Matti Eklund has now replied to Gabriel Broughton’s critique of his conception of Carnapian frameworks. Broughton is more than capable of fending for himself, but in the course of addressing Broughton’s critique, Eklund also takes a little swipe at my post about it, so I’ll respond briefly to that. He accuses me (as he accuses Broughton) of having unfairly attributed to him what Broughton calls the “Natural language thesis: Carnapian frameworks are natural languages.” Now of course I’m very glad he wants to back away from that, and I’m glad he now thinks the “natural language thesis” absurd, but I think he’ll have trouble convincing anyone he didn’t hold it previously. In his own self-quotation on pp. 8-9 of his reply, he explicitly conceives of a Carnapian framework as a specific natural language (English). Still, it’s never too late for repentance; no soul is beyond redemption (the Carnap Blog adheres to a broadly Christian attitude in such matters).
Eklund waxes indignant at having a restrictive interpretation of the “natural language thesis” attributed to him, in which it is understood as “Carnapian frameworks can only be natural languages.” — Fair enough, but he does (or did, at least) clearly think they can be natural languages, as in that passage self-quoted on pp. 8-9.
So okay, yes, Broughton should perhaps have been a little more precise and stated his “natural language thesis” more loosely as something like: “Carnapian frameworks can be natural languages.” Just about everything he says in his paper holds up with that looser version, since the problem with Eklund’s conception of frameworks (or his previous one, anyway, of which he now repents) is that he doesn’t (or didn’t) understand that they can’t be natural languages, since natural languages leave open the question whether their terms are to be regarded as explicated in some framework (i.e. are to be taken as internal) or are to be understood in some absolute, language-transcendent (external) sense. To allow that a natural language could be a framework is circular, against that background.
Eklund also claims that Broughton’s conception of frameworks “won’t work.” But his only critique of Broughton’s conception (which looks to me remarkably like Carnap’s) is his claim that “internal” and “external” in Carnap’s sense are not exhaustive of all existence questions (which would make Broughton’s definition of “external” as “not internal” too broad). This is a much more plausible objection than his complaint about the “natural language thesis.” However, he doesn’t give concrete examples, so it’s not entirely clear what he has in mind. I’ll guess that what he may have in mind are existence questions framed vaguely, i.e. for instance in natural language. But this gets us back to the same circularity, since the ESO distinction between internal and external questions has to be understood against the background of Carnap’s ideal of explication. The problem with natural language is that it underdetermines whether it is to be understood internally or externally; it doesn’t come with an index attached, telling us which of those is meant; a decision is required.
But perhaps Eklund could argue that there are more than two choices at this decision point; one could e.g. think up pathological natural-language examples that are so vague and underspecified that, while clearly not internal, also do not even qualify as external, such as “are there woozles?” (the tracks in the snow of the fearsome woozle, readers will remember, turned out to be Winnie-the-Pooh’s own tracks); the entity in question here is fictional (i.e. nonexistent) even within the context of a particular work of fiction. I think Carnap himself would have classified fictional existence, along with everyday historical existence, as internal (he’d have thought the question whether Pooh exists to have a well-defined context just as the question whether King Arthur existed has its — different — well-defined context, and the woozle was established in that well-defined fictional context not to exist); historical and fictional questions of existence were sub-headings under questions of existence within the everyday thing-language. Outside the Pooh context, the question would be an external question. Well, how about completely nonsensical or unknown entities? Shouldn’t Carnap’s own earlier diagnosis of Heidegger in “Überwindung der Metaphysik” (1932) lead us to classify those as simply nonsense? Perhaps, which would mean that “internal” remains well-defined, but “not internal” contains two categories of nonsense, “total nonsense” and “attempted language-transcendence.” Admittedly, only (some of?) those in the second category are candidates for translation into practical questions, but my sense is that Carnap would probably still have regarded internal and external questions as largely exhaustive of existence questions (perhaps disqualifying the total nonsense from even amounting to “natural language”).
Another possibility is that Eklund might have had fuzzy frameworks in mind (frameworks whose defining language is a fuzzy logic); here the boundaries of concepts are not Fregean but defined as vague or stochastic. But in that case “internal” and “external” are defined by that same vague or stochastic criterion; there is some ambiguity about borderline cases, but “internal” and “external” are still exhaustive.
Or was he thinking of something else? If so, I hope he lets us know, either as a reply on this blog or in some other forum.