The brogue that isn't

J.C.Wells, University College London

[This article was originally published in the Journal of the International Phonetic Association, vol. 10 (1980): 74-79. Versions of it were presented

With hindsight, there are various things in this article that I would prefer to have presented slightly differently. However, for the present electronic version (1996) I have restricted myself to the correction of typographical errors, the insertion of cross-headings, and the addition of a very small number of extra notes, shown, like this one, in square brackets.]


We have all heard of the remote community in the Appalachians (or is it somewhere in the North of England?) where the locals still supposedly speak pure Elizabethan English, unchanged for centuries. But this is not the only dialectological myth which persists tenaciously in the popular imagination and will no doubt continue to do so despite its lack of factual basis. [See now also Pullum, 1991]

'In a remote corner of the Leeward Isles', we read in a 1979 issue of the British Radio Times, 'there lurks a tiny British dependency which not only boasts the normal West Indian complement of blacks (around 12,000 currently on the strength) but where the ethnics are called by Irish names, where local hideaways include Cork, Kinsale and Sweeney's Well, and there is a shamrock carved over the door of the Governor's island replete with such typically West Indian names as O'Garra, Riley and O'Connel, which prints stamps with the Irish harp on and whose funky inhabitants speak a charming Irish brogue, the like of which would do no shame to the hills and dales of Kerry.'

This is an account of an enquiry into the accuracy of such a claim. Does the local speech of Montserrat have brogue-like characteristics? Is the 'Emerald Isle of the Caribbean' dialectologically Anglo-Irish? It is known that Montserrat was first settled not by the English but by a group of disaffected Irish Catholics from the nearby St. Kitts in 1633 (Le Page, 1957-8: 55; Fergus, 1975: 7-12; [Fergus, 1994: 16-35]). What I have attempted to do is to investigate the linguistic characteristics of Montserrat Creole, as I have observed it at the present day, and to look for possible Irish influence. My observations are based partly on a five-week visit to Montserrat in 1977-8 and partly on extensive acquaintance with Montserratians in the United Kingdom (where, incdientally, there are proportionately more West Indians from Montserrat than from any other territory (Philpott, 1977: 90)).

Montserrat Creole

Montserrat Creole has a great deal in common with other Caribbean English-based creoles. As in the other British or ex-British territories such as Jamaica, Guyana, Trinidad, Barbados and Antigua, there exists a continuum extending from the broadest Creole up to a local variety of Standard English. In searching for Irishisms our attention will naturally focus particularly on the broad Creole end of this continuum.

[Current browsers cannot display IPA phonetic symbols except as graphics. Accordingly, non-ANSI phonetic symbols in the text have been replaced by the corresponding SAMPA (or X-SAMPA) characters. The graphics show the corresponding IPA characters.]

In phonology, it is true that there are several important characteristics in which Montserrat Creole (MC) agrees with popular Southern Irish English (IE). An example is the use of plosives, [t] and [d], where Standard accents of English have dental fricatives /T, D/, as in thin, bath, then, breathe. It would, however, be unreasonable to attribute this MC characteristic to Irish influence, since -- as is well known -- it is shared by all Caribbean English creoles. Furthermore, most Irishmen use a dental plosive in the th words as opposed to an alveolar plosive in words where standard accents have alveolar plosives, and thus retain contrasts such as thick [t_d_hIk] vs. tick [t_hIk], breathe [bri:d_d] vs. breed [bri:d], characteristically lost in broad West Indian accents.

Similar considerations apply to the agreement between MC and IE in the use of unrounded vowels corresponding to RP /Q/ and /O:/, in words such as lot, stop, thought, talk (MC /lat, stap, taat, taak/), if we use the notation of the Dictionary of Jamaican English). Here again we note that precisely the same usage is found in Jamaican Creole (JC) and in Guyanese Creole, not to mention many varieties of American English; and once more it is crucial that the Irish retain the oppositions exemplified by tap vs. top, farm vs. form, which MC has lost in popular speech (IE /tap, tAp, fa:rm, fA:rm/; MC /tap, tap, faam, faam/).

The merger of the PRICE and CHOICE diphthongs, RP /aI/, /OI/, leading to the homophony in popular pronunciation of vice and voice, is again common not only to MC and IE but to many other local varieties of English, both in the Caribbean and elsewhere.

At this point I should mention that the MC vowel system is identical, qua system, with that of JC. It iis as follows:

i	u		ii	uu
e	o		ie	uo
    a			    aa		ai       ou

bit	put		beat	boot
bet	cut		late	boat
  pat,pot		  bath,talk	bite	bout 

In the open vowels there is sociolinguistic variability in Montserrat just as in Jamaica: pot may non-basilectally be distinct from pat by having [Q] as against [a], and similarly talk with [Q:] or choice with [QI]. Also as in Jamaica, non-basilectal /ie/ and /uo/ are usually monophthongal, [e:] and [o:].

