The Ulster-Scots Dialect Boundaries in Ulster *

Robert J. Gregg

The experts — and laymen — have long been aware that, linguistically considered, the province of Ulster is divided into three parts: the Ulster Anglo-Irish (UAI), the Ulster-Scots (US), and the Gaelic-speaking. Professor Heinrich Wagner's Linguistic Atlas and Survey of Irish Dialects (1958) has provided us with the location and approximate limits of the scattered fragments of the dwindling Ulster Gaeltacht. As late as 1960, however, the boundaries separating the other two language types had never been fixed with any degree of precision in spite of their strikingly contrasted features.

With this objective of boundary-drawing in mind the present writer undertook a detailed survey of the US dialects between 1960-63. It was felt than an accurate delimitation of the speech area concerned was a necessary preliminary to any future work in this field, and further, that Ulster, where three language types are in sharp confrontation, was the perfect arena for trying out experimental discovery procedures for boundary mapping. Apart from its intrinsic interest the enterprise was also intended to provide information for the various Irish dialectology surveys as well as the linguistic surveys of Scotland and England and — more remotely — to make some contribution to the understanding of North American English dialect features, especially for certain parts of New England, Pennsylvania, the Ohio Valley, and the Southern Highlands where, according to Dr. Hans Kurath and his colleagues, Ulster-Scots settlements as far back as the 18th century contributed to local speechways. The same claim could be made for more recent times in Eastern Canada, especially the Ottawa Valley and neighbouring areas where mass migratory movements from Ulster have introduced a strong Ulster-Scots element, markedly in contrast with the surrounding typical Ontario speech patterns. It should be emphasized, however, that as used in North America the term 'Scotch-Irish' has a rather wide frame of reference, including all types of Ulster speech, which stand in sharp contrast with the well-known southern Irish brogue. In this paper, the preferred term, Ulster-Scots (US), is restricted to the rural Ulster dialects of an archaic broad Scots type, stemming mainly from south-west Scotland. US is set off against Gaelic and Ulster Anglo-Irish (UAI),(1) the latter being based on north and west Midland English dialects as spoken during the Plantation period in the 17th century. Within the US districts the towns have developed their own version of Standard English, which may be called Ulster-Scots Urban (USU),(2) important because it is also the second language of educated US dialect speakers and because, when US forms drop out, they tend to be replaced by USU equivalents.

Even a casual observer from outside will notice that all the dialects of English spoken in Ulster have features that contribute to what might be called their 'Irishness'. The task of the boundary seeker is to try to recognize these features and then to ignore them, aiming rather to collect the data that will polarize the systematic differences between the dialect groups. Thus no capital can be made of the universal frictionless continuant [ɹ](3) occurring even in word- and syllable-final position, so different from the Scottish trilled or flapped [r] and linking up rather with the [ɹ] of other parts of Ireland and North America — doubtless a common Elizabethan or Jacobean English colonial legacy of the 16th and 17th centuries; nor of the ubiquitous Ulster 'light' [l] which contrasts rather with the laterals of most English and Scottish dialects in Great Britain, and probably derives — all over Ireland — from the Gaelic substratum; nor of the front-central rounded /ü/ found everywhere in Ulster, whose allophones constitute one phoneme vis-à-vis the two English phonemes /u/ and /ɷ/. Here we have, of course, a parallel with Scottish speech in general, but the monophonemic /ü/ belongs just as much in UAI as in US, and in standard speech of all types as well as in all the dialects. This central /ü/ also belongs to the Gaelic of Ulster, apart from west Donegal which, like the southern varieties of Irish, has /u/.(4)

In fact, even with the limited data provided by Seamus Ó Searcaigh,(5) and by Nils Holmer's two studies of county Antrim Gaelic,(6) coupled with the information gleaned during joint field research the writer made with Dr. Emrys Evans in the Fanad Peninsula and Inishowen in county Donegal,(7) it becomes quite clear that there is a very close link in general between the Ulster Gaelic phonological material on the one hand and that of all the Ulster dialects of English on the other.(8)

