CRISSP Seminar: Jan Koster

CRISSP is happy to announce another installment in the CRISSP Seminar series:

Lecturer: Jan Koster (University of Groningen)

Title: The revolution reconsidered

Date & time: 28 March 2011; 17.00-18.30

Location: CRISSP/Hogeschool-Universiteit Brussel, Stormstraat 2, room 5212

Participation: free


The revolution reconsidered

It is commonly believed that linguistics underwent a revolution in the 1950s. In retrospect, that seems more questionable than was realized at the time. The development of transformational-generative grammar between 1955 and the 1970s was not a gradual expansion of revolutionary conceptions but a somewhat disguised retreat to some of the major tenets of certain forms of pre-revolutionary structuralism. At the time, nevertheless, the revolution was believed to involve both perpetual technical innovation and a new metatheoretical perspective.
The technical innovations were largely based on the practices provided by the study of formal languages and their recursive definitions. According to this perspective, a language L can be seen as a set of sentences generated by some algorithm. The algorithms consisted of the “rules of formation” and the “rules of transformation” reminiscent of the work of Carnap (1937). The empirical rationale for this kind of dual model was derived from Zellig Harris’s notion of normalization, also known as “reconstruction”, which was the basis for “transformations” and the related notion of multi-level theories (with “reconstruction” levels like “deep structure”, next to “surface structure”).

The analogy with formal languages was highly misleading, and in the long run counterproductive, as was also reflected by the less and less prominent role of mathematical linguistics after the 1970s. In retrospect, transformational-generative grammar did not survive the development of a lexicon since Aspects (1965). The ensuing X-bar theory was, conceptually, an almost complete return to the “Wortgruppenlehr” of the older (pre-)structuralists, also because it became increasingly clear that the arguments for multi-level theories were extremely weak. Structure-preservingness (Emonds 1970) had suggested that grammars were “monostratal” after all. The more general use of empty categories since the late 1960s and the related use of anaphoric strategies completely undermined the multistratal conception of grammar. It is studies like Higgins (1973) that, in my opinion, gave the death blow to the Harrisian reconstruction paradigm and to transformational grammar.

I will argue that the idea that there was a “second revolution” after the early demise of transformational grammar is a myth and that Chomsky’s attempts to save the revolution with transformational residues like “move alpha” or minimalist “internal merge” have been a failure. In fact, the very notion of sentence generation by “merge” is an ill-advised return to the pre-lexicalist models of grammar that were rejected in the 1970s (on the basis of redundancy arguments).

At the metatheoretical level, the allegedly revolutionary paradigm has been even less convincing. The essence of the shift in the 1950s was the transition of a Saussurian sign-based linguistics to a syntax-based linguistics, in which the grammar was interpreted as a psychological “reality” or even as a language organ. This has led to the current fad of “biolinguistics”. Apart from the trivial fact that all our capacities are biologically-based, analogies between grammars and organs or between our conceptual system and the immune system, are hopelessly misguided and based on the “panglossian” error of mixing up the non-agentive functionality of biological structures and the agentive functionality of the intentional objects of human culture. There is no language without the signs found in the record of human agency and, therefore, the Saussurian conception of language was not so bad after all. No matter how narrowly construed, language is not biology but applied biology at best.