A Window on the Past: "Colonial Lag" and New Zealand Evidence for the Phonology of Nineteenth-Century English

The American linguist Albert H. Marckwardt (1958, 80) wrote:

These post-colonial survivals of earlier phases of mother-country culture, taken in conjunction with the retention of earlier linguistic features, have made what I should like to call a colonial lag. I mean to suggest by this term nothing more than that in a transplanted civilization, such as ours undeniably is, certain features which it originally possesses remain static over a period of time. Transplanting usually results in a time lag before the organism, be it a geranium or a brook trout, becomes adapted to its new environment. There is no reason why the same principle should not apply to a people, their language, and their culture.

In this paper I confirm that there is at least one sense in which "colonial lag" (see also Görlach 1987) is, or at least in certain situations can be, a demonstrable linguistic reality. It is in this particular sense, moreover, a linguistic reality that can indeed be explained in terms of the transplantation of colonial societies, although I personally would hesitate to extend the notion to the social and perhaps especially the cultural characteristics of such communities.

Colonial lag in the linguistic sense in which I will use it here does not require explanation in terms of adaptation to a new environment, exactly. I use the term here rather to refer to a lag or delay in the normal progression and development of linguistic change that lasts for about one generation and arises solely as an automatic consequence of the fact that there is often no common peer-group dialect for children to acquire in first-generation colonial situations involving dialect mixture. It must therefore, I would submit in agreement with Marckwardt, have been a feature of those early varieties of North American English that developed in dialect-contact environments. I propose in this paper that this form of delay can on occasion have, as in the case of the New Zealand data presented here, interesting methodological consequences for the study of the recent linguistic past.

The conventional sociolinguistic wisdom is that young children speak like their peers rather than, for example, like their parents or teachers (Trudgill 1986, 220). This conventional wisdom is necessarily correct, since otherwise regionally distinct dialects would never have survived in the face of the increased geographic mobility of modern societies. In any case, the evidence for the thesis is overwhelming: in the context of families moving from one dialect area to another, the phenomenon of total childhood accommodation is the object of so much and such widespread observation and comment on the part of nonlinguists that it does not really need scientific confirmation. American parents moving to London know very well that before too long their younger children, at least, will sound like Londoners. No one expresses any surprise, though they may express regret, if a Welsh-accented family moving to East Anglia quite quickly comes to consist of adults who still sound Welsh and young children who sound as if they have lived in Suffolk all their lives.

It is true that occasionally individuals may be found of whom this is not so, but they are usually socially maladjusted, nonintegrated people whose lack of linguistic accommodation to their peers is a sign of social pathology (Newbrook 1982). Payne (1980) has also suggested that after a certain age children may not master perfectly all the intricate details of phonological conditioning in a new variety they are exposed to, although the children she studied were in an area with an unusually large number of incomers. Chambers (1992), too, has shown that older children do not necessarily accommodate so completely or so successfully as younger children. And Kazazis (1970) has warned that the degree of accommodation normally demonstrated by children in the anglophone world may be at least partly culture-specific rather than universal. There are also a number of situations where children may become bidialectal in a peer-group dialect and a family dialect (Trudgill 1986, 32). But the general trend, at least in most European style communities, is very clear: up to a certain age, normal children accommodate rapidly and totally, or almost totally, to the speech of any new peer group of which they become long-term members.

However, there are certain situations where this does not happen. In such situations it does not happen because it cannot. These are admittedly unusual situations where children are unable to accommodate to a peer-group dialect because there is no common peer-group dialect for them to accommodate to. Berthele (1998), for example, has investigated the Swiss German speech of a group of children at a private school in Fribourg in which, for religious and historical reasons, there has been a tradition of speaking Bernese German rather than the local Fribourg German, and in which the students come from a very wide range of nonlocal linguistic backgrounds, and arrive at the school speaking many different varieties of German, or no German at all. Here, it emerges from Berthele's pioneering study, individual children adopt individual strategies whose eventual linguistic outcome is determined in part by their integration into the social structure of the class.

I report here on a similar but much larger-scale situation, which turns out, I argue, to have useful methodological consequences for the study of the recent phonological history of English. The English of New Zealand is the most recently formed major variety of natively spoken English in the world, with only the English of the Falkland Islands postdating it as a colonial native variety of English. The English language, obviously, arrived in New Zealand from the British Isles-census figures for 1881 show that approximately half of the immigrants were actually from England-and the crucial formation period was almost certainly between 1840 and 1860 (Gordon et al. forthcoming). This paper argues that an unusual set of circumstances due to the colonial new-dialect formation situation, together with a possibly unique data set, enables us to gain accurate information about the way in which at least some forms of British English were pronounced long before the advent of electronic recording techniques.

