There is nothing controversial about Keller's assertion that human beings operate according to a powerful maxim which he renders as “Talk like the others talk.”¹

This is self-evidently true (although we could of course discuss precisely who “the others” might be). If it were not true, there would be, say, no local dialects. The fact that everyone who has grown up in the same community speaks in the same way therefore, in a sense, needs no discussion. It is always the case, as near enough as makes no difference.

Local dialects maintain themselves, subject to linguistic change, even in the face of the arrival of outsiders in the community – up to a certain demographic point. Adult arrivals accommodate to the dialect to varying degrees, ranging from almost total to hardly at all; but in the long run this is unimportant, since young child arrivals accommodate totally, and contribute together with the rest of the community to transmitting the dialect to future generations.

Since this is what always happens, we can suppose that it is the result of something like a universal of human behavior. I have suggested that this universal is even broader than that posited by Keller, namely the more general principle of behavioral coordination which is found not only in humans but also in other primates. This is due, as Pelech says, to “the innate biological basis of interactional synchrony.” A number of the commentators have alluded to a link between identity and accommodation; but normal coordinatory behavior, including linguistic accommodation, is so automatic – children do not operate it “at will,” as Holmes and Kerswill believe – that I do not believe that identity is required as an explanatory factor. Nik Coupland agrees with me “that there are dangers in running too freely to causal explanations around identity;” and my emphasis on the automaticity of linguistic accommodation and its link to primate behavior generally is indeed intended to warn against the too ready recourse to identity as our “first explanatory resource,” as Coupland calls it. If “identity” explains everything, it explains nothing.

Because it is innate, this kind of behavioral coordination is the unmarked case. Where exceptions to the norm of “Talk like the others talk” occur – for example, Labov's New York City Nathan B, or Newbrook's (1986) Scots Scouser – they do genuinely need discussion and explanation, and it is possible that identity factors may well be found to have played a role. This is nicely alluded to by Donald Tuten when he points out that in recent formulations of accommodation theory, “convergence is taken as the norm and group identity is generally emphasized only when explaining cases of divergence.”

New-dialect formation by children in tabula rasa, new-dialect formation scenarios is a fascinating phenomenon, but it is clear that it too is totally predictable in the same way. There is no known case of it not happening. Dialect contact of this type always leads to dialect mixture, which always leads to new-dialect formation. It is obviously then the result of the same kind of automatic and/or universal coordinatory aspect of human behavior which leads to local dialect maintenance. We need look no further for explanations of why this is so.

The fascinating difference with these scenarios is that there is no already-established dialect for speakers to converge on. The question for us linguists is therefore not at all why this form of linguistic convergence happens – it is inevitable – but why the linguistic outcomes are precisely what they are. Janet Holmes and Paul Kerswill believe that we must look for the social reasons for the nature of these linguistic outcomes. There must be, they say, social reasons for why one language variant is selected over another. I do not believe that there are any in colonial new-dialect formation (see below). And in any case, we do not need social reasons in order to explain what happens. I have argued, and shown on the basis of data from New Zealand English, that there are linguistic and arithmetical explanations readily to hand: The numerically superior variant – other linguistic things being equal – is always the survivor. “But what if there is no such variant?” they ask. This is only extremely rarely the case in the New Zealand data I have examined. But I have actually answered this question: Generally, where there are two or more major variants at approximately the same level of occurrence in the dialect mixture, then both variants survive, according to the principles of reallocation as explored in some depth in Britain & Trudgill 1999, 2005.

Holmes and Kerswill have rightly noted from my other writings the importance I attach to the role of children in these scenarios. And Salikoko Mufwene has also understood that “the agency of children in selecting some variants into and out of their idiolects” is very significant. But I did not highlight in my essay as I should have done the crucial role played by children in the kind of colonial dialect mixture situations I have been focusing on. The Berwick-on-Tweed and New Zealand adult accommodation experiences of Laurie Bauer (a very “good guy” from whose work – not least the current commentary – I have learnt a great deal on this topic!) are very important and revealing, but they are not relevant to the issue at hand as I see it. All the data I analyzed in my book (Trudgill 2004) indicated that the initial “work” of new-dialect formation in these situations is performed by the small children of the first two generations, which is why the social factors which Holmes and Kerswill would like to look for do not come into play. The adult settlers in any colony, of course, bring all the baggage of “the social marking of linguistic variants” (Holmes & Kerswill) with them, although Belich (1997:330) and Britain 2005 make it clear that in New Zealand they discarded a very great deal of it of it. But this kind of baggage is not relevant to 7-year-old children in the colonial situation – which is precisely why the mixing of variants from different dialects of different regional origin and different degrees of social status, with a mass of different associations and connotations, always takes place. The challenge is to work out why, in the children's idiolects, some features are acquired and preserved, and others not. Holmes and Kerswill argue that “children would have been linguistically socialised in specific ways.” But we know very well that children do not speak like their parents, and there is no more reason to suppose that parents were more influential in colonial New Zealand than they are anywhere else, especially since they came collectively from a range of different regions and backgrounds. Crucially, we also have in the New Zealand data siblings who differ considerably from one another in the variants they have adopted out of those available. (Of course, if other scholars can come up with a feature-by-feature social-reasons account – of the type I have provided for my deterministic accounts – for why a new colonial dialect has the phonological and grammatical shape that it does, then I will have to rethink my position.)

The point about the role of children is also relevant to a comment of Edgar Schneider's: I did not intend to imply, as Schneider supposes, that the role of identity was not relevant in the Martha's Vineyard situation. It obviously was. What I pointed to was Labov's assertion that identity is “not often” involved in this way, and that we should always think of the “mechanical view” as our first explanatory resource. Important for me here, in any case, is the fact that on Martha's Vineyard small children were not the crucial locus of linguistic change.

Enormously helpful on the role of children is Tuten's exposition of the interactive alignment model of Pickering & Garrod, who propose that “participants in dialogue automatically align their cognitive representations and production on a number of levels”; Tuten tells us that there “is evidence that the model is very highly predictive of the behavior of younger children” (my emphasis).

Holmes and Kerswill also ask, “Why would one person accommodate to another's speech rather than the other way around?” But the fact is that in the tabula rasa situation everyone (i.e., all the children) is accommodating to everybody else. To demonstrate this fact was one of the purposes of my focus on Irish English, Colonial German, Southern Iberian Spanish, Valencian, and Old English, where I directed attention as much as I was able to actual linguistic details, illustrating the fact that the new dialects were obviously the result of multidirectional accommodation. This is why, say, Valencian colonial Catalan was from the very beginning a mélange of Eastern Catalan, Western Catalan, and Aragonese forms. (To the – in my view perverse – objections to my invocation of the uniformitarian principle, by the way, I concede that “using the past to explain the present” may be more risky than vice versa, as Schneider rightly says, but if the principle operates in one direction it must necessarily operate in the other as well.)

In the end, however, I am very willing to concede that the actual complexities and subtleties of what really happens when speakers of different dialects come into contact with one another on a long-term basis in a new land still largely continue to elude me; and I am very happy to acknowledge that my thinking and writing have benefited, and will continue to benefit enormously, from the contributions of my seven kind and expert commentators.