This volume (hereafter OBG) offers ‘guidance on the pronunciation of names, words and phrases . . . which are tricky, much debated, curious, or exotic’ (p. 7). It employs what is optimistically referred to as an ‘easy-to-read phonetic respelling’ (p. vii) alongside ‘the system familiar to users of larger Oxford dictionaries’ (p. 7) (though not the system in the present second edition of the great Oxford English dictionary (OED). It is very similar in its compact, large-pocket-size form to the 1983 second edition of the (unfortunately no longer available) BBC pronouncing dictionary of British names (Pointon Reference Pointon1983), with which it very occasionally goes over the same ground. OBG's 16,000 or so entries constitute about eight percent of a database of approximately 200,000 items available electronically within the British Broadcasting Corporation.

OBG is clearly aimed at a rather more popular readership than the more academic Longman pronunciation dictionary (Wells Reference Wells1990, 2000; hereafter LPD), Cambridge English pronouncing dictionary (Jones Reference Jones, Roach, Hartman and Setter1997, 2003; hereafter EPD) and Oxford dictionary of pronunciation for current English (Upton, Kretzschmar & Konopka Reference Upton, Kretzschmar and Konopka2001; hereafter ODP). It contains many interesting asides, notably those at the entries for Althorp, Davos, Moog, Häagen-Dazs, jacuzzi, Nabokov, Ncube, Porsche, Purcell, quark and ski. For example, one learns that Nelson Mandela has a middle name recommended to be pronounced (when Anglicised) /rəʊli ɬɑː ɬə/. One encounters numerous words one is glad to have advice on and many one even had no idea existed.

Twenty ‘information panels’ provide a page or two each of informative notes on the interpretation of the orthographies of a dozen languages among which are Arabic, Czech, Latin and Welsh though not Russian. The other panel topics are Accents (referring to those which appear over alphabetic characters), Anglicisation, BBC English, Clicks, Tone and three ‘top ten’ lists. These last are said to ‘appear in the appropriate place in the alphabet’ (p. xii) but locating them using this criterion is not straightforward: they are ‘mispronunciations’, ‘most frequently asked pronunciations’ (both found under M) and ‘pronunciation complaints’ (under C). We are not told what form the mispronunciations take.

The main entries are set out in two columns per page, each item having its headword followed in the same line by a ‘brief definition’ (often extending one or more lines further). In the next line after this gloss a single recommended pronunciation is given firstly in ‘respelling’ in a sanserif font with its tonic syllable distinguished by being printed in bold, and secondly between phonemic slashes in a slightly simplified version of the set of IPA symbols the ODP uses (for a discussion of which see Windsor Lewis Reference Windsor Lewis2003, Reference Windsor Lewis2004). The pronunciations are said to ‘reflect . . . in most respects . . . the standardized [sic] accent of British English known as Received Pronunciation’ (p. ix) but with the bracketed inclusion of the r-sounds of higher-rhoticity varieties as in previous BBC publications. At interrogative a bracketed letter r precedes the single /r/ sound that the word actually contains and another which I believe to be unsuitably bracketed appears at Surya. It would be odd to omit the /r/ from this word, in my opinion. A few brackets around r's are missing at Abertawe, Caerdydd, Ponchartrain, Villiers (at Stanford). No other brackets are used nor are italicised or superscript letters or subscript syllabicity marks etc. This makes the IPA transcriptions very comfortable to read. So-called ‘optional’ schwas are quite regularly included. This is useful in that it tells us where schwa can occur, but gives no information about the frequency of its inclusion. However, we do find nugatory as /ˈnjuːgətri/, Canaveral as /kəˈnavrəl/ and Niagara as /nʌɪˈagrə/ as well as the more usual treatment, as at e.g. comparable /ˈkɒmpərəbəl/. The French items matelot /ˈmatləʊ/ and Miquelon /ˈmiːkəlɒn/ are respectively without and with such schwas.

