternatives (admittedly none of which is ideal) have been proposed including my own suggestion, 'General British', first put forward in 1972 in my Concise Pronouncing Dictionary (CPD:xiv). This suggestion was alluded to in Gimson (1980:303) thus: "'General British' (GB) has been used and may in time supersede the abbreviation RP". It will be used below. In the light of the above discussion we shall now seek to set forth the pros and cons of each of the Upton departures from EPD14.

5. /ɛ/ versus /e/

The first departure from EPD14b, viz /ɛ/ for /e/, represents a perfectly acceptable relatively narrow transcription of the mainstream GB phoneme. Also it provides the benefit of instant recognisability as a phonetic symbol. As with /ӕ/, Gimson (1962/89) again diagrammatically represented /e/ with a closer value for it than Jones had assigned to it. Jones (1950), a book which was in the main a contribution to the dialectology of British English, showed /e/ in a relatively narrow transcription of 'one type of Received English' (ie no doubt his own) as clearly nearer to open-mid than to close-mid. His choice of symbol in that dialectological context was /ɛ/. In Jones (1932/64), the EFL textbook, and in the general-purpose EPD he used /e/. He was acting according to the just-mentioned principle he had advocated in Jones (1949 §20) "When a vowel is situated in an area designated by a non-roman letter, it is recommended that the nearest appropriate roman letter be substituted for it." Again, very justifiably, the Cruttenden/Gimson (1994) diagram does not show /e/ as necessarily as close to CV [e] as Gimson had put it. Although I now regret that EPD14 did not originally incorporate /ɛ/, at this stage of things an alteration of EPD14 from /e/ to /ɛ/ is probably on balance better not adopted.

6. /a/ versus /ӕ/

The second departure from EPD14b can readily be understood to aim to reflect the notably lowered and retracted value that has become obvious over the last three or four decades for mainstream GB /ӕ/. Anything front and noticeably closer than [ӕ], ie less than halfway from Cardinal Vowel 3 to CV 4, is now generally perceived as tending to sound old-fashioned or regionally marked. Note the Cruttenden/Gimson (1994) re-siting of the /ӕ/ of Gimson's diagram. This change, which in fact turns it back to how Jones had shown it in EPD and elsewhere, seems very necessary and even perhaps somewhat too cautiously conservative. Gimson's (1962) comment that /ӕ/ was "only very little more open than half-open" is definitely dated and indeed didn't seem to apply to his own usual (at least later) pronunciation as witness the various recordings he made. Current mainstream /ӕ/, being neither fully front (cf. the Wells 1990:xv diagram and the Wells 1982:292 reference to the lowering and centring of /ӕ/) nor fully open, may be positioned where many would presumably either prefer to show it with the 'a' symbol or at least to find that notation equally feasible (compare the IPA Handbook §6). It is perhaps worth recalling in any discussion of the suitability of [a] as a dictionary symbol what Jones (1949 §38) referred to as his impression that "authors and printers still generally regard a and ɑ as variants of the same letter". The older (Jonesian) principle of IPA symbol choice would point to the selection of the cardinal symbol rather than the non-cardinal. However, such a procedure is, wisely, not enjoined upon IPA users in the new Handbook. There is also the benefit of the instant recognisability of æ as a phonetic symbol. Finally, there is the point well made in Wells (2001:7) that "A further argument in favour of retaining the symbol [æ] is that it preserves the parallelism with American and Australian English, in which the movement towards an opener quality has not taken place". In summary it seems that this alteration of EPD14b, in which ӕ is changed to a, is not worth adopting.

7. /əː/ versus /ɜ:/

The third departure from EPD14b is the substitution of /ɜː/ with /əː/. This has the advantage of reducing the total number of unfamiliar symbols to be assimilated by the general user by an undeniably "exotic" item. However, ə itself is hardly less exotic. At any rate, ɜ is a distinctive and legible symbol and seems to be fairly easily recognised by most readers. A possible advantage that can be argued for having /ɜː/ is that the contrast with /ə/ serves to symbolise the much narrower range of allophonic variation of the longer vowel than the shorter. Another practical convenience is that retention of /ɜː/ helps those who might wish, at least at times, to employ a version of EPD14 which dispenses with the redundant length-marks. Not feeling unduly disturbed by the /hit/ and /but/ type of clashes with orthography mentioned above, I often make handwritten notes omitting the somewhat tiresome-to-write redundant length-marks.

