Generative complexity and psycholinguistics: divorce American style

In the American scientific context of the 1950s, the confluence of information theory and behavioristic views seemed to hold the promise of a truly interdisciplinary psycholinguistics. However, the prospects opened up by this confluence were soon ruined by the advent of transformational grammar (TG). For reasons detailed in this paper, such was the attraction of TG that it became the nearly exclusive source of psychological hypotheses on linguistic processing. Correlating transformational complexity with measures of performance, such as response times, set the new methodological trend. Problems quickly crept in: psychologists could not solely rely on linguistic theory to account for their data and had to make room for heuristics and biases. Evidence for transformational theory was difficult to come by and TG was evolving at a pace psychologists had difficulty sustaining. On their part, linguists were reluctant to submit their constructs to psychological testing, so that the relationship of psychology to linguistics justifiably seemed to be one-sided. In all likelihood, countering the threat posed by psycholinguistics and defending the autonomy of linguistics underpinned Chomsky’s affirmation that linguistic theory, with its own methods, lays a claim to psychological reality and does not need an auxiliary science. Divorce was inevitable, but this short-lived episode of interdisciplinary research had significant consequences both in linguistics and in psychology. These consequences are dealt with in the last part of this paper.