Thread:User talk:CodeCat/Frankish vowel shifts/reply

Yes I just noticed that. I fixed it now in the other entries. I think the change ai > ei is part of umlaut, but there is other evidence (like *warjan) that umlaut didn't occur yet. It was probably purely phonetic, so the diphthong was [æi] like you said and was "heard" by Romance speakers as [ai] rather than [ɛi]. I don't know about *au. It shifted to ou at some time, and to ō in Old Dutch after that, and Old French borrows it as o too. I suppose it's "neater" if the change ai > ei is paired with au > ou, so we can just say neither of them happened yet. We can be pretty sure that it wasn't ō yet, because the original PG *ō became uo first; in the same way, PG *ē2 was still ē, not ie.

I wonder what Proto-Germanic *eu became. In Old Dutch, Old High German and Old Saxon it is io, but I think the earliest texts show eo and Old English still has eu in some early texts. So Frankish probably had eo or eu too.

I think that Sievers' law disappeared in West Germanic after the West Germanic gemination had taken place. The law determined whether a word had -j- or -ij-, depending on the weight of the root syllable. But the gemination caused all light roots to become heavy anyway (except those with -rj-), so the alternation no longer had any phonological trigger and probably disappeared. I'm not sure whether the end result was -j- or -ij-, but there are some hints.
 * -ij- was already the usual alternant for heavy stems before the gemination, and it was also the more common one, so it would make sense that it became generalised once all stems were heavy.
 * The fact that ja-stems have the ending -ī and never -i probably speaks in favour of the "long" alternant -ij- becoming the general form. This apparently also happened in the imperative of class 1 weak verbs, which always have -ī and never -i (which would have disappeared).
 * I think it's also more likely that the -ij- would disappear in Old High German and Old English if it was a syllable, and some early Old High German texts actually treat it as a vowel or a kind of diphthong (willeo, willio), so it was probably syllabic and therefore -ij- rather than -j-. This actually fits with how Old Norse treats it. Old Norse had no gemination (except for k and g) but it lost -ij- while keeping -j-; this is exactly what happens in Old High German and Old English, too, except that -j- only occurred after -r-.