This article was published in
Speech Hearing and Language: UCL Work in Progress, volume 8, 1994, pages 259-267

Transcribing Estuary English:

a discussion document

John Wells, University College London

  1. Introduction

    Many of our native-speaker undergraduates use a variety of English that I suppose we have to call Estuary English, following Rosewarne 1984, 1994, Coggle 1993, and many recent reports on press and television. So do many of the patients our BSc students will have to deal with.That is, they use the popular speech of the southeast of England (based on that of London, and thus supposedly centred on the Thames estuary). This means that their accent is located somewhere in the continuum between RP and broad Cockney (= the broadest London working-class variety).

    I would really prefer to call this variety simply London English, although obviously its ambit is much wider than the GLC area, covering at least most of the urban south-east. Other names we could refer to it by might include General London (GL), McArthur's New London Voice, and Tebbitt-Livingstone-speak. (Note that Rosewarne seems to use the term 'London speech' to refer to what I call Cockney, since he refers to 'a continuum with RP and London speech at either end', with his Estuary English speakers 'grouped in the middle ground'.) Nevertheless, it has to be acknowledged that the term 'Estuary English' has already achieved some degree of public recognition. As with the equally unsatisfactory term 'Received Pronunciation', we are forced to go along.

  2. Phonetics of Estuary English

    Estuary English (EE) is like RP, but unlike Cockney, in being associated with standard grammar and usage; it is like Cockney, but unlike RP (as traditionally described), in being characterized by tendencies towards, for example,

    • vocalization of preconsonantal/final /l/, perhaps with various vowel mergers before it (miwk-bottoo 'milk-bottle')
    • striking allophony in GOAT (> [QU] before dark /l/ or its reflex), leading perhaps to a phonemic split (wholly holy)
    • use of [?] for traditional [t] in many non-initial positions (take i' off)
    • diphthong shift, particularly of the FACE, PRICE and GOAT vowels (wotshor nime?)
    • yod coalescence even before a stressed /u:/ (Chooseday)
    Phonetically EE differs from Cockney in usually not being characterized by, for example,
    • h-dropping ('and on 'eart)
    • TH fronting (I fink)
    • monophthongal realization of the MOUTH vowel (Sahfend).

    It is not entirely clear whether EE is to be regarded as a variety (lect, dialect) in its own right, or whether it is simply the formal style/register for which Cockney is the informal one. A decision depends on two empirical issues:

    • Is there a casual style of EE that is unquestionably distinct from Cockney? Tentatively, yes: there may well be speakers who avoid stigmatized h-dropping even in their most casual style (as RP speakers do; NB we are not dealing here with /h/ in unstressed pronouns.).
    • Is there a formal style of Cockney that is distinct from EE? Tentatively, yes: Cockney is arguably the speech of the uneducated, who are unable to achieve standard grammar even where it might be called for; while EE speakers are those who can consistently use standard grammar with ease and fluency.

    The boundary between EE and RP is also hard to establish. Presumably it rests on the degree of localizability: EE is localizable as belonging to the southeast of England (see Coggle's impressionistic sketch-map, 1993:28), whereas RP is not. (Many of Rosewarne's comments surely relate to style or to change over time, rather than to the decline of RP, to localizability or to the Thames estuary area. Things like cheers for thank you/goodbye are surely part of contemporary casual RP/StdEng -- at least I use them, and no-one has ever suggested that I am a speaker of EE! Some commentators seem not to appreciate that RP can be spoken in informal situations.)

  3. Principles of transcribing EE

    We need a standardized phonetic transcription for the EE accent. Although Rosewarne and Coggle have arguably done a public service by drawing attention to it, neither they nor anyone else (as far as I am aware) has attempted a serious phonetic description of it; nor indeed have they ever transcribed more than the odd sound in isolation. Once we have standardized and codified EE, we could teach it to students, ask them to do transcription exercises using it, and give phonetic dictation in it. Quite apart from its importance for EE's own native speakers and their imitators, it would furthermore attract great interest and attention among EFL teachers disenchanted with RP. Some might wish to teach it to their EFL pupils. ('Because it obscures sociolinguistic origins, "Estuary English" is attractive to many' - Rosewarne 1984.) The present document aims to highlight some of the decisions we would need to make in arriving at a standardized transcription.

    One basic question we must face is: how comparative should the EE transcription system be? Should we use phonetically explicit symbols, often different from those we use for RP (with the resultant risk of confusing the beginner)? Or is it more sensible to use the same symbols as for RP wherever possible, but with different conventions of interpretation? E.g.,

    • Should we write EE face as fVIs or as feIs?
    • Should we write EE price as prAIs or as praIs?
    • Should we write EE loaf as lVUf or as l@Uf?
    • Should we write EE goose as g}:s or as gu:s?

