Donald Davidson
The Philosophy-Linguistics Connection 1967-76

Gilbert Harman
Princeton University

March 26, 2004

Donald Davidson was at Princeton University from 1967 to 1970 and at Rockefeller University in New York from 1970-76, occasionally teaching a seminar at Princeton. During this time he and I talked regularly, ran two conferences, and published two collections of essays.

I first met Davidson at the American Philosophical meeting in December 1963 when he presented his paper, “Action, Reasons, and Causes,” a paper which contains the germ of many of the ideas he developed in the following years.

I next saw him in the summer of 1964. I was teaching a course at Berkeley that summer. Hearing that Paul Grice was running a weekly seminar at Stanford, Thomas Nagel, Barry Stroud, Thompson Clark and I drove down to attend. In the seminar, Grice presented an early version of “Logic and Conversation,” with Davidson regularly asking for clarifications, “because otherwise what you say will just go in one ear and out the other.” Various other philosophers were at the seminar, including Michael Dummett, who later presented his own “antirealism” in opposition to Davidson’s “realism.” (However, I believe that at this time the big issue between them was whether Dummett would go surfing with Davidson.)

Back in Princeton we were discussing Davidson’s ideas about semantics as presented in his paper, “The Method of Intension and Extension,” (in The Philosophy of Rudolf Carnap) and especially in “Theories of Meaning and Learnable Languages.” Davidson’s ideas connected in an interesting way with Noam Chomsky’s views about what is needed for languages to be learnable.

In 1966 Davidson gave a lecture at Princeton on “How Is Weakness of the Will Possible?” One very important idea in the paper concerns the interpretation of seemingly conditional statements involving ought, like, “If you promise to do something, you ought to do it.” Davidson argued that such statements are not conditional statements of the form, “If P then Q,” but are statements of conditional oughts by analogy with statements of conditional probability. Since the word if is not functioning as a sentential connective in these contexts, the logical principle of modus ponens does not apply.

There are other instances of this sort of if. In “Adverbs of Quantification,” David Lewis observed that if clauses are often used to put restrictions on the interpretations of quantifiers rather than as sentential connectives. It may be that if never functions as a sentential connective, so strictly speaking modus ponens never applies.

Davidson came to Princeton in 1967, a year after his talk. He chaired the department the year after that and spent the following year at the Stanford Center for the Behavioral Sciences. Then he left Princeton for Rockefeller University and moved to New York. So, he actually spent only two years in Princeton!

As soon as he arrived in Princeton, Davidson and I met regularly to discuss how his ideas about semantics and logical form fit with what what was going on at the time in linguistics. Davidson’s ideas at the time were being developed in papers like “Truth and Meaning,” “The Logical Form of Action Sentences,” “Causal Relations,” and “On Saying That.”

A central aspect of these ideas was a stress on the importance for semantics of providing a certain sort of theory of truth conditions. In a way he was defending a version of Quine’s translational approach to meaning: your semantics for a language L should provide a systemic way to translate from L into your own language with the following restrictions: The system of translation was to take the form of a theory of truth for the other language; it was to have at most a finite number of axioms (because of considerations of learnability); and it had to be expressed in a language of first-order quantification theory. The use of second-order quantification or substitutional quantification was ruled out in part because it would trivialize the requirement that the rules of translation should take the form of a theory of truth. Appeal to possible worlds was ruled out as well for somewhat obscure metaphysical reasonss.

Like many other philosophers at the time, I was completely unconvinced by Davidson’s meta-view about the connection between truth and meaning. On the other hand I was (also like many others) impressed with Davidson’s particular accounts of logical form, for example, his suggestion that sentences containing some sorts of adverbial modification were best analyzed using quantifications over events, where the verbs in such sentences were treated as predicates of events and the adverbs represent further predicates of events indicating time, place, manner, etc. There was also his claim that causal statements are often statements of causal relations between events and his proposal that sentential that clause complements, as in “Galileo said that the Earth is flat” be treated as performances that the rest of the sentence comments on.

Davidson recruited students and colleagues to work on various problems, such as how to handle adjective noun constructions, like large mouse, which cannot be analyzed as simple conjunctions, since a large mouse isn’t very large; and how to treat pronouns in a theory of truth so as to avoid the result that someone else’s remark, “I have a headache,” is true if and only if I have a headache.

