Should pregroup grammars be adorned with additional operations? (Q2467222): Difference between revisions

The author of this article initiated, 50 years ago, in the categorial tradition of Ajdukiewicz and Bar-Hillel, an interesting viewpoint in the mathematical approach to language, the calculus of syntactic types, opening in this way a new chapter at the interface of logic, linguistics, algebra and computation. Along the years, he published a long series of papers enriching the linguistic relevance and the mathematical accuracy of his approach. Ten years ago, he proposed a new computational algebraic approach to language, based on the notion of `pregroup', a partially ordered monoid with two additional unary operations, `left adjoint' and `right adjoint'. The main idea is to work with the free pregroup generated by a partially ordered set of some basic types. This approach was challenged by another one, proposing to return to residuated monoids or even residuated semigroups, with two binary operations of division, claimed to be better than two unary operations of adjoining. More recently, \textit{M. Moortgat} [``Labelled deduction in the composition of form and meaning'', in: H. J. Ohlbach, et al. (eds.), Logic, language and reasoning. Essays in honour of Dov Gabbay. Dordrecht: Kluwer Academic Publishers. Trends Log. Stud. Log. Libr. 5, 377--399 (1999; Zbl 0968.03033)] proposed to supplement the pregroup approach with some additional operations. The present paper is Lambek's reaction to this claim; clear reasons are proposed for resisting the introduction of additional operations. Three appendices are concerned with a historical look, Capulet semantics and other look at crossed dependencies in Dutch. Let us mention also the recent book by \textit{J. Lambek} [From word to sentence: a computational algebraic approach to grammars. Milano: Polimetrica (2008)].