Kaplan's Indexicals

In language, there seems to be a bunch of small words with some special meanings. These words, such as 'I', 'here', 'now' and 'tomorrow', seem to refer to different things depending upon the context in which they are spoken.

In his theory of demonstratives, David Kaplan gives the semantics of pure indexicals. 'Pure indexicals' are distinguished from 'true demonstratives', which include words such as 'he', 'she', 'his', 'that', etc. The distinction between these two groups is drawn according to the way in which the intended referent for the word is arrived at. In the case of demonstratives, the word, say 'he', is usually accompanied by some kind of pointing, nodding, eye contact, or other means of indication. These actions – the demonstration – supply the means by which the hearer can understand the intended referent. Pure indexicals, on the other hand, are determined automatically, simply by the facts of the context of utterance.

For instance, in the context where Frankie Roberto is speaking on the seventh of January 2005, and utters 'I am 21 years old today', 'I' refers to Frankie Roberto and 'today' refers to the seventh of January 2005. However, the meaning of the words 'I' and 'today' are not Frankie Roberto and the seventh of January 2005, nor can it be said that 'I' refers to all the possible people in the world and that 'today' refers to all the possible days in the world. Instead, the meanings of 'I' and 'today' can roughly be described as being 'the utterer of the sentence' and 'the day of utterance'. These meanings cannot, however, play a part in the truth conditions of an utterance, or else we would get strange interpretations of meaning such as follows:

'The utterer of this sentence' is 21 on 'the day of this utterance'.

Kaplan solves this problem by separating meaning into two components; content and character. Under Kaplan's terminology, content is the meaning contributed to the truth conditions of a sentence by an indexical in a given context. The character, on the other hand, is the context-independent meaning which describes the means by which you can evaluate the content against a given context. In the circumstance given above, for instance, knowing the character of 'I' allows us to determine that when it is spoken in an utterance by Frankie Roberto, the content is Frankie Roberto. The character and content together describe the meaning of pure indexicals. Kaplan also states that all expressions must have character and content, however for non-indexicals these are identical.

There are, however, a few circumstances which seems to pose a problem for Kaplan's theory of the semantics of pure indexicals. The most common example is where someone might record a message for their answerphone, uttering 'I am not here now'. If 'I' and 'here' and 'now' refer to the speaker, location and time of an utterance as Kaplan suggests, it would be logical to think that sentences of the type 'I am not here now' would necessarily always be false. However, in the case of the answerphone message, when you listen to a message from someone saying that they aren't there, the proposition expressed seems pretty true. In this circumstance, 'now' refers to the time at which you are listening to the utterance and not the time at which the utterance was recorded. The fact that you can derive this meaning seems to blow a big hole in Kaplan's story.

Kaplan himself does shows awareness of this possible problem however: There are certain uses of pure indexicals that might be called 'messages recorded for later broadcast', which exhibit a special uncertainty as to the referent of 'here' and 'now'. Kaplan (1989a, fn. 12)

This answer seems vague to say the least though, and makes no attempt at explaining how the uncertainty arises and is resolved.

There have been many attempts from other philosophers to save Kaplan's semantics from these kinds of examples. One such attempt comes from Sidelle (1991), who makes a distinction between the context of encoding and the context of decoding. Under Sidelle's picture, the person who utters 'I am not here now' into the recording mechanism of their answerphone is 'arranging to make an utterance at a later time, or if one likes, deferring their utterance' (Sidelle 1991: 535). The time at which the message is played back triggers a 'genuine utterance', and it is in this context that the pure indexicals are evaluated. Thus, when someone listens to the answerphone message, the word 'now' refers to the time at which the recording is played back, and not the time of the original utterance.

The distinction between encoding and decoding seems to nicely capture a variety of intuitions. If, for instance, you were in the same room as someone whilst they were recording 'I am not here now' into their answerphone, you might well joke with them that they weren't being truthful. Sidelle would be able to explain this as being because you were decoding the utterance at the same instant as it was encoded (as happens in natural speech every day), and in this context the utterance is false. Even if no-one hears you recording the message, you could argue that as you were uttering the message, your utterance is false. However, as there is no-one but yourself there to decode the utterance, the truthfulness of the statement is irrelevant. This analysis also explains situations where someone may either forget or deliberately avoid turning their answermachine off when they get home – leading to their message incorrectly telling callers that they are still away.

Sidelle's solution may run into difficulty when applied to other types of recorded message however. If, for example, someone is away from their family for a period of time and decides to record onto a tape a message to be sent back home, they might well record a message saying something like 'I am now getting ready to go out…'. In this circumstance, by the time the tape has gone through the post and is played back at home, the sender may well be involved in some other activity. It is clear, however, that the family at home would take the meaning of 'now' to be the time of encoding, not decoding. This poses a pretty big problem for Sidelle, as his theory clearly cannot generalise against all possible circumstances. It would seem that there are times in which indexicals are evaluated against the context of encoding, and times in which they are evaluated against the context of decoding.

