Semantic Innocence
0 Followers
Recent papers in Semantic Innocence
Many recent semantic theories have involved explicit acceptance of the following two theses: 1. DIRECT REFERENCE (DR): The utterance of a simple sentence containing names or demonstratives normally expresses a "singular proposition" -- a... more
Many recent semantic theories have involved explicit acceptance of the following two theses:
1. DIRECT REFERENCE (DR): The utterance of a simple sentence containing names or demonstratives normally expresses a "singular proposition" -- a proposition that contains as constituents the individuals referred to, and not any descriptions or conditions on them.
2. SEMANTIC INNOCENCE (SI): The utterances of the embedded sentences in belief reports express just the propositions they would if not embedded, and these propositions are the contents of the ascribed beliefs.
Such theories face a well-known difficulty: they seem to conflict with a third, and apparently obviously true, thesis:
3. OPACITY (O): Substitution of coreferring names and demonstratives in belief reports does not always preserve the truth of those reports.
In order to meet this difficulty, two different strategies have been proposed.
A. Conventional Implicature: The first suggestion, adopted by Salmon and Soames, is to claim that O is false: despite appearances, the substitution of coreferring names and demonstratives in belief reports does preserve the truth of those reports.
Now, of course, this suggestion leaves us with a puzzle, namely: why do we ordinarily suppose, and speak as if it were the case, that O is true? Here, Salmon and Soames suggest that belief reports carry conventional (or generalised) implicatures which can change under substitution of coreferring embedded names and demonstratives.
Neither Soames nor Salmon has given any details about the nature of these implicatures. However, they do say that what gets implicated is information about the mode of presentation under which a singular proposition is grasped by a subject. Consequently, it seems hard to resist the suggestion that what they require is a compositional theory according to which implicated modes of presentation associated with sentences are composed of implicated modes of presentation associated with the words which make up those sentences.
But then the question naturally arises why it should not be supposed that these allegedly implicated modes of presentation are actually part of the semantic content of belief reports. Given that we need a theory which associates modes of presentation with words, won’t all of this talk about implicated modes of presentation be just a pointless and unmotivated complication in the theory?
B. Unarticulated constituents: The second suggestion, adopted by Crimmins and Perry, is to deny that the theses DR, SI, and O are inconsistent. This response seems most unpromising. Consider the following quasi-logical principle which connects the notions of truth and semantic value:
4. FUNDAMENTAL SEMANTIC PRINCIPLE (FSP): If the substitution of expression E1 for expression E2 in sentence S (in context C) leads to a change in (literal) truth-value, then this change is due to the semantic values of E1 and E2 (in context C)
I take it that this is more or less a definition of what it is to be a semantic value: semantic values are whatever it is that words contribute to the determination of the literal truth-values of sentences in which they occur (upon particular occasions of utterance, or more generally, of tokening of those sentences). Moreover, I take it that it is obvious that FSP entails that if the substitution of a name or demonstrative E1 for a coreferring name or demonstrative E2 in a sentence S (in a context C) leads to a change in the literal truth-value of S, then it follows -- contrary to DR, or SI, or both -- that E1 and E2 do not have the same semantic content.
Not surprisingly, Crimmins and Perry are at least tacitly committed to the rejection of FSP. In their words, their view is as follows:
"It is very common in natural languages for a statement to exploit unarticulated constituents. When we consider the conditions under which such a statement is true, we find it expresses a proposition which has more constituents in it than can be traced to expressions in the sentence that was spoken. Each constituent of the content that is not itself the content of some expression in the sentence, is an unarticulated constituent of the content of the statement. ... The important principle to be learned is that a change in the wording can precipitate a change in propositional constituents, even when the words do not stand for constituents."
In other words, the "important principle" to which Crimmins and Perry wish to draw attention is that the substitution of an expression E1 for an expression E2 in a sentence S (in a context C) can lead to a change in the semantic content of the sentence S (in the context C) even though E1 and E2 have exactly the same semantic content -- and this is the denial of FSP.
In order to illustrate the notion of "unarticulated constituency", Crimmins and Perry consider the example of utterances of:
(1) It's raining.
As they note, if one utters (1), one will be understood to be claiming that it is raining at the time of one's utterance at some place which is indicated by features of the context of utterance. (Often this place will be the place of one's utterance, but it needn't be.) Moreover, there is no (surface) expression in (1) which has the place in question as its content.
