Text

theses

What I communicate with ‘Give me that’, pointing at a book, differs from what I communicate by it pointing at a cup. Your actions and my expectations likewise differ in these two cases. At the same time, the referent of any particular use of ‘that’ is typically unambiguous and recovered effortlessly. What determines the referent of ‘that’ and which resources permit us to recover it so easily? Though everyone agrees that what we can communicate is constrained by grammar, most believe that the role of grammar is very limited, and that interpretation largely relies on general reasoning about the communicative situations and intentions of the speaker. Philosophers frequently identify such (purported) context-sensitivity within philosophically interesting expressions like ‘know’, or ‘good’, and appeal to it to shed light on problems involving the concepts these expressions denote; but they also assume that these expressions get their values in context by way of common-sense reasoning about speaker intentions. This has lead to many radical conclusions. To take just one concrete example, the behavior of context-sensitive expressions describing possibilities and necessity (‘must’ and ‘might’) prima facie gives rise to apparent failures of classical patterns of inference, like Modus Ponens, which has led many philosophers throughout the history of analytic philosophy, operating under the traditional assumptions about the resolution of context-sensitivity, to conclude that there is a deep incompatibility between classical logic and natural language. Against this tradition, I argue, drawing on resources from philosophy, linguistics, mathematical logic and computer science, that the reason we can interpret context-sensitivity so effortlessly is because grammar itself is much more subtle and pervasive than has been assumed, and that resolution of context-sensitivity is entirely a matter of linguistic convention. Thus, linguistic rules render a particular object prominent in a given context, and this is what determines what ‘that’ picks out in that context. Moreover, in recovering this referent it is this narrow set of linguistic cues that we exploit. The conventions that I argue govern the resolution of context-sensitivity have gone unnoticed because their principal domains are entire discourses and not just their constituent words and sentences. While it is universally accepted that the way individual sentences are constructed depends on conventions of syntax and semantics, which specify the rules by which individual expressions combine, I argue for rules–discourse conventions–that specify how individual sentences combine to form a discourse. These conventions govern how speakers organize utterances into larger units that address particular topics and answer questions about them; I argue these rules determine the resolution of context-sensitivity. The move to investigating discourses has far reaching consequences: I show that a host of contextualist arguments that resort to context-dependence rest on a flawed conception context, particularly in the debate about the relation between the natural language and logic, and argue that context-dependence will have to be invoked quite differently than has become customary.

ETD_7742

Ph.D.

Includes bibliographical references

by Una Stojnic

doi:10.7282/T3X069BZ

ETD doctoral

The author owns the copyright to this work.