Thursday, January 25, 2007

Paraphrasing! (Huh, yeah) What is it good for?

Metaphysicians and other philosophers are often uncomfortable with the supposed ontological commitments of many seemingly true ordinary sentences of English. One approach to blunting these commitments is to offer paraphrases of them. I am interested in the details of the approach. In particular, I would like to know whether someone who offers a paraphrase of an ordinary English sentence should take that sentence to be true.

Consider the following English sentence:
(F) There are fictional characters.
Many metaphysicians are uncomfortable with (F)'s apparent commitment to fictional characters. Since the paraphrase approach is intended to help one avoid such commitments, many metaphysicians are likely to offer a paraphrase of (F). Should such philosophers take (F) to be true? If so, then the following claim is true:
(T-F) "There are fictional characters" is true.
However, the following instance of the Tarski bi-conditional also seems to be true:
(I-TB) "There are fictional characters" is true iff there are fictional characters.
However, (F), which is just the claim that one would've thought someone employing the paraphrase strategy would like to avoid, follows from (T-F) and (I-TB). That is, the following argument is valid:
1. "There are fictional characters" is true.
2. "There are fictional characters" is true iff there are fictional characters.
3. Therefore, there are fictional characters.
So, anyone who believes (1) and (2) should also believe that there are fictional characters. Since someone who is concerned with the apparent ontological commitments of (F) would like to reject the claim that there are fictional characters (after all, they are concerned about the apparent ontological commitments of (F) for the very reason that they believe that there are no fictional characters), they need to reject either (1) or (2). But (2) seems pretty obviously true. So, they should reject (1). That is, they should reject the truth of the very sentence that they are offering a paraphrase of. And it seems like such a result can be generalized: Whenever someone hopes to avoid ontological commitment by paraphrasing an ordinary English sentence S, that person should reject the truth of S.

13 Comments:

Blogger rock* said...

I should note that, despite the suggestive title of my post, nothing I said there shows that paraphrasing is good for absolutely nothing. What I take myself to have shown is that someone who is inclined to apply the paraphrase approach to an ordinary English sentence S should deny that S is true. (Alternatively, they might deny the instance of the Tarski bi-conditional that applies to S. However, I take it that this move is less attractive than the other.)

I should also comment on why I take the fact that someone who applies the paraphrase approach to a sentence should deny that it is true is important. It is important because then someone who is strongly inclined to believe that the sentence in question is true cannot avail him- or herself of the paraphrase approach in order to avoid the ontological commitments of that sentence. I believe that I am in that position vis-a-vis many of the sentences that those applying the paraphrase approach attempt to paraphrase (including sentences concerning holes, for instance). So, at the very least, I'm inclined to believe that the paraphrase approach really is good for very little, if not absolutely nothing.

12:48 PM  
Blogger Joshua said...

I will now speak on behalf of the paraphraser. (1)-(3) are either written in ordinary english or they are written in our theoretical langauge. If they are written in ordinary english, then the argument is sound. But the conclusion is not as crazy as it seems. This is because the ordinary english sentence (2) and (3) express something different then what we might think they express. (2) should be translated in our theoretical discourse as

(2*) "there are fictional characters" is true iff [insert theoretical langauge sentence with the same content as the ordinary english sentence Here}

But this theoretical language sentence does not have the ontological commitment that the ordinary english sentence seems to have.

Moreover, the things that we are ontologically committed to are the things quantified over in our best theories (which will be expressed in a theoretical language.

On the other hand, if (1)-(3) are expressed in our theorietical languag, then (2) is false. It seems to be true because we are confusing ordinary english with theoretical english.

All this hinges on the lciam that the ordinary english sentence just emans the same thing as the its theoretical paraphrased. This is, or course, where I usually get off the boat.

8:49 AM  
Blogger rock* said...

G

6:10 AM  
Blogger rock* said...

[My last comment was merely a test to figure out how to bold and italicize words in the comment function. Please ignore it.]

I have two responses to Joshua's comment: a short response and a long response.

Short response: I would have thought that the motivation for paraphrasing (F) was that one did not believe that there are fictional characters. However, on move Joshua is suggesting, the paraphraser does believe that there are fictional characters (that is, he or she does believe the content of "There are fictional characters") because he or she believes (1) and (2) and that there are fictional characters follows from these.

6:17 AM  
Blogger rock* said...

