A few days ago I argued that Chalmers’s proposed replacement of Carnap’s internal-external distinction (in ESO) bears little resemblance to its Carnapian original.  Today I will go on to claim that this proposed replacement (like other related proposals from the new metaontologists) not only doesn’t resemble that original, but is actually incompatible with it.

One strand of discussion within the supposedly “Carnapian” metaontology literature sees the internal-external distinction as concerned primarily to distinguish substantive or genuine questions from those that are merely “verbal.” To be fair, this can perhaps be seen as not entirely inaccurate, merely somewhat anachronistic. The Vienna Circle of the 1920s thought of meaningful discourse as delimited by the limits of language described by Wittgenstein in the Tractatus. In that conception, discourse “internal” to the language — there was only one — had meaning, while “external” discourse was meaningless pseudo-language that failed to qualify. Carnap himself sympathized with a version of this conception for a while, though he always had his doubts (see my book, Ch. 6-8). In 1931 he dropped it entirely, however (Ch. 9), and went on to embrace tolerance in 1932. What is not widely appreciated is that even in the Syntax, where the principle of tolerance was first programmatically enunciated, Carnap still thought of each possible language, once it was chosen, as completely specifying all meaningful discourse. There was still no possibility of “external” or “pragmatic” meta-discourse about language or about choices among languages (though Thomas Uebel sees in the “bipartite metatheory” of the left Vienna Circle at this time a precursor to the explicitly external meta-considerations after 1939). The language was freely choosable, as was its meta-language and every further level up in the hierarchy, but once such a framework had been chosen, all possible discourse was confined to it. This vestige of early Wittgenstein lasted until the late 1930s, when Carnap finally began to admit practical discourse external to such language frameworks, and thus set the stage for the internal-external distinction in its familiar ESO form.

So maybe there’s a sense in which you could see the early-Wittgensteinian criterion of meaning, even such later forms of it as the original Syntax conception, as distinguishing between genuine questions and merely “verbal” ones.  But even this isn’t quite right; what the Tractatus criterion distinguishes is genuine sentences from those with no meaning at all. The latter aren’t the material of (meaningful but pointless) verbal disputes, as in Chalmers or Hirsch; they are just empty verbiage. So when Chalmers reaches a “Carnapian conclusion” in his “Verbal Disputes” paper (pp. 563-4), he is terminologically on very thin ice.

Far more problematic, though, are the criteria for the verbality of a question or dispute propounded both in that paper (pp. 522-6) and in an earlier one by Hirsch (in the Metametaphysics volume, pp. 238-40). The details of these proposals are of less interest than the approach they exemplify. Hirsch proposes something like a definition of a verbal dispute, though the reference language isn’t fully specified. Chalmers suggests that a precise definition may not be appropriate in this case, and proposes something he sees as merely heuristic, not as stipulative. “We can instead see the characterization as pointing us toward a salient and familiar phenomenon, rather than delineating its contours precisely.” (“Verbal disputes,” p. 525) But what he proposes to rely on to identify this phenomenon is, ultimately, a shared intuition that there are pointless discussions, and that these fall into certain distinguishable categories: “Of course there are pointless disputes that are not verbal. . . But verbal disputes have a familiar and distinctive sort of pointlessness. I will not try to define this sort of pointlessness, but I will use it as a heuristic guide to the presence of a broadly verbal dispute.”

Within a Carnapian context, these are astounding words. First, Hirsch proposes a definition that is internal to something, though it is unclear to what. But then Chalmers backs away even from that to rely mainly on shared intuitions. The intuitions in question, though, are notoriously not shared among philosophers (never mind the population at large, who mostly — perhaps correctly — intuit just about all philosophy to be pointless). For Carnap it was intuitively obvious that ontology talk (taking external questions literally instead of as pragmatic questions of language choice) was pointless, but this is not intuitively even plausible to the new metaontologists. So why should either side accept Chalmers’s characterization?

It was precisely the shortcomings of intuition that drove the logical empiricists to seek precise criteria instead, as Rick Creath has emphasized. When this move is reversed, and the criteria (here for “ordinariness,” Chalmers’s misleading translation of “internality”) are once again replaced by intuitions, Carnap’s distinction goes out the window.

UPDATE (9 May): This post now has a sequel.