Page 246 - Encyclopedia of Philosophy of Language

Reference
not to t, (the circumstance of evaluation) but to to, the context of utterance. In other words, the indexical takes wide scope over the operator. The realization of this point was an essential ingredient of Kaplan's Logic of Demonstratives formulated soon after.
The standard model-theoretic treatment of index- icals was not without its critics; for example, Cresswell (1973) argued against the parametrization of con- textual features. He holds that an indefinite number of contextual aspects can be relevant, and that we cannot tell in advance precisely which one will be needed: should we postulate a 'previous drink par- ameter' for sentences like 'Just fetch your Jim another quart' ? Instead of postulating a fixed list of contextual parameters beforehand, we should speak in terms of the properties a context may have or not have; for example, a context may have the property that John is a speaker there. This, he claims, allows for a more flexible and natural treatment of context-dependence.
3.2 Kaplan's Theory of Direct Reference
David Kaplan's theory (most fully stated in the semi- nal 'Demonstratives: an essay on the semantics, logic, metaphysics, and epistemology of demonstratives and other indexicals' 1977, published in Almog, et al. 1989) is widely accepted as the standard treatment; it also stands at the basis of several important new treatments of related problems.
3.2.1 Basic Notions
The single most important aspect of Kaplan's theory, which also distinguishes him from all his predecessors, is that he calls all indexicals directly referential. By this he means that the linguistic rules for indexicals directly provide us with a referent for each context of utterance (e.g., the time of speaking), without the mediation of a Fregean sense: they do not yield an intension that still needs to be evaluated at a cir- cumstance (i.e., a possible world-time pair) in order to yield an extension.
The term 'direct reference' is likely to cause confusion. First, it does not mean that indexicals do not have a conventional meaning or a descriptive con- tent. The conventional meaning of / gives us the speaker for every context; but as Kaplan stresses, its extension is not something to be settled by the cir- cumstances of evaluation: rather, given a context of use, a directly referential term will have its content determined by that context, whatever the cir- cumstances may be. Secondly, it does not mean that the referent (a physical object) is part of the abstract proposition, but only that the object referred to deter- mines what the proposition expressed is, rather than the other way round; put differently (and against Frege), there is a road back from reference to sense. Thirdly, it does not imply that we must know the referent of an indexical for it to refer directly. In his work 'The logic of demonstratives' (1979), however,
Kaplan formally treats indexicals as expressions with a constant intension, which implies that there is an individual concept (albeit a stable one) involved instead of an individual; but he really intends them to be simply independent of circumstances (cf. Almog, etal. 1989:497,507).
Kaplan distinguishes two aspects of meaning for indexicals: character and content. Content cor- responds to Montague's intension: it is a function from circumstances to extensions. A content is fixed or stable if it is a constant function: for example, proper names have a stable content. Character (which resembles Perry's notion of 'role') is the function that determines the content in different contexts of utter- ance; it captures what we informally call the 'meaning' of indexicals (in fact, in Kaplan's logic, sameness of character is the closest approximation of true synonymy, i.e., sameness of meaning). Nonindexical terms have a stable character, but their content may vary; indexicals have a nonstable character, but a stable content: given a context of use, they refer to the same individual in every circumstance.
Reichenbach and Russell, he claims, did not acknowledge the directly referential character of indexicals, and therefore mistakenly think that the rule identifying / and 'the person uttering this token' captures the meaning of the indexical, whereas in fact thedescriptionmerelyfixesthereferent ofthedirectly referential term. Descriptions depend on cir- cumstances for their interpretation and are therefore not directly referential. Against Russell, Kaplan also argues that wecannot completely eliminate indexicals, because we will never have enough names available to refer to every possible object under the right character.
Kaplan also distinguishes pure indexicals, i.e., expressions such as / or now, the referent of which is entirely determined by linguistic rules, from demon- stratives proper, expressions that require an act of— usually nonlinguistic—demonstration, like pointing, in order to determine their referent or demonstratum. One object can be demonstrated in different manners, so a demonstration may, in Fregean terms, be seen as a manner of presenting a demonstratum.
Kaplan criticizes Montague's treatment, which con- flates context and circumstance into a single index, and thus blurs the distinction between context- dependent and context-free expressions. This approach, he holds, cannot adequately deal with
I am here now. (3)
This is true, but not necessarily true: in other cir- cumstances, I could have been elsewhere. However, in a sense, (3) cannot be uttered falsely. We cannot adequately account for this unless we strictly separate context and circumstance, and add the rule that index- icals refer directly. (Here, Kaplan does not do full justice to Montague (cf. Bennett 1978:3): the latter would not accept if p_ is logically valid, then so is
224