    My view is that for the time being, at any rate, we should take a minimalist position. We should aim to make the notation as similar as possible to that used for RP. So we would continue to transcribe the above words in the usual way. Just as the symbol /{/ in cat can already in RP cover a range of possibilities from around cardinal [E] to cardinal [a], so /eI/ in face can be permitted to cover a range of possible first elements for the diphthong, from conservative RP [e] to qualities in the area [E ~ 6 ~ V ~ {].

    Exceptions to this principle might perhaps be made, though, for

    • the diphthongs of price and mouth (RP /praIs, maUT/), for which it seems most intuitive to write EE /prAIs, m{UT/ (for the latter, compare also Cockney [m{:f]);
    • explicit notation of the glottal stop as a realization of /t/.

  4. Competing solutions

    This means that the most important remaining decision is how to symbolize the product of l vocalization. For RP /l/ in non-prevocalic positions, EE uses a vocoid in the area [7, o, U, u]. This may be non-syllabic, typically forming the second element of a new diphthong in milk, shell etc.; or it may be syllabic, as in middle. Let us consider a selection of possible solutions, looking at the pros and cons of each.

    • first solution: write U, thus mIUk, SeU, "mIdU
      • pro: familiar symbol;
      • con: implies phonetic (and for the syllabic vocoid, phonemic) identity with the vowel of put, which is wrong. Articulatorily, the l-vocalization product is considerably further back than the /U/ of put, which can nowadays be very central.
    • second solution: write w, thus mIwk, Sew, ("mIdw)
      • pro: reasonably intuitive for the non-syllabic vocoid; but we should need something else for the syllabic;
      • con: implies phonemic identification with prevocalic /w/, which is questionable; violates the general English phonotactic constraint against final semivowels. (To overcome the objection to syllabic w, John Maidment suggests using an omega, w. But I find this unacceptable on general transcriptional and typographic grounds.)
    • third solution: write o, thus mIok, Seo, "mIdo
      • pro: phonetically reasonable, visually distinctive. On balance my preferred solution;
      • con: in various other accents (Scottish, GenAm) this symbol is associated with the vowel of goat.

    I therefore suggest that we adopt o as the symbol for the vocalized-l vocoid, whether as the second element of a diphthong or as an independent weak vowel. Phonologically, a case can be made for regarding [o] as still a member of the /l/ phoneme; but clearly in this regard our standard notation must allowed to be allophonic. (In view of the phonological indeterminacy here, though, I think we should permit people to write "/o/", and for that matter "/?/".)

  5. Vowel problems

    The mimimalist assumption is that all vowel contrasts are preserved before /o/. So, unless there are reasons not to, we could expect to be able to write: field fi:od, build bIod, wealth weoT, canal k@"n{o, Charles tSA:oz, doll dQo, call *kO:o, pull pUo, school *sku:o, bulk bVok, world w3:od, sandals "s{ndoz, fail feIo, roll *r@Uo, mile mAIo, howl h{Uo, oil OIo, real *rI@o. (There seem to be no items with RP nonprevocalic /e@l, U@l/.) This assumption appears to be incorrect, though, with at least those items that are asterisked in the list: we must now explore them.

    Words like roll. As we know, EE speakers (and many others) intuitively reject the identification of the vowel of goal, told with the phoneme /@U/. For some time we have been offering them the symbol QU. It is clear to them, furthermore, that goalie does not rhyme with slowly, nor roller with polar. By our minimalist principle, we keep @U for go g@U, slowly "sl@Uli and polar "p@Ul@. We can write QU in goalie "gQUli and roller "rQUl@. The question then arises with goal and told, do we write gQUo and tQUod (preserving the reflex of the lateral as o), or simply gQU and tQUd ? (I assume that -- in spite of what may happen in Cockney -- gQo is not correct for goal, since it does not rhyme with doll dQo.) In general, does the former lateral get fully absorbed after back vowels? Or does it leave o behind? The conservative inference is that it does leave some phonetic residue, or that we must at least allow for this possibility. Hence I recommend goal gQUo (although I would change this recommendation if gQU proves adequate).

    Words like cool. The same question arises with cool, and perhaps with pull and call. The vowel of EE cool [kou, ku:] is phonetically very different from that of coo [k}:, k1}]. In cool the quality is back, in coo central; both vowels may be somewhat diphthongal. If we continue to write /u:/ in coo /ku:/ (rather than, say, the phonetically more explicit [}:]), we need something else for cool.

