The Daniel Jones English Pronouncing Dictionary, so well known in the EFL teaching world as to need only to be referred to simply as "the EPD" or even simply "Jones", was first published in 1917. J. M. Dent, its original publishers, repeatedly hyped as new "editions" various quite minor revisions but not, one may be sure, at the instance of Jones himself. In its heyday, in the two or three postwar decades, it had a remarkably wide circulation for a relatively specialist publication though by the eighties its popularity had waned a good deal. Not long after the appearance in 1990 of the remarkable 800-page Longman Pronunciation Dictionary by J. C. Wells (hereafter "LPD"), Cambridge University Press acquired the copyright of the EPD. The present long-awaited 15th edition of "Jones" is most certainly a major revision, indeed the most fundamental in its whole history. It has been much enlarged so that it now has "over 18,000 new entries", and although it has a similar number of pages, they are 3cm taller and more than 2cm wider. The principal architect of this very ambitious re-working of the EPD has been Professor Peter Roach of Reading University.
My provision in A Concise Pronouncing Dictionary of British and American English (OUP 1972 "CPD") and in the Oxford Advanced Learner's Dictionary of Current English (OUP 1974 "ALD") of the then fairly novel feature of British and American pronunciations shown side by side soon became the normal practice in British EFL dictionaries of some size, but it has taken till now for the EPD to be brought into line. The contribution of the American editor is generally effective but it is not without some rather eyebrow-raising elements. For example, for a whole set of words containing the spellings -oss and -ost, LPD, in harmony with previous authorities, lists them as having in General American predominantly a rounded vowel and less widely an unrounded one. EPD15 for its US model gives only the LPD second choice and no alternatives. Among other problems are the understanding of expressions like "being sensitive to the traits of the individual word", the lack of a systematic account of just how "Network English" differs from General American, and the acceptability of some judgments at particular entries. For example it's hard to imagine any US speaker regularly using the first version given of Malaga. See also below at §5 remarks on the notations of the US vowels.
EPD twice received major revisions by its originator, one in 1937 and another in 1956. An excellent very full account of the origins and development of this dictionary is to be found in Collins and Mees (1999 pp 167-173 and 397-398). Jones's successor as the most highly regarded authority on the pronunciation of British English, A. C. Gimson, in 1967, the year of Jones's death at the age of 86, brought out a quite extensive revision of the text. A decade later he converted the transcription into the now very widely used set of symbols which incidentally he knew very well Jones would have considered over complicated. Professor Roach has rightly kept these symbols but additionally incorporated the minor improvement of colonless /i/ and /u/ to identify weak vowels such as the final one of the word happy, a practice in effect originated by Gordon Walsh, the then Pronunciation Editor of the Longman Dictionary of Contemporary English, in 1978 and subsequently employed by Wells in the LPD.
The twenty-two pages of introductory matter of the previous edition have been completely re-written as a dozen two-column pages. In the Editors' preface, which occupies p.iv, it is remarked that "an enormous amount of work on the transcriptions" was entrusted to Professor Jane Setter, quaintly billed not as co-editor but as "Pronunciation Associate", who began her work in 1992. She is thanked for "keeping us from drifting into inconsistency" but not very surprisingly the prodigious numbers of transcriptions contain quite a few examples of failure to resist that drift.
The principles on which the recommended versions of each word are ordered are clearly explained with various candid and judicious observations including the one that they are offered "with some alternant forms rivalling the first-given in perceived frequency while others may be a more distant second". It would be a welcome improvement if in future this problem were able to be given more consideration. Gimson had the seeds of a better handling of it at a limited number of items. Jones had provided for the indication of three degrees of frequency of distribution at each entry: predominant forms, common but less frequent forms which followed enclosed within square brackets and a third category labelled specifically either rare or old-fashioned. Gimson decided at certain entries to show two co-variants outside of the square brackets neither of which was to be taken to predominate over the other but both of which were to be taken as more commonly occurring than any of the subvariant forms placed within the square brackets. This could be seen in EPD14 eg at controversy. He didn't make this procedure explicit in the explanatory matter one guesses for the reason that he never found the time to carry it out in regard to many words to which it might have been desirable to apply it. No doubt a similar explanation accounts for the absence of more subtle frequency grading in EPD15.
