UPTON, C., KRETZSCHMAR, W. A. Jr & KONOPKA, R. (2001, 2003)

The Oxford Dictionary of Pronunciation for Current English. Oxford: OUP

pp xx 1, 208. ISBN 0-19-863156-1(h/bk) 0198607725 (p/bk 2003)

The original form of this review appeared in the Journal of the International Phonetic Association Volume 34 Number 2 of December 2004. The present version has undergone slight modifications.

1. This volume (ODP) is certainly a weighty contribution to English pronunciation lexicography. It automatically invites comparison with the only two previous works of closely similar aims and dimensions viz the Longman Pronunciation Dictionary (LPD) and the Cambridge English Pronouncing Dictionary ({C}EPD). LPD has 869 pages in two columns and EPD has 606 pages in three columns. ODP has 1,208 pages of text in four columns per page. In hardback ODP has the same footprint and weight as LPD but it is 1.5 cm thicker and consequently rather less convenient to handle. Despite this disadvantage ODP immediately scores heavily over the other two as regards convenience of use in offering less condensed and therefore much more comfortably readable transcriptions.

2. The ODP preliminary matter begins with a single page (p.vi) headed Dictionary team which lists five additional editorial staff and eighteen "Foreign language consultants". These cover Afrikaans, Brazilian Portuguese, Lusitanian Portuguese, Czech, Danish, Dutch-and-Flemish, French, German, Hungarian, Irish Gaelic, Italian, Norwegian, Polish, Russian, Spanish, Swedish, Turkish and Welsh. Even so, ODP offers only extremely limited information on original-language pronunciations of its loanword entries. EPD jettisoned in 1997 the rather few foreign pronunciations it had once contained. LPD has excellent, if neither exhaustive nor quite infallible, coverage of foreign language versions. Their value is, of course, relative to the user's knowledge of the IPA alphabet.

3. The single-page rather effusively, not to say egregiously, expressed Preface (p.vii) proclaims that the authors offer models "in what we believe to be a new manner for the new millenium" having "developed" models "not as incremental improvements upon some prior practice but as the product of our long research experience as students of language variation". They see themselves as having been "called upon [sic] to decide what pronunciations are held in common as national habits or norms by British and American speakers". They refer to their belief that their "wide experience with variation" gives them "standing [sic] to create pronunciation models which avoid slavish imitation of the dictates of self-appointed arbiters [sic!] of taste or style in language in favour of patterns which reflect the actual speech of real people". Some lines later they talk of having "ascertained [sic] particular models". Further on, at the beginning of their Introduction, at p.viii, they refer to models having to be "sought" and "decided upon", expressions that do not seem to be in perfect harmony with others they use. They return at length to this models topic at pp xi to xvi of their Introduction.

4. They acknowledge indebtedness to certain pronunciation lexicographers of past generations (UK's Jones and Gimson and US's Kenyon and Artin) but make no mention at all of any to their contemporaries. They don't even include either LPD or EPD in their Bibliography at p.xvii. To have studied these might have saved them from many regrettable omissions. In a work claiming to provide pedagogical models for learners of English, their reference to debts to the editors of the Leeds University Survey of English Dialects, which dealt exclusively with largely 19th-century long obsolete dialectal  usages, strikes one as very curious.

5. After this single-page Preface, the following dozen pages (viii to xix) are titled Introduction. In its first section, Use of the dictionary, we meet the abbreviations by which the American and British entries are identified viz BR and AM (in the subsequent dictionary text these appear always in small capitals). No reference is made to any of the many other varieties of English than those of England and the USA except very briefly to Canadian at p. xiv. They say of their BR and AM transcriptions that the "choice has been for that model which, if reproduced by users of this dictionary in accordance with the principles [sic] of the International Phonetic Association, will enable them to be understood by native speakers of English without being categorized as belonging to any narrow class, age or regional grouping". This seems somewhat at variance with the remark that their text has been designed to be used "without careful study of a set of complicated conventions"? The only illustrations given of this last feature is their use of BR and AM "rather than … symbols". The one striking advantage for the user of ODP over LPD and EPD, is the showing at every entry "complete … rather than abbreviated transcriptions ... for each alternative pronunciation" (p.viii).

