Lewis Powell
In Insensitive Semantics, Chapter 7: Objections to Radical Contextualism (I), Cappelen and Lepore give three tests for context sensitivity.
Test 1: Blocking inter-contextual disquotational direct reports
Take a clearly context-sensitive term like, "she" in the sentence "She is a banker".
Utterance made in context C1, by Jones, while in a bank, demonstrating a female bank employee: "She is a banker."
Indirect report in C2, made by Smith, who knows Jones, demonstrating a lawyer: "Jones said that she is a banker."
This is what test 1 blocking looks like. The indirect report is unacceptable because there was a context sensitive term and so Smith's disquotation is insufficient for a true report.
Test 2: Blocking collective descriptions
Staying with "she", consider the following:
Utterance 1: Jones, demonstrating the manager of the First Bank of Fredonia: "She is a banker"
Utterance 2: Smith, demonstating the manager of the Second Bank of Fredonia: "She is a banker"
The following is clearly false: "Jones and Smith both said that she is a banker" (note that it really doesn't matter who is indicated, since there is no one female who both Jones and Smith said was a banker.
Test 3: Admit of Real Context Shifting Arguments
The final test is, I think, slightly more confusing.
Consider:
U: "She is a banker"
(1) There can be false utterances of "She is a banker" even though she is a banker.
(1*) There can be true utterances of "She is a banker" even though she is not a banker.
If one imagines a demonstration accompanying the non-quoted she, it should be clear that (1) and (1*) are true.
Ok, so that's how the tests are supposed to work. In the chapter, Cappelen and Lepore argue that quantifiers do not pass the test, using the quantifiers "at least one" and "at least two". My intuition is that "at least one" is able to be used disquotationally, and so that is evidence against its context sensitivity (test 1). My intuitions are substantially weaker for test 2, where it is alleged that the following is acceptable:
If there is a true utterance in some context of "Jill bought at least two penguins" and a true utterance in some context of "Jill bought at least two ducks" then it is perfectly natural to say: "Jill bought at least two penguins and ducks". Set aside that worry for a moment though. In fact, set aside the quantifiers they've chosen for a moment, and consider the following:
(2) There can be true utterances of "every bat is flying" even though not every bat is flying.
(2*) There can be false utterances of "every bat is flying" even though every bat is flying.
(2*) strikes me as obviously false. (2) on the other hand, is far less objectionable, and were my intuitions not so skewed, I might find it utterly unproblematic. Consider the utterance as made by a Zookeeper who is showing a tour group the bat cage, and none of the bats in the cage are at rest. Now say sentence two stressing the non-quoted "every" so that it reads like this:
(2) There can be true utterances of "every bat is flying" even though not every bat is flying.
This sounds fine to me. Stressing the every seems to be a fair counterpart to imagining the "she is a banker" case with a demonstration for the non-quoted "she".
For the sake of completeness, I'll put "every" through the first two tests.
T1:
Beth, a zookeeper standing in front of the bat cage, in which no bat is at rest: "Every bat is flying".
Report by Steven, who is in a cave filled with many sleeping bats and who just got off the phone with Beth. He turns to his fellow spelunkers and says: "Beth said that every bat is flying"
Evaluation: "Every" seems fine to use disquotationally here, and I intentionally chose a case in which the context of the report was one in which not every bat was flying. This type of result is claimed by Capellen and Lepore to indicate that the word tested (in this case, "every") is not context sensitive. I suspect that contextualists would have something to say about this case, but I hardly think its necessary in light of the following two tests.
T2:
Beth, a zookeeper standing in front of the bat cage, in which no bat is at rest: "Every bat is flying".
Steven, having just dropped his flashlight, startled the bats in the cave so that none of them remained at rest. He says, "Every bat is flying."
Ming the Merciless, watching both Beth and Steven on some sort of monitors turns to an advisor and says, "they both said that every bat is flying"
Evaluation: Something seems wrong about this attribution.
So, T1 indicates non-context-sensitivity, while T2 and T3 indicate context sensitivity for "every" (or, arguably, "bat", but my money is on "every").
Most people probably don't see it as a startling discovery that there is evidence for context sensitivity for quantifiers, but Cappelen and Lepore claimed that the tests came down on the side of quantifiers not being context sensitive, so I thought I'd see if my reasoning above seemed right to others.
(this post is due in part to discussion I had with Geoff Georgi, a USC Graduate Student and all around good guy)
