Trudgill argues cogently against identity and in favor of automaticity as the primary basis for accommodation in koine formation. I agree with the general thrust of his arguments, and below offer qualifications and evidence that support them, but I also present caveats and alternative views, particularly with regard to the ordering relations between identity (or identity formation) and accommodation.

First, Trudgill posits that a new national identity is largely irrelevant to shaping the outcomes of dialect mixing in colonial contexts, and that any linking of new national identity to the newly formed dialects occurs only after the formation of said dialects. To the extent that we limit Trudgill's discussion of identity to new national identity (understood as a pre-imagined goal), it seems to me that his assertions are correct. The evidence of premodern dialect mixing that he reviews makes clear that these cases occurred in varied contexts and periods that preceded the rise of national ideologies/identities, and that national identities could therefore be linked to these varieties only after their development. Indeed, one of the goals of Tuten 2003 was to explore cases of dialect mixing in contexts where nationalism was unlikely to have had any significant effect on the process of dialect mixing and koine formation.

Second, Trudgill also argues that new colonial identities are irrelevant to new-dialect formation. If identity here is again understood as a preconceived goal, then I agree. However, we should consider the possibility that a new colonial (or local/regional community) identity could begin to arise at the same time that the first immigrants arrive and dialect mixing begins (i.e., simultaneously). All the premodern cases of dialect mixing that Trudgill describes occurred in frontier zones, where the newly arrived inhabitants often faced common human and natural threats. Such common threats to survival could well favor contact, cooperation, and the negotiation of common practices for purposes of defense and/or trade, within towns or regions, and thus could favor the development of a “frontier” identity in opposition to local “others” (e.g., Muslims in medieval Spain) as well as to distant “others” who remained in home communities. I would not want to claim that the early creation of such a rudimentary community identity is necessary to the regional dialect leveling that often accompanies dialect mixing (similar inputs, contact, and originally weak social networks may suffice to explain this phenomenon, at least for most phonological features), nor that it necessarily happens, but it may. And if it does, this early sense of community may reinforce accommodative behavior within a colonial community, first by adults and later by older children and adolescents.

Third, Trudgill claims that accommodation – the process underlying dialect mixing and koineization – is a largely automatic and mechanistic process, and I am inclined to agree. Thus, I too would reject my earlier “explanation” that accommodation occurs primarily “in order to promote a sense of common identity” (Tuten 2003:29). Of course, the source for this view is to be found in early Speech (or Communication) Accommodation Theory (e.g., Giles 1973), first borrowed into sociolinguistics by Trudgill himself (Trudgill 1986). As Auer, Hinskens & Kerswill (2005:7) point out, accommodation theory was originally grounded on theories of “social action” and “rational action,” in which accommodation by speakers – whether convergent or divergent – was analyzed as the outcome of conscious goal-oriented linguistic choices on the part of rational social actors. In the context of dialect mixing, speakers were understood to avoid salient features of their dialects of origin and adopt those of their interlocutors in order to gain acceptance or cooperation – and to present a common or at least less divergent identity. However, in more recent formulations of accommodation theory (e.g., Gallois, Ogay & Giles 2005), convergence is taken as the norm and group identity is generally emphasized only when explaining cases of divergence. As Trudgill rightly points out, social and cognitive psychologists have called into question intentionality as the dominant explanation of convergent behavior and have argued increasingly in favor of a more mechanical or automatic view of processes of accommodation. A particularly interesting proposal in this line (not mentioned by Trudgill) is the interactive alignment model of Pickering & Garrod 2004, who propose that participants in dialogue automatically align their cognitive representations and production on a number of levels (situational, semantic, lexical, syntactic, and phonological-phonetic). Such automatic alignment is favored because it allows dialogue participants to establish an “implicit common ground,” favoring mutual understanding, and because it reduces cognitive load, favoring rapid interaction. Pickering & Garrod (2004:183; also Clark 1998) propose that the interactive alignment process may be responsible for fixing routines in the language or dialect of a community of speakers, since it has been found in experimental research that groups of adult maze game players show high levels of alignment. Although the applicability of this model to adult speech may be more limited than what Pickering & Garrod claim (e.g., because adults may not be able to alter their production or because they are heavily socialized to perform in certain ways), there is evidence that the model is very highly predictive of the behavior of younger children. Indeed, Garrod & Clark 1993 found that 7-year-old children who participated in a maze game experiment could not circumvent the automatic alignment process, whereas older children and adults could adopt a strategy of nonalignment when appropriate. Pickering & Garrod (2004:180) explicitly claim that strategic processes of nonalignment are overlaid developmentally on the basic interactive alignment mechanism, which tends to dominate all dialogic interaction, but which does so most clearly in children. It is, therefore, probably no accident that children have been identified as the primary agents of koine formation (though “agency” here is primarily nonintentional). Still, Kerswill 1996 and Kerswill & Williams 2000 have shown in their Milton Keynes study that it is not very young children who first display the changes associated with new dialect formation, but rather older children and adolescents. As Kerswill explains, this is due to the changing social orientation of the children. The 4-year-olds tended to orient toward their parents and reproduce variants found in their parents' speech. However, 8-year-olds and especially 12-year-olds, increasingly oriented toward their peer groups, showed movement to a new convergent norm. Is the fact that children change their social orientation away from their parents and toward their peers (found in many studies) independent of identity factors? Why do older children choose to follow peers rather than parents? We might ascribe this only to changing frequencies of interaction, but it also seems to be linked to children's growing awareness of physical and other similarities between themselves and other children, and a growing identification with these similar others. Even though identity-as-preconceived-target is likely irrelevant to accommodation, it may be that accommodation (or automatic interactive alignment) and identity formation (among older children and adolescents) are closely linked. The clear preference for accommodation by older children to peers favors their integration into groups and even the creation of the very groups into which they integrate, and these groups in turn help to shape their personal and social identities. In this sense, a universal tendency toward accommodation would seem to be part and parcel of the identity-formation process. If this is so, then we may want to claim not that the formation of community identity is a consequence of koine formation, but rather that community identity formation and koine formation are simultaneous and mutually dependent processes.