Jones described the forms of speech that he originally set out to record in the EPD as the usages of those who attended the "great public schools". The kinds of people he largely mixed with were no doubt so educated, though the description didn't fit either of his own two senior schools. His earlier public school, Radley College, was hardly one of the "great" ones, and his later one, University College School, even he himself seemed to regard as not a public school at all. See Collins and Mees (1999 pp 7-11). Incidentally, Wells attended a minor public school in Surrey and Gimson a minor London public school (Emanuel School, Wandsworth which in the 1970s became a state comprehensive). For at least thirty years the majority of BBC newsreaders have not been public school products. Roach who, like your reviewer, was also not so educated, remarks at p. 5 (in a decision that he has arrived at a good deal later than quite a few of his colleagues) that "The time has come to abandon the archaic name Received Pronunciation".
Gimson (1973 p.116) very reasonably changed the emphasis in defining his "somewhat fictional standard" model accent for the EPD from "social criteria" to geographical ones. It would have been a suitable moment for a new name instead of having "cowered behind the term received pronunciation" (ibid). (The origins of the expression RP are discussed at length in this website under the heading "British Non-dialectal Accents".) However, Gimson fought shy of such a step, as has Wells. In my CPD in 1972 I introduced the term "General British", the subject of a comment in Gimson 1980 (p. 303). See in Professor Alan Cruttenden's 1994 admirable re-working of that text p. 273. In these days of heightened socio-political sensitivity the term "Received Pronunciation" sounds, on account of its obvious implied downrating of other varieties, more regrettably complacent than ever. RP was Jones's third attempt at finding a term for his kind of pronunciation. He couldn't bring himself to try out any other after 1926 but a decade later in an 800-word note issued by the International Phonetic Association he, somewhat ironically in the light of his propagation of the term RP, commented "Above all it appears to me important that no person should ever disparage the pronunciation of another".
It would be agreeable to report that the invidious term Received Pronunciation has been replaced by a substitute that will meet with general approval but unfortunately the new term employed in EPD15 is 'BBC English'(p.v). It is true that forty or more years ago the then very few official voices of the BBC were almost without exception what Jones would have classified as "RP" though Collins and Mees (1999 p.369) justifiably remark that the unpublished comments on them that the BBC employed him to make revealed less liberal views than his publications presented to the world. There has never been any declared uniform BBC policy of requiring a particular accent from their announcers and in recent decades all branches of the BBC have ever-increasingly expanded the range of accents to be heard in newsreading and the presentation of programmes, making the always questionable term 'BBC English' less and less appropriate.
In LPD the educated accents of the west, the midlands and the north of England, though they're ungracefully labelled "Non-RP", are given valuable attention. EPD15 hardly mentions them but stays conservatively with its traditional policy of exclusively recording the least regionally marked type of accent of England. Likewise, as in LPD, there is no attention to Irish, Scottish, Welsh or Commonwealth pronunciations. There is, however, a certain amount of indication of locally preferred versions of place-names. For example the information on the local stressing of Carlisle has been retained though that on Newcastle has somehow been lost. Barnstaple and Rievaulx have local versions pointed out but not eg Bridgend, Coatbridge or Chester-le-Street.
The "Symbols used for the Transcriptions" in EPD15 appear in the main text and mainly even in the introductory matter in a smaller type size than LPD employs. This is not surprising when one notes that the EPD text is now printed in three columns instead of the previous two per page. It is a pleasure to report that even so its legibility is excellent. The one slight blemish as far as the printing of the symbols is concerned is the fact that the two elements of the voiceless postalveolar affricate /t∫ / are not as closely approximated as one should like.
Although the choices of symbols for the British phonemes, being exactly as in the previous edition, equate precisely with the "de facto standard in EFL work" on British English (LPD p.xviii), those for the American phonemes differ slightly from the LPD ones. Most interesting is the treatment of the schwa vowels. The American final vowel of mother is glossed as an "r-coloured schwa" (on the front endpaper) and the IPA-sanctioned right-hook schwa symbol [ ɚ ] is used for it. By contrast LPD seems by its use of a superscript "may be inserted" schwa [ əʳ ]to be desirably more non-committal about the possibility of the US vowel not being necessarily rhotacised.
