On the Computational Character of Semantic Structures

Introduction

The exact nature of semantic interpretation as a property rather than as a psychologically grounded process is still barely understood or grasped despite the fact that we use language everyday and understand an enormous number of linguistic expressions with no a priori bound. This paper aims to make sense of the nature and form of semantic interpretation by tracking the differences between logical form, which is used in logic and Logical Form (LF), which is a part of the Minimalist architecture of language within Generative Grammar. Throughout the entire paper, the phrase ‘logical form’ will be used to signify the logical representation and LF will denote the syntactic component in Generative Grammar. Both logical form and LF represent aspects of semantic structures of natural language. But what is it about semantic structures that they can be represented as such by logical form or LF? Note that logical form is a metalanguage that can be employed to express properties of semantic interpretation, while LF is a syntactic component within the architecture of the language faculty. When we say that LF represents aspects of meaning, we mean that LF is a syntactic system which is interpreted semantically. Within Generative Grammar, the model that interprets LF objects is a mental organization called ‘Conceptual-Intentional (C-I) system’, which interfaces with LF, while the model for logic is an abstract model in which interpretations of logical forms are couched. On another view, LF represents what may be called structured meanings in the sense of Cresswell (1985). Structured meanings derive from the meanings of the component expressions and from the meaning of the whole structure the component expressions are components of. Suffice it to say, these formal representations express properties of natural language meaning, and so they can uncover, one may believe, much about the form of semantic representation. There are well-known differences between logical form and LF many of which have already been noted (Bach, 1989; Heim & Kratzer, 2002).The significant question for us is whether semantics or meaning in language can be computational in its character, given that syntax is generally believed to be computational in Generative Grammar, and that descriptive generalizations about semantic facts can be made with reference to both logical form and LF. The important proviso to be made is that the aim is certainly not to merely zoom in on differences between logical and LF. Rather, the aim is to understand how and in what ways logical and LF can tell us something about the ontological status of semantic interpretation. Given this goal, much of what will follow in essence derives from Mondal (2013).

With this in mind, first an introductory sketch of what logical form and LF are will be drawn up. Then a range of pertinent but unexplored differences between logical form and LF that may have serious consequences for the ontology of semantic interpretation will be highlighted.