WHAT THE AUGUST 30 POST SHOULD HAVE SAID!

In the August 30 post I attributed to Noam Chomsky views he disavowed in an e-mail to me (the  text of the message is my September 9 post). What follows is

HOW THE AUGUST 30 POST SHOULD HAVE READ:

In my next post I will say how the mistake came about, and why some of Chomsky's formulations did strike me as suggesting that he believes the "language organ" that he posits has to contain representations of all learnable meanings (which is the feature of Fodor's original "mentalese" hypothesis that I find unsupportable).

Innate language?
 I do not believe in “innate language”, (e.g., Fodor’s “mentalese”, which, when he wrote The Language of Thought, was supposed to have a built-in truth-conditional semantics, that is a semantics which fixes the meanings of all the sentences a human is capable is thinking (later Fodor moved to the position that “mentalese” has only a built-in syntax, and developed a causal theory of reference to account for content).  Here are my reasons for this rejection.

The decisive objection, in my view, is that the innate language that Fodor posited was supposed to be able to express all learnable meanings. If all possible meaningful terms were definable from a number of basic terms that might have been selected for by evolution (such as the logical positivists’ “observation terms”) , this would be compatible with Darwinian evolution.. But Fodor rejects logical positivism, and he suggested no alternative account of what the basic terms of “mentalese” might be. Nor did The Language of Thought tell  us what mechanism could have endowed the brains of primitive men and women with terms with such meanings as “quantum potential”  and “macroeconomic”, or with terms by means of which they could be defined, if, indeed, there are more elementary terms in which this could be done.

I am well aware that Fodor rejects Darwinian evolution. (Since that theory is highly successful, and constitutes the beating heart of population genetics, to mention just one part of biology, I would think that is already reason to dismiss talk of innate language.) But if we are to suppose that“mentalese” appeared in the brain serendipitously, that the makes the story even crazier. The story that the hundreds of thousands of terms in human languages, and the millions of terms in possible human languages, all correspond to terms in a master vocabulary of ‘Mentalese’ supplied to us by serendipity is precisely analogous to supposing that monkeys typed out Proust’s Remembrance of Things Past. Of course, to reject that story is not to deny that some aspects of language might be innate; that would not be incompatible with evolution.

MESSAGE FROM NOAM CHOMSKY

Someone sent me this, which surprised me:

"I am well aware that  both Chomsky and Fodor reject Darwinian evolution. (Since that theory is highly successful, and constitutes the beating heart of population genetics, to mention just one part of biology, I would think that is already reason to dismiss their talk of innate language.) Chomsky apparently thinks that the “language organ” appeared in the brain serendipitously, but that the makes the story even crazier. The story that the hundreds of thousands of terms in human languages, and the millions of terms in possible human languages, all correspond to terms in a master vocabulary of ‘Mentalese’, or to a a set of "switch"-settings, supplied to us by serendipity is precisely analogous to supposing that monkeys typed out Proust’s Remembrance of Things Past. (Of course, to reject that story is not to deny that some aspects of language might be innate; that would not be incompatible with evolution.)”

In my entire life, I have never once even hinted at rejection of the modern theory of evolution, Darwinian in origin though of course greatly extended and crucially modified since.  You’re quite right in dismissing talk of “innate language,” just as I have always done.  I have never once even hinted at your story about the terms of human language, though I have discussed the fact that the modern adherence to the referentialist doctrine, abandoning a long philosophical tradition, has led to great confusions in modern philosophy and is quite inconsistent with basic facts about human language.  As for the origin of the terms of human language, as I’ve repeatedly discussed, that remains a deep mystery, for which no one has any sensible account.

Sorry that our efforts to get together have so far failed.  Hope we’ll have a chance to talk about these (and many other) things one of these days. 

(comment to come - HP)

I first criticized the "Innateness Hypothesis" in 1967, in a paper I collected in Mind, Language and Reality, vol. 2 of my 1975 Philosophical Papers.  That paper was delivered in a debate with my friend, Noam Chomsky, and we have been debating the issue ever since. The following remarks will be included in a lecture I will give via Skype to a conference in Mumbai in October.

