Content and Character
The following is a tutorial essay for Philosophy of Logic and Language with James Kirkpatrick. In effect, it is a quick agenda for an hour-long discussion on a guiding question, which for this week was ‘According to Kaplan, indexicals and demonstratives have both content and character. Explain this distinction and how it might help capture the meaning of these terms.’ A free-standing and much more readable version of some of the material may eventually go up on my Substack.
I. Kaplan’s distinction between content and character
Suppose that Ann says, ‘I smoke’. In doing so, she simultaneously says /aɪ sməʊk/ (a vocalisation), says ‘I smoke’ (an English sentence), and says that she smokes (a proposition).
The content of what Ann says is the proposition that she smokes. This proposition might be true or false if different possibilities were the case—call these possibilities the circumstances of evaluation.
However, suppose that Ben now also says, ‘I smoke’. In doing so, he simultaneously says /aɪ sməʊk/ (a vocalisation), says ‘I smoke’ (an English sentence), and says that he smokes (a proposition). Ben’s utterance is a different event from Ann’s—for one thing, they happened at different times. But the content of Ben’s utterance is also different from the content of Ann’s: it might be that Ann’s utterance is true while Ben’s is false, because it might be that Ann smokes but Ben does not. Nevertheless, it seems that Ann and Ben in some sense said the same thing: they uttered the same English sentence. If a sentence’s meaning is just its content, it’s puzzling how the same sentence can have two different meanings when said by two different people. How could we learn a language if sentences commonly change meaning? How would we know what sentences mean?
Sentences are typically complex expressions, being built out of words. The meaning of a sentence is determined by the meaning of the words. We can replicate the problem at the level of individual words. When Ann and Ben say ‘smoke’, we may suppose that they mean exactly the same thing: it’s not that Ann means smoking cigarettes, while Ben means smoking something else. The content of ‘smoke’ in either case is, roughly, the function which maps any individual x to the proposition that x smokes. Meanwhile, in typical cases, when Ann and Ben say ‘I’, they mean different things: the content of ‘I’ when Ann utters it is Ann herself, while the content of ‘I’ when Ben utters it is Ben himself. Combining the contents of ‘I’ and ‘smoke’ by function application yields the right result for the content of ‘I smoke’ in each case.
Now, then, we’ve localised the problem to the expression ‘I’. Here are two possible routes that one might go down.
The first is extreme homonymy, in Dorr’s (2014) terminology. The thought is that Ann’s ‘I’ and Ben’s ‘I’ are two different words. We have a whole family of words, all written (and more or less pronounced) identically, just like ‘bank’ (meaning a riverbank) and ‘bank’ (meaning a financial bank). We will return to this option later.
The second option is to say that the same word ‘I’ can have different meanings depending on the context of utterance. The word ‘I’ has a fixed character which, roughly, is a function from contexts of utterance to contents (Kaplan 1989; see also Braun 1995). At first pass, ‘I’ maps to the speaker of the utterance. Of course, there are anomalies—what if Ann is acting as a translator for someone saying
II. Indexicals and demonstratives with the distinction
As we saw with ‘smoke’, sometimes the ideology of character is idle: most words have a constant function as their character, so we could get by without invoking character or context at all.
However, for indexicals (which must be supplied an index by the context) and demonstratives (which must be supplied a demonstration by the context), the context does seem to change the content, and so we must invoke character. The list Kaplan (1989) provides includes:
the pronouns ‘I’, ‘my’, ‘you’, ‘he’, ‘his’, ‘she’, ‘it’, the demonstrative pronouns ‘that’, ‘this’, the adverbs ‘here’, ‘now’, ‘tomorrow’, ‘yesterday’, the adjectives ‘actual’, ‘present’, and others.
Each of these terms seems capable of meaning very different things in different contexts of utterance: ‘I’ can refer to any individual speaker, ‘you’ any audience, ‘he’ any male (person), ‘she’ any female (person), ‘that’ and ‘this’ anything demonstrated, ‘here’ any speaker’s location, ‘now’ any speaker’s time, and so on. Formally, there is no issue having the content of these terms vary with context, rather than being constant across all contexts.
Notably, however, these terms are not equivalent to the corresponding definite descriptions (‘the speaker’, ‘the audience’, ‘the relevant male’, …). The content of the indexicals and demonstratives is simply the individual; the wide-scope readings of the definite descriptions are not available for the indexicals. To use a version of Kaplan’s (1989) example, (1) but not (2) has a reading on which Ann smoked the day before yesterday.
(1) The day before, the day before, Ann smoked.
(2) Yesterday, yesterday, Ann smoked.
In any case, the ideology of character gives us an answer to the puzzle with which we started. We can learn a language by learning the character of each word, and figure out the content “on the fly” for each utterance of that word. We know what sentences mean by figuring out what the individual words mean; and we know what the words mean because we can calculate the content from character and context.
III. Indexicals and demonstratives without the distinction
The extent to which this distinction helps capture the meaning of indexicals and demonstratives depends on how well we can do without such a distinction. As noted above, the main alternative is to posit extreme homonymy. This is what Kaplan himself posits for directly referential proper names, like ‘David’. Although Kaplan’s treatment of demonstratives and indexicals unifies a wide variety of phenomena, unifying them further with proper names would be a theoretical virtue.
We already posit a map from vocalisations to sentences that depends on context. For instance, consider (3).
(3) Flying planes can be dangerous.
An utterance of the words in (3) is ambiguous between one sentence which means that it can be dangerous to fly planes, and another which means that planes which are flying can be dangerous. Similarly, an utterance of ‘bank’ is ambiguous between one word meaning a riverbank and another meaning a financial institution. However it is we figure out which word is being uttered, by inferring from context, we could also figure out which of the many ‘I’s are being uttered. Each ‘I’ then, like a directly referential proper name, has a constant character, and so we have no need to invoke character. Allowing for both context-dependence in the move from vocalisations to expressions in a language and in the move from expressions in a language to contents looks like messy redundancy.
Of course, it intuitively feels like there’s some sense in which ‘I’ means the same thing in every context of utterance. But that goes too for the proper name ‘Mom’, which typically means the speaker or hearer’s mother, depending on context. The context dependence here is best captured as homonymy between different lexical items all spelled and pronounced ‘Mom’. And the move from context to lexical item is better captured with a cognitive heuristic than a strict semantic rule: it really does depend messily on context whether ‘Mom’ is the speaker or the hearer’s mother (consider a child and her father talking about her mother). We can bring similar behavior out of expressions like ‘I’, as the case of the translator showed; similarly for projected uses of ‘here’ (say, on a transcontinental phone call) or ‘now’ (say, on a recorded message).
We conclude, then, that the distinction between content and character might initially seem to help capture the meanings of indexicals and demonstratives; but a better treatment of their meaning might ultimately assimilate them to directly referential proper names, where the map from utterance to word depends on context and the map from word to content is constant, rather than (as in Kaplan’s theory) where the map from utterance to word occasionally still depends on context and, for indexicals and demonstratives, the map from word to content depends on context as well.