Long response: Suppose we distinguish between ordinary English (OE) and ontologically serious English (OSE) in something like the way Joshua suggests. Now, when we are speaking OE, let's indicate that using italics, and when we are speaking OSE, let's indicate that using bold case. Then Joshua's suggestion seems to be that the paraphraser believes each of the premises of the following argument and thus believes its conclusion:
1. "There are fictional characters" is true.
2. "There are fictional characters" is true iff there are fictional characters.
3. Therefore, there are fictional characters.
That is, the paraphraser believes the proposition expressed by (F), assuming that (F) is a sentence of OE. However, he or she doesn't believe the proposition expressed by:
(OSE-F) There are fictional characters.

Since the paraphraser supposedly believes the proposition expressed by (F) and fails to believe the proposition expressed by (OSE-F), (F) and (OSE-F) must express different propositions. Since (F) and (OSE-F) appear to have the same syntactic structure, though, then if we assume compositionality holds for both OE and OSE, at least one of the words in (F) differs in meaning from one of the words in (OSE-F). Now suppose that it is “there are” that differs in meaning between OE and OSE. This has the strange consequence that the paraphraser is no longer concerned with what things there are, but rather with what things there are. In other words, or so it seems to me, the paraphraser has simply changed the subject.

On the other hand, suppose that “there are” does not differ in meaning between OE and OSE, but “fictional character” does. Then we may presumably introduce an expression into OSE that has the same meaning as “fictional character” does in OE. Let “fixshonal character” such an expression. Then (F) means the same thing as:
(OSE-F*) There are fixshonal characters.
So, since the paraphraser believes the proposition expressed by (F) on the view under consideration, the paraphraser believes the proposition expressed by (OSE-F*). Thus, the ontological commitments of the paraphraser are the same as the ontological commitments of the anti-paraphraser. It’s just that the paraphraser describes them differently.

A final option would be for the paraphraser to deny that compositionality holds for both OE and OSE. Such paraphraser would claim that “there are” and “fictional character” mean the same thing in OE and OSE, but (F) and (OSE-F) mean different things. Probably such a paraphraser would want to accept compositionality for OSE but deny it for OE, which they would claim is a much less well-behaved language than OSE.

All of these options for the paraphraser of the sort under consideration seem problematic to me. They have the consequence that the metaphysician is speaking a different language than an ordinary speaker. Speaking from my experience as a metaphysician, I am not aware of any change in what I mean by my words (or sentences) when I move from ordinary life to doing metaphysics, so this consequence seems implausible to me. Also notice that, on this view, when a metaphysician attempts to motivate the need for paraphrase when speaking to an ordinary speaker, such questions as ‘Are there really fictional characters?’ shouldn’t do the job, since these will be understood by an ordinary speaker to be asked in OE, in which case the paraphrasing metaphysician will agree that the answer is “Yes”. However, it is exactly by asking such questions that metaphysicians often motivate the need for paraphrase. Finally, there is the tough question of how OSE gets learned, given that its rules differ from OE to such an extent. How does a speaker of OE come to learn OSE? I, at least, would like an account from the paraphraser.

7:32 AM  
Blogger rock* said...

So what do the rest of you guys think of my argument? Relatedly, does the paraphraser have to deny the truth of the ordinary English sentence he or she is offering a paraphrase of?

4:54 AM  
Blogger Joshua said...

As I mentioned before, I believe that the argument from compositionality is a strong one. Moreover, I believe that using the strong argument from compositionality to bolster your Tarski truth schema argument is a good idea. But I want to be clear taht it seems to me that the whole strenth of the argument lies in the compositionality portion of the argument and not in the Tarski portion. This is what I said in my original post.

Now, I think that the suggestions you made on behalf of the paraphraser are good ones. I guess if I were a paraphraser, I would say that ordinary english "there are" is different from technical english "there are". Moreover, I am not sure I see what is so wrong with this. English just has a phrase that is similar to, and perhaps even related to, a phrase of technical english that has ontological commitment. Unfortunately, I think that most people who use paraphrase strategies accept the other option that you displayed. The one that denies a strong version of compositionality. They would probably accept some kind of weak from of compositionality for all languages and their goal is to use a technical langage that obeys strong compositionality so that the ontological commitments can be easily read.

6:27 AM  
Blogger rock* said...

