@JoonasIlmavirta I had to check De Vaan to make sure I was right: clavis is likely an independent derivation from the same root. It's possible claudo was formed from it, but not likely. You do have claustrum, which was derived from claudo, too, and that does mean "bolt, bar" like sera, but the derivation is the other way around.

So: sera -> reserare (1st conjugation because that's typical for deriving verbs from nouns), claustrum from claudere, but claudere and clavis both from *kleh[sub]2[/sub]u-.

De Vaan does indicate that it's possible claudere comes from clavis, but that would have well-preceded Proto-Italic, so I'm not sure how immediately clear the connection would be to a Roman. Same with clavus, another cognate of clavis.

@cmw Thanks! I'm glad to have dispelled that misconception of mine.

@cmw Not much, turns out. Lots of people discussing early article-like behaviour of demonstratives and a couple of overviews of bilingual interference (e.g. in Plautus's Miles Gloriosus, which has more demonstratives than you'd expect under the influence of its Greek model), as you'd expect, but seemingly nobody explicitly discussing a Greek origin of the Romance definite article recently.

Bunch of people talking about the Peregrinatio Aetheriae, a 4th-century travel narrative that "vividly shows the nascent definite article".

There's supposed to be a 1975 article in the Transactions of the Philological Society, "Greek influence on Latin syntax" by R. G. G. Coleman, that discusses it, but I don't have access to it through my university. I'm told that he concludes influence was likely, but I don't have any details.