La preuve de Gaifman: Réflexions sur la méthode de construction des grammaires catégorielles
In 1959, Gaifman proved that category grammars were the equivalents of phrasal grammars, so much so that they could never have the expressive power for representing the complexity of natural languages. As a consequence, no new category model was proposed for more than thirty years. Nevertheless, we are now witnessing a renaissance of the category perspective, which makes up the main syntactic current of study in formal linguistics. But the problem, which this article proposes to examine, is that contemporary category grammars continue to be subject to the limitations of Gaifman’s proof, since the increasingly complex formalisms that underlie them paradoxically elaborate according to the Lambeck syntactic calculus (1958) whose expressive power is limited to that of context-free grammars. What sense is there in refining a model that has at its core a theory that is inappropriate for representing natural language?