I have previously discussed Devitt’s claim that definite descriptions are ambiguous as between a referntial and an attributive meaning. In my earlier post I gave an argument from Kripke that to my mind is pretty convincing and for which I can’t see a plausible reply.
One of the positive arguments that Devitt gives is a comparison to demonstratives like ‘that’ (argument 3). His point is that the convention by which we express singular thoughts using ‘that’ is straightforwardly a semantic convention (‘if anything is’ as he puts it). Add to this that ‘the’ and ‘that’ are completely interchangable. Anytime that we could use one we can also use the other. So, to take the classic example, if I were to say ‘the murder of Smith is insane’ to express my singular thought about Smith’s murder I could just as well have said ‘that murder of Smith is insane’. This, together with the previous point suggests that ‘the’ has a referential meaning (just like ‘that’).
But if we take the comaprison seriously we will have to postulate an abiguity for ‘that’ as well since there are attributive uses. For insatnce if I am looking at the bloodly murder scene and I say ‘that murder of Smith is insane’ it is clear that I could mean ‘that murder of Smith –whover he is– is insane’. But we have no reason to posit this kind of ambiguity for ‘that’ so we have no reason to postulate it for ‘the’. If anything, what the comparison to demonstratives show is that it isn’t obvious that ‘the’ has the Russellian attributive meaning.
Richard Brown 
Devitt’s primary argument for the ambiguity thesis stems from the fact that both attributive and referential uses of descriptions are regular, systematic, and cross-linguistic. That is, such uses constitute semantic conventions. Kripke’s (1979) and Neale’s (1990) methodological arguments don’t address this basic fact. Neale (2004) has even said so himself. I don’t believe that many people will maintain that attributive uses of complex demonstratives are regular, systematic, and cross-linguistic. And, so, Devitt’s primary argument for the ambiguity thesis doesn’t extend to ‘that’, unless one endorses the conventionality of attributive uses of complex demonstratives. Therefore, your argument isn’t really relevant to the discussion at hand.
Now, I do know that Ludwig and Lepore, King, and Neale maintain that complex demonstratives are really univocal Russellian descriptions. That is, they think the logical form of ‘that F is G’ is [The x: Fx & x = that] Gx. Some considerations pertaining to this position are:
(1) One can bind into complex demonstratives and (supposedly) such a fact cannot by accounted for on a completely referential account of complex demonstratives — e.g. ‘Every man1 responded to [that person he1 liked most]’. (Ludwig and Lepore, and King)
(2) Complex demonstratives have complex syntactic structure and every complex noun phrase in English is a quantifier phrase, empirically speaking. (Neale)
(3) There are empty but meaningful uses of complex demonstratives. (King)
This, I think, shows that even if one maintained that there were regular, systematic, and cross-linguistic attributive uses of complex demonstrative, one wouldn’t have to postulate an ambiguity a la Devitt. One could have a purely referential analysis of ‘that’ but hold that complex demonstratives are purely quantificational — this is Ludwig and Lepore, King, and Neale’s view. And, so, your argument, even if cleaned up, wouldn’t really touch Devitt’s argument.
However, there is an out. You could, like Neale, maintain that referential uses of definite descriptions have the following logical form: [The x: Fx & x = that] Gx. And, then, you could maintain that the elliptical completion ‘in question’ provides ‘x = that’. (i.e. A referential use of ‘the murderer is insane’ is shorthand for the use of ‘the murderer in question is insane’.). But, then, you’d be saddled with a non-compositional semantics. And, I think you don’t want that. I, myself, don’t mind much about it though.
Hey Frank thanks for the comment! Very helpful.
Devitt’s primary argument for the ambiguity thesis stems from the fact that both attributive and referential uses of descriptions are regular, systematic, and cross-linguistic. That is, such uses constitute semantic conventions.
Right, no disagrees that we regularly and standardly use definite descriptions to express singular thoughts but without further argument this, by itself, doesn’t establish the existence of a semantic convention. It may be a pragmatic convention, or what is most likely pretty much the same, Bach’s notion of standardization that is at work. The comparison to demonstratives is what is supposed to get us to the conslusion that the convention in question is really a semantic convention.
I don’t believe that many people will maintain that attributive uses of complex demonstratives are regular, systematic, and cross-linguistic. And, so, Devitt’s primary argument for the ambiguity thesis doesn’t extend to ‘that’, unless one endorses the conventionality of attributive uses of complex demonstratives.
Well, I am not so sure about this. It seems to me that we do regularly and systematically use complex demonstratives to express singular thoughts (I can’t say anyhting about cross-linguistic uses)…But either way the point is that if it is true that we can literally substitue ‘that’ for ‘the’ in every case and vice versa then we should be able to do it in the attributive uses as well, in fact it seems perfectly natural to do this; just as natural as in the referential cases.