Also the reverse of the LPD approach is the EPD15 choice of symbol for the American vowel of bird. It receives the notation of a long schwa [ ɜʳ ] that is basically not r-coloured after which an [ʳ] is added when it is necessary to represent the corresponding "retroflexed" vowel. Indeed non-rhotic long schwa occurs surprisingly often in the text. It is given as the most usual or even only US value at virtually all the items with no r in their spelling eg boeuf, pas de deux, Peugeot, and vingt-et-un. This constitutes an assertion that a generation has emerged in America of whom the majority are so sophisticated as to be able to resist the traditional tendency to use rhotic values in many such words. LPD, perhaps more convincingly, prefers a specifically rhotic-value symbol (right-hooked) for this long-schwa vowel [ ɝ ] and offers no American even variant transcriptions with the non-rhotic equivalent symbol, using it only for its British entries. At a couple of EPD15 items containing oeuvre eg chef d'oeuvre the usual rhotic value of US long schwa is in fact recorded for some variants, in which an /r/ paradoxically appears before the /v/ but not after it. At at-least one entry, Monsieur, a non-rhotic US variant is given for a word at a place where it has an r in its spelling.
The EPD15 American symbol set represents only one long rounded back vowel phoneme whereas LPD has separate symbols for three different types using the Cardinal Vowel symbols 7 and 13 as well as EPD15's 6. One imagines that one of Wells's reasons for embracing such complexity was the thought that the American element in LPD was essentially going to be most often considered by (not necessarily native) speakers of British English who might be glad to have some explicit information on the qualities of the sounds of American pronunciation. Such a motive certainly seems to underlie the LPD choice of the specific symbol (a t-letter with a subscript v) for the American voiced t-sound. This EPD15 also adopts and, what is more, offers the distinctly odd-sounding explanation that it has been used because "speakers of British English find it difficult to apply the rule which determines when phonemes are flapped" (p. viii)!
At p.x of EPD15 it says of the diphthong of boat that "it can be argued that" Cardinal Vowel 7 [o] could have been used to begin the representation of its British pronunciation as reasonably as it does the American variety but that schwa was used for the British sound "in order to preserve compatibility with other works". In my view a schwa first element for the US vowel would represent their most typically heard value very nicely but I suppose that the traditional choice has remained in favour because it reflects better the centre of gravity of the spread of person to person variants which is much further back in General American than in General British. Such matters get little discussion
The values of the symbols for the British vowels are demonstrated by the inclusion of three diagrams at pp viii and ix. These are not so clumsily executed as were the ones in the previous edition. Nevertheless they are not an improvement on the EPD14 diagrams because they are of the old-fashioned small-dot type. The corresponding LPD diagrams are markedly superior. What is more LPD provides diagrams for the American values as well. The glaring omission of such items from EPD15 is especially highlighted by the p.viii remark "It is standard practice in phonetics to represent the quality of vowels and diphthongs by placing them on a four-sided figure…". The text says of the dot-type vowel indicator used "it is misleading to think of this as a precise target". If it is misleading to show vowel values in this way then why do so one asks.
Where EPD15 does score over LPD is in its effective identification of American usages by preceding them with the letters "US" in quite small type but enclosed in a "bubble" which neatly facilitates their instant recognition. In the Cambridge International Dictionary of English (1995), whose treatment of pronunciation matters considerably drew upon what was being prepared for EPD15, my CPD device of identifying American and British pronunciations with $ and £ signs was adopted. Either of these procedures is clearly preferable to the miserably unimaginative use in Longman publications of "||" to herald an American pronunciation.
A situation where alternating values occur that has never been satisfactorily dealt with in EPD is where the text shows omittable sounds with italic letters. This is a difficult matter but EPD15's resort to this device is less satisfactory at eg anodyne or geographical than it is at eg notation or opaque. LPD differentiates such items thus avoiding the charge that the diphthong shown in the first syllable of the latter words in EPD would tend to sound heavy-footed in the former.
Another surprising difference between EPD15 and LPD is the decision that, because it "is not felt to be acceptable in BBC pronunciation to pronounce … with a vowel in the second syllable" (p.xv) words like bottle and wrestle as opposed to words like lightening, it was decided to show them only with an /l/ having the subscript syllabicity mark [ l̩ ]. It is far from certain that all observers will completely concur with this judgement. Any who do are surely just as likely to reject schwa in eg hadn't, wooden etc which are given with superscript schwa in EPD15.