The IE vowel system agrees as far as the monophthongs are concerned with Barbadian or with non-basilectal MC/JC: it has six short vowels including pat=/pot and six long vowels including last=/lost and monophthongal late, boat.

There are many respects in which MC agrees with JC while both differ from IE. An example is the typical West Indian /w/ allophone [H], a labial-palatal approximant, used in the environment of a following non-open front vowel, as wheel, wet (Wells, 1973: 11).

It is clear, then, that if we are to search for traces of Irish influence in MC we must concentrate on those characteristics in which MC differs from other Caribbean English creoles. I shall restrict the comparison to Jamaican Creole, since it is the variety which is best documented and with which I am most familiar.

MC and JC

Montserratians do not drop historical and standard /h/ in the way characteristic of some kinds of JC (Wells, 1973: 93). As in RP and IE, and differently from the popular speech of most of England, phonemic /h/ consistently remains in MC: hand is consistently distinguished from and, /han/ vs. /an/, etc. But in this respect MC is merely in agreement with the majority of accents of English, including American English and Scottish English; there is no justification for invoking Irish influence. Indeed, it appears that at the time of the settlement of Montserrat in the seventeenth century not even Englishmen dropped /h/; in England, and presumably also in Jamaica, it reflects a subsequent innovation.

One of the phonological characteristics in which MC differs from JC is in the use of /u/ as a reduction vowel (a vowel in weak syllables). All West Indian creoles tend to avoid [@], and JC has /a/ in most environments where RP has /@/, as in attack, letter, breakfast (JC /aták, léta, brékfas ~ brékfaas/; RP /@"tæk, "let@, "brekf@st/). MC is unusual in that under certain circumstances -- and I have been unable to discover precisely what those circumstances are -- the vowel /u/ is used instead of, or as an alternative to, /a/. Examples include /krísmus/ Christmas, /brókfus/ breakfast, /fu pórpus/ on purpose, /áatu/ after, /mushíin/ machine, /kòpukáafi/ cup of coffee. But this must be an independent development in Montserrat; it is certainly not Irish.

The same must be true, I imagine, for another very striking phonological characteristic of Montserratian English. This is the shortening of historically long vowels in word-final position. Thus tea is /ti/ (and similarly free, see, key, etc.); play is /ple/ (and similarly stay, pay, day, etc.); Ma is /ma/, a homophone of basilectal maw (and a rhyme for straw, jaw, law, etc.); show is /sho/ (and similarly blow, know, grow, etc.); two is /tu/ (and similarly true--through, blue, new, etc.). Historically, therefore, MC must have added the rule

(1)	V -> [-long] / _ #		FINAL SHORTENING
to its phonology. The consequence is that, in addition to the forms exemplified above, lay is now a homophone of let (which has lost its final /t/), while lawless /lális/, a frequently used word, rhymes with palace /pális/.

As will be demonstrated in a moment, rule (1) is not a true generalization about surface phonetics. It does, however, apply to all lexical items which met its structural description before the loss of final /r/, and in this respect differs sharply from the comparable development of JC, which affected only a small number of words.

While (1) constitutes an unusual development, indeed an idiosyncratic one, a rule vocalizing or deleting non-prevocalic /r/ is very widespread in English. Montserratian, like RP (but unlike Irish English) is a non-rhotic accent; that is to say, /r/ has been eliminated as such in the envioronment of a following consonant or major boundary. As far as basilectal MC is concerned, the historical rule is simply (2):

(2)	r -> 0 / _ {C, #}		R DELETION

Thus part is /paat/ and north /naat/; both rhyme with bath /baat/ and cloth /klaat/. As elsewhere in the West Indies, (but not in Ireland), historical /i:/ and /e:/ have merged in the context of a following /r/, giving MC /ie/, while /u:/ and /o:/ in the same environment have merged to give MC /uo/. Thus beard, having lost its /r/, is /bied/, homophonous with bathe; court /kuot/ is homophonous with coat (though non-basilectally such pairs remain distinct). These developments are identical with JC. But whereas Jamaican has not extended the rule to final position in the word, Montserratian has. Thus star is MC /staa/ (and similarly bar, jar, etc.); basilectal war rhymes with it as /waa/. Sure and shore are /shuo/ (and similarly poor, four, etc.). In the case of the front vowel, the phonetic output is [i_o6], which would naturally be written /ia/ in the Cassidy-Le Page notation; but it can be considered as a co-allophone of /ie/, since the former ocurs only finally, the latter only non-finally. Thus steer and stare are /stia/ (/stie/), and similarly fear--fair--fare, near, etc.