Re 1-3 can’t one say the very same thing in each case about ‘the x’? So doesn’t that argument cut both ways?
“Right, no disagrees that we regularly and standardly use definite descriptions to express singular thoughts but without further argument this, by itself, doesn’t establish the existence of a semantic convention. It may be a pragmatic convention, or what is most likely pretty much the same, Bach’s notion of standardization that is at work. The comparison to demonstratives is what is supposed to get us to the conclusion that the convention in question is really a semantic convention.”
Fair enough. But, this is a distinct argument. One, imagine, we couldn’t have in this format. But, for controversy’s sake, I’ll say something intentionally provocative. Your position, Dr. Brown, seems to be that once God gives a word its meaning, that meaning is unchangeable. That is, you seem to hold the thesis that meaning change inhabits the same possible world as 2 + 2 = 5. This, I think, is an unrealistic position.
“Well, I am not so sure about this. It seems to me that we do regularly and systematically use complex demonstratives to express singular thoughts (I can’t say anything about cross-linguistic uses)…But either way the point is that if it is true that we can literally substitute ‘that’ for ‘the’ in every case and vice versa then we should be able to do it in the attributive uses as well, in fact it seems perfectly natural to do this; just as natural as in the referential cases.”
Again, fair enough. In that case, however, one could have a decent reason to posit an ambiguity within ‘that’. The reason will be Devitt’s. In particular, complex demonstratives regularly admit of both quantificational and referential uses. In your original post, you held that there is ‘no reason to posit this kind of ambiguity for ‘that’’. And, so, you concluded that there was no reason to conclude that ‘the’ is ambiguous. But, now, we have a decent reason. And, moreover, you’ve actually provided it! Of course, Nunberg does provide numerous cases in which ‘that’ and ‘the’ aren’t semantically interchangeable.
“Re 1-3 can’t one say the very same thing in each case about ‘the x’? So doesn’t that argument cut both ways?”
Exactly! And this, again, is why one could have a decent reason for postulating the required ambiguity.
In fairness though, I, myself, don’t think that one really needs to postulate an ambiguity within ‘that’ in order to explain the attributive use of complex demonstratives. These, I think, can be handled by maintaining the referential nature of ‘that’ and embedding it within a complex quantifier phrase a la Neale. That is, an attributive use of ‘that F is G’ amounts to [The x: Fx & x = that] Gx. One would, I think, only need to postulate an ambiguity within ‘that’ if one could regularly use ‘that’ alone as a quantifier. But, that seems unlikely. In this respect, ‘the’ is different. It is ungrammatical to use the definite article as a stand-alone phrase but it’s perfectly grammatical to use ‘that’ as a stand-alone phrase.
So, in short, the analogy that you, following Devitt, are drawing between ‘the’ and ‘that’ doesn’t really seem apt. These phrases have distinct syntactic features and, as Nunberg demonstrates, aren’t always semantically interchangeable, and so aren’t really analogous.
Hey Frank, I tried to find that Nunberg paper but I couldn’t can you give me the reference? I am interested in looking at it.
I’ll say something intentionally provocative. Your position, Dr. Brown, seems to be that once God gives a word its meaning, that meaning is unchangeable. That is, you seem to hold the thesis that meaning change inhabits the same possible world as 2 + 2 = 5. This, I think, is an unrealistic position.
Ouch! Et tu, Dr. Pupa? I do think that there is a resistance to change inherent in natural languages. This is necessary, it seems to me, given the Gricean view about communication. Given that communication is a coopertive endevour that relies on what I, the speaker, can reliably expect you, the hearer, to infer about my mental states based on what I utter then we need a relatively stable structure. This doesn’t mean that meaning change is impossible –it is probably much more likely for nouns and maybe even verbs than it is for things like pronouns and determiners– it just means thatit will be slow and that the language will resist it.
Another consideration here is that meaning of lexical items is arguably simply whatever a competant speaker would need to know in order to use the item correctly. If so then it is obviously an advantage to have a relatively stable set of things that people need to know.
And, so, you concluded that there was no reason to conclude that ‘the’ is ambiguous. But, now, we have a decent reason. And, moreover, you’ve actually provided it!
Well, I don’t think that this follows. The point I was trying to make was that the regularity of use bit doesn’t by itself give any support to the claim that there is a semantic convention at work, so the regularity of attributive uses of ‘that’ doesn’t support an ambiguity. What we would need is some independant argument that showed that the regularity is due to a semantic convention. This is exactly the purpose that was originally served by the comparison to ‘that’; there is no such argument for ‘that’ and so the strategy breaks down.
