I shall begin this brief contribution with a small amount of self-justification, since I appear to have become picked out as one of the “bad guys” in Trudgill's thesis. I shall then go on to agree with some of Trudgill's points, but not with all. In particular, I shall conduct a small thought experiment on the link between accommodation and the development of new dialects.

I am quoted by Trudgill as saying that “New Zealand English derives from a variety of English spoken in the south-east of England” (Bauer 1994:391), and also as saying that “New Zealand English is derived from Australian English” (Bauer 1994:428), something which Trudgill interprets as meaning that “New Zealand English arrived as a ready-formed entity transported in its entirety from Australia.” In more recent papers (Bauer 1999, 2000) I have retreated slightly from the first claim here, showing that there are many phonological and lexical features of New Zealand English (NZE) that come from places other than the southeast of England. But even so, it would clearly have been absurd if I had claimed that NZE was derived from Tyneside English or Lowlands Scots. The whole phonological system of NZE is so close to the systems found in the southeast of England and so distant from that of lowland Scots that it seems unlikely that similarities on such a scale could arise purely by chance. Of course, the two are not identical. And it is for that reason that my text says “derived from” rather than “is”: There are changes to the system, and to the phonetics of the system, caused by a number of factors, including the ones that Trudgill discusses. For the same reason, I reject Trudgill's interpretation of what I say about Australian English (a comment, incidentally, made as much on the basis of vocabulary similarities as on phonological choices): “Derived from” does not mean “transported in its entirety from.” This excuse may not be sufficient to remove me from the list of people who got it wrong, but it may clarify the situation slightly.

While I am not an expert in any of the various examples of colonial expansion which Trudgill adduces as indicating that identity was not a major factor in the formation of the new varieties, I am, in general, happy to concede that point. Trudgill provides no evidence for his claims on this point, but I think that some is probably available, at least in more modern instances. In New Zealand, for instance, Britain was referred to as “Home” (note the initial capital), even by people whose parents and grandparents had been born in New Zealand, well into the second half of the 20th century. Any people who consider a country 25,000 kilometers away to be “home” are unlikely to feel a strong pull to create a new identity based on a different country. I suspect that social historians could provide many examples pointing in the same direction. While it is difficult to know what percentage of the population in the 1950s really thought that their home was half a world away, the fact that such phraseology persisted over a century after settlement indicates that unity with Britain rather than difference from Britain had been a defining perception for many New Zealanders of an earlier generation. Trudgill cites Moore 1999 as claiming that identity is an important factor in the development of “colonial dialects,” but Moore is specifically talking about “post-colonial societies” rather than about the society at the time of settlement, a very different situation. Perhaps Moore is not one of the bad guys, either.

Trudgill talks about dialect mixing as beginning in accommodation (Giles 1973), where speakers attempt to “talk like the others talk” (Keller 1994:100). We can begin by accepting the reality of accommodation, and the force of trying to talk like others. But not everybody is equally good at accommodation. We do not know what might cause the differences, but we all know people who accommodate a lot, and other people who appear to accommodate very little. At one time in my life, the moment I traveled north of Berwick-upon-Tweed I became rhotic and used monophthongs for the face and goat vowels, and yet after nearly 30 years in New Zealand, I still show remarkably little accommodation to the New Zealand accent (despite a host of favorable social reasons why I might have accommodated). In one place, at one time, I accommodated greatly; in another place at another time I accommodate relatively little. The same must be true for other speakers (the uniformitarian hypothesis demands it).

So what does accommodation lead to? It cannot lead directly to a mixed dialect, as Trudgill appears to believe, because different speakers will accommodate to different degrees, and, in any case, will accommodate to the interlocutor of the moment. What it does is provide variation between alternative versions of the “same” linguistic unit within a single speaker. I suspect that accommodation is sporadic and starts with places where the systems of the two interlocutors differ in salient ways and in ways that speakers can imitate to some degree. What causes the salience and what speakers can imitate are surely questions worthy of further investigation, but this is not the place to look into them. I would like to suggest, as a part of this thought experiment, that variation between alternative realizations of the same element within a single speaker gives children who are learning the language around them something of a problem, and something of a solution. The solution part comes about because they can tell from repetitions provided by the same speaker what the range of options available to them is at any point in the system; the problem is in determining which of those options to adopt. The problem here may well be solved by adopting the numerically superior form, as Trudgill proposes elsewhere. But now we have a situation where the numerically superior form is not numerically superior simply because more of the people in the founder generation happen to have it as part of their normal repertoire, but because some people for whom it is not a part of their normal repertoire are also using it in a (more or less accurately) copied form.

In other words, I should like to suggest that it is not the accommodation as such that leads to dialect mixing; rather, it is the use that accommodation is put to by the next generation that leads to dialect mixing. However, if we accept this view (and ideally we would like some experimental evidence to support it), we can come back to the question of identity, and ask with whom the new generation is going to identify. The obvious answer, where there are sufficient children together, is with peers. That is, the numerical superiority of a given form has to be present not only in the experience of the individual child, but also in the group experience of a set of children (who may all have parents from different original dialect areas) in order to be passed on. So any calculation of numerical superiority really requires a computational model of great sophistication if it is to provide a realistic answer to why one variant rather than another becomes adopted in a particular community. And at the same time, no calculation can totally ignore the notion of group identity; we just have to ask what complex kinds of identity are being expressed in the choice of a particular phonetic variant. It will not be as simple as feeling that one is “British” or “New Zealand”; it will be much more local and much more specific.