The data on which this paper is based derive from the Origins of New Zealand English (ONZE) project, directed by Elizabeth Gordon at Canter-bury University in Christchurch. Much of the work of this project is based on a rather remarkable database. In 1946, a Mobile Disc Recording Unit was set up by the New Zealand National Broadcasting Corporation, and until 1948 it was sent around the country to collect pioneer reminiscences and local music. In 1988, Gordon had the enormous foresight and insight to arrange for the purchase of copies of these recordings. The recordings have now been organized, cataloged, and rerecorded onto digital tape, and colleagues working on the project are transcribing them orthographically. I have been concerned with analyzing the phonology of the oldest speakers on these recordings in order to trace the new-dialect formation processes (Trudgill 1986) involved in the development of the New Zealand accent with reference to the British Isles (and other) input (see, for example, Trudgill et al. 1998; Trudgill et al. 1999). I have so far analyzed the speech of 83 informants from 34 different places in both the South Island and the North Island of New Zealand.

While the purpose of the ONZE project is obviously to look at the genesis of New Zealand English, it has also become clear as our work has progressed that these very valuable recordings give us some important and possibly unique insights into earlier forms of British Isles English. The point is that our speakers, who were recorded in the late 1940s, when they were already elderly, were born between 1850 and 1890. They represent the first generation of New Zealand-born anglophones. Now, typically, as already mentioned, children acquire the dialect and accent of their peers. However, in early anglophone New Zealand, just as in the Fribourg school mentioned above, there was no single, established peer dialect for children to acquire. This, of course, is typical of all situations involving the dialect contact, dialect mixture, koinéization, and new-dialect formation (Trudgill 1986) that occur in colonial and other similar situations, such as the development of new towns (see Kerswill 1994). It would also therefore presumably have been a feature, as I indicated above, of many early anglophone North American settlements.

In our New Zealand data, we see that this situation can lead to two different scenarios. First, in some of the New Zealand cases I have analyzed, it is clear that certain elderly speakers, as children 110 to 150 years ago, acquired, unusually, something very close indeed to the English, Scottish, or Irish English dialects of their parents, for the very good reason that there was nothing else for them to acquire. These are people who were raised in very isolated rural communities. One of our informants, for example, grew up in a community that consisted entirely of herself, her many siblings, and her mother and father 40 miles by rowing boat from the next settlement. Thus, some such people who never set foot in Scotland sound totally Scottish because their parents were Scottish. Interestingly, in another case, an informant whose parents were from Kent sounds utterly Irish, apparently because he was brought up by an Irish washerwoman. The speech of these isolated people is therefore to a considerable extent fossilized in the sense that it represents forms of speech typical of a generation older than would normally be the case.

Second, in other cases, the lack of a single peer-group model led, as in Fribourg, to fascinating and varied individual dialect mixture processes, which I have described elsewhere (Trudgill 1998). This is what happens in the case of people brought up in mixed, nonisolated communities where dialect contact took place early on. It emerges that the mixed speech of these nonisolated speakers is also interesting in precisely the same way as that of the isolated speakers because, even though they do employ innovative combinations, these turn out to be combinations of CONSERVATIVE features (for further discussion and argumentation, see Trudgill et al. 1999). Their speech features, then, when looked at individually, are equally fossilized. It is also noteworthy that in the case of these speakers, there seems to be no particular connection between their speech and that of their parents.

Thus, the New Zealand recordings available to us give important insights into the way in which vernacular English was spoken in the British Isles by people born a generation earlier. The colonial situation, that is, enables us to push back the time depth available to us for historical linguistic studies in the sense that we are able to investigate speech that is, as it were, one generation earlier than would be the case with British-born speakers and for which we do not have large-scale sound-recorded evidence. I am saying that speakers born in New Zealand in 1850 use forms typical of their parents' speech and that in listening to people born in 1850 it is therefore AS IF we were actually listening to recordings of people born in the 1820s-something we are not actually able to do. For instance, members of the ONZE team have already argued, on the basis of these recordings and on the supposition of colonial lag in the sense I am using here, that the short front vowels /I, ε, æ/, which are often phonetically very close in the ONZE recordings, were probably much closer in the nineteenth-century English of England than they are today (Trudgill et al. 1998).

