I end by briefly answering our first two questions, and then lingering a bit on the third.
Indexicals are like definite descriptions in that they denote; they are like names in that they refer.
Indexicals differ from other shifters in the role that context plays. In the case of indexicality, context does not affect designation by providing evidence for what word is being used, with what meaning. Context plays its role after the words, syntax and meanings are all fixed, for in the case of indexicals meaning determines content relative to contextual factors.
This point invites an important distinction, between reflexivity and indexicality.
Return for a moment to our example of the note saying ``Cet homme est brillant,'' with which I introduced the concept of relative truth-conditions. I'll alter the note slightly, to get in a definite description: ``Cet homme est l'homme le plus brillant dans cette salle''. We could construct a whole hierarchy of relative truth-conditions for such a message, of the form, given that such and such, m is true iff so and so:
) Given that m is in French, the words are
``Cet homme est l'homme le plus brillant dans cette
salle'', that in French these words mean that the man the speaker
directly intends to refer to is the most brillant man in the room, m
is true iff there is a man the speaker of m directly intends
to refers to and that man is the most brillant man in the room.
) Given that m is in French, the words are
``Cet homme est l'homme le plus brillant dans cette
salle,'' that in French these words mean that the man the speaker
directly intends to refer to is the most brillant man in the room, and
that the speaker of m directly intends to refer to Henri, m
is true iff Henri is the most brilliant man in the room.
) Given that m is in French, the words are
``Cet homme est l'homme le plus brillant dans cette
salle,'' that in French these words mean that the man the speaker
directly intends to refer to is the most brilliant man in the room, and
that the speaker of m directly intends to refer to Henri, and given
that the most brilliant man in the room is Jacques, m is true
iff Henri is Jacques.
We have reflexivity at point m. At m we get content ; the
meaning is fixed, but not the context and other facts relevant to
designation and truth. That is, even given the meaning, we need context
to get official content. That is indexicality.
But we have reflexivity at every stage up to and including m. That is, the truth-conditions, given what has been fixed, are still conditions on the utterance itself. That is reflexivity. Indexicality is, one might say, simply the highest form of reflexivity, reflexivity exploited by meaning.
Now the relative concepts of truth-conditions at each of the stages lower than m--the reflexive but pre-indexical stages--can give rise to a species of content, and all of these kinds of content can be put to good use in the epistemology of language. The epistemology of language is not just a matter of understanding how people who know all there is to know about the language in which a given utteranace is couched go on from that point. It needs also to deal with how languages are recognized and learned, how new words are learned, how poorly pronounced or indistinctly heard words are recognized, how ambiguities are resolved and the like. In all of these inquiries, the proper kinds of content to represent the knowledge of the agent are reflexive.
One often hears that indexicality is pervasive, that practially every bit of language has a hidden indexicality. This is not quite right. Indexicality is widespread, but much of what passes for discoveries of new instances of indexicality are actually discoveries about the utility of reflexive content at a pre-indexical level in understanding how we understand language. The importance of indexicality is really that, as the highest form of reflexivity, it is the gateway to the riches of reflexivity.