Presentation on theme: "Ambiguous contents? Arvid Båve, Higher seminar in Theoretical Philosophy, FLoV, Gothenburg University, 8 May 2013."— Presentation transcript:
Ambiguous contents? Arvid Båve, Higher seminar in Theoretical Philosophy, FLoV, Gothenburg University, 8 May 2013
Last time, I gave an argument for: (7) The LF C of ”All Fs are G” is the same as that of ”a is F” To say that the LF C of sentences S and S’ are the same is to say that the propositions they express have the same syntactic structure. (7) thus entails that there is a syntactic category at the level of thought including both quantifiers everything, every F, some F, etc., and individual concepts like Aristotle, etc.
(7) is just an expression of a common stance in contemporary linguistics, according to which it was wrong to suggest that sentences with the same surface grammars have ”different logical forms”. Focussing now on the present, specific sense of logical form, however, we can see that a potential problem involving structural ambiguity immediately arises for (7).
Objection against (7): (a)”Everyone greets someone” (E) is ambiguous. (b)Ambiguity consists in the expression of distinct contents. (c)If (7) were true, (E) could only express one content. Hence, (7) is false. ((b) is supported by the idea that contents cannot be ambiguous)
The idea is that if (7) were true, (E) would express something like (C) P 2 (Greet, ) P 2 is a function from relational contents and pairs of nominal contents to propositional contents, and, importantly, the order of the elements of the pair indicates only agent-patient relationships. So, P 2 (Greet, ) is the proposition that John greets Mary, and so on. But (C) would be ambiguous! For although it is determined what agent-patient roles the quantifiers take, their scopes are left undetermined!
Reply 1: We can deny (c), i.e., the claim that if (7) were true, (E) could only express one content. That is, we could claim that nominal quantifiers are indexed or marked, so that (E)’s ambiguity consists in its expressing contents like (C), differently indexed, as per: (C’) P 2 (Greet, ) (C’’) P 2 (Greet, ), where the subscripts indicate scope.
Reply 2: We deny (b), and claim that ambiguous sentences can express single contents. It follows that contents can be ambiguous! But what could it mean to say that a content is ambiguous???
Whether on a truth-theoretic or conceptual role theoretic framework, the idea of ambiguous contents is, in one respect, straightforward: On the former, the ambiguity of (C) consists in its having different extensions (intensions, truth- conditions, etc.) on different ”readings”. On the latter, the ambiguity of (C) consists in its figuring in different inferences depending on the ”reading”.
Since the different truth-conditions or inferential patterns can be described unambiguously (using first-order logic), at least part of what it is for contents to be ambiguous can be described.
However, someone might very understandably object that if sameness of content does not entail sameness of inferential role or coextensiveness, it is unintelligible what ”content” means. Let us grant that this is true in one sense of ”content”: let’s stipulate that sameness of A- contents entails sameness of inferential role and coextensiveness (etc.).
But I have been speaking of propositional contents as thought contents, i.e., as the objects of propositions attitudes, and I have described them as a kind of structured entities, involving essentially a number of non-composite contents/concepts. Let us call contents in this sense ”B-contents”. Reply 2 is thus the claim that the B-content of an attitude does not determine determinate A- contents. Similarly, that a sentence expresses a given B-content does not determine which determinate A-contents the sentence has.
Were all these philosophers from William of Occham to Jerry Fodor just wrong, then, in holding that contents cannot possibly be ambiguous? No, for it seems plausible that sameness of non- composite B-contents entails sameness of A- contents. It would be absurd to propose that, e.g., there is a single yet ambiguous concept bank. It is possible that exclusive focus on this kind of example has sustained the idea that contents cannot possibly be ambiguous.
I am proposing merely that (certain cases of) structural ambiguity consists not in the expression of distinct contents, but in the expression of single, yet ambiguous, contents.
