The Distribution of Raised and Lowered Diphthongs as Reflexes of M.E. ī in Two Ulster-Scots (US) Dialects*

Robert J. Gregg

Since the advent of generative phonology, the Great English Vowel Shift has become once again the focus of much attention. We hear — as a modern refinement — a great deal about α-switching rules whose elegance derives mainly from the front-back symmetry apparent in Standard English (whether Southern British or General American), a symmetry paralleled — as far as diphthongization is concerned — by Standard German and Standard Dutch. This symmetry is missing, however — and a-switching as a consequence intuitively less elegant — in the northern dialects of English, where ME ū does not diphthongize, while in continental Germanic languages and dialects other than those mentioned there is no diphthongization whatever, although peninsular Scandinavian shares with most Scottish dialects an anticlockwise (or upward/frontward) rotation of the close back vowels.

All modern dialects of English thus show diphthongal reflexes for ME ῑ, ranging from the narrow or more 'raised' type [əi] to the wide or 'lowered' type [aι]. It is, incidentally, fashionable in current discussions about these diphthongs, especially among American linguists, to refer to the use of [əi] in Canada as 'Canadian raising'. As a Canadian linguist I feel that a much more appropriate term to describe what happens in the North American context would be 'American lowering'. By examining as widely as possible the whole problem of the distribution, including the possible opposition, of these two diphthong types, this paper hopes to demonstrate that in view of the evidence from Scottish and US as well as Canadian English the [əi] can be established as the underlying form and the [aι] as derivative.

These two types of diphthong, [əi] and [aι], occur then not only in Canadian but in Scottish standard speech, although in both these varieties of English the two diphthongs are subsumed in one phonological unit. This paper will, however, go on to explain how in contrast with the Canadian or Scottish situation — in two dialects of present-day US speech (the urban standard speech of Larne and the rural dialect of its East Antrim hinterland) the /əi/ and /aι/ have become polarized and are now found in meaningful contrast. This seems to be a somewhat rare phenomenon in modern English, yet the rules and constraints which govern the selection of /əi/ or /aι/ are well motivated.

In order to explain the systemic distribution of these two diphthongs it is best to start with the current situation in Scotland. Some Scottish dialects have remained at the stage where /əi/ is the only reflex of ME ῑ, although most have added a rule which gives [aι] — allophonically — in certain environments. In Standard English as spoken currently in Scotland, however, it could be stated that [əi] occurs exclusively in the environment of a closed syllable whose final segment is any one of the set of consonants except the subset of voiced fricatives and /r/. In all other environments [aι] occurs.

For Canadian English a very similar situation obtains, except that some constraints to the rule that [əi] → [aι] are removed. Thus [aι] will occur before all voiced consonants in Canadian English, not only before voiced fricatives as in Scotland. It is of course by a further and complete removal of constraints that we get the rule for Standard English as spoken either in England or the U.S. whereby /aι/ has taken over the complete territory.

To turn now to the various types of English spoken in the northern part of Ireland, Ulster Anglo-Irish has in all environments the /əi/ diphthong — very similar and closely related historically to the southern Anglo-Irish /ʌi/, both deriving from Elizabethan English. All the Ulster-Scots dialects, however, have not only /əi/ but /aι/, and further, as a result of their history and linguistic contacts, these two diphthongs are nowadays found in meaningful contrast: they are, in structural terms, different phonemes.

In order to interpret the somewhat complex system that has arisen it is best to start with the rural US dialect. Here, as in the broad Scots dialects that were its source, [əi] (< ME ῑ) could be regarded as the general underlying form, while [aι] occurs only in the environment of a following voiced fricative (including /ɹ/, or in open syllables, whether final or in hiatus). This seemingly simple allophonic distribution, however, is complicated by the survival of a small set in the lexicon of short open syllable forms with /əi/ derived from ME ei or ai. This leads to meaningful opposition between pairs such as:

/bəi/'bay'versus/baι/'by'; 'buy'
/stəi/'stay'versus/staι/'sty (for pigs)'

With the rather limited set involved it is clear that this opposition is quite marginal within the phonological system. On the other hand it is well and securely established and allows for no deviation. For example, any use of /əi/ for the expected /aι/ would be a kind of shibboleth, identifying the speaker as coming from another dialect zone, specifically from the Ulster Anglo-Irish area.

The most complicated situation of all has arisen in the Ulster-Scots urban centres, for example in my own home town, Larne, a small town with about 20,000 inhabitants, in East Antrim. To unravel the phonological complexities of this modern urban type of speech we must know something of the town's history and geography and the relevant sociolinguistic factors:

(1) Larne and many other towns with similar urban dialects have grown up in the heart of a solidly Ulster-Scots rural environment in the course of the last three or four hundred years.
(2) Demographically their increasing population until quite recent times was drawn in from the surrounding US-speaking areas.
(3) For social reasons the US speakers in the urban setting found it necessary and desirable to modify their dialect drastically in the direction of a model closer to the standard language.
(4) Before the advent of modern communications the only immediately available model was, of course, Ulster Anglo-Irish (UAI), which has thus left many traces in the urban US speech of today.

