It’s rather pleasing that the conversation here has moved past the article, proceeding more or less onto the territory of A Nice Derangement of Epitaphs. We worked through that essay in the Other Place](A Nice Derangement of Epitaphs - The Philosophy Forum Archive).
Here’s a summary of my account from that archive. Davidson posits three principles that might allow for a “first meaning”; that it be systematic, shared and governed by learned conventions. The paper’s subject is not malapropism but whether these three principles jointly describe linguistic competence. On the face of it, malapropisms, by their very nature, cannot be subsumed and accounted for by the conventions of a language. But there are complications. Malapropism can only register as a deviation, and so is only interesting, because it presupposes the very convention it breaks. They occur against a background of convention.
That Davidson’s radical interpretation can cope with a malapropism is shown using both the “flamingo/flamenco” and “soup latrine” examples. Rather than one fixed rule, the interpreter constructs a T-sentence empirically, holistically, using background knowledge and the Principle of Charity to select among live candidates:
- “That’s a nice soup latrine” is true IFF that’s a nice soup tureen (mispronunciation reading)
- “That’s a nice soup latrine” is true IFF that soup looks like shit (disparagement reading)
The T-sentence apparatus itself is not threatened by malapropism, because the interpretation is not word-for-word; it is holistic. The semantic theory survives even though naive convention-following does not.
The structural problem is that if language use really were governed exhaustively by convention, encountering a violation should leave us unable to proceed, in a way analogous to moving a pawn backwards ends a game of chess. But instead, we find that we often can indeed proceed, almost seamlessly. The mooted criteria do not serve to decide what is to count as part of the language, and what is to count as nonsense, since malapropisms provide interesting ways of misapplying the rules that nevertheless count as language.
Davidson holds that principles (1) and (2) survive, albeit in rather unusual ways, but (3) — that first meanings are governed by learned, prior convention — cannot stand. This is not to deny any role for convention; conventions are necessary but not sufficient for understanding. Language is more than following conventions; rule-based conventions are part of language, but not fundamental to it.
At fault is the assumption that there is a thing that can be called the meaning of the sentence; we are better off considering what is being done with the sentence, whether it be admiring a tureen or disparaging a soup. This ties the paper back to a Wittgensteinian commitment that meaning is use/action, not object.
The Principle of Charity is not a formula for reaching agreement but a practical, unavoidable stance: interpreters have little choice but to make charitable assumptions.
I’ll here point again to my An approach to aesthetics .