OK, but what are the ”readings”, relative to which contents can have different A-contents? Related question: what is it for a speaker to ”interpret” a content one way rather than another?
One proposal is to say that readings of contents are speakers’ dispositions to infer with them (cf. Carruthers). Thus, one way to ”read” (C) is to be disposed to infer with it the way one would infer with ” x y (x greets y)”. Another way is to be disposed to infer with it as one would with ” y x (x greets y)”. E.g., on the former reading, the speaker would infer that John greeted someone from (C), but not that someone greeted John, and conversely for the latter.
Thus, it seems fully coherent to maintain that contents (in our sense) can be ambiguous, and reject premise (b). There is also a consideration indicating that contents probably are ambiguous: In silent reasoning, people are just as prone to commit the fallacy of equivocation as when reasoning aloud. But thinking involves operations on contents. So how can they possibly be unambiguous? Note, however, that people are not prone to equivocate between the different senses of ”bank”. On the present proposal, this is because they are operating on different contents, and there is no risk of conflating them.
There is a response to this argument on behalf of the defender of “Reply 1”, according to which there are no ambiguous contents, but (7) is still true, because nominal quantifiers are always marked. They might respond that when equivocating, we are merely “forgetting” or “confusing” markings, so that contents are always unambiguous after all. The assessment of this response must depend on what “markings” are supposed to be.
Could it be that a content’s being “marked” merely consists in our being disposed to infer with it a certain way? But if so, it does not seem to make sense to say that we can “conflate” or “forget” markings. Suppose a speaker first infers with a content “as if” it were marked one way, and then begins to infer with it as if it were marked another. How, then, is it marked? If we say that its markings changed, it seems that we have said nothing other than that the speakers started inferring with the content differently.
Perhaps the proponent of marked contents should say instead that having a certain marking is distinct from our having certain inferential dispositions, but that our inferential dispositions still depend on the markings of contents. Markings might be some categorical (perhaps physical) properties of contents that merely help guide inferential practice. But the fact that many speakers are prone to equivocating rather ”consistently”, i.e., they have no stable inferential dispositions, it still seems arbitrary to say that the content such a speaker is inferring with is marked one way rather than another.
In general, it seems that the fact that some speakers equivocate so much shows that no ”markings”-solution is possible: on the one hand, ”markings” must somehow connect to inferential practice (otherwise, it would be unable to explain the ambiguity of the relevant sentences); on the other hand, people equivocate to such an extent that to explain how equivocation happens, the markings theorist is forced to take the markings of contents to be very disconnected from speakers’ inferential practice.
It seems that the ”traditionalist” has the same problem. He explains the ambiguity of ”Everyone greeted someone” as consisting in its expressing either of: (1) the proposition that x y (x greets y), (2) the proposition that y x (x greets y). But when people equivocate in silent thinking, e.g., inferring from (1) what can only correctly be inferred from (2), how can this be explained? Are they ”confusing” (1) and (2)? But what grounds the claim that the speaker was first inferring with (1) and confusing it with (2), rather than inferring with (2)? There must be something about the speaker’s way of inferring with the proposition that makes it (1) rather than (2), but in cases of equivocation, both candidate propositions are equally well- (or, rather, ill-) fitted with the way the speaker inferw with it. So the pehnomenon seems to be better explained on the hypothesis that contents/propositions can be ambiguous.
Recap: There are at least two possible replies to the objection against claim (7). Both replies open up new, interesting possibilities. In particular, it would be interesting to investigate which types of linguistic ambiguity are more plausibly handled by recourse to distinct contents (plausibly ”bank”) and which might be more plausibly dealt with by appeal to ambiguous contents (some or perhaps all cases of structural ambiguity). Equivocation, we have seen, is likely to occur in the latter type of case, but not in the former. Thus, equivocation in thinking might be explained by ambiguity in contents. The alternative explanation, which eschews ambiguous contents and opts for markings that can be conflated, requires an account of how markings are supposed to work.