These, I think, can be handled by maintaining the referential nature of ‘that’ and embedding it within a complex quantifier phrase a la Neale. That is, an attributive use of ‘that F is G’ amounts to [The x: Fx & x = that]
How is it that the embeded use of ‘that’s retains its referential meaning? What does it refer to?
One would, I think, only need to postulate an ambiguity within ‘that’ if one could regularly use ‘that’ alone as a quantifier. But, that seems unlikely.
This depends on how you feel about ‘those’. Suppose that there are five boxes and you ask me ‘how many boxes are we supposed to take?” abd I say ‘those’ [gesturing at three of the boxes]. Or, said by a student ‘How many of us should do the extra credit?” Response: ‘those who want to pass the class’. Or, if you don’t like these how about something like this ‘philosophy is that by which we understand our own concepts’. I think we could even get ‘that’ as an answer to the how many boxes question…
In this respect, ‘the’ is different. It is ungrammatical to use the definite article as a stand-alone phrase but it’s perfectly grammatical to use ‘that’ as a stand-alone phrase.
Is this really relevant? I thought the point was that the point was tht the two were interchangable in referring phrases not tout court
Nunberg’s paper is called ‘Descriptive Indexicals’. It’s in Reimer’s collection ‘Descriptions and Beyond’.
“I do think that there is a resistance to change inherent in natural languages. This is necessary, it seems to me, given the Gricean view about communication. Given that communication is a cooperative endeavor that relies on what I, the speaker, can reliably expect you, the hearer, to infer about my mental states based on what I utter then we need a relatively stable structure. This doesn’t mean that meaning change is impossible –it is probably much more likely for nouns and maybe even verbs than it is for things like pronouns and determiners– it just means that it will be slow and that the language will resist it. Another consideration here is that meaning of lexical items is arguably simply whatever a competent speaker would need to know in order to use the item correctly. If so then it is obviously an advantage to have a relatively stable set of things that people need to know.”
No argument here.
“The point I was trying to make was that the regularity of use bit doesn’t by itself give any support to the claim that there is a semantic convention at work, so the regularity of attributive uses of ‘that’ doesn’t support an ambiguity. What we would need is some independent argument that showed that the regularity is due to a semantic convention. This is exactly the purpose that was originally served by the comparison to ‘that’; there is no such argument for ‘that’ and so the strategy breaks down.”
Fine. But, this cuts both ways. As I’ve shown in my dissertation, one can provide a perfectly respectable pragmatic account of attributive uses of descriptions from a purely referential semantics. For that matter, Sainsbury showed that one can account for attributive uses within a negative free logic if one assumes a purely referential semantics. So, in response, I’d say this. If Devitt’s argument doesn’t establish an ambiguity, then one cannot postulate a univocal quantificational account of ‘the’ unless one can demonstrate that the pragmatic explanation really runs from quantificational to referential as opposed to referential to quantificational. For my part, I just don’t see how you can do this. This, I call, the ‘argument from arbitrariness’. That is, the choice between Russell and Strawson becomes irrelevant from a pragmatic perspective.
So, then, why be a Russellian? The answer: because, well, I read Russell first.
I’ll have to think about the ‘those’ cases. My hunch, however, is that ‘those’ and ‘these’ are bit more intricate than simply the plurals of ‘this’ and ‘that’. But, I must say, each case you give is irrelevant. The first use of ‘those’ is referential and so irrelevant. The second use is quantificational but complex and so irrelevant. Again, the Anselm use of ‘that’ is quantificational but then again its complex — the DP is ‘that by which we understand our own concepts’ not ‘that’ — and so irrelevant.
No argument here.
Ah, good. So, you’ll drop the charge of Platonism?
As I’ve shown in my dissertation
Nice name drop! 🙂
one can provide a perfectly respectable pragmatic account of attributive uses of descriptions from a purely referential semantics.
Yeah, I agree; part of the point I was making was that what the comparison to ‘that’ might actually show is that this is the better way to go.
If Devitt’s argument doesn’t establish an ambiguity, then one cannot postulate a univocal quantificational account of ‘the’ unless one can demonstrate that the pragmatic explanation really runs from quantificational to referential as opposed to referential to quantificational. For my part, I just don’t see how you can do this.
Bach tries to do this as one of his arguments against Devitt. He argues that we have to exploit the quantificational meaning of ‘the’ in order to figure out that the person was making a referential use; but I take it you probably don’t like this argument.
So, then, why be a Russellian? The answer: because, well, I read Russell first.
That’s not entirely fair, is it? It may be teh case that you can choose which ever semantic theory that one likes (Russell or Strawson) and then explain the other pragmatically but the Russellian approach has the virtue of solving some puzzles that the other approach doesn’t.
the first use of ‘those’ is referential and so irrelevant.