6. This intention is not quite fully realised because parentheses are constantly used around "optional" sounds (often more than once in the same word and on the odd occasion even separately around two symbols in succession). Happily this needs no elucidation for the non-specialist reader though the italicising of every "intrusive" /r/ may do so. Use is made of the traditional "vertical bar" to signal the point to which a word is to be cut back to add an inflectional ending (though it fails to appear at house) and of a reverse oblique stroke to indicate alternative forms of the same ending. The example of this given at p. ix is of spell whose past tense may end as either /-ld/ or /-lt/. Unfortunately, at the spell verb entry in the text no such variation is in fact recorded. One other way in which they may perhaps be said to fall slightly short of their aim to avoid abbreviated transcriptions is their use, not explained though easily comprehensible, of a plus sign to avoid repetition of a transcription already given in the previous line eg as at c'est la vie (twice), at ceteris paribus (on 6 of the 8 lines devoted to the entry) and at Plaid Cymru where the "+" is obviously used to avoid giving again the same transcription for Plaid. At that entry Cymru is intended to appear in a different form, with [ʊ] instead of [ʌ], though it unfortunately doesn't do so. The variant intended can be found at the separate entry for Cymru.

7. In the second section of the Introduction, headed The text explained, they say that "The ordering of variant pronunciations does not imply that one form is more desirable or 'correct' than another" but don't say whether or not first-given forms are to be taken to be the more frequent ones. They say that "optional elements" enclosed in parentheses have equal "acceptability" but again without comment on their relative frequency. "A most basic entry", they point out, "consists of a headword in bold type and identical transcriptions for BR and AM". Surely there would be no loss of clarity in such cases if they used a single line beginning e.g. "BR/AM".

8. So far things are for the most part what we find in LPD and EPD but next we come to two symbols, new to British lexicography, which they have introduced. In regard to them they refer to what they allege to be "the IPA convention of barring to signify centralization of high vowels and retraction of front vowels". The IPA alphabet does indeed contain the barred symbols corresponding to [i, u] and [o] but no such convention has ever in fact been formulated by the Association. Nevertheless, symbols denoting approximately equal frequency of occurrence of /ә/ or /ɪ/  and of /ә/ or /ʊ/ ie [ᵻ & ʊ̶̵]  seem a perfectly reasonable innovation in this lexicographical context and their barred small capitals certainly look less ungainly than the Longman Dictionary of Contemporary English notation which puts symbols one on top of another or the Webster's Third New International Dictionary schwa surmounted with a dot for /ɪ~ә/ variation. They might well have ventured also to employ barred ordinary lower-case [ɨ] to convey the freely varying possibilities of /i(:)/ and /ɪ/ in items like cityscape, Furies, Scillies, sundriesman, undies etc and in the plurals of the many words like city, abbey, safari, coffee etc which they unsatisfactorily show with only /ɪ/ for BR and only /i/ for AM. Sadly, however, purely because the size of type used in ODP is so small, their barred letters may cause problems of legibility for many readers.

9. For the unstressed endings of eg business, challenge, climate, palace, -itis words, the superlative suffix -est and the middle syllables of words like villages the barred small capital /ɪ/ (legibility apart) serves well enough because both versions are very common in BR. However, with many other items, e.g. most -ness-ending words and all -less-ending ones, where /ә/ is markedly predominant in BR, the reader is being denied the information that that is so. Moreover their use of barred /ɪ/ for the final syllable of plurals like villages and past forms like chatted is likely to give the wrong impression that there is very often no difference between them and villagers and chattered when schwa ending the first pair is only a very untypical GB usage. There's a similar problem with the -ful ending where their barred /ʊ̶̵/ is appropriate for adjectives but definitely not for nouns. The claim that environmental factors (preceding high front vowel or so ending diphthong) "frequently govern the choice of vowel-sound" is no doubt defensible for American usage but for BR -less (p. xviii) it is quite unsustainable.

