by Dave Maier
Last time, in part 1, I distinguished two strategies for combating philosophical modernism of a certain dated kind: a pluralistic post-empiricism (the exact nature of which I left open for now), and a more narrowly focused post-phenomenological approach which regards the former (and/or its main components) as merely another form of the supposedly mutually rejected picture. In sections I and II, I discussed Charles Taylor’s and Hubert Dreyfus’s phenomenological criticisms of Richard Rorty and John McDowell; today I continue with a look at Taylor’s analogous criticism of Donald Davidson. As before, the point is not to reject phenomenological approaches, but instead merely to understand why Davidson looks to Taylor even less like an anti-Cartesian ally than do Rorty and McDowell, and thus why Taylor will not be impressed by a pragmatist strategy of multiple philosophical tools in which Davidsonian semantics plays a major role. Let me also say that in reading a lot of Taylor’s work recently, I was quite impressed with the scope and rigor of his overall project, and I think that what I present as his drastic misreading of Davidson’s philosophy may most likely be detached and discarded without threatening that project. Or so it seems to me at present.
III. Taylor vs. Davidson: truth-conditions in the theory of meaning
In Taylor’s telling, modern naturalism is again the explicit target. The story begins the same way: the Lockean/Cartesian epistemological turn led to important advances in our understanding of the natural world and our knowledge of it, but by the twentieth century the cracks in this picture were starting to show. As we’ve seen, Dreyfus and Taylor take their cue mainly from Heidegger, but each credits Wittgenstein as well with important insights. Wittgenstein’s earlier philosophy, along with the advances in logic and philosophy of language of Russell and Frege, led to what we now call “analytic philosophy,” and it is this tradition that Davidson is most naturally seen to follow: a direct line connects the (early-) Wittgenstein acolytes of the “Vienna Circle,” especially Rudolf Carnap, with Carnap’s student Quine and Quine’s student Davidson (full disclosure: my own teacher, Akeel Bilgrami, was Davidson’s student).
This line of descent from Wittgenstein thus passes through Carnap’s “logical positivism.” Like its psychological cousin behaviorism, positivism seems from our present perspective to have been an overly simplistic reaction to the modern dualistic picture, doomed from the start. This makes it easy to overlook its lasting contributions (indeed, another cousin to positivism is pragmatism; later Carnap can sound almost like Rorty in spots). Still, it is empiricism of just this sort that Taylor rejects as “the modern epistemological tradition” – only two steps away from Davidson, and we’re still on the wrong side of the fence.
Quine is the pivotal figure here. A dedicated naturalist, Quine is committed to purging the philosophy of language of what fellow naturalist Patricia Churchland calls “spooky stuff”: the Cartesian mind and its contents, whether meanings, beliefs, intentions, or whatever, when considered as swinging free, conceptually speaking, from the physical world. As was Rorty at the beginning of his career, Churchland herself is an “eliminative materialist,” rejecting as illusory whichever mental items cannot be reduced to the physical, but Quine famously rejects the reductionist fantasy as being one of, as his 1951 article calls them, “two dogmas of empiricism” (the other being the analytic/synthetic distinction). For Quine, while we may reject “meanings” as unacceptably unempirical posits, linguistic meaning itself can be accounted for without reduction if we take careful note of the process by which we attribute it to the utterances and inscriptions of actual speakers. This is the point of his account of “radical translation,” as given in his 1960 book Word and Object.
Even though its target is a recognizably Cartesian picture, Quine’s project retains the naturalist and empiricist character of his analytic heritage, and Davidson seems at first look to follow Quine very closely. It is thus not surprising to see Taylor regard Davidson’s modifications (referring to “radical interpretation” rather than “radical translation,” for example) as only minor and insignificant tweaks, making “the Quine/Davidson view” serve as a foil for Taylor’s contrasting views rather than as an inspiration. Let’s see how that works.
One aspect of analytic philosophy of language Taylor regards as characteristic of its unacceptably “external” focus, and for which he criticizes Davidson explicitly, is its tendency to explain meaning in terms of truth conditions. As with many of his intended targets, Taylor traces it to the naturalist and empiricist turn characteristic of modernity, and ties it to the conception of language as an instrument of representation, the point being to predict and control nature by obtaining true beliefs about it. If Davidson really does fall into this category, it will be a mystery why Rorty and other pragmatists take him as an inspiration rather than, as Taylor does, an opponent. As Quine’s student, Davidson certainly has the proper pedigree; and that his theory of meaning is indeed truth-theoretic is beyond dispute. So what’s going on here?
The basic idea of explaining meaning by reference to truth conditions is clear enough. We have learned the correct meaning of the word “apple” when we apply it to apples and not (except derivatively) to other things. We understand what “apples” are, that is, when we reliably use the word only to denote actual apples. “That’s an apple” is true only when the indicated object really is an apple. We may thus attribute that meaning to that utterance when it reliably occurs only in the presence of apples – its “truth conditions.” If it does not, the speaker may not be using the word with the meaning we’re thinking of attributing. If the speaker is a child learning English, we may say in such cases that the child has not yet learned the proper meaning of the word.
