How can the developments in linguistics benefit neurolinguistics? An Instrumentalist approach

Chia-Hua LIN, University of South Carolina

There is an ongoing interdisciplinary debate over the question of combining linguistics and neuroscience for the purpose of understanding the biological basis of the human faculty of language. On the one hand, a case that supports such an interdisciplinary marriage can be made for the sake of practical efficiency. On the other hand, it has been argued that the effort to locate the linguistic units of study in the brain, such as ‘the syntax,’ is leading towards hasty or even false conclusions. In this paper, coming from the philosophy of science perspective, I argue that it is not the application of linguistic units in neuroscientific investigations that leads to unwarranted conclusions. The culprit is, rather, the way in which they are applied. Using the example of Hauser et al.’s proposed research schema, which they base on the developments in the minimalist program in linguistics, I discuss how a linguistic model may be used as an essential tool to advance neurolinguistics.

The faculty of language, or ‘FL’, is originally hypothesized to account for the human ability of acquiring and using languages. Traditional research questions regarding such a faculty concern the nature, the neural substrate, and the evolution of the faculty. For instance, what makes up the FL? How does the brain implement the FL? How has the FL evolved in the species? These three questions are closely related. A proper understanding of what constitutes the FL is necessary for a productive investigation of its neurophysiological underpinnings. And a proper understanding of what underlies the FL is, in turn, vital to a suitable evaluation of its phylogenetic history.

However, because of the joint breadth of these three topics, it may appear unrealistic for an individual discipline, say, linguistics, biology, or neuroscience to tackle all three of the inquiries independently. In fact, interdisciplinary collaboration appears to be desirable in the name of efficiency. Linguists have generated a body of literature on the nature of the FL, neuroscientists have gained a wealth of knowledge on how the brain works in giving rise to other cognitive faculties, and finally, biologists have accumulated a great deal of insights on how a mechanism may have evolved. It would be reinventing the wheel for, say, neuroscientists or biologists to start their research by defining what language is. Arguments along these lines can be found in Hauser, Chomsky, and Fitch (2002)[1] and Fitch (2010)[2].   Their suggestion is to adapt core ideas of the minimalist program in linguistics (such as generative grammar, discrete infinity, and recursion) to the broader research areas of language including not only neuroscience and evolutionary biology but also psychology and anthropology.

Although not directly challenging the use of the minimalist program, some prominent linguists and neurolinguists’ work is in conflict with Hauser et al.’s view. For example, Poeppel and Embick[3] point out that the fundamental elements in theories of linguistics, such as morphemes, syllables, and noun-phrases, are defined for the purpose of analyzing and formalizing structures in specific languages. In contrast, the fundamental elements in neurobiology, such as dendrites, neurons, and cortical columns, are defined based on their anatomical features. They argue that such linguistic and neurobiological elements are incommensurable.  Moreover, there has been no empirical evidence for a direct mapping between linguistic and neurobiological elements, but only some suggestive correlations. Hence, they conclude, “there is significant danger of (long-term) interdisciplinary cross-sterilization rather than cross-fertilization between linguistics and neurobiology, or, for that matter, linguistics and other empirical disciplines.”[4] Moving forward, Poeppel and Embick call for “a substantive alternative research program, by which neuroscientists can explore how brain mechanisms form the basis for linguistic computation.

In this paper, I extend Poeppel and Embick’s discussion on the comparability of the elements across the disciplines to the minimalist program of Hauser et al. I argue that the incomparability of elements across disciplines becomes a problem when one attempts to create a directly map between the linguistic and neurobiological elements. However, such a direct mapping reductionism is required neither for the minimalist program nor for Hauser et al.’s proposed research schema. More importantly, reductionism is not necessarily based on such ‘maps.’ Therefore, what is hindering progress in neurolinguistics is not the linguistic model per se but how it is used.

To demonstrate how the elements in linguistics may be essential to investigations in empirical sciences, I first show that the Hauser et al.’s proposed research schema provides a conceptual tool or a so-called ‘what-description’ which specifies what is constitutively necessary—in terms of functional elements and structures among these elements—for a linguistic capacity to be realized. The alternative called for by Poeppel and Embick, in contrast, would be a ‘how-description’ which specifies how the linguistic capacity in question is implemented—in terms of neurobiological elements. Under this distinction, I argue that Hauser et al.’s schema and the alternative suggested by Poeppel and Embick are compatible. Then, I point out that due to the variety of neuroimaging techniques, current research is producing many interesting, but not necessarily showing a clearly converging picture. Thus, I argue, a conceptual tool such as Hauser et al.’s proposed research schema is essential to neuroscientific investigation, as it provides a framework that helps one to see convergence where it may not have been visible before. Indeed, to evaluate and compare discoveries from multiple disciplines, a what-description, such as Hauser et al.’s schema, is indispensable because it provides a common reference to compare variety of how-descriptions side-by-side. Thus, I will conclude that what is needed may not be a neurolinguistic alternative to the linguistic models but an alternative, non-reductionist way of using them.