    Some speakers may be happy with the notation ku:o, which implies the phonemic identification of this vowel with that of coo. Others will not be, but it may be the case that none of them preserve the distinction between fool and full, so that they would accept the notation kUo. Actually, for some EE speakers kO:o might be appropriate, namely those who have a cockney-style three-way homophony pool-pull-Paul.

    Note, though, that EE pooling is distinct from pulling. So cooling would presumably have to be "ku:lIN. Does this give rise to a problem in the light of words such as truly "tru:li and clueless "klu:l@s ? (Compare the goalie-slowly contrast discussed above) If it does, we shall have to think of something else.

    Words like call. Then there is the question of words with RP /O:l/. Is the product of l vocalization absorbed into a preceding /O:/? If it is absorbed, board and bald come out homophonous, as do pause and Paul's; if it is not fully absorbed, they remain distinct. We should be prepared to write bald as bO:d or bO:od respectively.

    There is, however, another issue. Although Paul's and pause may be homophonous, Paul and paw are certainly not. Nor are board and bored, in spite of their homophony in RP. For which of these vowels should we reserve the familiar notation /O:/?

    • If we write /O:/ for paw-pour-pore /pO:/ and bored /bO:d/, we need some other symbol for pause and Paul's. A possibility would be /o:/ in pause, thought, board; Paul's would have the same notation if homophonous, otherwise /Oo/.
    • If on the other hand we reserve the notation /O:/ for pause, thought, board, we could recognize the tendency towards a centring-diphthong quality in paw-pour-pore by writing /O@/, thus paws /pO@z/, bored /bO@d/. The notation /Oo/ would still be available for Paul's if distinct from pause.

    Thus in any case the best notation for the vowel of EE call may be /Oo/. Therefore some EE speakers can write cool, too, as /kOo/. Others will have call /kOo/ but cool /kUo/.

    Words like real. EE real is a homophone of reel and should accordingly be written ri:o. (In Cockney real and reel are homophones of rill and can all be written rIo.) Presumably really can acceptably be written "ri:li, implying a rhyme with freely -- or should it perhaps be "ri:oli ?

    There may other vowel neutralizations in the context of a following dark /l/ or its reflex. They are probably confined to Cockney and therefore irrelevant for the EE we are considering here. Examples of possible 'new' homophones include feel-fill, well-whirl, child's-Charles, veil-Val-vowel, doll-dole, pool-pull-Paul. However EE regularly makes rhymes of oil-royal ("OIo, "rOIo), owl-vowel ("{Uo, "v{Uo(?)).

  6. The glottal stop

    To revert to the question of the glottal stop: although it is surely no more than an allophone of /t/, I think we could well consider writing glottal stop for /t/ in certain positions. I would suggest that we prescribe "/?/" for traditional /t/ in the following environment:

    when BOTH preceded by a vowel (including /o/ from /l/), or /n/;
    AND followed by the end of a word or by a consonant other than /r/.

    t > ? / [+son] _ {C, ##} (ordered later than the rule making /tr/ an affricate!)

    Examples: bit bI? football "fU?bOo
    belt beo? Cheltenham "tSeo?n@m
    bent ben? Bentley "ben?li

    This applies particularly where the syllable in question is unstressed:

    Examples:stop it "stQp I? bullet-proof "bUlI?pru:f

    With a preceding obstruent, however, [?] is not possible: compare best best (never *bes?). Glottal replacement does not happen in EE mattress "m{tr@s (cf. possible Cockney "m{?r@s). And glottalling word-internally before a vowel is well-known as a 'rough' pronunciation variant: thus EE water "wO:t@, but Cockney "wOU?@.

  7. Other points

    We have already decided to adopt the symbols /i, u/ for RP in happy, influence. They can be applied to EE, too, where they will tend definitely to imply tense vowels more similar to /i:, u:/ than to /I, U/. The relationship between EE /O:/ and /o/ (normal /"nO:mo/) is comparable to that between /i:/ and /i/ (weedy /"wi:di/), namely that the first in each pair belongs to the strong vowel system, the second to the weak.

    If these ideas are accepted we end up with a standardized systematic transcription system for EE. (It is 'systematic' rather than 'phonemic', because of the use of /o, ?, i, u/.) Once we have agreed it, I intend to prepare some classroom materials using it. Before that, though, -- colleagues (particularly native speakers of EE), do let me know your views.


Placed on the Web 1999 Jan 15
Back to the Estuary English home page
J.C. Wells home page
UCL P&L home page