10. The third section of the Introduction at p.x, entitled Technical discussion: transcription sets, remarks that these are "broadly phonetic" and "do not hide potential variation. For instance, both [ru:m~rum] and [rʊm] are possible pronunciations for room". Surely [rum] suggests a common Scottish version of room rather than one common in England. They say "A limited symbol set results in broad transcriptions, and may suggest de facto phonemicizations to some readers, but our intention is always to indicate actual sounds to be produced". It's difficult to see much point in such a remark: their transcriptions for room etc are simply what one finds in LPD, EPD and elsewhere. And they immediately proceed to talk in terms of phonemes saying "No single set of vowel and consonant phonemes can represent all varieties; the following sets are appropriate to the BR and AM models used." Then beneath this appear two groups of vowel symbols set out in diagrammatic spacing with the vertical titles Front, Central and Back and horizontal ones High, Mid and Low. The first of these, "Vowels (BR)", has 12 single letters four of which are accompanied by length-marks but ɛ, which is used in the text at the dress and square vowels and accompanied by a length-mark in the latter case, is missing. The letter "e" supplies the earlier element of the first of five diphthongs given below the diagrammatic grouping but not diagrammatised themselves. At least two further BR diphthongs occur in the dictionary text of which, puzzlingly, no mention is made here viz /ɪә/ and /ʊә/.

11. For detailed discussion of their not completely happy choice of vowel symbols for their BR entries (these have already appeared in certain other OUP reference books) the reader is referred to Windsor Lewis (2003). As to the choices of symbols for the AM vowels, one would have thought that ordinarily the fewer unhelpful contrasts made between BR and AM items the better. After all, the purpose of putting both these lists in one book rather than two separate volumes has obviously to be at least in part to facilitate comparison of the two varieties. Accordingly, to have harmonised the representation of /i(:)/, /u(:)/, /ɑ(:)/and /ɔ(:)/, would have (as in LPD and EPD) met this desideratum and provided desirable economies of space: for example feet shown as BR /fi:t/ and AM /fit/ on separate successive lines seems wasteful of space if nothing else.

12. Incidentally there is very surprisingly nowhere in ODP any convenient listing together with explanatory keywords of all the phonetic symbols used. This has rightly long been a de rigeur feature of most dictionaries using IPA symbols. The use of a different AM symbol /ә/ corresponding to the BR /ᴧ/ is especially unfortunate partly because it requires stress marking for AM in non-tonic syllables that constantly suggests AM/BR differences which, like the length differences mentioned, are either non-existent or totally trivial at least as far as the EFL user is concerned. LPD and EPD sensibly use /ᴧ/ for both BR and AM.

13. In a work that incorporates as this does (at least in respect of the category of words such as bath, dance etc) the predominant forms of non-demotic pronunciation of northern England, it seems strange that the notation does not recognise the fact that most sophisticated northern speakers do not employ the same vowel quality in e.g. foot and cut while yet not using in such words anything from the range of values generally heard from GB speakers in cut. The values they do have in "educated" north-of-England usage for cut etc are in the closer mid central range. These values are likely to be fairly immediately perceived by most GB speakers as not their type of accent, I should think more so than most General American values of the phoneme would strike them.

14. Symbols for five diphthongs are given below the "Vowels (AM)" quasi-diagram and, as with the BR diphthongs, not diagrammatised. No reference is made to a sixth AM diphthong /oә/ which occurs in the text but is confined to monosyllables and only the final stressed syllables of polysyllabic words where /r/ follows it e.g. in fork /fo(ә)rk/ with its schwa element always bracketed. It seems to be being suggested that speakers who have the diphthongal variant in eg fork do not have it in eg forkful etc. Further AM diphthongs not mentioned here are /ɪә/ and /ʊә/, as shown at fear and sure etc, and /ɛә/ and /ӕә/, as at yeah, all of which appear in the text with bracketed schwa.

15. In the list of consonant symbols for BR and AM the first fricative item shows a gross deficiency in proofreading by having [æ] where a bilabial consonant symbol is required. ODP does include a bilabial-fricative phi symbol [ɸ] at the entry pshaw. A lateral-fricative symbol [ɬ] is, as in LPD, unfortunately employed at the ODP first BR versions of e.g. Llanelly, Llangollen, Pwllheli. Such transcriptions indicate sounds which are in fact only within the competence of a speaker with native-equivalent ability to pronounce Welsh. EPD's /hl/ is a much more realistic representation of the native English-speaker's performance in such items. The p. x entry "l=[ʎ]"is absolutely mystifying.

16. The fourth and longest section of the Introduction (pp xi to xviii), headed Pronunciation models, is divided into two major parts BR (pp xi to xiii) and AM (pp xiii to xvi) but ends with two short subsections which deal with AM and BR side by side (pp xvi and xvii) on matters of stress marking and vowel reduction. The former of these contains the vague assertion that AM "has a heavier stressing pattern than BR" which is illustrated by the curious claim that the second syllable of baseball has secondary stress in AM but tertiary stress in BR. Whatever may precisely be meant by this comment, what learner can possibly benefit from the display of a secondary-stress mark in the AM version? The other example is dictionary in which the subordinate stress marking is equally superfluous because the different phonemic structures of the BR and AM versions of the word make matters perfectly clear.

