Quasi-Realism and Natural Language Semantics
Surely this has been remarked upon elsewhere by someone clever, but if quasi-realism is committed to the view that sentences containing moral terms do not have truth-values (i.e. don't express propositions) but rather truth-like-values (i.e. express accept-o-sitions or some such), then wacky things might be expected to happen to the semantics for natural language.
Compare "Killing Rob softly with my song is fun" and "Killing Rob softly with my song is very wrong". The standard view of all of the expressions (ignoring "wrong" for the moment) in these sentences is that their meaning consists in their contribution to determining a truth-value at different situations. So, for example, the meaning of "Rob" is such as to get together with "Killing __ softly with my song" to determine a value that then gets together with "__ is fun"
to yield, as it happens, TRUE as the value.
Somehow, however, once you enter a moral term, like "wrong" into the sentence, it had better turn out (if quasi-realism is true) that these words function to contribute to determining a truth-like-value (an accept-o-sition) instead.
One way for this to happen would be if every term in the language had two semantic values: one that contributes to determining truth-values and one that contributes to determining truth-like-values. This is cool, I suppose, because it makes work for semanticists and, as we all know, idle hands are the devil's tools. The point about idle hands applies especially to semanticists, who are a motley bunch.
Another way for this to happen, I suppose, would be to retain the normal truth-determining values for terms, but for words like "wrong" to have a type-shifting semantic value: they convert truth-value-determining meanings to truth-like-value-determining meanings. They'd have to take wide-scope over the whole sentence in order to manage this trick... Actually, I am not sure that "scope" is the right term, but they'd have to be last in the computation. All the non-moral terms do their work, making a partial determination of a truth-value and then the moral terms convert this semantic value into a truth-like-value.
Opaque, no? Obviously, I need to sit down and do the math.
I might have expected that if either of the previous two possibilities were actual, then linguists would have stumbled across the quasi-realist phenomenon independently of moral philosophers.
*A third possibility, and one I suspect, not having read his book, that Gibbard advocates: There is no ambiguity. Sentences do not have truth-values in the first instance, but instead express a kind of function that in some circumstances yields a proposition (and thence a truth-value) and in other circumstances yields an accept-o-sition (and thence a truth-like-value).
Compare "Killing Rob softly with my song is fun" and "Killing Rob softly with my song is very wrong". The standard view of all of the expressions (ignoring "wrong" for the moment) in these sentences is that their meaning consists in their contribution to determining a truth-value at different situations. So, for example, the meaning of "Rob" is such as to get together with "Killing __ softly with my song" to determine a value that then gets together with "__ is fun"
to yield, as it happens, TRUE as the value.
Somehow, however, once you enter a moral term, like "wrong" into the sentence, it had better turn out (if quasi-realism is true) that these words function to contribute to determining a truth-like-value (an accept-o-sition) instead.
One way for this to happen would be if every term in the language had two semantic values: one that contributes to determining truth-values and one that contributes to determining truth-like-values. This is cool, I suppose, because it makes work for semanticists and, as we all know, idle hands are the devil's tools. The point about idle hands applies especially to semanticists, who are a motley bunch.
Another way for this to happen, I suppose, would be to retain the normal truth-determining values for terms, but for words like "wrong" to have a type-shifting semantic value: they convert truth-value-determining meanings to truth-like-value-determining meanings. They'd have to take wide-scope over the whole sentence in order to manage this trick... Actually, I am not sure that "scope" is the right term, but they'd have to be last in the computation. All the non-moral terms do their work, making a partial determination of a truth-value and then the moral terms convert this semantic value into a truth-like-value.
Opaque, no? Obviously, I need to sit down and do the math.
I might have expected that if either of the previous two possibilities were actual, then linguists would have stumbled across the quasi-realist phenomenon independently of moral philosophers.
*A third possibility, and one I suspect, not having read his book, that Gibbard advocates: There is no ambiguity. Sentences do not have truth-values in the first instance, but instead express a kind of function that in some circumstances yields a proposition (and thence a truth-value) and in other circumstances yields an accept-o-sition (and thence a truth-like-value).
posted by Idris at 1:53 PM
0 Comments:
Post a Comment
<< Home