Only relatively recently has the term advanced learner's dictionary (hereafter abbreviated to ALD) acquired the special sense of ‘a specific genre especially characterised by its providing phonetic and grammatical information which ordinary dictionaries do not include’. It was first adopted in 1952 by the late Albert Sidney Hornby (1898–1978) and his publishers Oxford University Press (OUP) in order to distinguish what they had previously called simply A Learner's Dictionary of Current English from lower-level dictionaries which Hornby had also edited. This ALD1 had been published in 1942 in Tokyo under the title Idiomatic and Syntactic English Dictionary. It had been compiled in Japan in the years 1937 to 1940 by Hornby, a Chester Grammar School alumnus who, after naval service in the First World War, had graduated in English Literature at University College London (UCL). He produced it in Japan in collaboration with two fellow researchers, Edward Gatenby and Harold Wakefield. The most innovative and distinctive content of the book, as Hornby always freely acknowledged, owed a very great deal to the influence of Harold E. Palmer (1877–1949).

Palmer was one of the most outstanding and highly prolific British writers on phonetics, grammar and language teaching in the first half of the twentieth century. He was a contributor to the m.f. (Le Maître Phonétique, the forerunner of this Journal) starting in 1910 and was for some years from 1915 a Lecturer in Spoken English on the staff of Daniel Jones's UCL Department of Phonetics. Among his numerous phonetics-concerned publications was a unique Grammar of Spoken English (Palmer Reference Palmer1924). His brief book English Intonation (Palmer Reference Palmer1922) was a radical new analysis which, despite its imperfections, sharply re-orientated British studies in that field. A Dictionary of English Pronunciations with American Variants (Reference Palmer, Martin and Blandford1926), a pioneering work compiled with J. Victor Martin and F. G. Blandford (Palmer, Martin & Blandford Reference Palmer, Martin and Blandford1926), gave American pronunciations, where they markedly differed from British usages, in it has to be admitted rather sparsely filled second columns. Palmer left Jones's Department in 1922 to move for thirteen years to Tokyo, where within a year he'd been invited to set up a new Institute for Research in English Teaching. In Japan by great good fortune he met, mentored and then employed ‘Ash’, as he preferred to be called, Hornby. When Palmer returned to England, Hornby succeeded him as the Institute's Director. Its research focused on matters previously neglected with regard to the needs of the advanced non-mother-tongue student of English.

These matters figured to a considerable extent in Palmer's 300-page book A Grammar of English Words (Palmer Reference Palmer1938). Its title-page summarised its contents as ‘One thousand English words and their pronunciations, together with information concerning the several meanings of each word, its inflections and derivatives, and the collocations and phrases into which it enters’. He went on to emphasise the book's ‘richness and abundance of examples’ and to provide explanatory introductions to its main features beginning with ‘Special Grammatical Categories’ and adding specific listing of noun countability, of verb-patterns (providing information ‘given in no other reference-book or textbook’) of adverbial particles and of twenty-four very important anomalous finite verb forms. These last were very common words with particularly complex phonetic and distributional characteristics. Unlike other reference works, his book devoted a great deal of attention to ‘collocations’ (successions of ‘two or more words that may best be learnt as if . . . a single word’) and what he called ‘phrases’, explained as ‘conversational formulas, sayings, proverbs, etc.’.

All the book's entries included transcriptions of their headwords with the set of IPA symbols employed in the Daniel Jones EPD (English Pronouncing Dictionary) of the day (Jones Reference Jones1917, edition of 1937). Palmer claimed at page x that ‘[i]n many cases the pronunciation of the collocation . . . is also given’. This turned out to be more noteworthy as an ideal than for its fulfilment in the book but it was nevertheless a significant pointer to the direction in which ALDs would develop later. Evidently Palmer was not inclined himself to undertake the production of a work of that kind going beyond the thousand words he'd selected for special treatment. A full dictionary with comparable detailed treatment for all its entries was an obviously desirable further step. The necessary intensive work to produce such a dictionary was embarked upon by Hornby and two of his colleagues two years after Palmer had left the Institute. They completed that very first draft in 1940 but, owing to the effects of the Second World War, it was not published even in Japan until April 1942 (ALD1; Hornby et al. Reference Hornby, Gatenby and Wakefield1942) and didn't become available elsewhere until six years later, when it was re-issued ‘reprinted photographically’ by OUP.