Because of the social composition of the immigrant population, moreover, our informants were speakers neither of Traditional Dialects, in the sense of Wells (1982), nor of Received Pronunciation, but of something in between, and are thus able to give us a more accurate picture of English as it was spoken by a majority of the population. The extent to which these New Zealand speakers can help to shed further light on what the nineteenth-century English of England was like can be seen from the following features.

The Pronunciation of /r/

In modern Britain, five different phonetic realizations of /r/ are extant:

1. 1. The sharply recessive voiced uvular fricative , which is confined to the northeast of England and some immediately adjacent areas of Scotland (Wells 1982, 368; Glauser 1994)

2. 2. The alveolar flap , which is usually associated with Scotland and parts of the north of England

3. 3. The retroflex approximant , which is most typical of southwestern England (Wells 1982, 342)

4. 4. The postalveolar approximant , most usually associated with RP and much of south and central England

5. 5. The labio-dental approximant , which is currently gaining ground very rapidly amongst younger speakers (Trudgill 1988)

There is no doubt at all that the labio-dental approximant is a new pronunciation. Of the other three widespread variants, we can suppose on phonetic grounds that the flap is the oldest and the postalveolar approximant the newest, with the retroflex variant being chronologically intermediate. We can suppose that even earlier forms of English may have had a roll (or trill). Bailey (1996, 99) indicates that "weakening of r from a trilled consonant was first reported in Britain at the end of the sixteenth century" (see also Wells 1982, 370). This gives us a history of lenition in the realization of /r/ as follows:

The relative chronology is clear. What is less certain is the absolute chronology. When, for example, did the variant become the most usual and widespread variant? Here the ONZE recordings are of considerable interest. The normal pronunciation of /r/ in New Zealand today is also, as in most of England, , although perhaps on average rather more retroflexed, that is, more conservative than in England. However, analysis of the ONZE tapes shows that the pronunciation of /r/ as a flapped is extremely common on these recordings. This, moreover, is not simply true of speakers who have a northern England or Scottish origin. In fact, it is employed, most often variably, by 47 of the 83 (57%) informants investigated. There is thus a strong suggestion that the weakening of the flap to an approximant in the Midlands and southern England is a very recent phenomenon, dating from about the middle of the nineteenth century, a development that has been followed more or less simultaneously in New Zealand (for further details, see Trudgill et al. 1999).

Rhoticity

Perhaps of even more interest is what the ONZE recordings tell us about rhoticity. Bailey (1996, 100) says of the English of England that "the shift from consonantal to vocalic r, though sporadic earlier, gathered force at the end of the eighteenth century." Strang (1970, 112) writes that "in post-vocalic position, finally or pre-consonantally, /r/ was weakened in articulation in the 17c and reduced to a vocalic segment early in the 18c." Lass (1992, 66) similarly says that "postvocalic /r/ began to delete systematically in the seventeenth and eighteenth centuries." I suggest, however, that there has to be some uncertainty about the dating of the development of nonrhotic accents in England. The dates given by Strang and Lass may very well be accurate for London, but it is very probable that most regional accents lagged much further behind this than sometimes thought. Bailey's (1996, 102) statement that "resistance to the spreading London fashion was, however, not long sustained" may be something of an exaggeration.

My evidence for this claim comes from our ONZE materials. Except for the well-known but small area of rhoticity in the south of the South Island, New Zealand English is today nonrhotic. This has often been explained-indeed I have done so myself (Trudgill 1986)-as follows. At the time of the anglophone settlements of North America, all forms of British English were rhotic, with widespread North American rhoticity, to this day, as a consequence. However, by the time of the main emigrations to the Southern Hemisphere, including New Zealand, much of England had become nonrhotic, with the consequence that nonrhoticity was exported to these parts of the world.

It is now very clear that this is not correct. Of the 83 ONZE speakers analyzed so far, an astonishing majority, 77 (93%), are rhotic. It must be stressed that this is just as true of speakers with English-born parents as of people with Scottish or Irish parents. The evidence is, then, that most of England was still rhotic, not just at the level of the Traditional Dialects so extensively investigated by Ellis (1889) but also at the level of urban and lower-middle-class speech in the middle of the nineteenth century, and that extensive spreading of the loss of rhoticity is something that has occurred subsequently in both New Zealand and England (Trudgill et al. 1999).