What makes you think that? As I told the story –or meant to– the speaker intends to be answering the question ‘how many boxes?’ and not to be refferring to the boxes.
As for the others, I guess I misread your point; but looking back at it I now see what you meant. But why should we think this? It is very unusual for a quantifier to be used by itself, why should ‘that’ be any different?
Oh and thanks for the reference. I definately should get that volume!
“Bach tries to do this as one of his arguments against Devitt. He argues that we have to exploit the quantificational meaning of ‘the’ in order to figure out that the person was making a referential use; but I take it you probably don’t like this argument.”
I admit it’s a respectable position. But, as I showed (more name dropping!), there is another respectable position in logical space: we have to exploit the referential meaning of ‘the’ in order to figure out that a person was making a quantificational use. This, of course, is not different from what one does to figure out that a speaker was making a quantificational use of a complex demonstrative or a pronoun. So, we need some independent reason to choose Bach’s story over mine. Again, the choice will be arbitrary.
“That’s not entirely fair, is it?”
Look, there is a sociological issue here. We philosophers, as a matter of training, are supplied the following items to read in this order: Russell, Strawson, Donnellan, Kripke, and, maybe, Neale. We, then, are supplied the following history: Russell was right, Strawson is irrelevant, Donnellan was confused, and Kripke sorted him out. So, I don’t think I’m being unfair; that’s the training. Interestingly, no one ever bothers to tell us the following juicy tidbits. Donnellan’s paper is mainly about Strawson; Kripke doesn’t really think that Russell’s theory survives because, well, not all incomplete referential uses can be cleaned up and, if you look at the footnotes, because through regularization descriptions will perhaps come to have two distinct meanings. Grice, himself, tends toward a Strawsonian approach at the end of his discussion on descriptions.
“It may be the case that you can choose which ever semantic theory that one likes (Russell or Strawson) and then explain the other pragmatically but the Russellian approach has the virtue of solving some puzzles that the other approach doesn’t.”
Now, any puzzle that a quantificational approach can solve — e.g. belief contexts, modal contexts, identity contexts, non-acquaintance contexts, negative existentials, empty descriptions, etc. — can also be solved by a referential approach. All the Strawsonian needs to do is take a cue from his direct-reference buddies and their discussions of the same exact problems with proper names.
“It is very unusual for a quantifier to be used by itself, why should ‘that’ be any different?”
I think there’s a better question. It’s very unusual that a quantifier can be by itself in a referential context. Why is ‘that’ different? Because it’s obviously no quantifier.
This has been good fun. I let you have the last word.
Now, any puzzle that a quantificational approach can solve — e.g. belief contexts, modal contexts, identity contexts, non-acquaintance contexts, negative existentials, empty descriptions, etc. — can also be solved by a referential approach. All the Strawsonian needs to do is take a cue from his direct-reference buddies and their discussions of the same exact problems with proper names.
Yeah, I really think that this is where the battle must ultimately be fought. I, for one, do not find any of these soulutions even remotely plausible….but that’s for another post!!
Re: “Grice, himself, tends toward a Strawsonian approach at the end of his discussion on descriptions.”
That IS plausible. I always love to read that ending passage in “Presupposition and conversational implicature” (in WoW, Studies in the Way of Words, if that is what is meant) as a nod to Grice´s OTHER paper by Grice, on this. That is: his “Definite descriptions”, which is Part IV, I think, of his “Vacuous Names”, and as repr. by Ostertag in the MIT reader, “Definite description”. So one has to be slightly serious here.
Some dates:
1969. Vacuous Names, for the Quine ´festschrift´. Repr. in part in Ostertag, Definite descriptions.
1970. Presupposition and conversational implicature. in WoW.
There are OTHER mimeos to consider here, too, such as “Definite descriptions in Russell and in the vernacular” by Grice, cited by Bealer in Concept and Quality.
—- In any case, Grice´s Presupposition and Conversational Implicature owes a lot to his conversations with Hans Sluga. When Grice had this essay repr. in Cole, 1981, Sluga came out as “Shuga”. WoW is Shuga-free, even — the footnote was dropped!
BUT I love the idea of the definite description alla “conversational dossier” in “Vacuous Names” (I cannot see how Ostertag got his cake and ate it by murdering the essay like that!). Grice in a foonote refers to Donnellan, but of course thinks that HIS identificatory-versus-non-identificatory uses is much better, and right he is too!
T. E. Patton has let me have a very intriguing critique of Kripke´s misuse of Grice to refute Donnellan, and I may have mentioned something along those lines at the Grice Club, where everybody is welcomed to drop a note, or two!
Cheers,
J. L. Speranza
for the griceclub.blogspot.com