It's perhaps worth noting that over the last decade or so Wells's decision to show (superscript) schwa in all such words is turning out to be increasingly anticipative of the tendencies of younger speakers. For much of the past decade, though it's not mentioned in Wells's 1997 article 'What's happening to Received Pronunciation?', there has been a startlingly rapid movement amongst younger speakers in the direction of replacing traditional syllabic consonants with schwa plus unsyllabic consonant. This has recently become so extreme that one commonly hears people like (non-Scottish-sounding) BBC television newsreaders putting in schwas that have either never at all or at least never within living memory been used in words like bubbling, simply and even occasionally in items like emblem and assembly.
If in those cases there may be some doubt, the apparent decision to let the spelling prompt a false differentiation in the representations of pairs of either perfectly homophonous or at least exactly rhyming words such as bridle and bridal, cycle and Michael, gamble and gambol, idle and idol, mantle and mantel, meddle and medal, peddle and pedal, principle and principal, rustle and Russell, nestle and Cecil etc seems undoubtably wrong-headed.
In earlier editions Jones used three dozen or so symbols from the International Phonetic Association's alphabet to offer information on how a certain number of the foreign borrowings in his text were uttered in their original language. Wells in a virtuoso performance in LPD expanded this feature greatly. Such information is fascinating to those with sufficient knowledge of general phonetics to appreciate it but it can hardly be deemed central to the aims of the EPD. Roach, with such a large expansion of the text in other ways on his plate, has prudently dispensed with such things in EPD15.
However, he does keep on half a dozen or so special symbols to represent sounds used in words of Celtic or Continental provenance that are attempted in a not completely naturalised English form by many people. One of these is the consonant symbol / x / for the voiceless velar fricative heard in eg Amlwch, loch, and Bach etc (respectively Welsh, Scottish and German).
To accommodate English speakers who imitate the Welsh speaker's initial "double l" sound as in Llandudno etc he has not used the devoiced-l symbol Jones employed to show Welsh-language pronunciations in earlier editions of EPD nor yet the IPA voiceless lateral fricative symbol adopted in the LPD but in order to cover two possibilities he has prefixed an italic ie "optional" / h / to an ordinary /l/ as in Miller (1971). This is a very happy solution because, when an English speaker does aim at an /h/ before an /l/ in the same syllable, the sort of h-sound usually produced is very much the same as the Welsh sound. The native Welsh speaker, however, normally produces his lateral fricative without the prop of a following non-fricative lateral. This procedure is therefore more realistic than the LPD representation which falsely suggests that a lateral fricative is being uttered independently of a following normal English /l/, something which would probably seem exaggeratedly authentically Welsh if it was actually produced by a native English speaker.
The remaining exotic symbols exemplified are mainly those for nasalised vowels said to be as in French 'vin', 'change', 'mon' and 'Lebrun'. However, others turn up in the text eg at blancmange, Besançon, São Paulo, etc; and a non-nasal IPA Cardinal Vowel No. 3 with length mark [ɛː] is used at Lefèvre and would have been desirable elsewhere eg at cortège etc. Also length marks come and go on some of the EPD15 nasal vowels. See for example garçon. In an unfortunate now quite out-of-date carry-over from Jonesian times there are still a number of representations of nasal vowels followed by italic velar nasal consonants ending syllables where no ng etc occurs in the spelling, eg at diamanté, dénouement, embonpoint, penchant, soupçon etc.
A neat device used for tagging pronunciations aimed at sounding as close as possible to the exotic original has been to list them with the brief phrase "as if (French, German etc)". Its use seems, however, to be forgotten from time to time eg at Amlwch, Auchtermuchty, Clydach, pibroch, Rachmaninof, Sassenach, Trossachs etc.
One of the most striking new features of the LPD when it came out in 1990 was the use of spaces to render explicit the boundaries between syllables in a way and on a theoretical basis that had never before appeared in a pronouncing dictionary. This move met with a mixed reception but anyhow now EPD too has syllable divisions though on a different principle (the "Maximal Onsets Principle") and with the divisions marked by what would be a full-stop in ordinary orthography but which is the "syllable break" symbol adopted in 1989 by the International Phonetic Association. That Jones's decision not to offer syllable divisions was a wise one is suggested by the many problems that surface especially in EPD15 in this area. That they frequently give the text an unfortunately cluttered appearance is perhaps the least of them.
