Formal and Functional Explanations: New Perspective on an Old Debate

As discussed by Newmeyer (1998), the debate between “formal” and “functional” approaches to explanation in linguistics has a long pedigree, and in some respects the two perspectives may seem almost irreconcilable. Here I suggest that, taking seriously certain aspects of Chomsky’s Minimalist Programme and, in particular, building on ongoing work proposing non-UGspecified, emergent parameter hierarchies (Roberts 2011, and work collected at http://www.mml.cam.ac.uk/dtal/research/recos), it becomes apparent that the old dichotomy is a false one. There is a small, irreducible formal core to Universal Grammar (Merge and a schema for formal features) which interfaces with aspects of cognition which are related to the functional aspects of language (expression/communication of thought and action). Both aspects of this “broad” design of language are required in order to account for almost any linguistic phenomenon of interest, and so the old debate dissolves simply into the question of which aspect of the overall design (form or function) is of most immediate interest for researcher; no real issue of substance hinges on the issue. I will illustrate this by arguing, following Biberauer, Holmberg, Sheehan & Roberts (2009) and Biberauer, Roberts & Sheehan (2013) that this kind of approach to cross-linguistic variation offers a suitably restrictive theory of the nature and limits of syntactic variation. My focus is one aspect of the proposed parametric hierarchies, the so-called Mafioso Effect by which certain formal parametric options are simply ‘irresistible’ for broadly functional reasons