Saturday, January 26, 2013

Breaking the spell of language

G. Randolph Mayes

Philosophy is a battle against the bewitchment of our intelligence by means of language.  ~Ludwig Wittgenstein

Since DoR is a philosophy blog I thought our first step into the blogosphere could be a meta-step, a post about the nature of the dance itself.

A lot of you know that Wittgenstein was the father of what we now call the Linguistic Turn in philosophy. He is known for claiming that many problems of philosophy are (what Rudolf Carnap later called) pseudo-problems.  Wittgenstein said that many apparently profound philosophical questions are actually confusions that result from a kind of illusion. We are tricked into believing that we are thinking about some property of the world when we are really thinking about some property of language. 

One of Wittgenstein's early mentors, Bertrand Russell, puts this beautifully in a 1923 lecture on vagueness.

Reflection on philosophical problems has convinced me that a much larger number than I used to think, or than is generally thought, are connected with the principles of symbolism... In dealing with highly abstract matters it is much easier to grasp the symbols (usually words) than it is to grasp what they stand for. The result of this is that almost all thinking that purports to be philosophical or logical consists in attributing to the world the properties of language. 

My absolute favorite example of this is the nature of truth, which certainly sounds like a profound question about the the relation of our thoughts to the external world. If you study the concept of truth in an introductory philosophy class, your teacher might motivate it by saying something like this: "If a statement is true, then clearly there must be something that makes it true. In philosophy we are trying to understand what that something is." 

Well, wait.  Way back in 1933 a logician named Tarski provided a famous definition of truth that doesn't particularly satisfy this intuition at all. Tarski successfully treated truth as a meta-linguistic predicate. What this basically means is that he saw 'true,' not as a word belonging to the part of our language that we use to talk about the world (the object language), but rather a word belonging to the part of our language that we use to talk about language (the meta-language). 

If this sounds all dry and technical and too demanding for a blog piece, hang in there. I'm trying to save your freakin' soul here. 

The basic idea is just that when we utter a sentence like "Snow is white," there are a number of practical reasons why we would choose to express the information contained in this sentence by talking about the sentence itself.  We might say '"Snow is white" is true,' or more simply "That's true." The important point here is that on Tarski's definition the meta-linguistic sentence containing the word 'true' does not contain any more information about the world than the original sentence. It's just a different way of communicating the very same information.

So one slightly bombastic way of understanding Tarski's work on truth is that he gave us a way to break the spell that the word 'true' has cast over philosophically inclined thinkers for thousands of years.  He showed that we can use the word in a perfectly straightforward and intuitive way while explaining the belief that there must be some profound metaphysical relation that makes sentences true as an illusion. It is an illusion that results from the surprising difficulty of distinguishing statements about the world from statements about language.

The idea that sentences of the form  "S is true" do not contain any more information than is contained in the sentence S itself is what we now call a deflationary view of truth.  This idea can be traced to Gottlob Frege, but it was first developed carefully by a lesser known and incredibly brilliant philosopher named Frank Ramsey, who died very young. (Both Ramsey and Frege were profoundly influenced by Wittgenstein, with whom they both corresponded.)

The deflationary theory of truth is popular, but it is not universally accepted. (Ironically, Tarski himself does not seem to have accepted it.) Those philosophers who reject the deflationary account may say something like this: "Well, yes, Tarski certainly provided an analysis of one meaning of the word 'true,' but that doesn't mean it is the only philosophically significant meaning."  Of course they are right about that. It is still possible that there is some completely distinct philosophically significant sense of the term. But if the deflationary account is not the whole story on truth, it has certainly shifted the burden of proof.  It is now up to people who believe otherwise to show what, beyond their own suspicions and intuitions, the deflationary view has failed to account for.

Every year I read (or at least start reading) lots of articles by philosophers who seem happy to go about their business as if the Linguistic Turn never happened. They practice their craft as"analytic philosophers" but without any real awareness of the concerns that brought this style of philosophy into existence in the first place. For me, Wittgenstein's account of philosophical problems completely changed the rules of the game. I do not think all philosophical problems are due to linguistic confusion.  But the first thing I suspect about any philosophical theorizing is that the people engaged in it are bewitched. And since I know how hard it is to detect, I suspect I usually am, too.