Innate language?

 I think it is worth while to say again, here, why I reject the “innateness hypothesis”. Since the most famous advocate of that hypothesis is Noam Chomsky, I will use his version as my target.

The decisive objection, in my view, is that the “universal grammar” that Chomsky postulates has to provide for all possible meanings. Chomsky likens his hypothetical “language organ” to a switch box, and he tells us that when the switches in a child’s brain are appropriately set, “the child has command of a particular language and knows … that a particular expression has a particular meaning and so forth.”[1]  If all possible terms were definable from a number of basic terms that might have been selected for by evolution (such as the logical positivists’ “observation terms”) , this would be compatible with Darwinian evolution.. But Chomsky certainly rejects logical positivism, and he has suggested no alternative account of what the basic terms of “mentalese” might be. Nor has he ever told  us what mechanism could have endowed the brains of primitive men and women with terms with such ‘particular meanings’ as “quantum potential”  and “macroeconomic”, or with terms by means of which they could be defined, if, indeed, there are more elementary terms in which this could be done.

I am well aware that  both Chomsky and Fodor reject Darwinian evolution. (Since that theory is highly successful, and constitutes the beating heart of population genetics, to mention just one part of biology, I would think that is already reason to dismiss their talk of innate language.) Chomsky apparently thinks that the “language organ” appeared in the brain serendipitously, but that the makes the story even crazier. The story that the hundreds of thousands of terms in human languages, and the millions of terms in possible human languages, all correspond to terms in a master vocabulary of ‘Mentalese’, or to a a set of "switch"-settings, supplied to us by serendipity is precisely analogous to supposing that monkeys typed out Proust’s Remembrance of Things Past. (Of course, to reject that story is not to deny that some aspects of language might be innate; that would not be incompatible with evolution.)

---

[1] Chomsky, Language and the Problems of Knowledge (Cambridge,MA, MIT Press, 1988), pp. 62–63.

Further remarks about Davidson

What I have said so far, in essence, is that although “A Nice Derangement” of Epitaphs did mark a radical change in some of Davidson’s views (the most famous change being his denial that there are “languages” as philosophers think of languages, namely as requiring conventions shared by a linguistic community that determine the meanings of the sentences), the central doctrine of Davidsonianism remains fixed, namely that the way to think of meanings (of sentences, again – why ‘sentences’ and not words is something I explain at the end of this post) is to think of them as truth conditions generated by a Tarski-style semantic theory. But in NDE the semantic theory is no longer held to be valid for a whole community, or even for an individual speaker: there is now a semantic theory that an individual speaker assumes when she begins to talk to another member of the community (her “prior theory”) and a theory that the speaker and the hearer ideally converge on in the course of a particular conversation or on a particular occasion (their “passing theory”). Thus Davidson became a “contextualist” (his semantics became “occasion sensitive”, in my terminology), but semantics was still a là Tarski.

I said in the previous post that I didn’t agree with Davidson that meanings should be identified with truth-conditions. I agreed, however, that truth-conditions are occasion-sensitive (so Davidson would be right that meanings are equally occasion sensitive if meanings were truth conditions). But I won't repeat all that.

I also argued that Davidson does not really say in what sense a particular speaker’s competence is “modeled” by a passing theory. He only emphasized that it is not known by the speaker.

But let us go a little farther.  For me the main locus of occasion sensitivity, as I explained in my June 2^nd post [“Occasion-sensitivity and productivity are not incompatible”] is the reference of names and predicates. Sentences,”E.g., there is milk on the table” have occasion sensitive truth-conditions because the words they contain have different references; e.g.,  a little spilled  milk may or may not count as “milk” for the purpose of evaluating the truth or falsity of “there is milk on the table” on a particular occasion.

Davidson, however, endorsed Quine’s unfortunate doctrine of the indeterminacy of reference.. That is why Davidson says so little about the semantics of individual words; words like "pig", "blue", "home", "coffee" are primary referring expressions in English, for example. And reference is not a dubious metaphysical notion, as Quine thought. Give me that, and it is not hard to say something (of course, not everything) about what a passing theory has to do to “model a speaker’s competence"): it has to tell us what his or her words referred to on the relevant occasion.