This ALD1 contained full phonemic transcriptions of all its headwords along with verb inflections, contracted spellings and weakforms (to use the spelling I consider most appropriate for this term). Hornby chose in it to show the incidence of stressed syllables differently from Jones's IPA style. At primary and secondary stressed syllables he used accents, acute over the simple vowels and initial elements of diphthongs for the former and grave over the latter. Thus better appeared as [bétə] and veneration as [vènəréiʃən]. Alternative pronunciations, including American variant ones, were supplied, though rather sparingly, as were American orthographic forms. After his repatriation to England in 1942 Hornby was engaged by the British Council to teach for a period at the University of Tehran. When the War ended he joined their London headquarters staff as a linguistic adviser. In that capacity he set up, and for its first three years edited, English Language Teaching (later re-named English Language Teaching Journal). In 1950 Hornby resigned from the Council to devote himself exclusively to writing textbooks and courses, etc. Ultimately he undertook single-handedly a complete revision of the earlier text so that in 1963 its second edition appeared, now entitled The Advanced Learner's Dictionary of Current English. This now abandoned the superscript-diacritic stress indications in favour of IPA practice. Its rather few American pronunciations were shown, in a space-saving style, with lower rhoticity than in General American (i.e. omitting the pre-consonantal /r/s most US speakers use) e.g. advertisement was [ˌӕdvəˈtaizmənt] and fertile [ˈfəːtil].

When initial plans began to be laid for ALD3, Hornby was by then into his latter sixties. For a while there were serious doubts about whether his health would permit him to work on it. During this time the present writer was approached by a representative of OUP to ask if he would be interested in editing the planned third edition. I explained that I didn't wish to give up teaching for the relative social isolation of entirely desk-bound work but I mentioned that I should be interested in assisting the next editor by taking responsibility for the whole of the new edition's phonetic content – which I was eager to see augmented. Happily, in the event, Hornby's health so much improved that in time he began preparing ALD3. I welcomed his invitation to assist him while continuing to teach at the University of Leeds.

The ALD3 new phonetic features included the major innovation of the provision, for the first time ever in a major dictionary, of a General American (GA) as well as a General British (GB) pronunciation for every headword. The transcription was entirely in symbols from the International Phonetic Alphabet but the then very unfamiliar symbol [ɒ] was avoided in favour of the eminently recognisable but otherwise unused letter [o]. This was employed to cover GA [ɑ] and GB [ɒ] conveniently coinciding with the words’ ordinary orthographies. This symbol choice didn't fall within IPA general-phonetics recommendations but was described as ‘defensible’ as a ‘convenient diaphonemic’ notation in Wells (Reference Wells1973). Gimson (Reference Gimson1974) said of such use of [o]: ‘This has much to be said for it on the grounds of simplicity and symbol economy’. Its selection may be compared with the way that phonemic notations of GB pronunciation of the past half century (including even those of the current Oxford English Dictionary (OED) and its non-EFL stable-mates) have practically universally rejected the dubiously distinctive symbol [ɐ], which in IPA general-phonetic use is allotted to representing a central vowel about half way between open and half-open. Instead, the symbol [ᴧ] has been preferred for the vowel in that range. In general-phonetic contexts, however, [ᴧ] is assigned to conveying a fully back half-open ‘Cardinal’ value. Similarly, now that mainstream GB /ӕ/ is undeniably nearer to Cardinal vowel [a] than the general-phonetic value of the symbol [ӕ], the symbolisation /ӕ/ is still so far generally preserved other than at the OED/ODP stable and in the forthcoming eighth edition of Alan Cruttenden's Gimson's Pronunciation of English.