Suprasegmental features such as pitch patterns and the realizations of juncture would bear out these widespread areal linguistic or substratum manifestations. In the same way syntax is a useless yard-stick, as even superficially very distinctive dialectal expressions turn out on closer examination to have an identical underlying syntactic structure, e.g. [ˌtɑːmz ˈɷ̈t we ðə ˈdʌg, ˈsnoːkən əbɷ̈t ðə ˈśʌxs, ˈlʌkən fəɹ ə ˈbɹɔːk] corresponds word for word — even without any transformational differences at the surface level — to the standard utterance that translates it: 'Tom's out with the dog, poking about the ditches, looking for a badger'. This underlying identity of patterning in all the dialects of English surely explains why so many dialect grammars stop short at the end of the phonology section, and why Wright's English Dialect Grammar in particular has 655 pages of phonology as compared to 41 pages devoted to the rest of the grammar. There is, in short, little of interest to be found at the grammatical, specifically morphological and syntactic, level when we are searching for polarized contrasts.

Thus neither the phonological material at the raw phonetic level nor the grammatical structures could be relied on in drawing up a questionnaire which would serve as the basic tool in the whole enquiry and provide suitable data to permit the clear separation of the dialect groups concerned. From various quarters the suggestion came that questions should be asked with the purpose of eliciting contrastive lexical items as had been the case with the Leeds survey and the first and second postal enquiry booklets of the Linguistic Survey of Scotland, but the results obtained from over a period of three months — mainly in Donegal — were disappointing, as the isoglosses failed to bundle and simply confirmed the well-known concept that every word has its own history — and, we could add, geography.(9) What did emerge, however, from these preliminary investigations was that in many cases where a given question did not elicit a variety of unrelated lexical forms, but rather variations of what was historically the same form, a clear-cut bundle of isoglosses did show up separating the typically Scottish historical-phonological developments from the equally typical north-west-Midland English forms reflected in the UAI dialects.

Following up this promising clue, it was easy to produce a questionnaire based mainly on items pinpointing these divergent historical-phonological reflexes of older English forms, relying for the most part on contrastive changes in the vowel and diphthong nuclei. The English consonants, as is widely known, exhibit much less systematic regional variation, the main contrastive situation among the Ulster dialects of English being the preservation of ME voiceless velar fricative /x/ versus its loss, or replacement by /f/, in such forms as daughter, night, laugh, high, etc.

To cope with the predictable variability among the US dialects themselves, it was felt necessary to set up a model with regard to which phonological segments could be judged as to whether or not they were representatives of normal Scottish developments, or their known variants — particularly among the south-west group of Scottish dialects spoken in Ayrshire, Galloway and adjacent areas from which settlement history shows the Scottish planters to have migrated to Ulster from the early-17th century onwards. This model was established on the basis of the writer's experience gained with the well-preserved east Antrim US dialects during a period of over thirty years of investigation and study, with their synchronic and diachronic phonology and their lexicon as the main focus of attention.(10) The model was, of course, subject to revision and amendment throughout the survey and later, when the results were under scrutiny.

The main concern of the project was to trace the spread of specifically Scottish forms, but naturally in what ultimately proved to be the neighbourhood of the dialect border equally specific UAI forms began to appear and thus gradually a contrasting model for UAI was built up whose 'Englishness' was for the most part immediately apparent, although some checking had to be done among the field records at Leeds, supplemented by special private investigations in parts of the English Midlands, to establish firmly the identity of certain English dialectal terms such as [ˈɛlːdəɹ] 'udder', [ˈɛːdəsəz] 'after-grass', [strit] 'farmyard',(11) etc., which had penetrated deep into the US areas, and markedly non-standard phonological forms such as [stroː] straw, [θoː] thaw, [joː] ewe, which take the place of US [streː], [θeü] and [jəü], once we cross the dialect border. These and other similar forms did point back to the north and west Midlands, thus confirming the story told by the Plantation historians, namely that Chester was the main port of embarkation for the English settlers who came not only from Cheshire but the wider hinterland.