Had EPD14 incorporated /әː/ from the start, as did some of its 1920s and later precursor variants to be seen in Le Maître Phonétique, then it would be doubtful that it was now worth changing from /әː/ to /ɜː/. But, as things are, it doesn't now seem worthwhile changing what we have.

8. /ʌɪ/ versus /aɪ/

The fourth departure from EPD14b, the substitution of /ʌɪ/ for /aɪ/, is the hardest to discern any justification for. None of the British authorities have ever diagrammatically or otherwise represented GB /aɪ/ with a more retracted beginning than /a/, though such relationships are found in varieties of Cockney and regional dialects. Indeed Wells (1982:292) has said explicitly "the starting-point of /aʊ/ is never fronter than the starting-point of /aɪ/". Gimson (1962/1980) differentiated the beginnings of /aɪ/ and /aʊ/ by representing the latter as /ɑʊ/. This was, and still would be, perfectly defensible. When Gimson published EPD14 he gave no reasons in it for his use of /aʊ/ rather than /ɑʊ/ but the choice was not made because he felt that any notable new development had become apparent. The overlapping indicators on the rather amateurishly drawn Fig.2 at p.xv showed that he considered the majority of GB speakers to have much the same (more or less central and somewhat raised) beginning to both diphthongs. From the many discussions that I had with him on matters of transcription it was clear that the essential motive involved was his view that, in a work of reference for a wide audience, simplicity of presentation was a very desirable aim. One way of achieving that end was, he felt, to minimise the use of unfamiliar characters. Hence his long-pondered decision to depart in EPD14 from the Gimson (1962) /ɑʊ/ and /ɛə/ in favour of /aʊ/ and /eə/ (cf Jones 1977:viii).

One need not dismiss out of hand [ʌɪ] as a possible way of representing the diphthong /aɪ/. Indeed its beginning in mainstream GB can reasonably be said (pace Wells 2001:7) to be just within the approved range of applicability of the CV symbol [ᴧ]. But surely, if /ᴧɪ/ instead of /aɪ/, why ever not /ᴧʊ/ instead of /aʊ/? One could have far better understood the reverse choice of /aɪ/ and /ʌʊ/. [Note added February 2012: One is inclined to wonder why, since Upton's preferences for a and ɛ appeared to stem from IPA purism, he was content to adopt a counter-IPA use of the symbol ʌ when the more strictly IPA-conformist policy would surely have been to make use of ɐ instead of ʌ and ɐɪ instead of ʌɪ.] The notation [ʌɪ] appeared in the Gimson (1962/1980) text, and has been incorporated into the relevant Cruttenden/Gimson (1994) diagram, but only to represent a distinctively Scottish value (in such words as side). Similarly Wells (1982:596) uses [ʌɪ] to represent Cockney and Australian values corresponding to GB /eɪ/. Wiik (1965) chose to use /ʌɪ/ and /ʌʊ/ for GB but on the basis of acoustic measurements using only five subjects of questionable typicality and in a non-lexicographical context. MacCarthy (1978) used both /ʌɪ/ and /ʌʊ/ but the oddness of this choice was plainly seen from its diagrams which represented the beginnings of both diphthongs as having almost fully front and fully open values. MacCarthy's motive for these representations was an ill-judged desire for the economy of avoiding employment of an extra symbol, viz /a/, in his notation, which would, unlike all his other diphthong symbolisations, introduce a letter not used for any of his simple-vowel items.

The apparent Upton suggestion that /aɪ/ and /aʊ/ have now reversed their relative starting positions in mainstream usage is not supported by my observations and I know of no-one else of such an opinion.

The Wells (2001:8) judgement on this matter was: "Upton's notation implicitly identifies the first element of price with the vowel quality of cut – an identification that accords with the habits neither of RP nor of southeastern speech (Estuary English). His choice of [ʌɪ] is really very unsuitable."

