Thirty years of research has not produced any really hard evidence that making people aware of formal features of the second language (L2) has any significant long-term effect on their grammatical development. However, people still have a persistent feeling that metalinguistic ability in the L2 is more than just a luxury extra or, viewed more pessimistically, more than a distraction and an encumbrance. It is surely a prerequisite for any proper research into such issues that we have a much more fine-grained explanation of the mechanisms involved in metalinguistic ability than has been the case so far. At the least, we need to develop a coherent theoretical model of this ability that we can use to generate interesting research questions about such issues as input enhancement (see Berent & Kelly, this volume) and focus on form (see Han, this volume). You might say that, although there has been no dearth of empirical research, not all that much has happened in this theoretical arena since the 1970s. The MOGUL framework being developed by Sharwood Smith and Truscott aims, among other things, to rekindle the search for more coherent conceptualisations of the problems involved.

MOGUL is a processing model that is devised in such a way as to engage coherently with research across a variety of domains. Following proposals by Ray Jackendoff, it involves a recognition of the existence of a separate, modular language faculty, containing the core phonological and syntactic systems. It also recognises the crucial importance of 'conceptual structure', which includes the vital semantic and pragmatic dimensions of language that, in this framework, lie outside this core and allow for the possibility of conscious introspection. In fact, it is in the conceptual domain that metalinguistic ability is anchored, allowing the language user to construct fragmentary or even quite sophisticated metagrammars that co-exist with the inaccessible grammar(s) processed inside the core language modules.