At Princeton, Davidson and I got involved in discussions of linguistics because these developments looked as if they might help with his ideas about logical form and because his ideas about logical form might help in linguistics. Katz and Postal had argued for a version of generative grammar in which syntactic transformations mapped the “deep structure” of a sentence into its “surface structures.” Semantic interpretation rules operated on the deep structure to arrive at a representation of meaning and phonetic interpretation rules operated on the surface structure to provide a representation of sound.

We discussed how a grammar of this sort might be used to help assign logical forms to sentences in a natural language. We looked at proposals by linguists like Emmon Bach, James McCawley, and George Lakoff, who were arguing for a version of the Katz-Postal theory that did away with the distinction semantic interpretation rules and syntactic transformations. In their view, sometimes called “generative semantics,” there was no separate level of deep structure. Instead, the base rules of the system built up a semantic interpretation which the syntactic transformation rules mapped into surface structure. The idea looked promising at the time, although linguistics soon went off in a different direction.

Davidson was going to be at the Stanford Center for 1969-70, so he and I set up a small conference there in 1969 that brought together a few linguists and philosophers of language. The philosophers included Quine, Geach, Bruce Vermazen, and David Kaplan; the linguists included Bach, Lakoff, McCawley, and Barbara Partee. (My memories of who attended this conference are different from Davidson’s. For example, in his autobiography Davidson says Richard Montague was there, but I am sure Montague was not there.)

Davidson thought that the conference went well enough that we should bring out a collection of new papers. Semantics of Natural Language was published the following year (1970) in Synthese. A year later we republished it as a hard-cover volume including also a transcript of Saul Kripke’s Princeton lectures on “Naming and Necessity.”

Then in 1971, with support from the Council for Philosophical Studies, we ran a six week “summer school” in Semantics and Philosophy of Language at UC-Irvine. The “faculty” included Kaplan, Partee, Quine, Grice, Peter Strawson, Kripke, and the linguist John Ross. The sixty or so “students” included Gareth Evans, Richmond Thomason, Robert Stalnaker, Carl and Sally McConnell Ginet, William Lycan, Peter Unger, and James McGilvray. After intense discussions, we would often drive to the beach where Davidson was teaching Quine to surf.

After 1971, serious semantics became an important topic in linguistics. Davidson’s ideas about adverbial modification have entered textbooks in linguistic semantics and have recently been developed in important ways by Terence Parsons, for example. The study of the semantics of natural language is flourishing.

During the decade I am describing, I was living in New York (and commuting to Princeton), so when Davidson moved to New York we were able to continue to meet regularly to discuss philosophy, play squash, and have meatless lunches. (Davidson had taken up vegetarianism for general health reasons.)

He had a water bed in an upper floor apartment near Rockefeller. He thought he needed to change the water regularly and worried about how to do this. At one point he got a hose and was siphoning water out the window when he almost fell out himself! He thought that, if he had fallen out, the headlines might have been interesting.

We often talked about the novels he was reading (Iris Murdoch, Trollope). He was enthusiastic about everything. He always spoke clearly, with excellent articulation, as if he was acting, and indeed he said he had been an actor earlier in life.

We sometimes traveled to conferences together. He wanted to rent a plane so he could pilot us to conferences, but his stories about how well he had able to handle various emergency situations led me to wonder why there had been so many such situations and I always insisted on driving.

We prepared an introductory anthology of readings in what we called The Logic of Grammar. Unfortunately, the publisher went bankrupt at the time of publication and hardly anyone has seen this volume who did not receive a free copy.

In addition to discussing issues of logical form, we talked a great deal about the theory of action, practical reasoning, intentions, planning, and doing something intentionally. Davidson had wanted to explain intentional action in terms of beliefs, desires, and causality, and then use the notion of intentional action to explain what intentions are. I thought this couldn’t work because it seemed to me that intentions are distinctive psychological states in their own right, not reducible to desires plus beliefs. Michael Bratman, who was a student at Rockefeller during this time, has since worked out in great detail a nonreductionist view of this sort.

During his six years at Rockefeller, Davidson wrote over twenty important papers on a wide range of subjects, including his Presidential Address to the 1973 Eastern Division American Philosophical Meeting in Atlanta, “On the Very Idea of a Conceptual Scheme.”

All good things come to an end. I left New York to move to Princeton in 1976 and at about the same time Davidson moved to Chicago, and then to Berkeley. From that point on I saw much less of him, alas.