Talk:mājur

@Womtelo Isn't the *maturuR variant attested in various subbranches of Oc, or would you argue that the vowel assimilation *i-u > *u-u occurred independently in several places? The i/u-alternation is also vexing us in the western MP area, and I can't think of a better approach than Blust's, viz. to reconstruct two forms from the start. –Austronesier (talk) 21:22, 23 December 2021 (UTC)

@Austronesier Hm, I see. I always had the idea that the proper etymon was *tiduʀ, because it's the maximally differentiated form (lectio difficilior), and because it's the one found in Malay. That said, it is true that Oceanic lgs tend to have double /u/, with POc *maturuʀ cited along with *matiruʀ (Some of the Oc languages I can think of are compatible with both etymologies; others, with only *maturuʀ). Also, Tagalog has tulog... so you might be right that *tuduʀ ought to be reconstructed all the way to PMP/PAN, with an indication that there's an alternation. OK, I'll change my edits to mājur. - Womtelo (talk) 22:04, 23 December 2021 (UTC).

@Womtelo There is this strange selective instability with the sequence *i-u. For *tiduR and *ijuŋ, the variants *tuduR and *ujuŋ appear all over the place; next to *inum, *unum occurs in the restricted range of Blust's CEMP (which I personally don't consider to be a valid subgroup); but then *ikuR rarely has produced a variant *ukuR (except for ukui in a few Bungku-Tolaki languages, or Ma'anyan ukuy), while *pitu and dozens of other words with *i-u never do so.

I actually find it hard to believe that pairs like *tiduR/*tuduR co-existed at so many different levels of intermediate proto-languages all the way up to PAN. But since there is no naturalness about the way *i changes to *u only in these specific words, reconstructions like *tuduR are – at the current state of "ignorance" – the least bad way of bookkeeping, I guess. Dyen probably would have added another vowel to the inventory... :) –Austronesier (talk) 23:11, 24 December 2021 (UTC)

@Austronesier, I had not realised the regular irregularity you're describing. If I were given just that set of examples as a corpus, I could venture a hypothesis (inspired by J. Lynch's reconstruction of stress in POc): namely, that *(C)iCuC were oxytone *(C)iCúC, so that unstressed /i/ tended to assimilate to a following stressed /u/: *tidúʀ ~ *tudúʀ, *ikúʀ ~ *ukúʀ (albeit rare), *ijúŋ ~ *ujúŋ, *inúm ~ *unúm…; whereas *pitu, being coda-less, was paroxytone *pítu, hence preserved its stressed vowel intact. That said, there may be many counter-examples that I'm not aware of.

But OK for the interim solution, to have separate entries for *tiduʀ vs. *tuduʀ, with mutual links. (I like your joke about Dyen!) Keep up the great work, — Womtelo (talk) 15:05, 26 December 2021 (UTC).