In search of conceptual frameworks for relating brain activity to language function

The focus of the current topic is the analysis and interpretation of second language (L2) and multilingual data. Looking at data from speakers who have learned their additional languages after the mother tongue has become well established is of special interest. It advances our knowledge about how different language systems share space in the same mind, a question to be asked of any kind of multilingual at any age—and secondly it can tell us more about potential differences between early and later learned languages (Kim et al., 1997; Kovelman et al., 2008). More recent research points to brain areas activated in late learners of L2s becoming more and more like those of L1 acquirers as their proficiency advances (Green, 2003). At earlier stages of acquisition, adults may simply adopt compensatory strategies, for example recruiting new cognitive resources that have become available with increasing maturity to complete communicative tasks that are demanding either because, unlike very young children, they personally want or are compelled by interlocutors to communicate complex ideas, or because the requirements of a given experiment simply make the tasks demanding. This will therefore implicate regions of the brain that are much less involved in young language users and these may stay involved even where higher levels of L2 ability render themmuch less important or even unnecessary. In any case, to get the full picture we need ways of tracking this strategic activity, one aspect of which is the deployment of explicit processes, both those involving conscious awareness and those that may be raised to awareness but can also operate subconsciously but there will surely be processes that only operate subconsciously as well (Sharwood Smith and Truscott, 2011). These will affect not only the spontaneous uses of L1 and L2 but also performance on experimental tasks. Tracking brain activity with sophisticated apparatus is not enough of course: the data needs to be analyzed and for this we need very sophisticated theoretical frameworks to guide interpretation. While research techniques such as brain imaging are gradually acquiring greater precision, helping to reveal much more about brain activity associated with linguistic processing, there still remain many problems interpreting results. This may not be an immediate problem in a given experiment because the research question may be suitably precise and focussed enough to guarantee an answer of sorts in the hope that answers to limited questions may gradually accumulate and provide the basis for wider explanations. In this way, for example, syntactic and semantic processing can be teased apart on the basis of participants’ differing responses to examples of, respectively, syntactic and semantic anomaly which then allows researchers to identify separate neural signatures and provide support for particular accounts of the status of language vis-a-vis other types of cognition. Issues of interpretation become more evident when trying to put results into a wider explanatory context. One problem concerns the choice of which theory and which concepts and categories to import from a neighboring research domain. Another one, related to that, is locating conceptual models and frameworks in related domains (neurolinguistic, psycholinguistic, theoretical linguistic) that can be combined in such a way as to promise the best possible explanation. Assuming the focus is on explaining language, and leaving aside sociolinguistic issues, if we get down to the basics, what do we have? 1.3 kilos (three pounds) of soft tissue and our current understanding of its functional architecture. Add to that theories about psychological function and, in many cases at least, entirely separate, well developed theories about linguistic structure. Each of these theories has, for very good reasons, its own conceptual framework and terminology, and its own favorite methods of investigation. For satisfactory explanations of how the brain stores and processes languages, we need somehow to coordinate findings in all these different disciplines. At the same time, it is not a straightforward job to bootstrap, for example, a Minimalist approach to explaining language structure to a model of human memory and make it into a real-time processing theory. This is true notwithstanding the obvious need, in the elaboration of theories of processing and development, for fine-grained accounts of linguistic structure. Standard generative linguistic approaches to language employ terms and concepts to explain abstract linguistic structure that are outside time and space. Without going into the details, these are notions like “move,” a structure-building operation changing the position of some item in a structure, “merge,” a combining operation, and “feature-checking,” the process whereby two associated items in a structure are assessed to see if they can be “licensed,” i.e., co-exist in their current position (in