It is an interesting fact that US speakers — knowing that the UAI speech of their neighbours generally approximates much more closely to Standard English than their own Scottish type of dialect does — tend to assume that UAI always uses 'correct' standard forms. For example, when asked about their name for a female sheep, they often responded: 'We ca' 'er a yow [jəü], but it should be a yoe [joː], should it no'?' Likewise, through some original taboo or avoidance of imagined coarseness, a cow's udder is frequently referred to by US speakers as her elder [ˈɛːldəɹ], the UAI word, rather than her bag [bɑːg], the older, traditional US term. The UAI dialect term (widely used in Ulster) for 'farmyard', namely street [strit], has come to replace the original Scottish term everywhere in the US part of the Laggan district in Donegal, though Antrim US has preserved the Scottish cassey [ˈkɑːse] and Down US mostly close [kloːs] in this sense.

The confrontation of US and Gaelic, of course, presented no problems of separation in general, as here we are dealing with one of the major historical linguistic boundaries between varieties of Indo-European speech in Europe, a boundary which has been moving slowly but steadily westward in the British Isles for the last millennium and a half and whose contemporary position the writer was able to confirm with some accuracy as far as the northern Donegal sector was concerned.(12) The only problem that arose occasionally with Gaelic forms was the question whether a given item such as [ˈgɛːlək] 'earwig', [ˈgɹiśəx] 'embers', etc., had been borrowed into US or UAI in situ from Ulster Gaelic or represented an earlier borrowing from Scots Gaelic imported by the 17th-century settlers. The complete disappearance of east Ulster Gaelic in the last thirty or forty years has made the task of tracing the history of such borrowings much more difficult, although O'Searcaigh's phonetic studies, Holmer's two monographs, Sommerfelt's work, and the more recent researches of Dr. Emrys Evans on Fanad, Glenvar, and Inishowen Irish (13) have made valuable contributions to the reconstruction of east Ulster Gaelic in general. The US/Gaelic boundary in Donegal was thus easily determined, and the line established on the basis of the writer's survey is confirmed by Professor Heinrich Wagner's Linguistic Atlas (14) as well as by the data worked out on the basis of the 1911 general census figures by G. B. Adams.(15) At one point (Termon, county Donegal) a bilingual informant (Mr. James O'Donnell, b. 1887) was found, who spoke Gaelic as his mother tongue, and as his second — now habitual — language, the Donegal version of US. With him the boundaries ran together in one person.

In these ways the nature of the border separating US from the other types of Ulster speech varies considerably, ranging from a major interlingual boundary with Gaelic in Donegal to a confrontation with a widely divergent type of English, namely UAI, an Irish derivative based on English Midland dialects, now in complementary distribution with US throughout the province outside the Gaeltacht.

The scope of this US boundary survey may be judged on the basis of some statistics. With the help of many well-informed local people, 125 informants were finally selected and interviewed: 34 in Donegal, 4 in the north-east corner of Londonderry, 23 in Antrim, and 64 in Down.(16) The density of coverage was purposely varied for a variety of reasons. County Antrim — well-known to the writer as his native county — was sampled round the perimeter of the US heartland, at points about 10 miles apart. Four points in county Londonderry were enough to link Antrim with Donegal. The latter county — unexplored territory — was given a more thorough survey. County Down was covered with a micromesh, not only for the sake of pinpointing the dialect boundary, but for the theoretical purpose of seeing what extra information would come up, for example, from checking the speech of every farmer within the transition zone in the South Ards peninsula, thus closing in the mesh to points one or two miles apart. The experiment was well worth while, for the micromesh certainly revealed linguistic facts and patterns that would otherwise have been missed, e.g. the detailed distribution of US variant forms for above — [əˈbïn] and [əˈbin], for dog — [dʌg] and [dəüg], and for farm — [fɛːɹm] and [fɔːɹm], as well as, e.g., the enormous phonetic variability in the forms of ant. This county Down micromesh investigation is probably one of the closest surveys ever to be carried out, certainly the closest in the British Isles.(17) The materials were collected during a period of over a year and a half, the distance covered within Ulster was about 25,000 miles, and the final form of the questionnaire included 683 items.