17. Their third point regarding stress marking refers to the adoption of Artin's Webster's Third New International Dictionary device of showing primary and secondary marks before the same syllable to indicate that it may bear either degree of stress. Thus the first and third syllables of overbearing are both so shown to convey, it is said, the possibility of the alternative stressings "primary-primary, primary-secondary and secondary-primary". In fact Artin's own use of the device referred only to the latter two of these choices. The word overbearing uttered with two tonic syllables would surely suggest an affective not a lexical pronunciation.

18. In the first major part of this fourth main section of the Introduction, in the two or three pages each of accounts of their BR and AM models, the discussion of BR, which is equated with "Received Pronunciation", refers to its mainstream variety as carrying "connotations of education and sophistication but no especially narrow regional overtones and certainly no serious negative judgements". It is remarked (p.xi) that "it might loosely be labelled broadcast RP if yet another label were to be thought desirable". Roach's explicit discarding of the term RP in the 1997 EPD 15th edition in favour of the unsuitable label "BBC" receives no mention. ODP's own BR is just such a new label which no other dictionary has used hitherto. They add "to correct a situation where the British model is the possession of a small minority restricted in terms of age, class and region, a younger unmarked RP is that which provides the model in this dictionary... that accent which will be most widely acceptable as well as most intelligible to native BR speakers".

19. It is possible to invent an artificial language as with eg Esperanto. Accents can hardly be considered as capable of being invented. Talking of "individuals who can legitimately lay claim to an RP accent", and of "RP" as a "possession", sounds less than felicitous. Despite the deploring earlier of "self-appointed arbiters", ODP's insertion of relatively regional items into the present work is by no means less prescriptive than other works but clearly more so. LPD in 1990 introduced the highly desirable feature of including many pronunciations widely used by people of education and sophistication which, although perfectly acceptable to intelligent and well-informed persons in general, are as it happens perceived by people who speak with this most general kind of accent to be extraneous to their habits and in most cases associated with varieties of English of northern England. LPD identifies them by the arbitrary symbol § (first edition †).

20. The decision by Daniel Jones in 1926 to label the type of accent he was accustomed to describe as "Received Pronunciation" was extremely unfortunate, ironically so for someone who in Jones (1937) went out of his way to insist that "no person should ever disparage the speech of another", because the term's inevitable implication is that those who do not employ that accent exhibit one in some way not accepted. Most of his pupils and their followers appeared to be either oblivious of the term's invidiousness or from ingrained habit unable to bring themselves to replace it. However, the most significant defining feature of the accent today is not its social acceptability but its geographical neutrality: it certainly tends to get thinner on the ground the farther one travels from its place of origin, London, but it can be met with anywhere in the whole of England and Wales, and even to a slight extent in Scotland, exhibited by persons who have spent the major part of their formative years where one finds them. See Windsor Lewis (1985).

21. My preferred replacement for the now increasingly rejected term "RP" is General British (in short GB) the term I introduced in Windsor Lewis (1972). This is not an ideal label but I have yet to see a more acceptable alternative proposed. GB has an identity which cannot reasonably be deemed able to be re-invented or modified. It can only, as in LPD and EPD, be identified, admittedly noting some inevitable fuzziness at the edges. ODP, with unacceptable arbitrariness, seeks to "correct a situation" from on high. Its designation as ODP-type "RP" of the vowels of brass, staff, bath, chance and sample is a self-contradiction acknowledged in the admission that "this [a] is the one specifically 'northern' BR feature which has regularly been incorporated into the transcription system" (p.xii). But in fact their inclusion of strong vowels as variant pronunciations of the first syllables of words like example, examine, exclude, expect etc is no less "northern"; and, seeing that strong-vowel variants are given for those, why have they not been admitted for e.g. admire, admit, advance, conclude, complain, compute, objection, observe, obtain, substantial, success etc? Other items widely employed by educated northerners which ODP fails to accept as BR include because and raspberry with /s/, decision with /∫/, gooseberry with /u:/, eighty and eighteen with /-tt-/, magazine with initial and yesterday with final tonic stress, mischief with latter vowel /i:/, the adjective perfect with /-ekt/, one, once, none and nothing with /ɒ/, us and various dis- words like dismiss with /z/, and with with /θ/. These usages, widespread in educated varieties of speech of various northern and some other areas of England and Wales, can hardly be classified as dialectal or demotic. [PS Indeed Upton himself embraces at least one of them.]

