Reference and Representation

Research on the semantics of inter-sentential anaphoric relationships in the eighties has led to a debate about the question whether some level of representation is indispensable in the analysis of the semantics of natural language. Within the logically oriented or model-theoretic tradition of natural language semantics, Hans Kamp is a prominent proponent of the thesis that it is; Jeroen Groenendijk and Martin Stokhof have explicitly argued against it. Interestingly, the two nowadays agree on this issue: that some level of representation is indispensable in modeling the interpretation of natural language. But what about the semantics of natural language? In traditional theories of meaning, the question what is the meaning of a sentence or discourse has been kept separate from the question how people actually assign meanings to texts when they interpret them. With the growing attention for the (inter-sentential) structure of discourse, this distinction has turned out to be not unobjectionable. Standard conceptions of what is the meaning of a sentence have been seen to be inadequate if the aim is to assign meanings to (structured) discourses in a compositional way. For instance, truth-conditions, or sets of possible worlds, have been seen to be insuuciently ne-grained means to model the semantics of inter-sentential anaphoric relationships, and therefore other, richer, notions of meaning have been developed for use in more elaborate frameworks of discourse interpretation. Of course, if additional means are called for to account for certain phenomena, they can, in principle, be assigned to the semantics proper, or to interpretation proper. Thus, e.g., some versions of situation semantics can be conceived of as locating the additional structure which is required to deal with anaphoric relationships in the semantics (Gawron and Peters 1990; Milward 1996). Alternatively, Cooper's E-type pronoun theory (cf. Evans 1977; Cooper 1979) assigns the additionally required means or processes to the area of interpretation (or to pragmatics, if you want). (Heim 1990 presents a combined E-type and situation theoretic account, one in which the distinction between meaning and interpretation can be left intact.) Yet, in quite an important branch of nowadays semantic theory meaning and interpretation have grown towards each other. adopt some sort of computational meaning assignment, according to which the interpretation of a sentence is conceived of as a process. In discourse representation theory (DRT, henceforth), interpretation involves operations on discourse representation structures (DRSs), in le change semantics (FCS) operations on les, and in systems of …