June 08, 2005

Bob and Carol and Howl and Alice

I thought of a somewhat PG-rated (for adult situations) example to illustrate one of my arguments against the knowledge account of assertion, and then I found something similar in Diana Wynne Jones' Howl's Moving Castle--now a major (G-rated, I trust) motion picture. Examples from common talk are nice, but I'm going to start with my version.

Suppose Alice, Carol, and Ted are talking about Alice's husband Bob. Alice says, "He sings 'We Suck Young Blood' in the shower in the mornings." Carol says,

(1) He does.

(i.e., Bob does sing "We Suck Young Blood" in the shower in the mornings.)

This, I think, would tend to stop conversation.

Why? Because Carol's assertion of (1) suggests that she has firsthand knowledge of what Bob sings in his morning shower. And Carol ought not to have firsthand knowledge of what Alice's husband sings in his morning shower.

Carol's assertion of (1) is improper no matter how you slice it. If she does have firsthand knowledge of Bob's shower songs, it's very rude for her to let Alice know like this. If she doesn't have firsthand knowledge, then it's even ruder--and she can be held to the charge of BSing. But the assertion of (1) wouldn't be problematic at all if it didn't represent Carol as having firsthand knowledge. (For the rest of this discussion, suppose Carol doesn't have firsthand knowledge.)

That's important, because Carol does know that (1) is true. Alice just said so, and she's in a position to know, etc. So the knowledge account of assertion in itself can't account for why Carol's assertion of (1) in the absence of firsthand knowledge is improper. If an assertion only represented the speaker as knowing what she asserted, then the dread implicature that Carol's knowledge was firsthand wouldn't arise; and it wouldn't be rude for Carol to assert (1).

Here the epistemic impropriety of Carol's assertion may get tangled up with all the other sorts of impropriety. So let's try and remove those other sorts. Suppose that Alice says, "He sings 'We Suck Young Blood' in the shower in the mornings," and then is called away to perform delicate heart surgery. Then Carol utters (1).

Suppose further that Ted has no particular loyalties to Alice or Bob, so it wouldn't be particularly rude for Carol to let him know that she'd been sleeping with Bob, were such the case. And let's suppose there's no general norm against lashon harah; wasn't it Timothy Williamson himself who said "gossiping is a way of caring about people"? (Though maybe he didn't mean this kind of gossip--anyway it's one of life's great pleasures, and I won't hear a word against it.)

Still, if Carol does not in fact have firsthand knowledge, Ted has the right to criticize her for asserting (1). He can't say, "He doesn't," because Bob does; he can't say "You don't know that," because Carol does (from what Alice said). There may be no concise way to express his complaint; closest perhaps is "You led me to believe that you knew yourself." In any case, the fact that Carol knew that (1) was true won't get her off the epistemic hook.

The explanation of how Carol's assertion implicates that she has firsthand knowledge is simple. Alice has just said that Bob sings 'We Suck Young Blood' in the shower, so all participants to the conversation can be assumed to know it. In fact, they have as good reason to believe it as can be got from Alice's testimony. For Carol's assertion not to be completely pointless, she must have something new to add. This will in general be another source of evidence than that provided by Alice's testimony; and that, most plausibly, will be firsthand observation. There are other ways that Carol might be adding to the conversation (she could say, "He does, Alice has often told me so"), but this is the most obvious; and the natural conclusion to draw is that Carol is adding the most obvious thing that she could be adding.

Note that it's perfectly OK for Carol to assert "Bob sings 'We Suck Young Blood' in the shower" to someone who wasn't involved in the conversation with Alice. They can't be presumed to already possess the evidence given by Alice's testimony; so Carol can add something new to this conversation even if she's merely passing along what Alice said.

(This argument parallels my explanation in "Must We Know What We Say?" of why you can't tell someone "Your lottery ticket didn't win" if you haven't heard the results of the drawing. What you assert is most likely true. But if you're basing it merely on the odds against the ticket winning, those are presumably grounds the lottery ticket owner already has. The ticket owner has the right to assume you're adding something new, and the most obvious thing you might be adding is that you've actually heard what the results of the lottery were.

Incidentally, that's the final version of "Must We Mean What We Say?" which I just now posted.)

Here's the example from Howl's Moving Castle:

"Are you sure your Lettie was telling the truth about Howl?" [Sophie] asked anxiously.
"Positive," said Michael. "I know when she's lying. She stops twiddling her thumbs."
"She does too!" said Sophie, chuckling.
"How do you know?" asked Michael in surprise. (p. 68)

Sophie wouldn't be able to get away with saying, "Because you just told me." She's got to have some sort of knowledge that isn't what Michael has already contributed. (But I wouldn't want to put too much weight on this, because I'm not sure what 'too' does in this context; and because I'm not exactly sure what kind of speech act Sophie's utterance is. It seems like assertion, but it doesn't seem like testimony.)

Posted by Matt Weiner at June 8, 2005 02:45 PM

Sophie's speech act is corroboration. Is corroboration testimony? It is to a lawyer, anyway.

In the Bob & Carol & Ted & Alice example, if Carol says "Uh-huh" or nods her head (or says "Yes?"), she's just acknowledging Alice's statement, but if she says "Yes" firmly, she's corroborating it. There is a difference even though we all know that nodding your head means "yes." It depends on what the meaning of the word "yes" is.

Posted by: Ben at June 8, 2005 03:30 PM

If Alice was in the habit of making false statements which were possibly true but then getting all "mind so fine" when called on it, Carol's statement could add additional information without being first hand evidance. Carol could have previously put Alice's statement into question and Alice could have earnestly defended it. Carol's statement would then help the others evaluate Alice's statement with risking the "mind so fine" trap.

Posted by: Joe O at June 9, 2005 05:01 PM

Carol could have previously put Alice's statement into question and Alice could have earnestly defended it.

Posted by: James Scott at June 11, 2005 12:23 PM

James and Joe--those are possible scenarios. (Wait--I retract the credit to James. He seems to be a spambot who just excerpts sentences from previous posts. Fortunately I've deleted his URL already. Has anyone encountered this phenomenon before?)

Joe, those are possible scenarios. Or if Alice is often joking, and Carol but not Ted is capable of telling when she's joking, that might make sense (maybe that's what you said). But in ordinary settings that won't be the most plausible explanation either--I think we need a lot of extra scene-setting to make it OK for Carol to say that without implying something unsavory.

Ben--Good point about the importance of the inflection on 'yes', and the difference between saying 'Yes!' and nodding. With Sophie's act, one of the things that worried me is that she seems to be talking to herself more than to Michael. But her act might be corroboration anyway. (Corroboration is epistemologically a lot like testimony anyway--it raises a lot of the same concerns about whether the speaker is a good source.)

Posted by: Matt Weiner at June 11, 2005 12:37 PM