Language: thought or communication?

With the premise that I have very little sympathy for the press, after the last couple of weeks, I'll attempt to comment briefly on an article called "Do we need language to think?" written by C.Zimmer in the New York Times on June 19, pandering to the usual clan of obvious suspects, known for their habit of periodically concocting straw man images of generative linguistics and charging on to windmills of their own creation. The latest turbine, whipped creamed into a New York Times article, has to do with the claim that "we do not need language to think." As such the statement is meaningless. It is completely obvious that there are many forms of thought that do not involve language (animals entertain many forms of sufficiently sophisticated thinking). So what is being claimed? Allegedly, this is "disproving" (how many times have I seen the word "disproving" invoked in vain!) Chomsky's well known statement that the fundamental role of language in humans is thought rather than communication.

It is immediately noticeable that something is badly wrong here with the use of logical implications and quantifiers, as "language is primarily used for thought" does not in any way imply "all thought is based on language." So if you set out to "disprove" the claim that "all thought is based on language" (big deal, this statement is so obviously false that you can just take a seat back and relax, or use your time for better purposes) you have said absolutely nothing about whether language is primarily used for thought or not. This should be the end of the story, but for the sake of it let's dig a bit deeper in the matter. 

There is the possibility that the statement "we do not need language to think" may be interpreted, not as the truism that "not all forms of thought require language" but as the astonishing statement that "no form of thought requires language." This would be quite an extraordinary claim, and as good old Carl Sagan used to say, "extraordinary claims require extraordinary evidence." So what is the extraordinary evidence? It is claimed that researchers scanned people "as they performed different kinds of thinking, such as solving a puzzle [...] but the language network stays quiet"... Duh. OK, so some forms of thought do not require language: this is the triviality we all knew already, which says nothing about what the primary use of language is. What next? 

Now we come to studies of people with brain injuries and here there is mention of the discovery that some people with aphasia can still do algebra. OK, here there is indeed something to discuss. Most mathematicians would immediately tell you that there is no way to do mathematics without language, even though many aspects of mathematical thinking are not directly language-related. Here one needs to be careful about terminology though. Mathematics is not number sense (understood as the capacity to perform elementary arithmetic), though a lot of neuroscientists seem to have enormous difficulties understanding this fact. Number sense (which is not mathematics) may well be a form of thought that does not engage language. In fact, there seems to be reasonable evidence for this possibility. On the other hand, I want to argue here that mathematics is a form of thought that does engage, and in fact requires, language. Before getting into this, though, note that the claim being referred to does not talk about number sense, but more specifically about algebra, which does make it, in principle, a more interesting claim. So let's look at the actual research paper being referred to here. The experiments reported in this paper do indeed involve something slightly beyond number sense, namely manipulation of elementary algebraic expressions (meaning still elementary arithmetic, but with numbers replaced by variables, such as verifying that 2a-b+3c-3a=3c-b-a). One can say that this is indeed algebra albeit of a very elementary form. So the specific statement made in this paper is that a patient with severe aphasia, affecting both lexical and syntactic aspects and both comprehension and expression, was however at least partially able to perform some of these algebraic manipulations. If one looks at the actual breakdown of scores by specific algebraic tasks, however, one sees (as observed in the paper) that performance is close to the normal range for the simplest tasks and drops significantly outside of the normal range for slightly more sophisticated algebraic manipulations, and that the score for algebraic manipulations is consistently somewhat lower than the score for the matching numerical expression (namely number sense works better than elementary algebra in the absence of language). It is also observed in the paper that "with respect to language, inner speech may be important in solving complex problems. Although inner speech may not be intrinsic to algebraic computation, it may provide a scaffold that permits offloading of information in working memory and thus assists in solution of problems involving multiple sequential steps." Well, this seems to be completely consistent with a view of language as primarily performing thought related task, and certainly does not disprove absolutely anything. I want, however, to spend a few more words on the role of language (specifically the "inner speech" referred to here) in mathematics. The more you move away from elementary manipulations like the ones considered here, and you enter into the realm of actual mathematics (where the goal is proving theorems), the more the role of language is primary, and the balance between the amount of time spent by a mathematician on inner speech as oppose to algebraic manipulations gravitates more and more towards the prominence of inner speech. 

The NYT whipped cream continues by rhetorically asking "If language is not essential for thought, then what is language for?" followed by the astounding resolution of this lame cliffhanger: "Communication"... Note again the same logical fallacy, "language is primarily used for thought" does not imply "language is essential for thought" and is therefore not "disproved" by "language is not essential for thought", but I am repeating myself: I just have an allergy to garbled logical implications! This is a magistral battle of the windmills, which would be funny to observe from the outside, if I could be in that position. 

If one is trying to argue (as it appears to be the case) that the language faculty in human exists for the purpose of communication, then what does our ability to do mathematics exist for? Communication? That would be very hard to argue, I believe. Mathematics (I mean mathematics) and language are distinct forms of thought, but they are highly intertwined systems, not at the elementary level of number sense and of the simplest algebraic manipulations, but more and more deeply so at anything resembling actual mathematics. So what is mathematics for, if it is not, like language, primarily a form of thought? 

Coming back to the matter at hand, unless one really is attempting to prove that no form of thought requires language, which appears to be easily disprovable, then the discussion and the evidence proposed seem to have no bearing at all on the statement that "language is primarily used for thought rather than for communication" and it is easy to suspect that all this is in fact just a game of smoke and mirrors, that made it into the press right now on the wave of a depressing two weeks sporting a whirlwind of remarkably bad journalism.