22. When a GB item is given with both /æ/ and /ɑ:/ variants in EPD or LPD one knows that they are being recorded as GB variants and that the second version is considered the less usual GB form but not simply a northernism. In ODP one is denied this information because their second version may well be a regionalism only. This applies to various items such as graph, autograph, cenotaph, photograph, telegraph, chaff, giraffe; lath, lather, McGrath; bastard, flabbergast, Madras, masque, Plaistow, Prendergast; Flanders, plantation, Sandra, tranche, circumstance and stance. Unlike castle and grass, the entries Glastonbury, Castleford, Castleton, Grasmere and Grassington are all given in ODP with /a/ first. These could perhaps be examples of an unstated policy of according precedence to the local form, yet Castlebar and Castlewellan are not so treated. The local stress value is only given second at Carlisle; and eg Newcastle and Chester-le-Street are not shown with local stressing at all. Some ODP items completely omit common variant BR forms with /ɑ:/ eg aftermath, Cassandra, elastic, masculine, mastiff, plastic. Sanders is given without initial capital or the /ɑ:/ listed as its only GB form in LPD and EPD. The ODP reader is not made aware that horrible, terrible etc with /ɪ/ are in unselfconscious or unpedantic speech exclusively northernisms.

23. The second major part of the fourth main section of the Introduction (pp xiii-xvi) deals with the AM 'pronunciation model'. It begins by saying that AM contains 'no identifiable variety widely spoken by well-educated, cultivated residents ... Regional varieties of pronunciation show few signs of giving way before the mobility of the population and the omnipresence of national broadcast media'. Yet later it remarks that 'the model adopted here follows the trend among younger educated speakers of exclusion of regional features' (p. xiv). It proceeds to refer to certain major regional variants 'not included in the transcriptions, both because of regional marking and because this volume seeks to present pronunciations typical of the model in slow to moderate speech' (p. xiv).

Acceptances and exclusions contain few surprises. The trio of marry, Mary and merry all have the same vowel because alternatives are judged 'recessive and regionally marked' (p. xv). Similarly with horse and hoarse and with halve and have. On the other hand calm and palm receive optional /l/ and eg mental is accorded optional /t/. There is also reference to possible diphthongisations of vowels being 'shown by an optional mid-central vowel ... except for' words like four (p. xv): yet at the AM entry for that word a bracketed schwa does appear.

24. No use is made of length marks or specifically retroflex symbols. EPD uses /ɚ/ for the US 'r-coloured schwa' where  LPD shows  /ər/. For American 'long schwa' LPD has/ˈmɝdər/ for murder and EPD has /ˈmɜr.dɚ/. I don't think ODP users lose much by being spared this complexity. 'Intervocalic /t/, as in latter, is often realized as a flap [ɾ] or voiced t [ t ̬] but is transcribed here as [d]' (p. xvi) so that latter and ladder have identical transcriptions.' (At p. xvi the symbol for the flap is much taller than the IPA authorised form being like an f with no crossbar.) So the claim to keep the transcriptions free from complications is considerably borne out as compared with the LPD and EPD American entries.

25. The former of the two final subsections, 'BR and AM stress marking', contains the vague assertion that AM 'has a heavier stress patterning than BR' (p. xvi), which is illustrated by the suggestion that the second syllable of baseball has a secondary stress in AM but tertiary stress in BR. Whatever may be meant by this dubious comment, what learner can possibly benefit from the display of a secondary-stress mark in the AM version? The other example given is dictionary, in which the secondary-stress marking is equally superfluous because the different phonemic structures of the BR and AM versions of the word make matters perfectly clear.

 The fifth main section of the Introduction is an eleven-item Bibliography which incidentally doesn't include Webster's Third New International Dictionary though that work is referred to at p.xvii.