There is no explanation of in what respects the policy of representing the General British accent (a term I prefer to ‘Received Pronunciation’, hereafter GB) is departed from but it is so at almond, Davies, nougat, tortoise and others, and also when schwa is recommended for the endings of Bowen, junket, octave, outage, Routledge, tallage, verbiage, visage and widget (which should appear with barred i) though suffrage, tutelage, ullage, umbrage and vestige are orthodoxly treated. Moor /mʊə(r)/ is followed by ‘also mawr’, after which the unelaborated comment occurs: ‘This difference in pronunciation depends on regional accent’. In addition various other words are shown with /ə/ when for GB one would have expected /ɪ/ or some other vowel, e.g. accursed, Beelzebub, Bethlehem, blessed (we find two transcriptions here, one (p. 3) with /ə/ and one at the expected place under B as /ɪ/), caret (homophonous with carrot), cursed, Dulles /-əz/, (Alec) Guinness, laches, unwonted, woebegone and others, too numerous to mention. Also less than usual GB (with /ɪ/ rather than schwa or /i/ in unstressed syllables) are the versions of all words ending with -ity (e.g. acuity, alacrity, laity, sodality), annihilate, anopheles (contrast Mephistopheles), coriander (contrast hacienda), homily, simile, irascible, risible, oogenesis (contrast genesis), pleurisy and others. Orkneys /ˈɔːkniːz/ and Sotheby's /ˈsʌðəbiːz/ are given only as indicated, and Ramillies only as /ˈramɪlɪz/, where /-iz/ might have been expected for all three. Similar discrepancies occur at atheism /ˈeɪθiɪzəm/ cf. pantheism /-i:ɪzəm/ where /-iɪzm/ would have been expected.

Some entries have an extra bulleted one-line comment and in many cases a longer note is set between ‘ruled lines’, giving usually only one alternative pronunciation and then only in ‘respelling’ not IPA, e.g.:

Very many items are undoubtedly eligible to have more variants listed than are accorded and when alternatives are included it seems that the choice of which words get them is extremely haphazard. The recommended pronunciation of Xanthippe has /t/ without any reference to the alternative form with /θ/ when it seems obvious to me that the /θ/ form is not only equally correct, but also arguably more common.

Sporadically, to offer transcriptions of loanwords’ original-language forms, IPA is used between square brackets. Occasionally, comparisons are made with American forms and with bygone usages. IPA native-language versions appear in some places between slants e.g. at Buenos Aires, Cervantes, déjà vu, fait accompli, Ludwig, Montevideo, pudeur, Teachta Dála and Ypres (presumably inadvertently, since these should have square brackets). The versions not representing original-language values sometimes (again presumably inadvertently) have /i/ instead of either /iː/ or /ɪ/ e.g. at /agəˈnistiːz/ for Agonistes (at Samson), /ˈalsiːbʌɪədiːz/ for Alcibiades, /daˈvið/ for David (Nalbandian), /ˈsɪsili/ for cicely, /franˈθiskəʊ/ for Francisco (Zurbarán), /ˈluːtviç/ for Ludwig (at Wittgenstein but not at Beethoven), /miŋˈgɛlə/ for Minghella, /mizəˈrɛrə/ [sic with the inappropriate final schwa and missing secondary stress] for miserere, /sɒˈbibʊə(r)/ for Sobibór and others; Domingo with the only recommendation /dəˈmiːŋgəʊ/ is possible but is much more often heard with /ɪ/. One finds other vowel representations which English-speakers would be unlikely to use in medial unstressed syllables, e.g. /ɛ/ in Alexandria, Seneca; /ɒ/ in Bogotá, Pergolesi, Tintoretto, Tivoli and in the first syllable of Stokowski, and the last of Don Quixote; /a/ in Salamanca and so on. Likewise with the final syllables of Angharad as /aŋˈharad/, Dyfed as /ˈdʌved/ and Gwynedd as /ˈgwɪneð/.