In any case EPD14's /aɪ/ and /aʊ/ are additionally preferable as satisfying the two recommended criteria of being (i) instantly identifiable as phonetic symbols and (ii) economical with the use of exotic letters.

Thus the verdict on /ʌɪ/ alongside /aʊ/ must be that it is a very regrettable departure from EPD14b that would be better abandoned in future.

9. /ɛː/ versus /eə/

The fifth and final major departure from EPD14b, the substitution of /ɛː/ for /eə/, is one with which one can have a great deal of sympathy. It can readily be granted that on the majority of the occasions on which this phoneme is heard from speakers of all but old-fashioned and/or socially conspicuous accents the value is certainly monophthongal. But the problem doesn't end there; and the comment in Kretzschmar (1994), an article which purported to give an account of the Upton views on British vowel values, that any diphthongal pronunciation of this item is now marked as old-fashioned is far from the truth.

The most usual basis for the choice of symbol for a phoneme is no doubt its form as least influenced by any allophone-inducing context. Unfortunately this phoneme is one of the most difficult British English sounds to observe not so influenced, largely because it is of such low occurrence particularly in such contexts. However, the following comments may help to show some of the problems involved.

There are now CD ROM sets of recordings of the texts of ALD and LPD. I have listened to each of them for all I could locate of the only fifty-odd words containing final stressed /eə/. ALD uses five named speakers (young actors, I gather) and LPD more than five, obviously not just the same people, but unnamed and in apparently most cases unknown relatively young speakers. There certainly were at least a couple who regularly used [ɛː], but the majority used fairly clearly diphthongal values. There were quite a few doubtful tokens and some speakers alternated between simple and diphthongal values, which may possibly have reflected the fact that they ordinarily habitually employed monophthongs but were trying to comply with the "instructions" of the texts they were reading from to produce diphthongs. There were one or two speakers whose voice quality to me sounded slightly socially conspicuous, or as Cruttenden/Gimson (1994) has it, "refined"; and a small minority on a few occasions produced unusually open schwa elements which likewise produced a socially conspicuous effect; but most of the speakers most of the time seemed to me to produce ordinary mainstream-sounding slightly diphthongal values for most of these words.

Böhn (1966) investigated the various realisations of this phoneme. She gathered almost 1,500 tokens from a variety of GB speakers and situations (twenty hours of recordings of BBC news bulletins, unscripted interviews and dramatic performances) and examined the various phonetic forms the phoneme took. Her data showed only 15% of the tokens as diphthongal, which was not surprising but at least was valuable evidence that the Jones (1932/64:§448) view of the phoneme as consisting of a single member (ie having no allophonic variation) was even then far from applicable, if it was ever truly unquestionable. Her data suggested that a monophthongal allophone was almost the sole one employed when /r/ followed and in unstressed syllables. Also she found that it predominated in some other situations. However, the data showed a slight predominance of diphthongal allophones in pre-pausal syllables.

I doubt if the state of affairs has changed much at all in the meantime. Jones's picture, though it did match that presented in the works of Ellis, Bell, Sweet and Murray, was in need of reconsideration. It seems most likely that the essential problem is one not of reflecting greatly changed usage but of interpretation of the facts. Over a century ago Soames (1891:27) showed a long vowel [ɛː] for words like Mary and fairy in transcriptions that would now be classified as allophonic. Note also the implications of the Jespersen (1909:422) comment that the schwa position is held only for a short time especially when a real [r] follows. Such archive recordings as I have examined for the first half of the last century have not suggested to me a different state of affairs from today's. Jones himself seemed regularly to use either a quite narrow or barely (if at all) perceptible diphthong. He showed no recognition of monophthongal variants of /eə/ until his (1950:63§204) remark ‘Occasionally one hears a monophthongal long ɛː’. This was accompanied by examples of it in both closed and open monosyllables with no reference to any social restrictions on its circulation. Gimson (1962) perceived a long pure vowel [ɛː]...especially in a non-final syllable as a newer usage ("advanced RP"). Wells (1982:293) showed some hesitation about labelling [ɛː] as perhaps a Near-RP northernism if in a stressed final syllable but by Wells (2001:8) had decided "What used to be a local-accent feature has become a part of the mainstream".