26. The sixth and final main section of the Introduction consists of a page or so headed Foreign pronunciations which is introduced with the remark: "When given in native form, a limited number of foreign headwords also have native pronunciations based on current national standards, but only where these differ significantly from the anglicized pronunciations." It is added that words "falling within specific subject fields, e.g. culinary and musical terminology, have not been given native pronunciations". Neither in fact have any items not linguistically European. No explanations are offered for these decisions. Quite what is meant by "significantly" also receives no elucidation. For example Murillo and Velasquez receive foreign native versions but not Cervantes, Goya or Picasso. One might have imagined that having [x] for the "x" Mexico and for the "ch" in Machynlleth would be considered a significant difference but neither is given a foreign-language version. Bolivar, Cadiz, Dali, Helsinki, Maastricht, Otranto, Trafalgar and Uppsala which have differently placed tonic stress from their native forms aren't either. Copenhagen, Trondheim and smorgasbord get no Scandinavian version; Paris, Besançon and Nez Percé no French; Dusseldorf, Pachelbel and Weltschmerz no German. By contrast Myfanwy and Pontypridd are given Welsh transcriptions differing only by the use of [ᴧ] in an unstressed syllable which for Welsh itself might at least as well have been shown as /ә/ as indeed it is in LPD. Mistakes occur at not a few foreign items eg Bergman, Brno, Clwyd, Hofmannsthal, Malmö, öre.

27. In their diagrammatic setting out of foreign-language vowel symbols a front vowel at Open-Mid or between that and Open (their three-only levels are termed High, Mid and Low) is omitted though there are plenty of examples of items employing [ɛ] in their foreign-language transcriptions. The item /a:/ as in Dutch Waal is not the symbol found in the text of the dictionary at that entry. At Beira the vowel exemplified is again not used but [a] appears instead of the suggested back vowel. At Douro they have underlined the right vowel but put it in the wrong syllable. At each of the simple-vowel types listed one or more languages containing it are specified; each language is illustrated by one or two words except that at Italian and Spanish "mid vowel" and "close back vowel" are used. The representation of the voiced velar fricative consonant is, by the common solecism, not IPA's proper gamma but its "ram's horns" vowel symbol. All these items except the diacritics are presented between slants not square brackets. The apostrophe symbol for the Danish glottal catch is given the definition appropriate to the slightly similar-looking "corner" diacritic [˺] for absence of audible release! If they meant to include a symbol corresponding to that definition, its appropriate location would perhaps have been with English sounds at p.x: yet one doesn't appear at Yep or Nope. A dental click symbol is included and given first position at the entry tut but appears not at all at tut tut. (At p. xiv the symbol for "flapped /t/" is much taller than the IPA recognised form.)

28. Despite the ODP declaration that it sets out to "reflect the actual speech of real people" (p.vii), it appears in this respect if anything to do so less than LPD and EPD. Perhaps what its authors have in mind are the many grossly outdated notations of its senior stablemate the great Oxford English Dictionary second edition of 1989. Since it comes from the "Academic" and not the "ELT Dictionaries" division of OUP, one wondered whether to expect to find in it numbers of those pronunciation variants which are in many cases well enough known to editors but silently eschewed even though extremely commonly heard from native speakers. Their inclusion would be of little value to EFL learners who are, on the whole, best offered pronunciations that accord most closely with words' ordinary orthographies. The fact is that recommending the most helpful models for learners and representing faithfully the actual speech of native speakers are mutually antagonistic procedures.

29. The chief reason that these thousands of frequent pronunciations are excluded from even the specialist dictionaries of pronunciations is the circumstance that English spelling has been very stable for several hundred years and literacy among English-speakers has been at a high level. Thus when words develop new forms these, so far from being adopted as norms, are often censured. Older forms represented by the orthodox spelling persist beside the new ones because the literate most generally favour faithfulness to orthography over oral tradition.

30. A notable set of words in this category are the huge numbers which have "weakforms". By this is not meant forms which are merely weakly uttered in terms of breath force or loudness but ones which have such phonological contrast with their "strong" versions they might well have been quite separate words. A small number of these are so constantly employed by all speakers that failure to use them in appropriate circumstances will suggest a markedly formal speech style. The few dozen such words, mainly certain verb forms, prepositions and pronouns such as have, to and her are quite carefully treated by EPD and LPD but inadequately so in ODP.