The respelling system is not the ‘in-house phonetic spelling’ of previous BBC publications which in fact the authors mention that they use in their daily work but one required by Oxford University Press to harmonise with the Press's Pocket Oxford English dictionary. This differs from the arguably superior BBC system (which is aimed at minimum departure from the normal orthography) and ones used in various less recent Oxford dictionaries by its complete avoidance of diacritics, italics, superscripts and underlinings except solely for italicised th for /ð/. It produces some rather dismaying-looking items as when Ieuan is given as ‘yyuhn’ and Severnaya as ‘syay-vuhrnuhy-uh’ (i.e. /ˈʌɪən/ and /ˌsjeɪvə(r)nəjə/). It seems a false economy to have used ‘y’ for both /ʌɪ/ and /j/. For /ʊ/ doubled ‘uu’ seems rather counter-intuitive. More on this notation will be found in Windsor Lewis (Reference Windsor Lewis2006).

With commendable realism, these OBG respellings do not attempt the [ø/ǫ], [y/ʏ] and [ç/x] native distinctions of foreign loanwords showing them as ‘oe, ue’ and ‘kh’, which (it is commented, p. xv) are ‘disambiguated’ by the IPA versions. The list of symbols ‘used for the main IPA transcriptions’ (p. xvi) includes ‘/i/ as in cosy’ but not the corresponding /u/, which, however, does appear in some words, e.g. conduit, congruous, contiguous, factual etc, and might better have been used than /uː/ at aluminium. It has the nasal vowels /, / but fails to include , which is in fact used in entries in the same volume for pronunciations recommended for English-speakers at e.g. arrondissement, bon (voyage), chateaubriand, contretemps, croissant, (filet) mignon, tendresse, ton and (vol au) vent, where, in my opinion, it provides a more suitable representation of what usually occurs in GB than does. One's impression is that is used often where would have been the better choice, e.g. compare OBG /bl/ with LPD /bl/ for (Mont) Blanc. It would similarly have surely been an advantage to use /ã/ rather than // for the French vin vowel in a text where /ɛ/ has been favoured for the dress vowel. This use of could perhaps mislead some readers into thinking that a suitable target for the vin vowel is a nasalised equivalent of the dress vowel. This list omits certain symbols employed only very occasionally for original-language transcriptions including the voiced palatal plosive (ɟ) used e.g. at Ligeti, the voiced velar fricative (ˠ) used at e.g. Nicaragua, the voiceless uvular plosive (q) and glottal plosive (ʔ) both used at e.g.Koran, and the voiced pharyngeal fricative (ʕ) used at e.g. Qaeda.

It is noted that ‘[t]he way the words are broken into syllables in the respelling is not an attempt to reflect actual syllabification [but] to reinforce [sic] vowel pronunciation’ (p. xiv). This results in some strange-looking sequences, e.g. ar-bit-ruh-ri. Secondary stresses are said to be ‘not given’ (p. xiv) in the respellings (though one is at Tir-na-nÓg). They are promised in the IPA versions but given very erratically. If Wuxi is /ˌwuːˈʃiː/, why is wonton /wɒnˈtɒn/ not /ˌwɒnˈtɒn/? The second tonic at Putrajaya is no doubt simply a misprint but many items like Aberfan /abə(r)ˈvan/, paradigmatic /parədɪgˈmatɪk/ and Rigoletto /rɪgəˈlɛtəʊ/ appear with no secondary stresses. Laboratory /ləˈbɒrəˌtəri/ on the other hand has a rather unorthodox stress shown at a syllable with a weak vowel. Forenames etc are regularly given secondary stress marks (though occasionally not so, e.g. at Zurbarán) and are so quoted in this review when mentioned separately from the main entry for the surname.

The authors say: ‘We have selected only those words and phrases which we believe a native speaker of English might be unsure how to pronounce’ (p. ix). This immediately makes one wonder how many intelligent enquiries they are likely to receive about their entries banquet, Diana, ferrous, Gad, Ham, Kidderminster, Norway, Paris, radium, remunerate, syllable, Warsaw, unanimous and many other commonly-heard items. The usefulness to their readers of entries such as deinonychus, dilophosaurus, phenylketonuria, strongyloidiasis, telangiectasia, uropygium and various other recondite items is questionable. Their admitted ‘magpie-like approach’ (p. ix) clearly sets out to provide entertaining demonstrations of the remarkable diversity of the enquiries they receive in their daily work rather than to maximise the usefulness of their text for anything like a wide audience.

