Exploring Meaning Composition: The Epistemology of Modeling Practices in Formal Semantics

Abstract

The present thesis attempts to offer an epistemological account of the modeling practices involved in linguistic model-theoretic truth-conditional compositional semantics. That is, it tries to answer the question of how formal semanticists can draw inferences when building and using scientific models, so knowledge or a deeper understanding of natural language can be gained. The methodological approach followed in this paper is to first examine case studies and then characterize the modeling practices in question by applying relevant insights from the contemporary philosophy of science literature which is focused in scientific practice and scientific modeling. Accordingly, an assessment of the inferences that semanticists can draw within models, between models, and from model to theory or model to data is provided. Finally, the following two claims are defended: that the modeling practices involved in formal semantics are exploratory in nature, and that the construction and application of semantic models license hypothetical inferences which provide modal knowledge about the syntax-semantics interface.