Thus far, a number of our interviews have alluded to what philosophers of language call indexical expressions. (In particular, episodes 12, 25, and 27.) Although the theory of what these words and phrases mean represents one of the major developments in philosophy over the past fifty or so years, it can seem counterintuitive at first. Since the questions these words raise are unusually subtle and difficult, we thought it might be worthwhile to go over some of the basics.

What’s an indexical expression? The term itself, originally coined by Bertrand Russell, is perhaps a bit unwieldy. But intuitively, an indexcial can be thought of as the kind of word or phrase that you typically make sense of by being around when the relevant person says it. If you’re visiting me for breakfast and I say, ‘It’s sunny today,’ part of what helps you figure out that ‘today’ means ‘Monday’–in addition to your knowledge of what day it is–is the fact that you’re there, which means that you know when I said it. If I say, ‘Please place the sofa over here,’ part of what helps you figure out that ‘here’ means ‘three feet to the left of the radiator’ is the fact that you’re with me in the apartment, and thus that you can see where I’m pointing. And if I say, ‘I am feeling tired,’ part of what helps you figure out that ‘I’ means ‘Matt’ is, once again, the fact that you’re there, and thus that you can tell who is speaking. Further examples of such words include now, there, yesterday, this, that, now, he, and she.

For a long time, indexicals fascinated philosophers of language because their meaning is so heavily dependent on context. ‘Please hand that to me’ might be a request for many different things, depending on what the requestee is sitting in front of. I could repeat those exact words twice in the span of thirty seconds, but be asking for a wrench on one occasion and a garden gnome on the other. So it would seem as though the meaning of ‘that’ were highly variable. Is there one single meaning that the word ‘that’ always has, regardless of when and where it is uttered?

Let’s think about this question vis-à-vis the word I.  Doesn’t the word I have a single meaning?  You’d think it meant something like, ‘the person speaking.’  (We’ll focus on speech for the time being, since written language raises additional complications.)  This seems like a good way to explain both what all uses of the word I have in common and how they can differ: although I might refer to Barack Obama when he uses it or to Angela Merkel when she uses it, it never seems to refer to anyone other than the person speaking.

Alas, things aren’t so simple.  As David Kaplan pointed out, that definition of the word I runs into trouble in cases where we use it to talk about what-if scenarios.  Here’s one of his examples.  Suppose I begin teaching, then say to my students, ‘I feel run-down today.  I wish I weren’t the one here talking to all of you.  I wish I were home in bed, and that Jessica were teaching the class on my behalf.’  Although that wouldn’t be the most pleasant thing to say, I would at least be expressing a coherent wish.  But our attempted definition of I falsely predicts that my wish is self contradictory: that I am saying, ‘The person speaking wishes that the person speaking were not the person speaking.’

This is a pretty serious problem.  To get around it, Kaplan proposed what is still the standard approach to such worries.  The basic idea is that in order to say what indexicals mean, we have to think of meaning in a different way.  We have to think of the meaning of indexicals as having two parts–as corresponding to two steps in a process.  The first part, which he calls character, is the rule that determines what the speaker is referring to on the basis of information perceptually available to the listener.  So the character of I is the ‘the person speaking’ part of its meaning.  The second part of the meaning of an indexical, which he calls its content, is the worldly thing that the speaker is referring to.  So the content of I is the ‘Matt’ part of its meaning, for example, in the sentence ‘I am sitting in my apartment’ (as uttered by me right now).

The details of Kaplan’s theory are a bit involved, so we won’t present them here.  The important thing for our purposes is that the theory, by breaking the process of determining what an indexical means up into these two steps, has as a consequence that the sentence ‘I am sitting in my apartment’ can be used to say very different things.  When uttered by me right now, the sentence is saying, ‘Matt is in Matt’s apartment.’  But when uttered by Barack Obama, the sentence is saying, ‘Barack Obama is sitting in Barack Obama’s apartment.’  Kaplan allows us to have our cake and eat it too: we get to grant that there is something common to all occurrences of the word I--that it doesn’t just refer to people at random–but we also get to say that the very same sentence featuring the word I can be used to make different claims, depending on who utters it.

This means that Kaplan’s theory also gets us out of the aforementioned difficulty, insofar as it predicts that mental attitudes like wishes are about contents, rather than characters.  Thus, it ultimately gives us the result we want: ‘I wish I weren’t the one speaking right now’ isn’t self-contradictory in the least, because (in the context we imagined) all it’s saying is, ‘Matt wishes that Matt were not the person speaking.’

At first it can seem surprising that philosophers like to go around claiming that sentences with indexical expressions in them express different contents, depending on the circumstances in which they’re uttered.  A natural response is ‘Huh?  Why can’t we just say that I means ”the person speaking”?’  Well, this is why.

Matt Teichman