Thus, OBG is plainly distinguished from the major pronouncing dictionaries which are designed with users of English as a foreign language very much in mind. It is said to be ‘intended for students, teachers, actors, journalists, broadcasters, and anyone interested in “saying things right”’ (p. vii). It is clear from the language panels and the usually impressively convincing transcriptions of a wide variety of native versions of foreign words that the compilers are well-equipped general phoneticians. However, many of the judgements involved in the details and even in the principles on which the book is based are unlikely to meet with general approval. One fundamental problem is that it does not seem to be accepted as axiomatic that, if a pronunciation is that most widely used by intelligent, educated people, then it is the most appropriate one to recommend. It is often difficult to guess whether this principle, if accepted at all, is being forsaken from inadvertence, ignorance of what usage really is or reformist zeal.

The authors say: ‘Our general aim is to recommend pronunciations that are as close as possible to the native language in question but modified slightly so that they will flow naturally in an English broadcast’ (p. xi). Can such a policy be said to have been carried out with recommendations such as Gijón /xiːˈxɒn/, Hugo /ˈhyxəʊ/ (at de Vries), Machynlleth /məˈxʌnɬəθ/, Krysztof /ˈkʃɪʃtɒf/ (at Penderecki), Przewalski /ˌprʒəˈvalski/, Pwllheli /puːɬˈhɛli/, São Tomé e Príncipe /ˌsãʊ tə ˌmeɪ ə ˈpr ɪ ˜ ːsɪpə/, Schiphol /ˈsxɪphɒl/, Szczecin /ˈʃtʃɛtʃiːn/, Tskhinvali /tsxɪnˈvɑːli/? They certainly have an unenviable task with such items. It is not surprising that one has never heard their recommendation implemented for the name of the late unfortunate Brazilian Jean Charles de Menezes as /ˌʒiː ˈʃɑː(r)lɪs dʒi mɛˈnɛzɪs/. It is hard to imagine exactly how they could best advocate its simplification. They only have, I suppose, to offer a transcription and expect readers to make the best attempt at it they can manage. For Tsvangirai, /tʃaŋgɪˈrʌɪ/ is recommended but /tsvaŋgɪˈrʌɪ/ would have been better judging from the Shona-speaker I have talked to. The authors make no attempt to modify much the same initial sequence in Tsvetovski. On the other hand, it is not at all surprising never to have heard anyone pronounce al-Qaeda as /əl kɑːˈɪdə/. The claim that this ‘recommendation is as close as possible to the Arabic pronunciation’ (p. 317) is strange. When heard from Arabic-speakers it has sounded to me far more like /ˈkɑːiːdə,kɑːˈiːdə/ or /ˈkaɪdə/ which last seems to be the predominant usage of GB-speakers, as EPD confirms (LPD and ODP currently lack the entry). It is possible that we have another misprint here of course. One has also never detected a completely successful attempt at their recommendation /svɛn ˈjœːran ˈeɪrɪksɒn/ for Sven-Göran Eriksson.

The authors’ expressed aim ‘is to reflect current spoken language’ (p. xii) but various names well known to English-speakers (often spelt exactly as in English) and generally used in anglicised pronunciations are recommended to be adopted in rather exotic versions, e.g. Adolph /ˈɑːdɒlf/ (Baeyer), Isaac /iːsɑːˈak/ (Albéniz), Louis /lwiː/ (Althusser et al. but not for French kings), Paul /pɒl/ (Éluard), Sebastian /zəˈbastiən/ (Bach), Juliet [sic for Juliette] /ʒylˈjɛt/ (Binoche), Jacqueline ˈʒakliːn/ (du Pré), Richard /ˈrɪçɑː(r)t/ (Wagner but not Strauss), Robert /ˈrɒbɛː(r)t/ (Schumann but /ˈrəʊbɛː(r)t/ Koch), Solveig /ˈsuːlveɪ/ (with no mention of the customary /ˈsɒlveɪg/).