Thus /jaa/ is the pronunciation of jar in MC (but of jaw in JC); /snuo/ is the pronunciation of snore in MC (but of snow in JC); /stia/ is the pronunciation of steer--stare in MC (cf. /stie/ = stay in JC). While jaw is /jaa/ in JC, in MC it is /ja/; while snow is /snuo/ in JC, in MC it is /sno/; and while stay is /stie/ in JC, in MC it is /ste/. (I use Cassidy & Le Page's symbol /j/ to denote /dZ/.)

It can be seen that logically rule (2) must represent a later trend of development than rule (1). If (2) had already been operative, it would have fed (1) with items such as jar, snore, steer, causing them to become */ja, sno, ste/. Whether (2) is an independent Montserratian innovation, or represents the adoption of a newly fashionable phonological rule from England, is not clear; but in either case it must have been added later than (1). It also makes (1) no longer a true generalization about surface phonetics, since it is violated by forms such as star /staa/.

Items such as stir, first, burn call for a special note. Basilectally they follow the rules given: /sto, fos, bon/. Non-basilectally they fluctuate between an r-coloured mid back or central rounded long vowel and the non-r-coloured equivalent, thus [stO_"`: ~ st3`_w: ~ st3_w:], etc. It is not clear how to phonemicize this, or how to write it in the Cassidy-Le Page notation; if we write it as /or/, then the environment in question must be made an exception to rule (2).

Final Shortening and final R Deletion are the most striking phonological differences between MC and JC. Neither is found in IE.


There are many typical characteristics of IE which are not to be found in MC. An example is the retention of the historical contrast of short non-open vowels before non-prevocalic /r/, as earn vs. urn, serve vs. curve (IE /Er/ vs. /O_"r/). These are merged in MC just as in RP and American English. Another example is the IE use of long vowels in cook, book, hook (IE /-u:k/); MC, like most varieties of English, uses short /u/.

I am not in a position to present an adequate analysis of the intonation patterns either of MC or of IE; but in terms of rhythm and intonation they do not sound similar.

In fact the only phonological characteristic of MC I have been able to find which could reasonably be considered an Irishism is the existence of /ou/ variants for the words old and cold. In most accents of English these words have a vowel phonemically identified with that of goat; so the expected MC forms are /uol/ and /kuol/ (cf. /guot/ goat). These certainly occur; but they are in competition with rival forms /oul/ and /koul/, which have the same vowel as mouth (MC /mout/). Forms with this vowel are well known as hibernicisms; old is sometimes spelt ould or oul' in comic or literary representations of Irish dialogue.


When we turn from phonology to morphology and syntax we find absolutely nothing (as far as I can see) that can be attributed to Irish influence. Constructions such as those in (3) are typical of Creole, but not of Irish.

a hu buk dat ?		'Whose book is that?'
wai i naa wok ?		'Why is he not working?' (FN 1)
mi hia shi mi tek sik	'I heard she had taken sick.'
mi no mi no yu get tikit 'I didn't know you had got tickets.' 


Vocabulary proves almost as fruitless a field for the seeker after hibernicisms. The word /bániklèva/ 'sour milk' is indeed of Irish Gaelic origin; but as well as Montserrat it occurs widely not only in the West Indies but also in the southern United States. In Montserrat it has also been applied as the name of a kind of tree, but this is not a meaning associated with the source form, bainne clabair. With one exception, I can find nothing in Montserrat vocabulary which is of Anglo-Irish or Irish Gaelic origin but not known in other varieties of English.

This one exception, this one clear instance of undoubted Irish influence in the lexicon, and a word restricted as far as I know to Montserrat, is /ménsha/ or /ménshan/, meaning 'young female goat'. The source of this word must be Irish Gaelic minnseach 'goat' ["m'iJS@x] [given in some Irish dictionaries as mínseach]; the final velar fricative is of course dropped in English, while the final /n/ which some of my Montserrat informants used in this word (which is a common everyday one in MC) is presumably attributable to contamination from mention.


In terms of linguistic influence, then, the Irish contribution to Montserrat has been vanishingly small. (FN 2). Of the vaunted 'soft Irish brogue' the Emerald Isle of the Caribbean retains barely the tiniest trace.


  1. Although Jamaicans spell /naa/ as nah or naa, Montserratians quite logically insist on nar (thus Fergus & Grell, 1976, passim).
  2. Particularly so in view of the fact that the French occupation of 1667, which lasted less than one year, was sufficiently influential to leave behind /krápo/ crapaud as the Montserrat word for 'toad, frog' -- or did this word come in from Guadeloupe or some other French-speaking island?


[I subsequently also published a short account of Montserratian pronunciation in my three-volume survey Accents of English (1982), Cambridge: Cambridge University Press.]
Converted to html and placed on the WWW 1996 04 22. Author's address: J.C.Wells, Dept. of Phonetics and Linguistics, University College London, Gower Street, London WC1E 6BT.