Probably the largest numbers of unsatisfactory entries in respect of syllable boundaries involve the undesirable positioning of the stress mark in regard to the beginning of the second syllable of a word which has an unstressed though not necessarily weak vowel in its first syllable. The words whatever, mistake and mistime are unexceptionable in this respect, their transcriptions giving the appropriate message in each case as to whether or not the consonant after the unstressed vowel may be said to begin the second syllable insofar as is indicated by the pattern of aspiration. However, very many others are in at least one of their versions misleading including eg Australia, atonal, Baku, echidna, eclectic, Estonia, Gestapo, gestation, gesticulate, Grappelli, macabre, Macao, maquis, metabolism, metallic, necrology, Nepal, occlusive, Ostend, Otranto, stochastic and Uttoxeter. Even where it is not a matter that aspiration would regularly highlight a false division there is the consideration that it is common to emphasise a syllable which begins with a vowel by prefixing a glottal plosive to it. Items like these so treated would sound rather odd: aphonia, Assam, Benito, Canaletto, cephalic, coquette, Emilia, erroneous, Gabon, heraldic, macabre, macaroni, Pegeen, pelagic, Penarth, Ravel, senescence.
Another consequence of going in for syllable boundary representation is that, in a small number of cases, where previous editions were merely uninformative through ambiguity, EPD15 is (in some cases despite its maximal-onset preference) positively misleading. Such items include bedroom, beestings, beetroot, noisome, peacock, quadrangle, rectangle, shyster and teapot. These are all appropriately transcribed in LPD. There are very many unsuitable syllabifications which, as they co-exist with more satisfactory ones, must presumably persist merely owing to inadequate proof correction. They include ceaseless, defenceless, graceless, priceless, senseless, sleepless, useless; creepy, sleepy, weepy; droopy, snoopy, snooty; flighty, mighty, nightie; dopey, ropy; bicycle, icy, icing; cyclone, Euclid, surplus, weakling etc.
Also a problem of rendering syllable boundaries explicit is the impression given that words like actually, agreeable, gradually etc have more syllables than they really ordinarily do in non-formal discourse. LPD avoids a good deal of this by a notation which has a symbol to signal "compressions". Cf CPD p.xvii.
There is perhaps one slightly countervailing advantage which earlier editions lacked but which, one fears, may escape the attention of most users of EPD15. This is that the transcriptions of words like fire, giant, layer, power and sour etc are, through containing no syllabic boundary points, to be taken to specifically represent such words as monosyllabic rather than disyllabic. The more explicit and less committed LPD style (ie conveying the existence at any rate of the monosyllabic variants by its compression signs) seems to have the edge here too.
The recognition of weak /i/ was long overdue because the majority of non-dialect speakers in England have over most of the 20th century come less and less to identify the final vowel of happy with the quality of the bit vowel. This development should not be confused with the related but distinct one that in the last three or four decades there has been an increasing tendency for a still relatively small number of General British speakers to strengthen that final vowel constantly to /i:/. Large numbers of mainstream speakers have for long exhibited strengthening of their habitual weak /i/ to fully strong /i:/ in certain particular contexts. Windsor Lewis (1990), a version of which may be seen at §3.2 on this website, had a discussion of these developments.
Embracing weak /i/ has brought certain problems in its train. Word-internally decisions may not always be easy to make. EPD15 recognises at p.xiv §2.9(b) that in words like busybody /-i-/ is a natural choice. There is no detailed discussion of the matter, but it plainly depends on how far the speaker tends to feel conscious of the constituents of the word. LPD seems perhaps to have gone a little too far in showing items like cauliflower, countryman, Derbyshire, handicap, marigold etc with the so-called "neutralised" /i/ notation only. EPD15 avoids /i/ in all of these. This seems a safer handling of them because the weak closer vowel possibility conveyed by the /i/ notation seems rather unusual in all the above items. The EPD American editor gives both possibilities at handicap. Another problem is that, although in most words like happy final weak /i/ probably now predominates in General British usage, for a very considerable proportion of speakers things like -ies plurals coincide unambiguously in sound value with the stressed verbform is. Yet such versions are not represented in LPD or EPD15.
Both LPD and EPD15 seem oddly reluctant to extend the appropriate recognition to the exactly equivalent weak vowel /u/ except with weakform words and before vocalic sounds word-internally. It's difficult to see any objection to according it to eg Andrew, argue, continue, cuckoo, menu, nephew, statue, value, Zulu etc. EPD15 does, however, give it as a value of the final sound in the name of the letter W and as the sole version of the last sound of the noun thank-you.