There is very frequent use of a term ‘Established Anglicisation’ and a whole panel is devoted to the topic but because such huge numbers of perfectly eligible entries are denied that classification, it would seem better to abandon it altogether as serving no useful purpose. Another superfluous expression occurs when variant forms of personal names are regularly described as used ‘depending on’ or ‘according to’ individual preference (though oddly not at Devereux or Featherstonehaugh with their five choices each). What else would affect people's pronunciations of their own names?

The authors say that they are ‘indebted to the editors of the English pronunciation dictionaries which are always at our elbow as we work’ (p. viii), meaning of course LPD, EPD and ODP. This gives the impression that OBG generally defers to their judgements, which has to be very far from the truth. For example, the editors of these three dictionaries all agree that the usual pronunciation of Maastricht is forestressed yet OBG flatly contradicts their evidence saying that ‘[t]here is no established anglicization, although pronunciations such as “mah-strikt” are sometimes used’ (p. 232). This use of ‘sometimes’ is arguably misinformation. So is the special warning to employ the forestress indicated for the only form of Helsinki they recommend when the predominant version Helˈsinki is denied any acknowledgment at all even though the three major dictionaries all give it first place. This kind of thing occurs also at Allah, an item only listed as /əˈlɑː/, a form which LPD and EPD merely give as a subvariant and ODP seemingly treats as not common enough to merit mention.

A number of recommended pronunciations include the velar fricative /x/, e.g. Bach (J. S.) at which there is no mention of the extremely common form /bɑːk/. As for Buchan, LPD and EPD give /x/ only as a subvariant possibility and ODP excludes it altogether. Similar disregard of these authorities occurs at many other words including, for example, Alsace, Casals, cerise, Chesil, Engels, Faro, Figaro, Gandhi, garage and a host of others from A to Z. The first name of the actress Ingrid Bergman is given only as /ˈɪŋrɪd/ though the three dictionaries all list it only as /ˈɪŋgrɪd/. Her surname is given only as /ˈbɛːgmən/ although the same three works show it only with /ɜː/. And so on.

The OBG stated aims and actual practice seem to be often at odds with each other. The authors say that they wish ‘announcers to use whichever form feels natural in their speech’ (p. xii) yet they remark at Cape Wrath ‘although [it] can also be pronounced [“roth”] . . . we recommend this Scottish pronunciation [/rɑːθ/] for the placename’. No explanation is offered for this strange decision or of why there are no similar recommendations at e.g. Glasgow, Newcastle or Carlisle. The use of ‘can’ above is symptomatic of the OBG actual constant prescriptivism referring to ‘permissible’, ‘correct’ etc pronunciations saying ‘we do not prohibit any acceptable forms’ (p. 42) when the omission of countless perfectly acceptable pronunciations is tantamount to prohibiting them. OBG even says at café latte that the pronunciation ‘lah-tay’ is ‘deprecated because it is felt to be less like the Italian’ (p. 59) but we are not told that /ləˈsanjə/ for lasagne departs equally from the mother tongue: indeed it is recommended. It is not necessarily clear that it is the job of an English pronouncing dictionary to recommend adherence to original language forms of loan words.

Another regrettable manifestation of this inclination to appease those who make arbitrary value judgements is the following comment ‘Our research indicates that pronunciations with’ /z/ of words such as [Muslim], ‘although common, can cause offence among Arabic speakers’ (p. 264). Intelligent Arabic-speakers with a reasonable familiarity with the English language will know that it would be, to say the least, incautious to assume that any slight is intended when an English-speaker conforms to a completely natural pattern observable in numbers of words of all sorts of origins to treat the kind of sequence found in a word like Muslim as in e.g. ASLEF, Breslau, Disley and many other words containing this sequence of sounds.

It is dismaying to read the remark at harass that end-stressing it is ‘disliked by many in the UK’ (p. 165). (Some Scots may indeed be irritated by the inaccuracy of describing its forestressed version as ‘the traditional British English pronunciation’.)

