Sunday, January 20, 2008
Language in Context: On Quantifier Domain Restriction (Part 1)
Why are philosophers, and not just philosophers of language, so interested in the phenomenon of quantifier domain restriction (QDR)?
1. One reason stressed, by Jason in Knowledge and Practical Interests, is that David Lewis based his contextualist semantics for knowledge attributions on QDR. The sceptic is right to suggest that in order to know that p we must be able to rule out all relevant alternatives to p; to adopt fallibilism, according to which one may felicitously utter 'p, but I haven't ruled out possibilities in which not-p', is madness. But the sceptic fails to recognize that this does not set an absolute standard on which alternatives need to be ruled out; it's 'all' of them, but in everyday contexts the domain of 'all' is seriously restricted. In particular, we can properly ignore sceptical scenarios. Indeed, Jason argues against Lewis's epistemic contextualism on the grounds that knowledge attributions don't behave like uncontroversial cases of QDR in the relevant respects (see Jonathan's recent post for discussion of Jason's argument). So a proper understanding of QDR is necessary to evaluate certain proposals in epistemology.
(I don't mean to endorse Lewis's characterization of fallibilism here. As most of you will know, Jason and others have resisted Lewis's argument against fallibilism on the grounds that it was never committed to the felicity of the problem utterances in the first place.)
2. QDR also plays an important role in defending various views Lewis held in metaphysics. For example, how are we to square the apparent truth of,
i. there are no talking donkeys,
with the truth of moral realism, according to which if there could be talking donkeys, then in some worlds there are talking donkeys? Likewise, how are we to square universalism--the doctrine that some things always compose some further thing--with our ordinary talk, which seems to pay no heed to objects like the fusion of my iPod and Jeremy Bentham's preserved corpse? In each case, it can be contended that in ordinary settings, we implicitly restrict our quantifiers, ignoring possibilia and gruesome fusions. (See Dan's paper for a critique of this as a defense of the claim that universalism is continuous with common-sense.)
3. Stephen Neale has argued that the puzzle associated with so-called incomplete definite descriptions, such as:
ii. the table is covered with books,
should not be thought of as providing a decisive objection to Russellian views according to which definite descriptions are really quantificational (and imply uniqueness), since the fact that the existence of a bare table in St Andrews does not entail the falsity of my utterance of:
iii. every table is covered with books
made in a room full of book-laden tables in Austin clearly does not impugn the claim that (iii) has quantificational form. The problem of incomplete definite descriptions is really one and the same as the problem of giving a satisfactory account of QDR.
iv. Stanley and Szabo (70) take QDR as a test case for determining the source and nature of an expression's dependence on context. That is, the process by which they think we arrive at a satisfactory account of QDR provides a model for discussion of other constructions. So one might hope that a proper understanding of QDR will lead to better understanding of context dependence in natural language more generally.
Of course, these do not exhaust the sources of philosophical interest in QDR. But I thought it was worth setting on the table some implications for epistemology, metaphysics and the philosophy of language before turning our attention to Stanley and Szabo's paper.
1. One reason stressed, by Jason in Knowledge and Practical Interests, is that David Lewis based his contextualist semantics for knowledge attributions on QDR. The sceptic is right to suggest that in order to know that p we must be able to rule out all relevant alternatives to p; to adopt fallibilism, according to which one may felicitously utter 'p, but I haven't ruled out possibilities in which not-p', is madness. But the sceptic fails to recognize that this does not set an absolute standard on which alternatives need to be ruled out; it's 'all' of them, but in everyday contexts the domain of 'all' is seriously restricted. In particular, we can properly ignore sceptical scenarios. Indeed, Jason argues against Lewis's epistemic contextualism on the grounds that knowledge attributions don't behave like uncontroversial cases of QDR in the relevant respects (see Jonathan's recent post for discussion of Jason's argument). So a proper understanding of QDR is necessary to evaluate certain proposals in epistemology.
(I don't mean to endorse Lewis's characterization of fallibilism here. As most of you will know, Jason and others have resisted Lewis's argument against fallibilism on the grounds that it was never committed to the felicity of the problem utterances in the first place.)
