A Psycholinguistic Model of Syntactic Complexity

Previously developed psycholinguistic notions of syntactic complexity, involving depth, degree of nesting and node/terminal node ratio are reviewed. The authors' own system, based on transformational generative grammar, is presented. This system employs a quantitative unit measurement based on the fundamental logical operations underlying phrase- and transformation-rules. Each rule is thus seen to comprise a given number of units, and any English sentence can be quantitatively " measured " in this way. Evidence is cited to support the contentions (1) that this system provides hypotheses which have been testable in experimental language performance tasks and (2) that it comes closer to common-sense notions of complexity than the three other approaches. Problems germane to both theoretical foundations and practical application of the system are listed and discussed. These include the total number of transformations, usage, depth, degree of nesting, non-syntactic variables and the question of linearity.