However, there are two points to note about this example. Firstly, this example does nothing to support the principle that a change in wording can precipitate a change in propositional content even when the words do not stand for the constituents. Rather, this example supports the principle that there can be a change in propositional constituents when there is no change in wording even in the case of sentences which contain no indexical expressions. (In other words: sentences can exhibit an indexicality which is not derived from the indexicality of their component expressions.)
Secondly, and more importantly, the genuine principle which can be derived from cases like (1) does nothing to support the account of belief reports which Crimmins and Perry wish to defend. In their view, the semantic contents of
(2a) Scott believes that Hesperus rises in the morning.
and
(2b) Scott believes that Phosphorus rises in the morning.
in a given context C, may differ because the context contributes different unarticulated constituents to (2a) and (2b). But it is incredible to suppose that cases like (1) lend any credence to such a view. In the case of (1), there is no word which can plausibly be connected to the place which (allegedly) forms part of the semantic content of (1). However, in the case of (2a) and (2b) there are obvious candidate words -- namely, "Hesperus" and "Phosphorus" -- which could be semantically associated with the (allegedly) unarticulated constituents of the semantic contents of (2a) and (2b). So why suppose that these constituents of the semantic contents of (2a) and (2b) are not (parts of) the semantic contents of the words "Hesperus" and "Phosphorus"?
This mystery is deepened when we note that Crimmins and Perry claim that "the whole utterance, the context and the words uttered, are relevant to identifying the unarticulated constituent". The point in the first example seemed to be that, since there is no word in (1) which could have the place in question as its semantic value, it is necessary to suppose that the place in question is an unarticulated semantic constituent; but now we are told that in (2a) and (2b) the words "Hesperus" and "Phosphorus" are "relevant to identifying" certain constituents of the semantic contents of (2a) and (2b), and yet that these words can't have those constituents as (parts of) their semantic values.
This time, I take it that the obvious question to ask is: Why not suppose that the semantic constituents which Crimmins and Perry claim are unarticulated semantic constituents in propositional attitude reports are actually (parts of) the semantic contents of the names and demonstratives which appear in those reports? Isn't all this talk of unarticulated constituents of belief reports a pointless and unmotivated complication in the theory?
In sum, then, the crucial question for the two types of semantic theory which I have discussed centres on the thesis of semantic innocence. Crimmins and Perry claim that it is "well-motivated by many considerations in the philosophy of language". However, I cannot see that this is so; rather, it seems to me that this thesis is completely unmotivated, and that, in virtue of the above considerations, it is obvious that semantic theory would be better off without it. The point of the rest of this note is to explain why this is so.
1. DIRECT REFERENCE (DR): The utterance of a simple sentence containing names or demonstratives normally expresses a "singular proposition" -- a proposition that contains as constituents the individuals referred to, and not any descriptions or conditions on them.
2. SEMANTIC INNOCENCE (SI): The utterances of the embedded sentences in belief reports express just the propositions they would if not embedded, and these propositions are the contents of the ascribed beliefs.
Such theories face a well-known difficulty: they seem to conflict with a third, and apparently obviously true, thesis:
3. OPACITY (O): Substitution of coreferring names and demonstratives in belief reports does not always preserve the truth of those reports.
In order to meet this difficulty, two different strategies have been proposed.
A. Conventional Implicature: The first suggestion, adopted by Salmon and Soames, is to claim that O is false: despite appearances, the substitution of coreferring names and demonstratives in belief reports does preserve the truth of those reports.
Now, of course, this suggestion leaves us with a puzzle, namely: why do we ordinarily suppose, and speak as if it were the case, that O is true? Here, Salmon and Soames suggest that belief reports carry conventional (or generalised) implicatures which can change under substitution of coreferring embedded names and demonstratives.
Neither Soames nor Salmon has given any details about the nature of these implicatures. However, they do say that what gets implicated is information about the mode of presentation under which a singular proposition is grasped by a subject. Consequently, it seems hard to resist the suggestion that what they require is a compositional theory according to which implicated modes of presentation associated with sentences are composed of implicated modes of presentation associated with the words which make up those sentences.
But then the question naturally arises why it should not be supposed that these allegedly implicated modes of presentation are actually part of the semantic content of belief reports. Given that we need a theory which associates modes of presentation with words, won’t all of this talk about implicated modes of presentation be just a pointless and unmotivated complication in the theory?