2. QDR also plays an important role in defending various views Lewis held in metaphysics. For example, how are we to square the apparent truth of,
i. there are no talking donkeys,
with the truth of moral realism, according to which if there could be talking donkeys, then in some worlds there are talking donkeys? Likewise, how are we to square universalism--the doctrine that some things always compose some further thing--with our ordinary talk, which seems to pay no heed to objects like the fusion of my iPod and Jeremy Bentham's preserved corpse? In each case, it can be contended that in ordinary settings, we implicitly restrict our quantifiers, ignoring possibilia and gruesome fusions. (See Dan's paper for a critique of this as a defense of the claim that universalism is continuous with common-sense.)
3. Stephen Neale has argued that the puzzle associated with so-called incomplete definite descriptions, such as:
ii. the table is covered with books,
should not be thought of as providing a decisive objection to Russellian views according to which definite descriptions are really quantificational (and imply uniqueness), since the fact that the existence of a bare table in St Andrews does not entail the falsity of my utterance of:
iii. every table is covered with books
made in a room full of book-laden tables in Austin clearly does not impugn the claim that (iii) has quantificational form. The problem of incomplete definite descriptions is really one and the same as the problem of giving a satisfactory account of QDR.
iv. Stanley and Szabo (70) take QDR as a test case for determining the source and nature of an expression's dependence on context. That is, the process by which they think we arrive at a satisfactory account of QDR provides a model for discussion of other constructions. So one might hope that a proper understanding of QDR will lead to better understanding of context dependence in natural language more generally.
Of course, these do not exhaust the sources of philosophical interest in QDR. But I thought it was worth setting on the table some implications for epistemology, metaphysics and the philosophy of language before turning our attention to Stanley and Szabo's paper.
Labels: Context Dependence, Jason Stanley, Language in Context
# posted by Aidan @ 1:26 PM 
Comments:
<< Home
For example, how are we to square the apparent truth of,
i. there are no talking donkeys,
with the truth of mo[d]al realism, according to which if there could be talking donkeys, then in some worlds there are talking donkeys?
Aidan,
How does QDF help here? Lewis has (i) not just mistaken in some contexts, but plain false, right? All possibilia and all actualia(?) exist simpliciter. We can freely quantify over them all. Do I shift domains to accommodate the truth of (i) when I realize that you mean by (i) that talking donkey's don't actually exist?
i. there are no talking donkeys,
with the truth of mo[d]al realism, according to which if there could be talking donkeys, then in some worlds there are talking donkeys?
Aidan,
How does QDF help here? Lewis has (i) not just mistaken in some contexts, but plain false, right? All possibilia and all actualia(?) exist simpliciter. We can freely quantify over them all. Do I shift domains to accommodate the truth of (i) when I realize that you mean by (i) that talking donkey's don't actually exist?
I had in mind the kind of move you suggest in your final sentence, which Daniel puts as follows:
'Lewis claims that we often (tacitly) restrict our quantifiers to only include actual things. When we say there are no giant gorillas who climb skyscrapers, or we say that cows never tap-dance, we are restricting our quantifier to the actual world. (We may not realize that we are doing this, but quantifier restriction does not have to be deliberate. Many of us do not realize until it is pointed out that in our ordinary talk about objects we seem to be restricting quantifiers.)'
(Nolan - 'David Lewis': 56)
'Lewis claims that we often (tacitly) restrict our quantifiers to only include actual things. When we say there are no giant gorillas who climb skyscrapers, or we say that cows never tap-dance, we are restricting our quantifier to the actual world. (We may not realize that we are doing this, but quantifier restriction does not have to be deliberate. Many of us do not realize until it is pointed out that in our ordinary talk about objects we seem to be restricting quantifiers.)'
(Nolan - 'David Lewis': 56)
Does Nolan suggest a similar strategy for resolving the vagueness of the counterpart relation? I'm guessing he does. It is impossible for the time traveler to kill his grandfather just in case there are no worlds in which a counterpart of him kills a counterpart of his grandfather. But on some resolutions of that vagueness, as the context varies, there are worlds in which a counterpart of him kills a counterpart of his grandfather. It depends on what worlds you're permitted to quantify over. So I'm guessing this exemplfies the same QDF phenomenon.
Post a Comment
<< Home
