A reaction to Jackendoff’s discussion note

It was with great pleasure that I read Ray Jackendoff’s discussion note ‘What is the human language faculty? Two views’, published in Language 87.3.586–624 (September 2011). Since it was not presented as an ordinary article but as a ‘discussion note’, it seemed appropriate to ask the editors of Language to print a short reaction, meant to make a positive contribution to the discussion.

I do not go into all of the details of Jackendoff’s analyses and arguments, with many of which I am happy to agree, but there is one basic aspect that I feel calls for comment, mainly for the sake of clarity. Jackendoff, correctly in my view, argues against Chomskyan minimalism, and especially against Merge as a basic operation in that theory. Yet something odd appears to have occurred in Jackendoff’s treatment of the issue. On p. 599 he correctly points out that in orthodox Chomskyanism a generative grammar was never explicitly presented as a realist production model. Remembering the old days, he writes: ‘When I was a graduate student in the 1960s, we were always taught that this has nothing to do with processing’—using the word processing but meaning production processing, as the text makes clear. Rather, the word was, a generative grammar should be looked upon as an ‘abstract’ characterization of speakers’ competence, much in the way mathematicians characterize mathematical notions. Jackendoff draws a parallel (p. 599) with the inductive definition of rational numbers, based on Peano’s axioms, as known in mathematics. Just as numbers are enumerated by an inductive system starting from a base, sentences are ‘inductively’ (we normally say ‘algorithmically’) enumerated by a grammatical rule system starting from an initial symbol. Here lies the origin of generative grammar, not in the empirical study of how humans produce and interpret utterances. Linguistic reality was thus relegated to mathematical reality, not to psychological, physiological, and social reality, which is where it belongs.

In fact, it would be absurd to claim realism for the various Chomskyan grammar models we have seen over the years, including minimalism. And Jackendoff rightly poses the question of the actual reality, if any, behind these models. Put bluntly, we want to know what the physiological correlates are of what we know as linguistic competence so that we may gain some understanding of how competence drives performance. There has been a great deal of beating about the bush with regard to this question, but Jackendoff takes the bull by the horns and asks (p. 599): ‘What does this inductive definition have to do with how the brain actually processes language?’ (emphasis mine).

Unfortunately, as Jackendoff will no doubt agree, it is proving very hard to make the brain yield its secrets. Linguists and psychologists, moreover, are on the whole not brain specialists, nor do brain specialists tend to have much linguistic or psychological knowledge, leaving aside a few precious exceptions. But even these have great difficulty tracing possible correlates between conscious actions and impressions on the one hand and brain processes on the other, let alone between psychologically or linguistically defined rule systems and brain structures. At the present stage, any quest for physiological correlates of language production and comprehension processes is bound to be seriously hamstrung, despite the sophisticated scanning machines now at our disposal. As a result, linguists and, to a certain extent, psychologists seek refuge in what I have called (Seuren 2009:9) nonhardware realism—that is, analyses and descriptions in terms that look as if they stand a better chance of being physiologically implemented than those formulated in terms taken from mathematical proof-theory. Clearly, there are no hard and fast standards for degrees of physiological implementability, but one can, to begin with, try to formulate overarching architectural principles of the system envisaged in terms that conform to what is known from direct intuitive observation and experience or as a result of brain research. The extent to which enlightening, explanatory generalizations are found and formulated in terms of a given architecture will then be a sensible measure of the possible reality value of one’s enterprise. In...