   (to be continued)

Apologies for the Interruption in this series of posts, and More on Davidson’s famous paper, “A Nice Derangement of Epitaphs”

In June (between June 2^nd and June 25^th to be exact) this blog consisted of a series of posts on contextualism[1], comparing my version and the version in Davidson’s famous paper (henceforth, NDE). I had planned to continue blogging on this subject at once, but my friend Mario DeCaro and I received word that Harvard University Press is prepared to publish another collection of my papers, to be edited by DeCaro, under the title, which is also the title of the paper whose appearance in the Journal of the APA I announced in my last post (and which will be the first chapter in the new collection),  Naturalism, Realism, and Normativity. Moreover, HUP asked for a file of the whole collection by yesterday, July 13^th! Mario and I did not quite meet that deadline, but we will have a file in the hands of the Press this week (and we are both happy that the Press is eager to expedite the publication). But making final emendations to the papers and formatting them to meet the style of the Press was a full time job, and so blogging had to be interrupted. But I will now continue this series of posts.

In those June posts, I said that I do not agree with Davidson’s identification of truth-evaluable content and meaning, but that NDE does contain a reasonable answer to a question [in my terminology rather than Davidson’s] about “truth-evaluable content”,viz.,”What information should a good semantic description of the truth-evaluable content of an utterance on a particular occasion provide?”

Davidson's answer is that a good description of the truth-evaluable content of the sentences that are uttered on a particular occasion is nothing more or less than a “passing theory” which has the form a a Tarskian truth-theory for those sentences (and the indefinitely large totality of sentences that can be derived from them by the familiar compositional devices.)

One might expect Davidson to add, “that describes the implicit knowledge of the speaker concerning his or her own words, but Davidson apparently rejects this. [I say, “apparently”, because NDE only mentions explicit knowledge.]  Instead, he says that the passing theory in question must “model the speaker’s competence”. Davidson’s reasons are stated briefly:

“To say that an explicit theory for interpreting a speaker is a model of the interpreter’s linguistic competence is not to suggest that the interpreter knows any such theory. It is possible, of course, that most interpreters could be brought to acknowledge that they know some of the axioms of a theory of truth; for example, that a conjunction is true if and only if each of the conjuncts is true. And perhaps they also know theorems of the form ‘An utterance of the sentence ‘‘There is life on Mars’’ is true if and only if there is life on Mars at the time of the utterance.’ On the other hand, no one now has explicit knowledge of a fully satisfactory theory for interpreting the speakers of any natural language. In any case, claims about what would constitute a satisfactory theory are not, as I said, claims about the propositional knowledge of an interpreter, nor are they claims about the details of the inner workings of some part of the brain. They are rather claims about what must be said to give a satisfactory description of the competence of the interpreter.” [NDE 256]

And my comment on this took the form of a question:

MY QUESTION ABOUT DAVIDSON: The  truth-evaluable content of a sentence on a particular occasion is, then, given by its truth-condition, as specified by a passing theory that does WHAT?

That is where my June 2015 series of posts left off. To resume, then, let me say why Davidson’s reasons for denying that speaker’s have propositional knowledge of the passing theory that gives the truth-evaluable content of the relevant sentences in a particular conversation seem off the mark to me.

Davidson’s reasons, just quoted, are that (1) no speaker “now has explicit knowledge of a fully satisfactory theory for interpreting the speakers of any natural language”; and (2) “, claims about what would constitute a satisfactory theory are not, as I said, claims about the propositional knowledge of an interpreter” (THUMP). Reason (1) ignores the fact that a passing theory is not supposed to be a truth-theory for a whole natural language, but only for a fragment large enough to contain the sentences used in a particular conversation, and (2) just repeats the claim that “modeling” a speaker’s competence is not describing his or her propositional knowledge (or what goes on in his or her brain), but does not say WHAT it is instead.

