Recent Posts

The Bertrand Russell Show

Feminist Philosophers

fragments of consciousness

Gender, Race and Philosophy: The Blog


Leiter Reports: A Philosophy Blog

Long Words Bother Me

semantics etc. highlights

Thoughts Arguments and Rants



Sunday, September 10, 2006

Cappelen and Lepore's Sensitivity Tests

In chapter 7 of Insensitive Semantics Herman Cappelen and Ernie Lepore offer a number of tests for context-sensitivity. Here are a couple of examples of the kind of test employed by Cappelen and Lepore. Consider the following utterances:

John: I am hungry.
Mary: I am hungry.

Brit: #John and Mary said/believe the same thing.

Since it is inappropriate to conclude that John and Mary said/believe the same thing, it cannot be that John's and Mary's utterances semantically express the same proposition. So, on the assumption that 'hungry' is context invariant, it must be that 'I' is context sensitive.

Consider another case:

Ernie: I am hungry
Herman (reporting): #Ernie believes I am hungry

If Herman aims at reporting Ernie's belief, he is clearly unsuccessful. This suggests that 'I' is context-sensitive. For if 'I' were not context-sensitive, then the semantic content of 'I am hungry' would be the same regardless of whether 'I am hungry' is embedded or unembedded.

After considering a number of such tests Cappelen and Lepore show that expressions like 'tall', 'know', and so on, fail the tests. This indicates that these expressions are not context-sensitive.

In his reply to Cappelen and Lepore John Hawthorne notes that expressions like 'nearby' and 'local' seem to fail the tests for context-sensitivity in spite of the fact that it is very plausible that they are context-sensitive. Consider, for instance, the following exchanges:

Ernie: I am at a local bar.
Herman: Ernie believes he is at a local bar.

Ernie: A nearby restaurant has good Thai food.
Herman: Ernie believes a nearby restaurant has good Thai food.

These exchanges seem fine. If Cappelen and Lepore's tests are good indicators of context-sensitivity, the felicity of the exchanges suggests that 'local' and 'nearby' are not context-sensitive. However, Hawthorne thinks denying that 'local' and 'nearby' are context-sensitive is absurd. For instance, it is widely agreed that the content of 'local' depends on the perspective salient in the discourse context. Thus, an occurrence of 'John went to a local bar' can mean that John went to a bar that is local to him, that John went to a bar that is local to the speaker, that John went to a bar that is local to the hearer, and so on. Expressions like 'local' and 'nearby' are what Francois Recanati and Anne Bezuidenhout call 'perspectivals'. Their content will depend, not on the speaker's location, but on a location that is relative to the perspective salient in the discourse context.

If perspectivals fail the tests for context-sensitivity, then it is open to argue that expressions that fail the tests (e.g. 'tall' and 'know') are perspectivals in Recanati's and Bezuidenhout's sense as well. This would be bad news for semantic minimalism and good news for moderate contextualism. The only concession defenders of moderate contextualism would need to make is that 'tall', 'know', and so on are perspectivals (in the mentioned sense) and not indexicals.


Mike said...

". . .an occurrence of 'John went to a local bar' can mean that John went to a bar that is local to him, that John went to a bar that is local to the speaker, that John went to a bar that is local to the hearer, and so on"

The bar might (it seems) also fail to be local to any of the parties to the discussion. If you are telling a story about the Smith's vacation to X you might describe his going to a local bar, even though Smith lives no where near the bar and neither does any other party to the discussion. I guess the bar is local to (what?) Smith at the time he was visiting X? No. At the time Smith was visiting he might have been staying at Y and going to the local bar in X. Local to the town X that Smith was visiting? But what could that mean? Local from all points in X?

Brit said...

Hi Mike,
Yes, thank you. I like the example. Maybe the bar in your example is local to the story-teller's narrative perspective (I think that's why Recanati calls 'local' a 'perspectival').

Mike said...

"Maybe the bar in your example is local to the story-teller's narrative perspective"

Oh, for some reason I was taking 'perspective' to be one's relative physical location. But here we have a sense of 'X is local to S' that does not entail that X's spatial location relative to S is sufficiently close to be physically local. It's narratively local, yes? Got to think about it.

Kent Bach said...

Since Cappelen and Lepore don't distinguish indexicality from other sorts of context sensitivity (or, for that matter, from semantic underdetermination) and thus implicitly assume that all context-sensitive expressions are indexicals, it's to be expected that their tests for context sensitivity are unreliable.

Question: Assuming relational terms like 'local' and 'nearby' are context-sensitive, why call them "perspectival"? I can see why terms like 'visible' and 'discernible', or 'plausible', 'memorable', and 'funny', are perspectival (assuming they're context-sensitive, which is debatable), because the (relational) properties they express are relative to someone's perspective (visual or other sort of point of view).

Brit said...

Mike and Kent: I hasten to say that by 'perspectival' I do not mean 'non-indexical context-sensitive expression' in the sense of MacFarlane.

Now, why do Recanati and Bezuidenhout call 'local' and 'nearby' perspectivals? The idea is that a particular perspective can be salient in the conversational context. So, if I say "John went to a local bar" and we are talking about John's trip to San Francisco, then the perspective would be focused on San Francisco. So, 'local' means 'local to San Francisco'. If, on the other hand, we are talking about John visiting his grandmother, then the perspective is the location of John's grandmother. And so, 'local' means 'local relative to John's grandmother's location', and so on.

As Recanati is using the term 'perspectival', an expression is a perspectival iff its content varies with the perspective of the conversational context (as described above).

A final thought: we actually do not need to give up minimalism, even if we grant that there are perspectivals in the language. The problem outlined in the post is only a problem for the tests suggested by Cappelen and Lepore.