The Lexical Set of Bath

It is well known that accents of English treat lexical items such as dance, laugh, path, and plant in two different ways. Accents in northern England and in North America have /æ/ in such words, while RP and accents in southern England have . This is sometimes explained by historians of the English language as having resulted from a lengthening of /æ/ (presumably [a]) and later phonologization to in the environment before the front voiceless fricatives /f, θ, s/, as in laugh, path, and grass and before certain clusters of nasals followed by obstruents, as in sample, demand, plant, dance, and branch, although it is, of course, recognized that it must have been more complicated than that. (For one thing, there are many exceptions, such as ample, grand, ant, and romance.)

One interesting problem here is that many Australians, as well as speakers with certain Welsh accents, are in a kind of intermediate position between English southerners, on the one hand, and Americans and English northerners, on the other: they have /æ/ in words such as sample, which involve a nasal, but in the fricative words such as laugh (see Trudgill and Hannah 1994; Hughes and Trudgill 1995). Other Australians, however, have the south-of-England system, which is also, crucially, true of modern New Zealanders. Wells (1982) points out that Leeward Islanders in the Caribbean also have the split system, while other West Indians do not. Wells speculates that the split system that we find in Wales, Australia, and the Leeward Islands-all, in a sense, colonial varieties-but not elsewhere "may well be because in eighteenth-century south-east England these dance-type words were still fluctuating between a short and long vowel; or indeed they may still generally have had a short vowel, and have gone over to the long vowel only later" (233).

In my view, the latter hypothesis is correct. I argue that we are dealing here with two separate sound changes. The evidence from the ONZE recordings is again very strong. We obviously have to exclude from our observations speakers who do not have a distinction between /æ/ and (e.g., they rhyme dagger and lager, presumably as a result of West Country and/or Scottish input), as well as those with obvious northern England accents, which have /æ/ throughout. Once we have done that, we observe that very many ONZE speakers consistently have the split system: they have in the lexical set of after, grass, and path but /æ/ in the set of dance, plant, and sample. In all, 38 out of an eligible total of 79 (48%) have this pattern. This strongly suggests that the change from /æ/ to in southern England (and in RP) was in fact two separate changes, the first-involving prevoiceless fricative environments, which, however, as is widely agreed, must postdate the settlement of North America-taking place a good deal earlier than the second, which involved combinations of nasals and obstruents and which cannot on this evidence have become general in England until the second half of the nineteenth century.

The Lot Vowel

North American English and Irish English are characterized by having an unrounded vowel in the lexical set of LOT, whereas, as Wells (1982, 130) writes, "in Britain the predominant type of vowel in LOT is back and rounded, ." However, he adds that we also find "the recessive unrounded variant in parts of the south of England remote from London." He further indicates that the vowel "often appears to be unrounded in the west [of England], being qualitatively , much as in the Irish Republic or in the United States" (347). Wells also quotes from my own work to say that "in Norfolk the LOT vowel has an unrounded variant" (339). This geographic pattern, with the east and the west of southern England being areas with unrounded vowels, separated by an intervening central area, including London, with rounded variants, strongly suggests that the area was formerly much bigger than it is now. This is confirmed by our New Zealand data. The fact is that unrounded in LOT, , is very common in these recordings, although it is totally absent from modern New Zealand English. Of the 83 informants, 38 (46%) use an unrounded vowel either consistently or variably. Since only a minority of our informants had Irish parents, we can suppose that unrounded vowels were geographically very widely distributed in the nineteenth-century English of England. It is therefore not surprising that at an even earlier stage this vowel quality was exported from England to Ireland and North America.

The Cloth Vowel

Paralleling the lengthening of /a/ before the front voiceless fricatives /f, θ, s/ in the set of BATH, there was also a change of short o to in the same environments (e.g., off, froth, cross), also in southern rather than northern England. Wells (1982, 203) refers to both these changes as PREFRICATIVE LENGTHENING and dates the beginnings of the lengthening to the end of the seventeenth century. Intriguingly, the change to , unlike the change to , is currently reversing, so that in England it is now mostly conservative RP speakers and speakers of low-status regional accents who retain this feature; other accents have gone back to in this set. The Survey of English Dialects (SED), however, shows the whole of the south of England as having . It is therefore interesting to note the extent to which this pronunciation is the norm amongst the ONZE speakers. We have to exclude from this computation those speakers who have the Scottish merger, also present in many forms of North American English, of the vowels of lot and thought-interestingly, this is as many as 14 speakers. Of the remainder, 70% pronounce the lexical set of off, froth, and cross with rather than , indicating that in the nineteenth century it was not only speakers of Traditional Dialects who had this feature in southern England. Interestingly, moreover, we can note that is still today a good deal more common in New Zealand than it is in England.