Joshua,

Let me address your point concerning where the strength of the argument lies first. You claim that the whole strength of the argument lies in the compositionality portion, as opposed to the Tarski portion. I disagree, for two reasons. First, it might be that some paraphrasers are disinclined to draw the distinction between OE and OSE. If so, then the Tarski portion of the argument gives them reason to deny the truth of the ordinary English sentence they are offering a paraphrase of. Second, the Tarski portion of the argument helps to motivate the need for paraphrasers who do not want to deny the truth of the ordinary English sentences that they paraphrase to draw the distinction between OE and OSE.

Perhaps the best way to view my argument, then, is as follows. I offer the Tarski portion of the argument. The Tarski portion then motivates a response based on distinguishing between OE and OSE. However, those paraphrasers who are disinclined to draw that distinction are knocked off the boat after the Tarski portion; they should deny that the ordinary English sentences that they hope to paraphrase are true. I then offer a reply to the response given to the Tarski portion. This reply relies, in part, on considerations concerning compositionality. The reply is meant to address those paraphrasers who were not knocked off the boat by the Tarski portion of the argument, and it is intended to show that they too should deny that the sentences being paraphrased are true.

So, the Tarski portion of the argument is, if anything, multipurpose. The first thing it does is to show that a certain sort of paraphraser ought to deny the truth of the ordinary English sentences he or she offers paraphases of. The second thing it does is to motivate the other paraphrasers to explicitly draw the distinction between OE and OSE and face concerns about compositionality. I take it that both accomplishments are important, given the state of the literature on paraphrase, in that each forces the paraphraser to be clearer concerning his or her commitments.

7:33 AM  
Blogger rock* said...

It is also important to note a distinction between the concerns about compositionality you raised in last week's class and the concerns about compositionality I raise in the second part of my argument. Your concern could be expressed as follows, I think: One claim a paraphraser might make is that the paraphrase he or she offers is synonymous with the sentence he or she is offering a paraphrase of. However, compositionality considerations tell against this synonymy claim. So, the paraphraser should not accept the synonymy claim.

On the other hand, I did not address the possibility that a paraphraser might take his or her paraphrase to be synonymous with the sentence he or she offers a paraphrase of. Rather, my concern was with the OE sentence "There are fictional characters" and the OSE sentence "There are fictional characters". My claim was that, since these sentences have the same syntactic structure, a paraphraser who accepts compositionality must hold that at least one of the words appearing in the first differs in meaning from at least one of the words appearing in the second. I then attempted to show that this difference of meaning claim was not a good option for the paraphraser, so that he or she either is not helped by the response to the Tarski portion of my argument or must give up compositionality for either OE or OSE. Finally, I attempted to show that giving up compositionality for OE (which I took to be a more attractive option for the paraphraser than giving up compositionality for OSE) was also not a good option. I concluded that, since all of the options available for one who responds to my argument by drawing the distinction between OE and OSE are bad options, that response fails and the paraphraser should accept the conclusion of the Tarski portion of my argument (i.e., that he or she ought to deny the truth of the sentences he or she offers paraphrases of).

7:55 AM  
Blogger rock* said...

Joshua,

As to your question asking what is so bad about the paraphraser taking the OE phrase "there is" and the OSE phrase "there is" to differ in meaning (while accepting that "fictional characters" means the same in each and that compositionality holds), I have a few comments. First, as I said in one of my earlier comment posts, it seems to me that a paraphraser who does this is simply changing the subject. But I admit that this is not a particularly compelling objection, so let me move on to some other concerns.

My second concern, which is also not very compelling, is the following: Suppose, as seems plausible, that "there is" in OE and "there is" in OSE express second-order properties. Since they differ in meaning according to the view in question, then they must express different second-order properties. But then what makes the property expressed by "there is" in OSE more interesting than the property expressed by "there are" in OE? (I take it that this concern is not very compelling because the paraphraser will claim that the property expressed by "there is" in OSE is more interesting than the property expressed by "there is" in OE because in attributing the former one incurs an ontological commitment but in attributing the latter one does not.)