B. Unarticulated constituents: The second suggestion, adopted by Crimmins and Perry, is to deny that the theses DR, SI, and O are inconsistent. This response seems most unpromising. Consider the following quasi-logical principle which connects the notions of truth and semantic value:
4. FUNDAMENTAL SEMANTIC PRINCIPLE (FSP): If the substitution of expression E1 for expression E2 in sentence S (in context C) leads to a change in (literal) truth-value, then this change is due to the semantic values of E1 and E2 (in context C)
I take it that this is more or less a definition of what it is to be a semantic value: semantic values are whatever it is that words contribute to the determination of the literal truth-values of sentences in which they occur (upon particular occasions of utterance, or more generally, of tokening of those sentences). Moreover, I take it that it is obvious that FSP entails that if the substitution of a name or demonstrative E1 for a coreferring name or demonstrative E2 in a sentence S (in a context C) leads to a change in the literal truth-value of S, then it follows -- contrary to DR, or SI, or both -- that E1 and E2 do not have the same semantic content.
Not surprisingly, Crimmins and Perry are at least tacitly committed to the rejection of FSP. In their words, their view is as follows:
"It is very common in natural languages for a statement to exploit unarticulated constituents. When we consider the conditions under which such a statement is true, we find it expresses a proposition which has more constituents in it than can be traced to expressions in the sentence that was spoken. Each constituent of the content that is not itself the content of some expression in the sentence, is an unarticulated constituent of the content of the statement. ... The important principle to be learned is that a change in the wording can precipitate a change in propositional constituents, even when the words do not stand for constituents."
In other words, the "important principle" to which Crimmins and Perry wish to draw attention is that the substitution of an expression E1 for an expression E2 in a sentence S (in a context C) can lead to a change in the semantic content of the sentence S (in the context C) even though E1 and E2 have exactly the same semantic content -- and this is the denial of FSP.
In order to illustrate the notion of "unarticulated constituency", Crimmins and Perry consider the example of utterances of:
(1) It's raining.
As they note, if one utters (1), one will be understood to be claiming that it is raining at the time of one's utterance at some place which is indicated by features of the context of utterance. (Often this place will be the place of one's utterance, but it needn't be.) Moreover, there is no (surface) expression in (1) which has the place in question as its content.
However, there are two points to note about this example. Firstly, this example does nothing to support the principle that a change in wording can precipitate a change in propositional content even when the words do not stand for the constituents. Rather, this example supports the principle that there can be a change in propositional constituents when there is no change in wording even in the case of sentences which contain no indexical expressions. (In other words: sentences can exhibit an indexicality which is not derived from the indexicality of their component expressions.)
Secondly, and more importantly, the genuine principle which can be derived from cases like (1) does nothing to support the account of belief reports which Crimmins and Perry wish to defend. In their view, the semantic contents of
(2a) Scott believes that Hesperus rises in the morning.
and
(2b) Scott believes that Phosphorus rises in the morning.
in a given context C, may differ because the context contributes different unarticulated constituents to (2a) and (2b). But it is incredible to suppose that cases like (1) lend any credence to such a view. In the case of (1), there is no word which can plausibly be connected to the place which (allegedly) forms part of the semantic content of (1). However, in the case of (2a) and (2b) there are obvious candidate words -- namely, "Hesperus" and "Phosphorus" -- which could be semantically associated with the (allegedly) unarticulated constituents of the semantic contents of (2a) and (2b). So why suppose that these constituents of the semantic contents of (2a) and (2b) are not (parts of) the semantic contents of the words "Hesperus" and "Phosphorus"?
This mystery is deepened when we note that Crimmins and Perry claim that "the whole utterance, the context and the words uttered, are relevant to identifying the unarticulated constituent". The point in the first example seemed to be that, since there is no word in (1) which could have the place in question as its semantic value, it is necessary to suppose that the place in question is an unarticulated semantic constituent; but now we are told that in (2a) and (2b) the words "Hesperus" and "Phosphorus" are "relevant to identifying" certain constituents of the semantic contents of (2a) and (2b), and yet that these words can't have those constituents as (parts of) their semantic values.
This time, I take it that the obvious question to ask is: Why not suppose that the semantic constituents which Crimmins and Perry claim are unarticulated semantic constituents in propositional attitude reports are actually (parts of) the semantic contents of the names and demonstratives which appear in those reports? Isn't all this talk of unarticulated constituents of belief reports a pointless and unmotivated complication in the theory?
In sum, then, the crucial question for the two types of semantic theory which I have discussed centres on the thesis of semantic innocence. Crimmins and Perry claim that it is "well-motivated by many considerations in the philosophy of language". However, I cannot see that this is so; rather, it seems to me that this thesis is completely unmotivated, and that, in virtue of the above considerations, it is obvious that semantic theory would be better off without it. The point of the rest of this note is to explain why this is so.