Semantic representation and natural meaning

Semantics is distinctive among the core linguistics disciplines in the general lack of agreement over the basic theoretical questions at its centre. Langacker’s (1987: 32) complaint about ‘the striking lack of consensus about the proper characterization of even the simplest or most fundamental linguistic phenomena’ applies to questions of meaning even more than it does in other linguistic subfields, and theoretical proliferation in the years that separate us from 1987 has only increased the diversity of approaches available. From one point of view, this diversity is a major source of semantics’ vitality and interest. But given most investigators’ goal – the construction of a naturalistic theory of language – the field’s inability to reach agreement on even the most basic questions is highly discouraging. In this talk I venture some suggestions about possible causes of, and repairs for, this situation. I suggest that the impasse in current semantic theorizing stems from two widely made assumptions linguists bring to the analysis of meaning. The first assumption compromises the representational naturalness of the semantic structures posited in linguistics. Many theorists claim that semantic structure and conceptual structure are identical, but do this without any serious effort to characterize non-linguistic conceptual structure independently. I suggest that research needs to bring about a closer alignment between the semantic representations assumed to underlie language, and the types of representation required to explain non-linguistic cognition. Only if this happens can the claim that semantic and conceptual structure are identical be non-vacuous, and constraints be placed on the types of possible semantic representation discerned in language. The second assumption concerns the discourse naturalness of semantic representations. Linguists’ tacit assumption that semantic structure is constant over different discourse contexts is, I suggest, implausible. Instead, it seems likely that semantic structure varies as a function of the factors (degree of attention, planning, formality, etc.) known to affect most other levels of linguistic structure. A fruitful direction for semantic research may, I suggest, entail greater attention being given to these two dimensions of naturalness.