Let's move on to my more serious concerns. The first has to do with motivating the need for paraphrase. Often metaphysicians who employ paraphrase often attempt to motivate the need for paraphrase by asking things like "But are there really fictional characters (holes, etc.)?" when addressing speakers of OE. (This happens, for instance, when philosophers teach undergraduate metaphysics classes.) Now the sentence uttered by such a paraphraser is either a sentence of OE or OSE. Either way, the uninitiated addressee will hear it to be a sentence of OE. But then, if the paraphraser is successful in getting the addressee to answer in the negative, then the paraphraser will not have managed to convince the addressee of the paraphraser's view, since the paraphraser (of the sort we are considering) believes that an affirmative, not a negative, answer to the OE sentence "Are there really fictional characters?" is correct. (I take it that a potential response on the part of the paraphraser is that it was not clear to him or her up to this point that asking "But are there really fictional characters?" was not an effective way to convince another of his or her view. But then one is left wondering how a paraphraser might go about motivating the need for paraphrase.)

Another of my serious concerns has to do with disagreement. Suppose that a speaker of OE utters the OE sentence "There are fictional characters" and a paraphraser utters "There are no fictional characters". According to the view we are considering, the paraphraser is speaking OSE and the proposition expressed by the sentence he or she utters is not the negation of that expressed by the sentence uttered by the speaker of OE. However, intuitively paraphrasing metaphysicians who utter "There are no fictional characters" are disagreeing with the man in the street who utters "There are fictional characters".

A third serious concern has to do with metaphysicians who, like me, are entirely unaware of any difference between their use of "there is" in everyday life and their use of "there is" when doing metaphysics. Suppose that I get into a debate concerning fictional characters with a paraphraser. I utter "There are fictional characters" and he utters "No, no, no! There are no fictional characters". Again, we are intuitively disagreeing. But according to the view in question we are not because the paraphraser is speaking OSE and I am speaking OE.

My final serious concern has to do with learning OSE. This concern is not as pressing for paraphrasers who postulate that "there is" differs in meaning between OE and OSE as it is for those who deny compositionality for OE. However, it seems to me that it is a concern for the former as well. How does one become a user of OSE? How do paraphrasing metaphysicians teach others to use "there is" with a different meaning than it has in ordinary English? These seem to me to be difficult questions which should incline us to be suspicious of the strategy under consideration unless an account is given by the paraphraser.

6:01 PM  
Blogger Chris Tillman said...

One variant of the paraphrase strategy, perhaps hardly worth the name, holds that "problem" sentences of English have surprising truth conditions (or express surprising propositions). It seems that if we allow that this is paraphrasing, we can allow that the "problem" sentence is true but its truth is not problematic--it does not commit us to ontological nasties. And furthermore, it is a metaphysical discovery that sentences of English have these surprising truth conditions or express surprising propositions. This seems somewhat more respectable to me, honestly, than what paraphrasers typically seem to do. So for instance I recommend that the Nihilist says 'Well, there are tables. It's just that that claim doesn't commit you to what you thought it did.' (Note that by this criterion counterpart theorists count as paraphrasers. That seems okay to me.) So I guess I think that if this is paraphrasing, then the conclusion of the main post is correct, but we still can see what paraphrasing might be good for.

9:58 AM  
Blogger rock* said...

Chris,

I'm not sure I understand the sort of paraphrase strategy you discuss. I would've thought that we were ontologically committed to Fs just in case we believed that there are Fs. If that is so, then if the paraphraser believes the conclusion of the argument in the post, then he or she is committed to Fs.

There are two other options I can think of. One is for the paraphraser to (i) draw a distinction like that I have drawn between ordinary English and ontologically serious English, (ii) say that although he believes that there are fictional characters when "There are fictional characters" is a sentence of ordinary English, he does not believe that there are fictional characters when "There are fictional characters" is a sentence of ontologically serious English, and (iii) say that one is committed to Fs just in case one believes that there are Fs when "There are Fs" is taken to be a sentence of ontologically serious English. But then the worries I expressed in the comments about this move kick in.

The second option see is for the paraphraser to deny that being committed to Fs has anything to do with believing that there are Fs, whether taken in its ordinary sense or in its ontologically serious sense. But then I really don't know what ontological commitment is supposed to be anymore, or why we should be interested in it.

6:00 AM  
Blogger Chris Tillman said...

On my proposal, believing there are Fs does ontologically commit you to Fs. It's just that Fs are not what you took them to be--they lack ontological nastiness. So a proponent of this strategy who was a nihilist would say that there are tables and he believes there are tables, but he's not thereby committed to the existence of a composite object. It's a metaphysical discovery that the word 'table' in English refers to some simples arranged table-wise.

12:33 PM  

Post a Comment

<< Home