The delightful Scottish lady who headed the BBC Pronunciation Unit for longer than anyone else could not resist including in the Christmas cards she sent to BBC announcers etc every year an appeal to avoid the English person's normal pronunciation of Auld Lang Syne with /z/ beginning its last word. She felt that the Scots-English word lang-syne, quoted in England from the great Scottish poet Robert Burns, being equivalent to south-of-the-Border ‘long-since’ should no more be uttered with that /z/ than that the English should say /lɒŋ zɪns/ for their version.

A general criticism is that OBG is too ready to categorise items as ‘mispronunciations’. Saying ‘The pronunciation bay-zhing is common but not correct’ (p. 42) is not helpful: everyone not born yesterday knows and understands this alternant. In the same vein we find: ‘Traditionally, English-speakers like to stress the first syllable of French words . . . The BBC sometimes gets complaints about this as it is perceived as less “French” . . . when we research French names for our broadcasters we often give last-syllable stress’ (p. 137). This comment is only too well illustrated by the solely end-stressed versions recommended for Abbeville, Balzac, Cocteau, Dijon, Flaubert, and so on through the alphabet. In normal GB, these are customarily stressed on the first syllable. Surely this advice would in any case be misplaced because it might make the broadcaster sound eccentric or affected.

Among other issues that arise are the extensive recommendation of /ʌ/ when the orthographies or transliterations of the Arabic, Dutch, Indic or Australian languages contain the letter a. Perhaps the OBG adjacent entries Tamil with /a/ and Tamil Nadu with /ʌ/ are symptomatic of this problem. We are recommended in OBG to use /ʌ/ in words such as Arundhati (Roy) /əˈruːnˈdʌti/, Jan Van Eyck /ˈjʌn vʌn ˈʌɪk/, Frans Hals /ˈfrʌns ˈhʌls/, Maghreb /ˈmʌgrɪb/, Zia ul-Haq /ˈziːə ʊl ˈhʌk/ and many others. The television Qatar airline advertisements have sounded much more like ‘catarrh’ then ‘cutter’. Similarly it seems difficult to understand why Arabic names beginning with ‘el’ are regularly recommended to have /əl/ when one rarely (if ever) hears el-Alamein spoken other than as /ɛl . . ./. Likewise al-Jazeera is normally pronounced /al/ but never as /əl/ as recommended.

Finally, although it is only a very trivial OBG consideration, one regrets the very perfunctory job that has been made of many of the glosses. These are mostly as promised ‘brief definitions’ but at times it seems that it has been forgotten that they should be so. For example, seraglio is ‘women's apartments in a Muslim palace’ when ‘harem’ would have sufficed. After all, items as different as allegretto, appoggiatura, embouchure and finale are simply identified as ‘musical term’. At Lothario, ‘man who behaves irresponsibly in his sexual relationships’ is ponderous for ‘philanderer’. At synagogue, ‘building where a Jewish congregation meets’ could be simply ‘Jewish temple’. Among others that could have been better expressed are augment (‘change'!), baksheesh, balaclava, caul, Dell'Olio, effete, esoteric, excelsior, hwyl, Jutland, kudos, Kurd, Lamont, logistics, retinue, rhotic, sapphic, sinfonia, sievert, Strine.

It is a sad fact that in producing this extremely interesting book the compilers seem to have been so poorly served by those who have advised them on its design and notably on its proofreading. It was released for publication before it had been adequately scrutinised. A revised edition could with advantage abandon the clumsy respellings. These are actually likely to be less familiar than the corresponding IPA symbols, which are almost only those found in a host of publications these days. The questionable ‘modernisation’ /ʌɪ/ might well be dropped in favour of the usual /aɪ/ and // in favour of //. It would greatly facilitate reference to lists of symbols to have them at the endpapers. It would help in finding forenames if they were all independently entered and cross-referenced with the surname. It would be as well to consider omitting most of the items already covered by the two major pronunciation dictionaries LPD and EPD. Probably most of those who buy OBG will possess at least one of them already. This would make space for more generous coverage to be given to the representations of the original-language versions of loanwords which OBG has done so interestingly but too sparingly in this edition.