As regards younger mainstream GB speakers, it is doubtful that it can be safely asserted that for the majority the least conditioned allophone of the phoneme is not diphthongal. The fact that a prepausal utterance of a word ending with stressed /eə/ is very often accompanied by lip-closure will mean that such tokens will not be totally steady-state in quality. This makes decisions often matters of some uncertainty. In fact, versions completely ambiguous between simple vowel and diphthong can often be heard.

The mainstream speaker a half century or more ago seems to me to have had much the same allophonic variations as today's mainstream speakers. However, there were no doubt more examples to be heard at that time of the surely always small minority who exhibited a conspicuously open value of its final schwa. This Jones (1932/64 §449) remarked could be represented by the notation ɛʌ. The famous Wykehamist BBC announcer and commentator the late John Snagge (born 1904) for example even had [ɛʌ] internally before /r/ as in Mary. I strongly suspect that it is not the slightly diphthongal occurrences of /eə/ that some of the proponents of the representation /ɛː/ have in mind as old-fashioned but that they unjustifiably associate the diphthongal representation with such archaic survivals. Windsor Lewis (1969:21) offered the following summary of the allophonic variation of mainstream GB /eə/ thus: a diphthong generally realised as a long simple vowel (a) before consonants (b) when unstressed (c) when stressed but in a structural word. I see no reason today to amend that judgement.

This is not to say that there aren't plenty of speakers who use a monophthong practically all the time. Peter Strevens was one as he indicated by his choice of the notation /ɛː/ to transcribe his own speech in Strevens (1954). Various of the recent BBC Radio 4 GB newsreaders can be heard to exhibit [ɛː] even on pre-pausal complex tones notably in saying the word fair in weather forecasts.

To sum up, in judging whether we should welcome the alteration of EPD14 /eə/ to /ɛː/ it seems to me that the essential criterion might as well be not whether we think the majority of mainstream speakers use [ɛː] in isolate situations or not but "Do most people find that a slightly diphthongal /eə/ sounds unusual?" If not – and I'm sure they don't – then it is surely best to leave EPD14 just as it is.

10. /i/ versus /ɪ/ and /u/ versus /u:/

Besides these five major departures from the EPD14b, there is the further matter that the editors of NSOED1993 and COD1996, like those of LDCE, have fought shy of, namely, extending the process of recognition of /i/ to the parallel rhythmically weak /u/. The original Longman introduction of /i/ was very tentative indeed. The LDCE explanations in 1978 and in the major revision of 1987 gave the impression that the editors considered that the norm of their British model was to be equated with /ɪ/, and the American model was to be equated with /iː/. They had adopted /i/ only as a space-saving device. This was not very convincing. It was even observable that where words were given full different American and British versions, as eg at worry, though the economy of space requirement did not apply, /i/ was still used in LDCE. One suspected that the real reason was the understandable reluctance to get involved in discussion of the problematic question of the phonological assignment of the final vowel of happy on the part of the 1978 LDCE pronunciation editor Gordon Walsh.

Even LPD and EPD15 seem to me to fail to fully accept the logical consequences of their recognition of /i/. Both Wells and Roach adopt the theoretical position that /i/ belongs neither to the /ɪ/ phoneme nor to the /iː/ (EPD15:xiv). One knows that some GB speakers positively identify their final happy vowel with their /ɪ/, and some speakers are unable to assign their version to either phoneme. But my strong suspicion is that a very great number of native English-speakers worldwide perceive /i/ as a phonologically (but essentially because rhythmically) distinct member of their /iː/ phoneme. Such speakers feel that stelae and steely are a variety of minimal pair: cf Wells (1982:166) which mentioned carrhae and carry as well. Bacchae and baccy and bases (from basis) and Basie's are others, but such pairs are very rare. Likewise many perceive a phonological contrast between the final vowels of pairs like pedigree and mimicry, Pharisee and fallacy etc. Such a difference was mentioned for General American usage in Pike (1945:79) between refugee and effigy. He didn't identify his /ɪ/ as ending either of them.