By this time the EPD had passed into the hands of A. C. Gimson, Jones's successor in the UCL Chair of Phonetics. He made no secret of his intention that its next edition should involve fundamental changes from the set of transcription symbols that Jones had used. The ALD3 new set of symbols were those employed two years previously in my CPD (Concise Pronouncing Dictionary of British and American English, Windsor Lewis Reference Windsor Lewis1972). Apart from the use of [o] already mentioned, that book anticipated almost exactly the symbols Gimson decided to adopt for the EPD in 1977 with one major difference namely the rejection of length marks. These became redundant in EPD when no longer essential components of the transcription. Length marks in Jones's previous transcription by contrast had been indispensable because they alone served to distinguish one phoneme from another. As Gimson (Reference Gimson1974) explained, he decided to retain such marks in order to preserve continuity and maximize legibility. One trivial ALD3 difference was the use of [ɑ] rather than [a] for the initial elements of the diphthongs which came to appear in EPD as /aɪ/ and /aʊ/. Gimson observed to me that his essential reason for departing in EPD from the differentiated representations of these two diphthongs as /aɪ/ and /ɑʊ/ which had appeared for eighteen years in his Introduction to the Pronunciation of English (Gimson Reference Gimson1962) was that, in a reference book, it improved simplicity and legibility. He felt that the mainstream ranges of the initial elements of the two diphthongs were so similar in GB as to make it hardly necessary to treat them differently. As in my CPD, the style of stress symbolisation adopted for ALD3 was ‘tonetic’ i.e. it used indicators which simultaneously represented the intonation appropriate for the lexical pronunciation of each word, e.g. /ə`gəʊ/ ago, /ˈӕkə`demɪk/ academic. The reasons for this choice were chiefly to accord greater prominence to the stress indicators by avoiding placing the secondary one at the base of the line of print, and having the primary one occupy a slightly wider space horizontally. This provided rather better legibility than the alternative IPA marks. Gimson (Reference Gimson1974) also acknowledged that this had ‘advantages, since it associates primary accent with pitch movement. . .’. He expressed the opinion that ALD3 exhibited ‘an eminently legible and unambiguous notation’ (personal communication, Reference Gimson1974).

Before the era of ALDs, dictionary phonetic transcriptions consisted almost solely of phonemic symbols accompanied by stress marks. EPD dealt with items like refractive and motherless by transcribing their suffixes alone within their headwords’ paragraphs. Occasionally only, in order to avoid duplicating segmental information, variant stress patterns were conveyed by replacing syllables with hyphens preceded where necessary by stress marks. No entries consisted solely of ordinary spellings accompanied with stress markings until the LPD (Longman Pronunciation Dictionary, Wells Reference Wells1990) first edition introduced such a practice. This extended coverage Wells applied to ‘a good selection of compounds and phrases’. These items were not given at their alphabetical positions but arranged in small miscellaneous groups after selected headwords. They contained few phrases and only a fraction of the coverage of combinations provided by ALD-type dictionaries but they supplied about 10,000 extra items. When Peter Roach and his colleagues took over the EPD in 1997, they too introduced such non-alphabetised groupings (Roach, Setter & Esling Reference Roach, Setter and Esling2011).

The increased phonetic content introduced into ALD3, aside from the American transcriptions, chiefly consisted not of segmental symbols but for the most part of stress indications for the extremely large numbers of items for which ALD2 had provided no stressing coverage at all. Information of this kind and extent was available absolutely nowhere else even in the largest of dictionaries. Much of it was incorporated in paragraphs like that headed bank, at which ALD2 had used the ‘swung dash’ to stand for headword repetition, giving e.g. ⁓clerk and ⁓holiday. ALD3 now regularly showed words’ stress patterns, by appropriate placing of the mark for principal stress, instead of doing so only occasionally. So these were now `⁓ clerk and ⁓ `holiday. The usefulness of this feature may be appreciated by comparing such numerous pairs of items as child `benefit and `child's play, `Christmas card and Christmas `day, `rice paper and silver `paper, `flying boat and flying `fish, `hay fever and yellow `fever, market `garden and `market town, `polling day and working `day, `walking stick and parting `shot.

Compound words and fixed phrases are all stress-unpredictable because of the various conflicting forces to which they're subject. An account of these influences is the subject of the article ‘Accentuation’ Windsor Lewis (Reference Windsor Lewis2012). Very large numbers of these word combinations and phrases have, however, particular stress structures each of which is customary in at least part of the English-speaking world. Many have changed these stressings over time. Henry Sweet, in his New English Grammar (Sweet Reference Sweet1891: §904), coming to ‘consider the use of even stress in noun-compounds’ took, as what he considered an ‘especially clear’ example, the stressing of sponge-cake. This he said had tonic stress on its latter element. There's no reason to think that this was anything but a perfectly accurate comment in his day. It was clear by the latter twentieth century, however, that its usual tonic stressing had pretty universally moved to its first element. Even the OED has never yet included any indication of its rhythmic character at its entry. Such changes have occurred to many common expressions such as `banknote, `countryside, `deckchair, etc. Most of these expressions eventually come to be written solid but they may also receive only hyphenation or remain frequently, or even regularly, represented as two separate words. Such are `fountain pen, `drinking water, `passion fruit, `pocket knife, `rose garden, `soda water, `water feature, etc.

