5. The Strong Competence Hypothesis

The Strong Competence Hypothesis

The application of a grammar in a model of the cognitive processes of sentence analysis entails implicit, if not explicit appeal to some form of competence hypothesis. A competence hypothesis is a conjecture that a grammar for a language determines, or at least imposes constraints on the procedures that might be implemented by the cognitive system to analyse sentences. In effect, a competence hypothesis is a claim that the sentence structure of a language, as represented by a grammar, provides evidence of the operations performed by the cognitive system to process sentences. Although the grammar and the operations of the human processor might differ in some respects, they nonetheless license and accept the same language.

The Strong Competence Hypothesis of Bresnan and Kaplan (see Ford et al., 1982) asserts that there exists a direct correspondence between the rules of a grammar and the operations performed by the human language processor. This conjecture therefore imposes the condition that the operations comprising a procedure implemented by the cognitive system to analyse sentences must, in some sense, be equivalent to the operations required to apply the rules of the chosen grammar G.

If G is a phrase-structure grammar with rules that express word classification, constituency, and order relations, then according to the Strong Competence Hypothesis, the human processor performs operations corresponding to these relations, such as the identification of different kinds or categories of word and the recognition that words can be combined according to an accepted order into more inclusive, collective categories that form the functional units or constituents of a sentence. Since the rules of G represent relations on lexical and syntactic categories, invocation of the Strong Competence Hypothesis also entails the assumption that objects corresponding to lexical and syntactic categories must be realised in some fashion by the cognitive system. A sentence analysis procedure implemented by the human language processing system therefore consists of sequences of operations on these realisations, where the operations themselves constitute the mental equivalent of the operations required to apply the rules of the chosen grammar.

If, rather than a phrase-structure grammar, G is one of the categorial grammars, then the objects realised by the cognitive system are assumed to correspond to the atomic and functional types associated with individual words, and the operations performed by the language processing system during its analysis of a sentence correspond to functional application. If, on the other hand, the cognitive system is assumed to implement procedures based on a tree-adjoining grammar, the human language processor is conjectured to perform operations corresponding to tree substitution and adjunction on realisations of the elementary trees associated with the words of the sentences being analysed.