With reference to the distribution of the /əi/ and /aι/, these diphthongs are for SI urban speech also in meaningful opposition, but involve a different and more complex pattern than is the case of US rural dialect. There is, to begin with, the same basic rule, namely that /əi/ seems to be in the underlying form which, however, becomes [aι] in specfic environments, namely, before the voiced fricatives and /ɹ/ and in final open syllables or hiatus. This simple rule turns out to have many constraints, some built in and some arising out of the languages-in-contact situation in the urban setting.

In US urban speech, as obvious counter examples to the final open syllable rule, we find the following:

/dəi/die 'cease to live'
/ləi/lie 'untruth'
which are in meaningful opposition to
/daι/dye 'change the color of'
/laι/lie 'be in a recumbent position'

In fact, when we examine the whole set of comparable monosyllabic forms we find what looks like a random distribution of /əi/ and /aι/. A careful, comparative study reveals, however, that the forms with /aι/ are identical with the corresponding US rural forms and thus represent an extension of the country dialect, whereas the forms with /əi/ coincide with UAI phonology and represent borrowings needed to replace US rural forms so divergent from the standard language that they would not be understood in the town. In the above examples US rural actually has /i:/ eye, /li:/ lie 'untruth', and /di:/ die 'cease to live'. In other cases — even when there is no minimal opposition — urban US has adopted the UAI /əi/ in final open syllables such as:

/fləi/flywhere rural US has/fli:/
/həi/highwhere rural US has/hix/
/səi/sighwhere rural US has/sæ̈x/
/ʃəi/shywhere rural US has/ble:t/

where the rural forms are clearly very deviant.

As obvious counterexamples to the rule which gives /aι/ before voiced fricatives we find in urban US that the underlying /əi/ is maintained in certain noun-class words which exhibit a voicing rule, such as:


With these forms we may contrast


which makes it clear that it is the pressure of the paradigmatic set that keeps /əi/ above, and that such items would have to be so marked in the lexicon.

A similar phenomenon crops up before voiced fricatives in the verb class so that we find

/dɹəiv/drive(past: /dɹo:v/)
/ɹəiz/rise(past: /ɹo:z/)

with /əi/ instead of the expected /aι/. A closer study reveals that these items belong to the strong verb subset where models such as:

/ɹəid/ride(past: /ɹo:d/)
/ɹəit/write(past: /ɹo:t/)

have clearly exerted enough pressure to block the application of the rule. A simple but interesting example which clinched the working of this rule and constraint is the verb thrive, whose vocalic nucleus was found to vary between /əi/ and /aι/. This is not an example of free variation, as it was discovered on closer checking that speakers who said /θrəiv/ had the strong past tense /θro:v/, whereas for those who said /θraιv/ the verb belonged to the weak class. All weak verbs, of course, have /aι/: /daιv/ dive, /pɹaιz/ prize, /ɹaιð/ writhe.

Suprasegmental features may also play a role in blocking the application of the rule /əi/ → /aι/ when followed by voiced fricatives, as in:


Here apparently the weakened, secondary stress on the final syllable acts as a constraint, for under primary stress we find for example:

/səɹˈpɹaιz/surprise/ədˈvaιz/advise, etc.

Morphology may also have some influence, at least in one example where both urban and rural US agree in making a meaningful opposition between /məin/ mine (sb.) as against /maιn/ mine (possessive). Here the first item has the expected /əi/ in a syllable closed by a consonant other than a voiced fricative. The /aι/ in the second item has to be explained in morphological terms as /maι+n/, with the juncture explaining the retention of /aι/ which would naturally occur in the open syllable form /maι/ my.

In summing up then the reasons why the US dialects have developed a meaningful opposition between /əi/ and /aι/ we must stress factors such as the survival of a diphthongal reflex of ME ei and ai now realized as /əi/ in the rural US dialects and thus opposed to /aι/ in a few open-syllable forms. Urban US speech based on the rural dialect has, however, discarded these survivals, but has added other forms with /əi/ borrowed from UAI when the US forms were too divergent for use in the town. The influence of paradigmatic sets in noun and verb classes blocked the shift /əi/ → /aι/, and, finally, reduced stress and morphological juncture act as constraints on the same rule. As a result in all types of US speech /əi/ and /aι/ have a complex but specific distribution and function as separate phonological entities, although in each language type it seems best to regard /əi/ as the underlying form and /aι/ as its development.

* Originally published in Dressier, Wolfgang U., and F. V. Mareš (eds.), Phonologica: Akten der 2 Internationalen Phonologie-Tagung (Wien, Munchen: Fink, 1972), 101-105.