Lambek vs. Lambek: functorial vector space semantics and string diagrams for Lambek calculus (Q388206)

This paper presents the authors' distributional compositional categorical model (abbreviated DisCoCat) of meaning [``A compositional distributional model of meaning'', in: Proceedings of the second quantum interaction symposium. College Press. 133--140 (2008); ``Mathematical foundations for a compositional distributional model of meaning'', Linguist. Anal. 36, No. 1--4, 345--384 (2010)]. ``The DisCoCat exploits the fact that finite dimensional vector spaces can also be organised within a compact closed category.'' The main technical objective in this paper is to emphasize that the choice of a compact or monoidal type-logic is not so crucial to the applicability of the procedure. In this paper the authors develop a homographic passage via a functor from a monoidal bi-closed category of grammatical types and reductions to the symmetric monoidal category of finite-dimensional vector spaces, inspiring an intriguing analogy with topological quantum field theory (abbreviated TQFT), which is also a monoidal functor from the category of cobordisms to that of vector spaces and linear maps. Different from [\textit{R. Montague}, Theoria 36 (1970), 373--398 (1971; Zbl 0243.02002)], meanings of words and sentences are expressed in terms of vectors and vector compositions rather than in terms of sets and set-theoretic operations. Information flows, which are the topology of the two-dimensional graphical representation of the operations producing the meaning of sentences from that of words, are mathematically expressed in the graphical language [\textit{P. Selinger}, Lect. Notes Phys. 813, 289--355 (2011; Zbl 1217.18002)]. Such graphical representations trace back to \textit{R. Penrose} [Combinat. Math. Appl., Proc. Conf. Math. Inst., Oxford 1969, 221--244 (1971; Zbl 0216.43502)] in the early 1970s. The ambition of the authors is to use this work as a starting point for providing vector space meaning for more expressive natural language sentences parsed with combinatorial categorical grammars or Lambek-Grishin calculus [\textit{R. Bernardi} and \textit{M. Moortgat}, Inf. Comput. 208, No. 5, 397--416 (2010; Zbl 1198.03027)]. The expressive power of these grammars is richer than that of Lambek grammars, which are only context-free. The starting point of the authors' approach is a Lambek calculus [\textit{J. Lambek}, Am. Math. Mon. 65, 154--170 (1958; Zbl 0080.00702)], in place of Lambek's pregroups [\textit{J. Lambek}, Lect. Notes Comput. Sci. 1582, 1--27 (1999; Zbl 0934.03043)], with a variant of proof theory of intuitionistic multiplicative linear logic. The grammatical structure of a sentence is represented as a derivation in a proof tree, which is depicted in diagrams similar to proof nets. These derivations are interpreted in vector spaces, regarded as a monoidal closed category, and the grammatical and semantical interactions are depicted via Baez-Stay diagrams [\textit{J. C. Baez} and \textit{M. Stay}, Lect. Notes Phys. 813, 95--172 (2011; Zbl 1218.81008)]. Although general linear logic proof nets are quite distinct from the Baez-Stay diagrams, their compact variants in [\textit{S. Abramsky} and \textit{R. Duncan}, Math. Struct. Comput. Sci. 16, No. 3, 469--489 (2006; Zbl 1099.03059)] highly resemble the compact closed string diagrams used in the pregroup derivations.