This interpretation can throw light on the recent increased inclination among GB speakers to make the final vowel of happy unstressed but rhythmically strong /iː/. Awareness of this is very probably what has led to the belated recognition of the long-present rhythmically weak final /i/. For a detailed discussion and historical comments see Windsor Lewis (1990). Those who have come to adopt rhythmically strong final /iː/ have been very likely influenced by aspiring after "clear" speech. Apparently all below the equator and many, some say most, in North America have rhythmically strong value as their usual form of their final happy vowel, probably as a further development of the same tendency.

It is notable that the dictionary compilers have been rather slow to recognise a corresponding distinction between word-final unstressed /uː/ and /u/. For example for probably a majority /`dʌblju/ ie w and /`dʌbl juː/ ie double u are distinct. And value for very many, perhaps most, ends the same as thankyou, ie with /u/ not /uː/. Wells (1990) showed /u/ as a subvariant form of continue, and Wells (2000) has subvariant /u/ at value and w, though in few if any similar cases.

11. Conclusion

The departures from EPD14 discussed above are freely acknowledged to be for the most part valid alternative transcriptions of GB, but the point should also be made that plenty of the EPD14b notations left unaltered by these changes can be as easily shown to have valid alternatives themselves. For example, Henry Sweet's and others' /ij/ and /uw/ notations for /iː/ and /uː/ have much to be said in their favour. The list could readily be extended but the point has been adequately made that reasonable alternatives exist to plenty of our long-familiar notations because the notations are often more customary than inevitable. It is the proper business of specialised students of phonetics to scrutinise notations with a healthy scepticism but the general public cannot be expected to be concerned with such matters. The arguments for the Upton innovations seem to me to rest on questionable presuppositions. The most important fact seemingly being overlooked is that no-one acquires a particular vowel quality/articulation from studying a dictionary. Dictionaries essentially only offer information on the distributions of phonemes within words – not on their articulatory realisations. The General British vowel phoneme set has hardly changed in a century. (The only inventorial change that dictionaries have needed to reflect is the practically complete disappearance from non-regional usage before the second half of the twentieth century of the diphthong /ɔə/).

In the interests of the overwhelming majority of dictionary users, the helpful consensus we have after so many years fortunately achieved is best preserved by all.

References and notes on pronunciation symbols used

ABERCROMBIE, DAVID (1964). English Phonetic Texts. London: Faber & Faber Ltd.

BÖHN, INGER (1966). The RP Phoneme /ɛə/. Unpublished thesis. University of Oslo.

Chamber's Universal Learner's Dictionary (1980). Pronunciations "supervised" by David Abercrombie. Pronunciation symbols: Jones's "Simplified" notation.

The Concise Oxford Dictionary of Current English (COD). Oxford: Oxford University Press (1990). Pronunciation symbols: EPD14. (1995) Pronunciation symbols: Altered EPD14b.

CRUTTENDEN, ALAN (1994). Gimson's Pronunciation of English. Fifth edition. London: Arnold. (Cruttenden/Gimson)

EPD14: See JONES, DANIEL (1977-91).

EPD14b. See Longman Dictionary of Contemporary English.

GIMSON, A. C. (1962/80). An introduction to the pronunciation of English. London:Edward Arnold

The Handbook of the International Phonetic Association. (1999). Cambridge: Cambridge University Press.

JESPERSEN, OTTO (1909). A Modern English Grammar. Vol.1. London:Allen and Unwin.

JONES, DANIEL (1917/66). English Pronouncing Dictionary. London:Dent (EPD).

Pronunciation symbols: Old EPD ie with /ɛ / but without use of ɪ, ɔ, ʊ, ɜ.

JONES, DANIEL (1932/64). An Outline of English Phonetics. Leipzig:Teubner; Cambridge:Heffer then CUP. Pronunciation symbols: Old EPD.

JONES, DANIEL (1949). Quasi-anonymously: The Principles of the International Phonetic Association. London:IPA.

JONES, DANIEL (1950,1958). The Pronunciation of English. Cambridge: Cambridge University Press.