31. An example of the many words which have weakforms which cause no stylistic problems is only. It is perfectly normal-sounding as given in EPD, LPD and ODP as containing an /l/ and it is invariably so pronounced as a prepausal item but, whether stressed or not, on every other occasion it is overwhelmingly most often uttered throughout the educated native-English-speaking world without any /l/. There is little gain for the learner to be given this information and the only admission of the existence of this majority form without /l/ is in LPD which wrongly categorises it as "non-RP". Some other cases where a form denied first place in (if not omitted completely from) almost all dictionaries is to the intensive observer clearly either a majority form or doubtfully of less currency than the dictionaries' preference may be exemplified by the following.

32. All three of our dictionaries record variants of the word government without the first /n/ suggested by its spelling. None of them, however, gives such a form first. Nevertheless, versions of any word with the possible sequence /-әnmә-/ are in fact only relatively infrequently to be heard with that /-n-/. This applies also eg to environment and to Tiananmen (Square) even though this last word was until recently very unfamiliar to most English-speakers.

33. ODP, which has no truck with the non-phonemic "flapped /t/" symbol of EPD and LPD, unexceptionably shows hospital as uttered by AM speakers with final /-dl/. Such a transcription would faithfully reflect the usage also of the majority of GB speakers but no trace of this even as a subvariant British form is to be found in any current dictionary.

34. Only in LPD but not at all in EPD or ODP is a version found of accept which coincides with first given version they have of except but such a value is commoner than the first versions they do give (which begin with schwa or /ɪ/).

35. Adverbs ending with -ly are particularly prone to reduction in unselfconscious speech very widely receiving a syllable fewer than is so often unrealistically indicated in dictionaries. Among examples of this are previously, seriously and invariably which generally lack any sound corresponding to the i of their spelling. Only in EPD of the three major pronunciation dictionaries would the first version given of usually not be relatively rarely heard.

36. The adjective meteorological was for many years to be heard daily from BBC weather forecasters who hardly ever said it as shown in the dictionaries.

37. Large numbers of words always indicated as having inflected forms ending /-sts/ are constantly to be heard without any final /s/ in all circumstances but especially when unstressed.

38. These remarks could be greatly extended but we must sum up our impressions of the ODP. Despite its rather numerous imperfections we must welcome this ambitious new work. It can only be an advantage to have an independent set of impressions from the ones available in the only two previously available works of this nature and scope. It must be salutary for students to be able to compare the contents of the three works and see that there is very far from unity of opinion on the prevalence or otherwise of very many variant pronunciations. Furthermore the presentation of ODP offers us a stimulus to reconsider how this kind of information should be presented in the future. It seems probable that in time ODP's more user-friendly way of setting out variants will become the norm because computer screens are likely to become the preferred mode of access to their content. In CD-ROM or DVD versions the type size can cease to be a problem. The great benefit of showing so many separate-line variant transcriptions uncluttered with the numerous italics and superscript miniature symbols etc which are so tiresome in other works will then be fully realised.

References

Jones, D. (1997, 2003) Editors: P. Roach, J. Hartman & J. Setter. English Pronouncing Dictionary (15th edn; EPD). Cambridge UK: Cambridge University Press.

EPD: see previous item.

GB: see reference to Windsor Lewis (1972) 

Jones, D. (1937) [Leaflet] On "Received Pronunciation". London: International Phonetic Association.

Longman Dictionary of Contemporary English (1978). Harlow, Essex: Pearson Education.

LPD: see Wells, J.C.

Cambridge International Dictionary of English (1995) Editor-in-chief : P. Proctor. Cambridge UK: Cambridge University Press.

Webster's Third New International Dictionary (1966) London, G. Bell & Sons Ltd. Editor P. B. Gove. Associate editor for pronunciation: E. Artin.

Wells, J. C. (1990, 2000). Longman Pronunciation Dictionary (LPD). London: Longman Group UK Limited.

Windsor Lewis, J. (1972). A Concise Pronouncing Dictionary of British and American English. London: Oxford University Press.

Windsor Lewis, J. (1985) British non-dialect accents. In Zeitschrift für Anglistik und Amerikanistik 33/8. 244-257 Germany: Leipzig. (For a version of this see Section 7 Item 3 on this website.)

Windsor Lewis, J. (1999) Review of Jones, D. English Pronouncing Dictionary edited by Peter Roach & James Hartman. Zeitschrift für Anglistik und Amerikanistik XLVII 3, 3 255-264 Germany: Tübingen. Stauffenburg Verlag.

Windsor Lewis, J. (2003). IPA vowel symbols for British English in dictionaries. Journal of the International Phonetic Association 33, 2 143-152.