scattered musings about the “because” locution

Take a look at the following sentence, from this statement from Mildred Loving, the plaintiff in Loving v. Virginia:

We didn’t get married in Washington because we wanted to marry there.

This sentence — and others like it — is kind of puzzling, though it’s hard to explain why looking at that sentence. Take a look instead at (1):

(1) We didn’t buy bread because the bakery was closed

(1) has at least two readings. On the first, we didn’t buy bread, and the reason for this was that the bakery was closed. On the second, we did buy bread, but our reason wasn’t that the bakery was closed — it was something else.

Both readings are somewhat puzzling. We could try to say that the first reading of (1) is a result of instantiating the following schema:

(S1) ~(φ because ψ)

But that doesn’t seem right, since making the negation take widest possible scope seems to defeat the factivity of the ‘because ψ’ locution (cf. my earlier post arguing that this was the case). The ‘because’ in the first reading of (1) is clearly factive. So maybe the first reading is a result of instantiating (S2):

(S2) (~φ) because ψ

This seems like the right thing to say about the first reading, though some puzzles remain. Explaining why the factivity appears to be affected by instantiating (S1) is a tricky matter, since the presuppositions of factives are, by definition, supposed to project under wide-scope negation. Maybe we could start by distinguishing between VP- and S-level negation, as Simon suggested doing in that earlier post, while holding that presuppositions of embedded elements project only under VP-negation. But this would just be a start, since:

(i) There are plenty of cases of S-level negation where a presupposition borne by an embedded presupposition-bearing element (e.g., a name or definite description) does appear to project.

(ii) There are plenty of cases where a presupposition fails to project under what appears (syntactically, at least) to be VP-negation.

(iii) To make this proposal work for (1), we would need to treat “[did] buy bread because the bakery was closed” as a VP-node. This doesn’t really seem plausible unless we treat “because the bakery was closed” as an argument of the predicate “buy bread”, and it’s not really clear what syntactic or semantic motivation there is for doing this (though I’m sure someone has written something about this question somewhere).

How about the second reading? The factivity of the ‘because’ locution has clearly been neutralized, so maybe this reading is just an instance of (1). But that is wrong, because the relevant instance of (S1) appears to be compatible with our not having bought bread at all, while (1) appears to entail that we bought bread. At this point I’m genuinely stumped. My hunch is that there’s some kind of metalinguistic negation targeting only the content of the “because” locution. This would explain both why the relevant reading of (1) entails that we bought bread, and also why the factivity of “because” is neutralized. But I have no idea how to tell any kind of formal story about this, especially given the apparent syntactic location of the negation. HALP.