The EPD15 chosen "model" of "BBC English" is explained as "…the pronunciation of professional speakers employed by the BBC as newsreaders and announcers on BBC1 and BBC2 television, the World Service and Radio 3 and 4, as well as many commercial broadcasting organisations…" among which professionals they say there is "a reasonable consensus on pronunciation". As we have suggested above, a considerable number of them do display some features showing regional affiliations or external influences in their speech which would not be particularly useful for a user of English as a foreign language to cultivate .
At any rate, to one who has for decades listened to newsreaders and the like as systematically as anyone else one knows of, the suggestion that EPD15 reflects their consensus of usages seems not very realistic. This is not meant as any strong criticism of what EPD15 offers as a "model" ie as a set of usages recommended for adoption. There would undoubtedly be at least some disadvantages for someone who did truly succeed in imitating the most typical habits of the media people in question. It seems clear to me that these have been sieved in the EPD15 model (as well as in the LPD one) through the intuitions of the compilers so that constantly relatively conservative usages are presented. These are typically closer to the the presumptions people naturally make about the pronunciations of words in the light of their traditional orthographical forms than to the ways in which they are currently most often uttered. Such forms no doubt in most cases make for easier learning and cause few problems of acceptability.
However, for anyone who would like to check out this claim, let us for example consider some common words in which the consonant sequence /-nm-/ is given as normally occurring in their unstressed syllables. The most easily observable word in this category is the term government which can be heard every day over and over again in countless news bulletins and current affairs programmes. Both EPD15 and LPD list first the variant which contains the /-nm-/ sequence. However, anyone who listens at all attentively to recordings will soon discover that this is not merely not the predominantly heard form of the word, even in situations of the greatest prominence or highlighting, but that it is actually even a relatively unusual form of it.
In fact it is difficult to mention any particular broadcaster who uses the form in question with any degree of regularity. I know of only one, BBC Radio's (current but about to retire by 2006) Chief Announcer Mr Peter Donaldson. In British public life in general I can think readily of only two persons in this category viz Baroness Thatcher and Mr Tony Benn -- which makes them linguistic as well as political extremists. The Queen is by no means so regular in her treatment of the word but at least on the solemn annual occasions when at the opening of Parliament she reads aloud information about her Government's forthcoming plans she usually unsurprisingly tries to utter the /-nm-/ sequence a proportion of the time. The same treatment has been given to environment etc and to Tiananmen when that name came suddenly to be much used.
Another example is the word only. This is about 90% of the time actually heard with no /l/ sound whatsoever yet the form with /l/ is the only version given in EPD15 and in LPD the form without /l/ is only included as a "non-RP" form. The fact is that the /l/ appears regularly in the word only when it precedes a break in rhythm. Similarly the word certainly is very often to be heard with no /l/. A further example is the word obviously. EPD15 gives the impression that the only form this word takes has four syllables. LPD indicates its potential to become three by compression of the middle vocalic sequence to yod ie /j/ plus schwa but neither dictionary acknowledges that it is more often than not to be heard (as are previously and seriously etc) with that sequence reduced to a simple short vowel. As to its /-bv-/ sequence, this is usually simplified either to /-bb-/, or to /-vv-/ or to a single consonant.
There is no trace in either EPD15 or LPD of the / d / heard in hospital more often than the / t / they show it with. EPD15 has no trace of the fact that accept is much more often heard with the pronunciation they show for except than with initial schwa. Nor is there any trace of the fact that -room in compounds like bedroom and newsroom is constantly to be heard with either schwa or syllabic /m/. Again temporarily is usually pronounced not with four syllables but with three. And BBC itself has normally not /i:/ but weak /i/ in its middle syllable which you don't find in EPD15 though it is at least recognised as a possible variant by LPD.
One could go on extending this list for a long while but the point of it is that there is no need to take literally the claim that the EPD15 notations faithfully reflect BBC etc usages. Indeed in one area Roach is fairly candid about preferring a conservative to a realistic approach to the question of whether to acknowledge the universal use of so-called "intrusive" linking /r/, leaving it out of his transcriptions because he thinks it "safer not to recommend it to foreign learners". I prefer the LPD acknowledgment of it because I tend to find items like concertina-ing, idea of it, Kafkaesque, sonata in C, banana-ish etc in danger of sounding precious without it. For a fuller account of this topic see Windsor Lewis 1975.