Further reading:  The best piece I ever read on the Linguistic Turn is the introduction to Richard Rorty's anthology by the same name.  The article is called: "Metaphilosophical Difficulties of Linguistic Philosophy." Check it out.

G. Randolph Mayes
Department of Philosophy
Department of Philosophy
Sacramento State 


  1. From Brad Dowden

    A deflationary view of truth implies that sentences of the form "S is true" do not contain any more information than is contained in the sentence S itself. It would be nice if we had a clear sense of what information is so that we could measure whether “S” and “S is true” have exactly the same information or only approximately the same information.

    Advocates of the deflationary theory pride themselves on not having the difficult problems had by their main rival, the correspondence theory of truth. According to the correspondence theory, truth is what a sentence S has when it corresponds to the facts. Advocates of the correspondence theory have a very difficult problem saying clearly what a fact is and what “corresponds” means. When they try to tell us what a fact is they usually talk in circles because they define truth in terms of facts, but then use the word “truth” in their explanation of what a fact is. The deflationary theory doesn’t have this problem.

    But the deflationary theory has other problems. I mentioned one, the problem of giving a careful explanation of information. Another difficulty for the deflationary theory is to tell us how to solve the liar paradox.

    The liar paradox is a piece of reasoning that begins with the liar sentence L.

    (L): The proposition (L) is not true.

    Then following the reasoning in the paradox, let’s ask whether (L) is true or not. Don’t we have to agree that if (L) is true, then it follows that it isn’t? On the other hand, suppose (L) isn’t true. Then doesn’t it follow that (L) is true? This result is unacceptable, and that’s the liar paradox. What is the way out of the paradox?

    Does deflating (L) help? Does this paradox show that we still need a positive theory of truth, not simply a deflationary rejection of truth? I’m not sure. I’d like to hear from the deflationists.

  2. Brad,

    Regarding the nature of information. I think there are probably a lot of ways to state the deflationist thesis without using that term, but I like it for precisely the reasons you give. The concept of information is critical to a variety of scientific endeavors, and achieving cross-disciplinary clarity about the nature of information is an important philosophical and scientific aim.

    Regarding the Liar's paradox. You know this area better than I do, but I think the deflationist I describe commits to the view that the liar paradox can be treated successfully by regarding truth as a metalinguistic predicate. Tarski actually did provide a solution to the LP, but to do it he had to insist that no language could refer to it's own semantic properties. Many philosophers think this is too extreme, since that is what natural languages appear to do. But the answer may just be that languages that do that will simply be stuck with a paradoxical notion of truth. It doesn't seem to prevent them from working.


  3. Randy, Thanks for explaining Tarski so clearly. I frankly admit that I've never really felt that I've understood his thesis (although I don't have that feeling so much when I read Rorty). I'm stil puzzled, though, over what the deflationist view implies (if anything) about statements like:
    your prediction was false
    your description is imprecise
    your account is inaccurate
    Can these terms also all be deflated? If not, then why aren't the properties they are ascribing to statements similar to truth and falsity?


  4. Emrys, thanks for stopping by! The blog is starting a little slower than I anticipated due to a busy beginning of the semester.

    Regarding your question. A sentence like "Your prediction is false" is easily deflated as "You predicted X, and not X." I think with the other two sentences, it depends on what we mean.

    By "Your account is inaccurate," I think we typically mean that something,though not necessarily everything, that was said was false. So, e.g., "You said X,Y and Z but it is not the case that X, Y and Z."

    By "Your description is imprecise" I think we are talking about the information content of a message. "There is an animal on the porch" is not as precise as "There is a pig on the porch." So we don't really have a putative word-world relationship to explicate as much as message-receiver relationship, where the basic phenomenon of interest is the degree to which the message reduces uncertainty in the receiver.

    So I'm inclined to think the answer to your question is generally yes.