Happy-Tensing

The phenomenon Wells (1982) calls happy-tensing-the incidence of rather than /I/ in the final syllable of words such as happy and money-is the norm in modern New Zealand English. Regional varieties of American English and the English of England accents are split on this point. Happy-tensing is, however, very rapidly becoming the norm in England and is even spreading into RP, which has traditionally not had it (Trudgill forthcoming). At the level of regional accents, it is an innovation that appears to be most characteristic of southern England accents but has been spreading northwards for many decades. For example, SED records show that many counties in southern England that now have had /I/ in this context in the speech of rural Traditional Dialect speakers in the 1950s and 1960s. And my own mother, for instance, has /I/ while I have , even though we are both from East Anglia. The relative newness of this phenomenon is confirmed by the ONZE recordings. On these tapes, happy-tensing is quite common but still a minority feature: 42% (35) of informants have this pronunciation. Here again, New Zealand English must have shared with the English of England the adoption of this feature over the last hundred years or so.

T-Glottaling

Wells (1982, 261) writes that the realization of syllable-final /t/ as in Britain "must have spread very fast in the course of the present century," and indeed there is plenty of evidence that this is exactly right. According to Bailey (1996, 76), observers have been commentating on the phenomenon only since 1860, and early references were almost exclusively to Scotland and to London. The SED records show hardly any instances of glottaling except in the London area and East Anglia. And there is convincing evidence that it reached western areas, such as Cardiff (Mees 1977) and Liverpool, only very recently. In many studies (e.g., Trudgill 1988), it has been shown that younger speakers demonstrate more glottaling than older speakers. It is hardly surprising, therefore, that there is hardly any glottaling in the speech of our elderly New Zealanders, a confirmation, if one were needed, of the relatively recent development of this phenomenon in England.

However, I believe the following is also of considerable interest. When I first started listening to the ONZE tapes, I noticed that there was something about even those speakers who sounded very English that was strange-something that gave their speech a distinctly un-British and/or old-fashioned ring, to my ears. I eventually realized what it was: the ONZE speakers also show very little evidence indeed of preglottalization (Wells 1982, 260). That is, they do not employ a glottal stop before /p, t, k, tƒ/ in items such as hopeless and match. Preglottalization of this type is today very usual indeed in very many-perhaps most-English accents in England, including RP. As Wells points out, however, it has attracted very little comment from either amateur or professional observers in Britain (although it may be one of the things that lead Americans to describe British accents as "clipped"). We therefore have no good information that might lead us to any satisfactory indication of its dating. Our New Zealand evidence, however, suggests that preglottalization in Britain, too, is a recent and probably late nineteenth-century phenomenon.

Dark Versus Clear /l/

It is well known that most forms of English in England have an allophonic distinction between syllable-initial "clear" /l/ and syllable-final "dark" or velarized /l/. I argue that this is a relatively recent feature, again probably no earlier than the late nineteenth century in most forms of English in England. The evidence of the SED is very clear: it shows many areas of southern England with clear /l/ in syllable-final position. Of course, I must acknowledge that we are dealing with phonetic continuum here, rather than an either/or distinction between "clear" and "dark," but the overall trend is apparent. For example, my grandmother, like myself from the county of Norfolk, had little or no allophonic differentiation between syllable-initial and syllable-final /l/, using a rather "clear" pronunciation in all environments. My mother's syllable-final /l/ is darker than that, but not as dark as mine, which is in turn not nearly as dark as that of people younger or from further south than I am. Our ONZE evidence also tallies very nicely indeed with the suggestion that "dark" /l/ may be a relatively new phenomenon. Of our speakers, 19 (23%) actually employ a "clear" /l/, and, for most of the remainder, "dark" /l/ is not very "dark" at all. There is also hardly any /l/-vocalization, even though this is now very common in modern New Zealand English.

Conclusion

I argue, then, that, because of a one-generation colonial lag phenomenon, which was due to the dialect mixture resulting from the settlement of New Zealand by English speakers from the British Isles, the New Zealand English spoken by elderly people in the 1940s offers many insights into what the English of England was like in the mid-nineteenth century. Moreover, as a result of the predominantly upper-working-class and middle-class background of the migrants, we have access through New Zealand English to information on the forms of English in England about which we otherwise know the least-the majority but neglected speech forms of those mid-nineteenth-century British people who were not speakers of Traditional Dialects or of standard English and RP. The fact that these ordinary elderly speakers were actually recorded en masse at a time when recording techniques were rudimentary and oral history was in its infancy was for us an extraordinary stroke of good fortune.