To return to the elaboration of the historical-phonological questionnaire previously mentioned as a necessary tool of investigation, the basic assumption was that the sets of vowels which are found nowadays in both US and UAI represent two distinctive and divergent lines of development of the underlying ME vowel system. Indeed, the very fact that the two main source areas from which settlers came to Ulster in the 17th century were located at relatively distant points in the continuum of English dialects on the island of Great Britain means that we find quite sharply contrasted developments in the two vowel systems.

The history of any range of phenomena can be considered diachronically from either the past or the present, the first method being more familiar to us. I feel, however, that as a matter of principle, historical dialectology should be dialect-oriented (18) and should take the contemporary state of the dialect as its starting point, mapping the present on to the past rather than vice versa. The current synchronic state is, after all, what is available for complete investigation: all past states are either more or less imperfectly or incompletely recorded or in some cases even quite unknown. Only for the present situation can we work out a coherent, viable, phonological system that can be thoroughly checked by unrestricted additional enquiries, and it is only on the basis of such systematized materials that we can fully understand the function and development of phonological units, and discover the ordered succession of rules added to and internalized into the grammar throughout the recorded history — where such exists.(19)

The same considerations naturally apply to the other levels of grammatical analysis as well: to morphology and syntax and, of course, to the lexical and semantic levels. We are in any case always scientifically obligated to seek 'un système où tout se tient', and the only systems completely accessible to us are the current ones, whether we wish objectively to collect so-called empirical data or subjectively to probe grammaticalness by introspective methods. On the basis of such arguments the final questionnaire was worked out, as already noted, on the results of a detailed study in great depth of an east Antrim US dialect, namely that of the Glenoe district spoken natively as a second language by the writer and as their first language by many of his relatives. Being located geographically close to Larne, one of the ports of entry for incoming Scots throughout the whole settlement period, the Glenoe dialect (20) has incorporated south-western Scottish innovations that came into force during the 17th century and perhaps even later. In this way Glenoe is more 'up-to-date', i.e. represents a less archaic type of Scots than, say, the US of Donegal or the Mid Ards. The lists in the questionnaire, then, represent groupings of items to be elicited, items incorporating in the dialectal forms of Standard English words, as well as occasionally in purely dialect words, first the consonantal phoneme /x/, and then in turn all the vocalic phonemes of Glenoe and similar US dialects,(21) which means the dominant forms of Antrim, north-east Londonderry and north-east Down US dialectal speech. The order is thus:

5/ü/12/əi/ and /ɑe/

These phonemes are linked back by phonological rules to earlier English forms which have given rise to a contrasting phonemic system in UAI.

Another reason endorsing the use of Glenoe as a model is that, according to Catford's schematic analysis,(22) Glenoe is a ten-vowel dialect, compared with which the Mid Ards/Donegal and north Antrim US have nine-vowel systems, in other words they under-differentiate as compared with the Glenoe type, having neutralizations in the phonological space (23) occupied by /i/ and /e/ respectively.

A detailed consideration of the questions — section by section — and an evaluation of the results obtained will now be embarked upon. For purposes of reference the full questionnaire is appended at the end of the paper, preceded by a tabulated summary of the relationships between US, UAI, and earlier English (see below).


* Originally published in Wakelin, Martyn F. (ed.), Patterns in the Folk Speech of the British Isles (London: Athlone Press, 1972), 109-139.

Introduction | Lists 1-7 | Lists 8-14 | Phonetic Systems | Questionnaire | Notes