As to English fixed phrases, little satisfaction is likely to be derived from attempts to find rules to predict their stress patterns. EFL students may well be inclined to expect that frequently-contrast-highlighting words like other, your or own, etc. are likely to receive tonic stressing. In fact, the tonic may or may not occur where they might expect it, e.g. on `other or on `time in some other `time which is much less often some `other time. It's not usually on other in the other way `round and never so in I saw him the other `day. Similarly the tonic may or may not be on your in it's none of your `business but in mind your own `business or stew in your own `juice the tonic stress, from no easily perceivable logical necessity, only ever comes on the final word. Large numbers of common expressions like the wrong end of the `stick, the boot's on the other `foot, the thin end of the `wedge, burning the candle at both `ends, turn over a new `leaf, and eyes in the back of one's `head also contain contrast-suggesting words which might well be expected to receive tonic stress but don't do so. Nor can it be presumed that tonic stress should usually fall on the last content word in a phrase. It doesn't in such common expressions as and `so on, an `awful thing to say, a `bee in his bonnet, ˈmakes your `blood boil, ˈneeds his `head examined, ˈget on like a `house on fire, ˈsee how the `land lies, all `over the place or a ˈsilver `spoon in her mouth. Yet again, the usual prohibition from re-accenting a repeated word which has just previously been highlighted doesn't apply for example to an ˈeye for an `eye, the ˈblind leading the `blind, ˈbusiness is `business, ˈboys ˈwill be `boys, or ˈlet ˈbygones be `bygones, etc.

In 1977 Gimson completed the 14th edition of the EPD implementing his long-pondered final decisions on a revised set of transcription symbols. His new choices were very widely adopted in a remarkably short time so nearly universally that a completely unprecedented harmony in choice of transcription became established, at least in the British EFL world that has remained essentially undisturbed for a generation or more. Oxford University Press of course decided that the future ALD4 should be brought in line with the new consensus. With both GB and GA transcriptions supplied for its 100,000 or so headwords and many thousands more items dealt with within paragraphs, it seemed that a good beginning had been made on the augmentation of the ALD3 phonetics content. A good deal more seemed desirable but would need a serious amount of time along with the conversion to the new transcription. It was something of a relief to me to hear in due course that Gimson himself expressed willingness to undertake for ALD4 the conversion to the new style in co-operation with his UCL colleague Susan Ramsaran. Some years before their preparations were complete, their collaboration was cut short by Gimson's untimely death in 1985. However, by the time of the appearance in 1989 of ALD4, progress in expanding phonetic information could be seen to have been excellently maintained. Since Dr Ramsaran's resignation from that work, it has been carried on to the present day with the same effectiveness by her UCL colleague Michael Ashby. To this day there still exist yet more opportunities for maximising of prosodic information in expanded new editions, particularly by further extending the prosodic annotation of its illustrative sentences and supplying more alternative pronunciations for individual words – a development all the more feasible when access to the next ALD is ever more likely to be chiefly electronic rather than via printed texts. This applies also to pronouncing dictionaries. Both of the two chief ones are far too tightly packed for anyone's comfort with space-saving but legibility-inimical presentation of their data. The third in the field may have disadvantages including greater bulk and less comprehensive treatment than the others, but the Oxford Dictionary of Pronunciation for Current English (Upton, Kretzschmar & Konopka Reference Upton, Kretzschmar and Konopka2001) is ahead of them as regards legibility. Another future step might be to provide intonation information at suitable items. Leech & Svartvik (Reference Leech and Svartvik1975) had considerable success in doing so in their Communicative Grammar. Examples that spring to mind might be:

• If you ˈask `ˏme. . . | `Never say ˏdie. | ˈYoo ˈhoo. | `There, ˎthere! |

  `You `ˏ said it. | You can ˈsay `that again | etc.

The first of various emulations of the ALD model, the long-gestated Longman Dictionary of Contemporary English, appeared in 1978 with Gordon Walsh as Pronunciation Editor. He introduced a pair of minor adaptations of the new Gimson transcription which proved to be the only noteworthy modifications it has ever undergone. The first of them consisted of using simple /i/, free of length-mark, chiefly to convey the weak rhythmic value normally heard at the final syllable of words like happy and (later) at medial ones in words like radiate. Subsequently a parallel simple /u/ was adopted mainly for medial syllables of words like punctuate.

