FFF CONFERENCE CTF07

Timothy Pritchard - The Character of Word Meaning

Do we mean the meaning of our words? Is word meaning the sort of thing that enters as a constituent of expressed content? A default assumption, both on denotational and conceptual approaches to word meanings, is that a word meaning can in principle provide a constituent of what is expressed. The content expressed is (in part) built out of the encoded semantic values provided by individual words. I shall suggest that linguistic content may better be represented as an intermediate construct, a basis from which expressed content is constructed but not itself a constituent of that content. I call this posited basis a semantic base. Motivations for this alternative account:

   1. Intuition that word meanings are ‘light’, allowing great flexibility of construal.

   2. Pronouns already provide a clear model for treating linguistic content as an intermediate construct.

   3. Philosophical puzzle over the nature of a ‘general idea’ – Locke/Berkeley/Hume/ Kant. Arguably there is no ‘general idea’ of triangle unless we think in terms of an underlying rule which enables us to construct triangles. This suggests a distinction between a constructed ad hoc concept ‘triangle’ that is relevant to an expressed content in a particular use, and the information encoded by the word ‘triangle’ which gives rise to this construction.

   4. Theoretical need to find something that is constant across contexts. If word meanings are concepts it may be difficult to isolate anything that is constant (cf. Relevance Theory loosening/enriching etc).

   5. The behaviour of ‘is true’ in natural language – it is only used of expressed content (with sentences in clause construction) not of sentences per se (suggesting that content directly relevant to truth is not encoded).

   Suggestion for how we might represent the constraint encoded by ‘river’.   Think of a highly schematic picture: a contoured surface, a band of relatively homogenous content moving across this surface within stable boundaries. This is meant to represent (as an idealized model) the semantic base. It represents the constraints under which we use the word ‘river’.   1. The constraint does not represent the content of what we express when we use ‘river’. It is though a necessary condition. There is no ‘feature cancellation’.   2. The constraint is specific enough to isolate an aspect of the environment such that an audience knows what we are speaking of (relatively few things fit the constraint).   3. The idealization overgenerates (e.g. streams, conveyor belts …), but apparent ‘error’ may also in part reflect merely contingent linguistic decisions. Sometimes (e.g. rivers of mud, rivers of people) I believe we can treat these as bona fide literal uses of ‘river’. Appeal to ‘figurative’ or ‘metaphorical’ meanings may often be artefacts of ‘richer’ theories of semantic content.   4. A clear separation of linguistic content from encyclopaedic knowledge is achieved.Two putative problems for the proposed account will be addressed:   1. The concept of context and whether it is uniform across word/pronoun meanings    2. Whether word-meanings are a distinct type, or merely watered-down concepts.