Predelli (1998) objects to Sidelle's view and offers an alternative explanation. Predelli's key argument centers around a supposed counter-example to Sidelle's story of encoding vs decoding contexts.

Suppose that Mr Jones expects his wife to come home at six, and leaves a message at four saying 'I am not here now', as he knows he will be out when his wife gets home. Mr Jones's wife gets home late, however, and only reads the message at 10. In this circumstance, Predelli argues, the content of Jones's message is 'to be evaluated with respect to the time of the intended decoding' (Predelli 1998: 110), which would make the meaning of the message something like 'I am not here at 6'. Such an evaluation allows Mrs Jones to come up with the right deductions if, say, the note had continued '…will be back in five hours', allowing poor Mrs Jones to figure out that her husband should be home at 11pm, and not at 9pm or 3am. In the case of answerphones, where you cannot usually predict when people will call, Predelli suggests that the intended context of evaluation is the time of the call, effectively giving the same results as Sidelle theory, but for different reasons. Predelli thus concludes that written and recorded messages are 'to be evaluated with respect to the intended context of interpretation' (Predelli 1998: p115).

It's clear that intentions must play some role in understanding indexicals in written or recorded messages, or else how would Mrs Jones be able to figure out the time of her husband's expected arrival in the example above? However, whether or not Predelli's argument is sufficient to explain the semantics of pure indexicals in all occurrences is another matter.

Powell (2003) is unconvinced with Predelli's example and offers an example whereby Jones returns home before his wife at 8pm, but forgets to destroy the note. Mrs Jones then reads the note at 10pm and is surprised to find her husband at home in bed. Powell points out that in these circumstances, Mrs Jones would be justified in believing Mr Jones's note to be false. As Mrs Jones would clearly be evaluating the note in the context of the time she reads it, this example seems to be a problem for Predelli. Predelli might argue back that this example is only a problem because discovering her husband in bed creates a confusion for Mrs Jones over what the intended context for the note is meant to be. The fact that the note is no longer true might block the intended context of 6pm as it is difficult to attribute intentionality of a context to a note which isn't true or relevant (unless you either assume that the writer is being deliberately misleading or that there has been a mistake). As these examples fail to test either theory by seeing what happens in circumstances where the theories might give the wrong result, we are left none-the-wiser when it comes to choosing between the two theories. Romdenh-Romluc (2002) complicates the matter further by outlining examples whereby the recipient of an utterance isn't able to retrieve the context intended by the utterer. For the example above, this might happen if Mrs Jones knew that Mr Jones consistantly mis-remembered what time she arrives home at, and always thinks that it's 5pm. However, on the day of writing the note, Mr Jones suddenly remembers the correct time. Mrs Jones would end up thinking that the 'now' referred to 5pm, differing from Mr Jones's intended context. Even if we build into Predelli's account the proviso that the utterer can only intend contexts which they can reasonably expect the hearer to interpret, we are still left with a problem for Predelli, as Romdenh-Romluc points out with the example of Fernando mistakenly bursting into a manor house, expecting to take part in a Norman-Conquest re-enactment but instead interrupting a wedding party, exclaiming 'now the French are invading England!' (Romdenh-Romluc 2002: 37). Here, the utterer expects his audience to be able to retrieve the correct intended context, but makes a mistake, leaving the wedding guests panically thinking the invasion is happening at the time of utterance.

Romdenh-Romluc's alternative account dives deeply into the realm of pragmatics, by suggesting that the referent of pure indexicals 'is determined by the cues that the competent and attentive audience would reasonably take the utterer to be exploiting' (Romdenh-Romluc 2002: 41). This suggests to me that the semantics of pure indexicals underdetermines the content of such words within an utterance, delivering only the basic Kaplanian meanings that 'I' refers to the utterer [in some context], 'here' to the location of the utterance [in some context] and 'now' to the time of utterance [in some context]. Whilst in regular face-to-face conversations, pure indexicals will be evaluated against the actual context of utterance, the rich pragmatic system of linguistic communication allows us to assume all kinds of alternative contexts. It is left to a theory of pragmatics to best explain how we are able to correctly derive these contexts, though I suspect that a relevance-theoretic approach may well be able to make inroads in this area.


  • Corazza, E., W. Fish and J. Gorvett. 2002. Who Is I? Philosophical Studies 107, pp. 1-21. Kaplan, D. 1989a. Demonstratives. In J. Almog et al. (eds.) Themes from Kaplan, pp. 81-563. Oxford: OUP.
  • Powell, G. 1998. The deferred interpretation of indexicals and proper names. UCL Working Papers in Linguistics 10.
  • Powell, G. 2003. Indexicals. Chapter 4 of Language, Thought and Reference. Ph.D. thesis, University of London.
  • Predelli, S. 1998. I am not here now. Analysis 58:2, pp. 107-115.
  • Romdenh-Romluc, K. 2002. Now the French are invading England! Analysis 62:273, pp. 34-41(8) Sidelle, A. 1991. The answering machine paradox. Canadian Journal of Philosophy 21. pp. 525-39.