JONES, DANIEL (1977-91). Editors A. C. Gimson and in 1988 S. Ramsaran. English Pronouncing Dictionary. London: firstly Dent; from 1991 Cambridge University Press. Pronunciation symbols: EPD14.

JONES, DANIEL (1997). Editors Peter Roach and James Hartman (with Jane Setter). English Pronouncing Dictionary. Cambridge: Cambridge University Press. Pronunciation symbols: EPD14b.

KENYON, J. S. & T. A. KNOTT (1944). A Pronouncing Dictionary of American English. Springfield, Massachusetts: G. & C. Merriam.

KRETZSCHMAR, WILLIAM A. (1994). The Oxford Concise Dictionary of Pronunciation. In RASK 1 pp 83-93. Denmark:Odense University Press. [The work described was eventually published not under this provisional title but as the Oxford Dictionary of Pronunciation for Current English.]

Longman Dictionary of Contemporary English (LDCE). Harlow, Essex: Longman.

Pronunciation editors: Gordon Walsh 1978, Dinah Jackson 1987. Pronunciation symbols:EPD14b.

MACCARTHY, PETER (1944). English Pronunciation. Cambridge: Heffer.

MACCARTHY, PETER (1978). The Teaching of Pronunciation. Cambridge: Cambridge University Press.

The New Shorter Oxford English Dictionary (1993). Editor Lesley Brown. London: Oxford University Press. (NSOED). Pronunciation symbols: Modified EPD14b.

Oxford Advanced Learner's Dictionary (ALD). Editors A. S. Hornby et al. Oxford: OUP.

1963: Pronunciation symbols: Old EPD. (No separate pronunciation editor.)

1974 Pronunciation editor: J. Windsor Lewis. Pronunciation symbols: CPD.

1980 etc Pronunciation editors: A. C. Gimson & S. Ramsaran. Pronunciation symbols: EPD14.

1995 etc Pronunciation editor: Michael Ashby. Pronunciation symbols: EPD14b.

The Oxford English Dictionary. First edition (1884 -1986). Editors J. A. H. Murray, R. Burchfield et al. Second edition. (1989). Editors J. S. Simpson & E. S. Weiner. London: Oxford University Press. Pronunciation symbols: EPD14 except /ɛ/ for /e/ and / ɛː / for /eə/.

SOAMES, LAURA (1891). Introduction to English, French and German Phonetics. (reprinted 1913) London: MacMillan.

STREVENS, PETER (1954). Representation of rate-of-change of tongue-position. Le Maître Phonétique 24-26.

UPTON, CLIVE, WM A. KRETZSCHMAR JR & RAFAL KONOPKA (2001) The Oxford Dictionary of Pronunciation for Current English. Oxford: Oxford University Press.

WELLS, J. C. & G. COLMAN (1971). Practical Phonetics. London: Pitman Publishing.

WELLS, J. C. (1982). Accents of English. (3 vols). Cambridge, UK: CUP.

WELLS, J. C. (1990, 2000). Longman Pronunciation Dictionary (LPD). London: Longman.

Pronunciation symbols: EPD14b.

WELLS, J. C. (2001) IPA transcription systems for English (first versions in Wells's personal website but printed in) P[honetics] G[roup] Bulletin 9, 3-8 Santiago, Chile (editor H. Ortiz-Lira).

WIIK, KALEVI (1965). Finnish and English vowels. Finland: Turku University Press.

WINDSOR LEWIS, J. (1969). A Guide to English Pronunciation. Oslo: Scandinavian Universities Press.

WINDSOR LEWIS, J. (1972a). A Concise Pronouncing Dictionary of British and American English (CPD). London: Oxford University Press. Pronunciation symbols: CPD.

WINDSOR LEWIS, J. (1972b). The notation of the General British English segments. Journal of the International Phonetic Association 2.2, 59-66.

WINDSOR LEWIS, J. (1985) British non-dialect accents. Zeitschrift für Anglistik und Amerikanistik 33/3, 244-257. Germany: Leipzig

WINDSOR LEWIS, J. (1990) HappYland reconnoitred: the unstressed word-final -y in General British pronunciation. In Susan Ramsaran (ed.) Studies in the Pronunciation of English. London: Routledge