What is presented by its empiricist proponents (including but not limited to Quine) as a virtue of this account, though – its careful avoidance of spooky non-empirical mental items like the agent’s internal intentions to mean this or that, or her subjective impressions – is the very thing which makes it unacceptable to Taylor. It’s not that he wants to include purely internal items, as the Cartesian does; it’s that he sees the recoil to the self-consciously external perspective as leaving in place the conceptual dualism – inner/outer – we were trying to overcome. For Taylor, the detachment necessary to carry out a scrupulously empirical investigation into the correlations between utterances and what seems to cause them is not what makes understanding possible, but what makes it impossible: “it is plainly impossible to learn a language as a detached observer. To understand a language you need to understand the social life and outlook of those who speak it.” Such theories as Quine’s and Davidson’s “can only seem plausible because of the hold of the epistemological tradition.” [“Theories of Meaning,” 1980]
Even when modifying Quine’s account in important ways (most of which we will leave to one side), Davidson retains its basic truth-theoretic form. It thus seems to Taylor that no real progress can have been made: we are still stuck in the empiricist mud. Not coincidentally, there are plenty of analytic philosophers of language (most famously Michael Dummett) who themselves reject truth-conditional theories of meaning on grounds at least analogous to Taylor’s: that the agent’s perspective has been elided in favor of an entirely external view of correlations between utterance and world (or, again, “word and object”). As Dummett points out, truth-theoretical accounts can leave it a mystery how one can use words correctly in stating falsehoods, which of course we do all the time. If I see a pear that looks to me like an apple, and I say “that’s an apple,” I get something wrong; but what I get wrong was something about the world, not about the concept of apples. Surely what happened, the thought continues, is that I correctly used the concept “apple” to express what turned out to be a mistaken impression, not that I misused the term. Dummett’s conclusion is that rather than explaining meaning in terms of truth-conditions, we should, in taking account of how things appear to agents, appeal instead to the notion of “warranted assertibility.” Just as I may be epistemically justified in believing falsely, so may I be semantically justified in correctly stating that false belief. The meaning of my utterance isn’t constitutively tied to the truth of the matter after all.
From this perspective, Rorty and Hilary Putnam, who also talk this way at least sometimes, will seem to Taylor as the more enlightened representatives of the analytic tradition, and indeed as we’ve already seen, it is Rorty and not Davidson (the former’s regard for the latter notwithstanding) with whom Taylor has had “n rounds” of debate. In any case, Taylor thinks the very idea of a “theory of meaning” of this sort threatens to leave in place a key assumption about language which he himself rejects: the primacy of its “designative” function. Here we will find another clue to the puzzle.
As we’ve seen, the empiricist heritage of analytic philosophy of language can be traced back to Wittgenstein’s early work. However, Wittgenstein’s philosophy underwent a radical change later on, and it is this later version of Wittgenstein which Taylor finds more congenial, and with reference to whom he explains his anti-Davidsonian attitude. (On my reading, Davidson takes more from the later Wittgenstein than from the earlier; but let’s stick with Taylor for now.) In the very first sections of Philosophical Investigations, as Taylor notes, Wittgenstein presents a seemingly plausible account of linguistic meaning which he attributes to Augustine. On this account, the meaning of a linguistic expression, nominal or otherwise, is that real-world thing or process which correlates to it. Wittgenstein’s own example is that of the shopkeeper manifesting his understanding of the meaning of the phrase “five red apples” by looking in the bin marked “apples,” counting up to five while removing an apple for each number, and so on. As with many of the ideas Wittgenstein examines critically, something is clearly right about this picture – certainly the shopkeeper filled the order correctly, and so can be seen in one sense to have understood it properly, as a monoglot Thai speaker would not – it should also strike us as somehow off, as if someone who literally acted like this were autistic perhaps.
Taylor’s construal of what exactly is off here, following Wittgenstein as he reads him, is that the Augustinian picture reduces language to its “designative” function, regarding it as an “intermediary” between ourselves and the world it is thereby taken to represent. As we saw last time, this criticism echoes Dreyfus’s and Taylor’s criticisms of Rorty and McDowell: Rorty’s alleged “anti-representationalism” notwithstanding, the role the latter attribute to language in structuring the content of our mental states reveals a characteristically empiricist focus on beliefs (i.e. their contents and justification) rather than the more primordial phenomenon of pre-conceptual engaged coping.
In this context, though, Taylor’s main task is not a takedown of Davidson, whom he mentions specifically only in passing, but instead an endorsement of his own radically different alternative to the “designative” view of language, one he finds for example in the work of J. G. Herder. Tracing as before the objectionable idea to its modern origins, Taylor cites eighteenth-century French philosopher Étienne Condillac’s theory of the origins of language, which sounds much like Augustine’s in the relevant respect. On Taylor’s reading, Condillac takes for granted what Herder’s deeper view explains: the pre-conceptual background necessary for linguistic meaning to take shape. This charge should sound familiar; on Taylor’s view, as we’ve seen, McDowell’s empiricism (however “minimal”), in emphasizing the unconstrained reach of the “space of reasons,” similarly takes for granted the more primordial phenomenon of engaged coping. “Condillac’s heirs today,” he says, are “the proponents of chimp language, talking computers, and truth-conditional theories of language.”
As one commentator explains Taylor’s view, in “put[ting] designation first in the order of intelligibility,” such accounts of language as Davidson’s is taken to be leave “no room for the expressive or constitutive power of language.” In fact, Taylor “does not simply claim that the expressive/constitutive capacity of language sits alongside the designative capacity [my emphasis]. The claim is that the power of expression – the power of disclosing and constituting a human “world” – is fundamental and originary. The capacity of language to designate things is one amongst a series of possibilities immanent to the power of expression itself.” Taylor attempts to make this clear by steering hermeneutics away from its traditional project of explaining meaning to a more productive engagement with philosophical Romanticism and related thinkers like Herder (which I have to say does sound interesting, but we’ll have to get back to that some other time). He also makes the explicit connection to the phenomenological rejection of “mediation” in epistemology: just as the very idea of designating/knowing is derivative of an originary expressive power of language, the very idea of propositional belief is derivative of a prior ability of engaged coping. As before, a seat at the table is not enough.
So far we have seen little reason to regard Taylor’s as the “drastic misreading of Davidson” I referred to above. For that we will have to wait for part 3. (This means I get to read more Taylor before I do a drastic misreading of my own!)