I say these reasons are off the mark, because they do not challenge the idea that the meanings of  the sentences in a particular conversation are the sort of thing that could be propositional knowledge. Maybe no one has “a fully satisfactory” version of that knowledge today, and maybe only a small number of speakers know the relevant Tarskian technicalities, but neither of these facts shows that the passing theory on which a speaker and a hearer converge couldn’t be someone’s  explicit propositional knowledge.

What I would say about all this depends on the “semantic externalism” (aka “anti-individualist” view) that Kripke and I independently arrived at in the late 1960s. On an externalist and anti-individualist view, knowing what the word “gold” means doesn’t require one to know a description of the metal of the kind a physicist would give today, and it doesn’t even require that the physics or metallurgy be developed beyond the point at which at least some people, people recognized by the linguistic community as experts, can distinguish examples of gold from other metals and alloys.  A truth-theory (“passing theory’) that describes what someone means by “gold” in a particular conversation might say that “gold” refers to gold, but to say of an English speaker that she knows what “gold refers to isn’t to say that she knows that theory; it is to say that she is able to use the word properly, and that involves, not knowledge of a description that is synonymous with “gold”, but social-know-how, including linkage to other members of the community who are in turn linked to samples of the metal itself and who can identify it. (Depending on the context, the appropriate passing theory might also says that "gold" refers to a color, or to a currency standard, etc.) The truth-evaluable content of sentences of a particular occasion can be formalized as a truth-theory; that is how we logicians formalize such things, but the “know” in “know what a word refers to” does not normally refer to propositional knowledge at all. It is because Davidson was unwilling to accept anti-individualist semantics that the contextualism of NDE has such a strange shape.

---

[1] In one of those posts [June 17^th] I wrote: The thesis of contextualism is that in general the truth-evaluable content of sentences depends both on what they mean (what a competent speaker knows prior to encountering a particular context) and on the particular context, and not on meaning alone.”     In NDE this becomes the claim that hearers’ theories of another speaker's first meanings (their “prior theories”) have to be replaced by different theories (“passing theories”) in the course of each conversation. The right passing theory  for a particular conversation will be one on which both speaker and hearer converge. In NDE,  The truth conditions of sentences depend both on their first meanings and on the way those first meanings have to be modified in a particular conversation; they are not not, in general, the truth conditions assigned by the prior theory. [Moreover, there is not, according to NDE  just one conventionally fixed prior theory shared by all speakers of a language. (“As speaker and interpreter talk, their prior theories become more alike”.) This leads Davidson to the famous “I conclude that there is no such thing as a language, not if a language is anything like what many philosophers and linguists have supposed.” (ibid. 265)
    In my view, as early as “The Meaning of ‘Meaning,” the prior knowledge a competent speaker has of the semantics of a sentence includes such properties as the categories to which the words belong – .e.g, “gold is the name of a metal”, and the “stereotypes” associated with the words, but (contrary to Davidson) not any such thing as one particular context-independent truth condition of the sentence. Nevertheless, the Davidson of NDE and I are in agreement that on any particular occasion of use, the truth-evaluable content of a sentence on that occasion is not fixed by the prior knowledge/prior theory of the speaker qua speaker of the language, but depends on the occasion (the “context”). That is the sense in which we are both “contextualists”, despite our disagreements on other issues.

My latest paper (with my views about truth)

It is titled “Naturalism, Realism and Normativity” and it has just appeared in the new Journal of the American Philosophical Association/Volume 1/Issue 02/June 2015/pp. 312-328.

It argues, inter alia, that Tarski’s results do not support deflationism about truth, contrary to what is often assumed. I write there,

“Tarski’s formal methods intuitively draw on and presuppose not just one property of truth, the T-Schema, or Disquotation, but on that property AND the further property that the extension of “true” depends on the extension of “refers”. The concepts of truth and of reference are intimately related, and his entire procedure exploits the relation.”

I also briefly describe and endorse Tyler Burge’s claim that reference is psychologically more primitive and more ubiquitous than language use.

If you are a member of the APA you will receive the journal automatically (it is free to members this first year of publication). If you aren’t, please get your library to subscribe, or at least to get online access!

This paper is directly relevant to my current posts on Davidson and on truth-evaluable content.