The marking of the stress patterns has been very carefully considered in EPD15. Things continue much as before as regards headwords but stress shift is a new area given attention though some of the examples are not as well thought out as one could wish and the words "British only" make a tiresomely obvious comment that would better have been dispensed with. Its tertiary stress indications are not followed but the LPD innovation of adding groups of compound words in ordinary spelling accompanied by stress marks is adopted.
One has the pleasure of pointing out that Roach goes a valuable step further than LPD. Not only does EPD15 show compound words but it makes a start on dealing with one of the most notable omissions from what is provided in English dictionaries of pronunciation by giving also various idiomatic phrases in which the headwords are involved. These in many cases may well be a great help to EFL users because of the frequently unpredictable idiomatic stressings they illuminate. These include a `pretty kettle of fish, have a `chip on one's shoulder, make someone's `hair stand on end, turn one's `back on something/someone etc. There is no song and dance made about the bold new move to include such information (it's not even illustrated in the Explanatory Notes at pp xvi to xviii) maybe because so far the coverage is fairly limited compared with what might eventually be attempted but it looks very promising. There are some doubts displayed in the examples about where non-tonic stresses should be given but that always will be a problem until people like Professor Roach realise the superiority of tonetic stress marks over the more traditional ones for such purposes -- not a matter one can go into here.
At this point I am very conscious of the many matters of interest on which I have made no adequate comment above, including especially the praiseworthy seriously thorough attempt to give full exemplification of the functioning of so-called weakform words but a line must alas be drawn somewhere.
That this very significant and substantial book should be offered to the public as merely a new edition of Jones's dictionary is really too modest a claim. It is importantly different from Jones's book in too many ways for such a description to do its compilers justice. EPD15 will be recognised, one may be sure, as an indispensable item in any substantial collection of books on the English language. No single observer escapes the problems of human fallibility so we must be grateful that we are given the opportunity to consider the inevitably somewhat differing opinions of various authorities. We have to be understanding about the blemishes that must of necessity accompany such a massive onward step as EPD15 constitutes and wish it well for its progress towards ever-improved succeeding versions.
References:
Gimson, A. C. (1973) Phonology and the lexicographer Annals of the New York Academy of Sciences vol. 211
----------- (1980) An Introduction to the Pronunciation of English. Third edition. London: Edward Arnold
Gimson's Pronunciation of English Fifth edition (1994) edited by Alan Cruttenden. London: Edward Arnold
Hornby, A. S. with A. P. Cowie and J. Windsor Lewis (1974) The Oxford Advanced Learner's Dictionary of Current English. Third Edition. Oxford: OUP ("ALD")
Jones, Daniel (1937) On "Received Pronunciation". London: International Phonetic Association
Collins, B.S. & Inger Mees (1989) The Real Professor Higgins: The Life and Work of Daniel Jones Berlin: Mouton de Gruyter
Longman Dictionary of Contemporary English (1978,1987) Harlow, Essex, England:Longman
Miller, G.M. (1971) BBC Pronouncing Dictionary of British Names. London: OUP
Procter, Paul (1995) Cambridge International Dictionary of English. Cambridge, UK:CUP
Windsor Lewis, J. (1972, 1979). A Concise Pronouncing Dictionary of British and American English. London: OUP ("CPD")
-----(1975). Linking /r/ in the General British pronunciation of English Journal of the International Phonetic Association Vol. 5 No.1 pp 37- 42
-----(1985) British non-dialect accents Zeitschrift für Anglistik und Amerikanistik 33/8
-----(1990) HappYland reconnoitred: the unstressed word-final -y in General British pronunciation. Studies in the Pronunciation of English ed. Susan Ramsaran. London: Routledge
----- (1993) Reviews of EPD14 and LPD ELT Journal Vol 47/2 pp 175-177
-----(2003) IPA Vowel symbols for British English in dictionaries Journal of the International Phonetic Association
Wells, J. C. (1990) Longman Pronunciation Dictionary. London: Longman Group UK Limited ("LPD")
----- (1997) 'What's happening to Received Pronunciation?' English Phonetics:Inaugural Issue of the Journal of the English Phonetic Society of Japan. Japan:ARUMU