ABSTRACT
An attractive - if not always clearly articulated - conception of the interest of truth-theory as an instrument in the study of the semantics of natural language may be formulated roughly as follows. We begin by stating and motivating a certain general requirement which an adequate interpretative truth-theory for a given natural language must meet; viz. that the theory should, upon the basis of a finite number of axioms, each treating of a seman tical primitive of the language, conform to a version of Tarski's Convention T with respect to each of the infmitely many sentences of the language. The thought then is that conformity to the general constraint will perforce reveal significant semantic properties of the expressions of the object-language; i.e., the kinds of axioms needed to fulfil these abstractly stated requirements will, almost algorithmically, determine the semantic category, ontology and contribution to truth-conditions of seman tical primitives and compounds thereof. Given the resources of the theory and the goal we set for it, a detailed semantics will be determined; we have only to contrive it. l The aim of the present essay is to test the plausibility of this idea against the issues raised for truth-theory by a class of grammatical modifiers to be found in English. In the current state of the art conclusions must be tentatively reached, and my remarks will be exploratory rather than definitive.