Latterly the expression ‘Advanced Learner's’, not adopted by the earlier emulations, has come to be treated as generic and begun to be incorporated in the titles of works from other publishers including Cambridge University Press, who in 2003 re-titled their Cambridge International Dictionary of English (1995) as the Cambridge Advanced Learner's Dictionary, Macmillan (Macmillan English Dictionary for Advanced Learners) and even Merriam-Webster (Merriam-Webster Advanced Learner's English Dictionary). During the final decade or so of the last century the arrival of digital compact discs and online facilities provided the possibility of audible spoken illustrations of pronunciations as never before. These are without the slightest doubt an invaluable development but their production has not been without problems. The limited first of these recordings appeared in 1993 as a disc deriving from the first ALD emulation, the LDCE (Longman Dictionary of Contemporary English). Some years later almost all ALDs came to be accompanied by extensive sets of recordings as have the two principal pronunciation dictionaries LPD and CEPD. The British and American English native speakers recruited for recording them, often university linguistics students, have not been the subjects of any biographical data. The phonetic sophistication and powers of concentration desirable to enable them to perform precisely the phonetic transcriptions which they read has certainly not been invariably manifested.

Among various less than completely satisfactory efforts have been items which have suffered from unnatural articulations. For example a performer, though certainly a native speaker of the accent being exemplified, may enunciate e.g. a final consonant so effortfully that the result is foreign-sounding or ambiguous. In other cases we hear inappropriately released plosives e.g. of the final /k/ of book club or when blind mice tends to sound like blinder mice. In a similar case the final /n/ in stone was uttered with such excessive force that a vocalic offglide was produced such as to make the performance suggestive of e.g. a native Italian speaker's imperfect attempt at the word. In another case a speaker says `blind side with a brisker than desirable uttering of the second word making it indistinguishable from the word sight. In one case one hears a clearly genuinely American speaker but one who has the regular low rhoticity that is found in GB speakers in contrast with the GA ‘r keeping’ that the transcription indicates. Intermittent rhoticity discrepancies between transcription and performance have occurred when American speakers, though having the usual GA-type high rhoticity, make various doubtlessly unconscious /r/ elisions. A lot of these are perfectly common among GA speakers medially in e.g. caterpillar, comfortable, farmer, government, particular, etc. and in the first syllables of proprietary, surprise, priority, etc. A lesser type of problem occurs when, instead of the transcription's weak vowel, the speaker may substitute a strong variant as for procrastinate with a diphthong instead of a schwa in its first syllable. In some cases a problem isn't entirely the fault of the speaker when they use their normal form in contrast with the less suitable dictionary indication. An American in one case says `castor oil in contrast with the text stressing. In another the American follows the text marking using the stress value inappropriately prescribed by saying olive `oil.

Some of the speakers have employed prosodies inappropriate for lexical items. These require a high-to-low-fall on a sole stressed syllable with any pre-tonic stressed syllables to be down-stepping upper-level tones. Instead of such normal lexical values some speakers at times reveal their consciousness of the repetitive nature of their task by employing pre-final intonations. Thus from one of them instead of `sponge cake one hears two level steps ˈsponge ˈcake. This ‘Drop’ tone can sound unsuitably casual and/or tend to make the position of the tonic less clear. One speaker saying a sequence of sponge compounds departed from the requisite lexical pre-final upper-level stress on the first word of sponge pudding by according it the low level pitch and reduced loudness of an unstressed syllable. This speaker at pudding was another to use the unsuitably non-final Drop tone. Some readers produce final low rises which even though quite narrow nevertheless inappropriately suggest continuity. Many of these tonetic irregularities are fairly trivial imperfections and they can't be said to occur wholesale but they must still count as having disadvantages as models for imitation. The foregoing comments should not be taken as anything more than a selection of random samples of the sorts of things to be heard.

Sound files, immensely useful though they are, will not at all make phonetic transcriptions superfluous. As students who begin using transcriptions must in many cases soon realise, they can be useful aids to recognition of what's been spoken and are also valuable visual supplements to the aural memory. By the way, it's as well for users of the free online versions to be aware that their benefactors are inclined to omit significant amounts of the contents that appear in the print versions. For example OALD (Oxford Advanced Learner's Dictionary) and CALD (Cambridge Advanced Learner's Dictionary) free online, currently at least, omit the valuable stress markings of very many expressions. ALDs will no doubt continue to develop but in the physical form of books their days seem likely to be numbered.