Reconstruction talk:Proto-Finnic/kevät

Citation needed
Where does this reconstruction *keväc come from? I've not seem any such thing referenced anywhere, and there is no t/s alternation in the singular (only in the plural). Most reflexes indicate simply *kevät. As noted at Wikipedia, I belive Veps keväs should be by default assumed to be simply a shift in declension class. Võro kevväi : keväjä- indicates something like perhaps *kevägä (and could be a similar development).

The Uralisches Etymologisches Wörterbuch reconstructs original *keŋä, and assumed secondary suffixation in Finnic, though this is kinda suspect, as this word is a completely sui generi stem type in Finnic. At some other place I've seen original *keŋäč(V), which might work better. --Tropylium (talk) 00:16, 9 November 2014 (UTC)
 * This really comes down to the question of whether there was a real distinction between consonant and vowel stems. As far as I can tell, if the two types were distinct at all, the only way they would differ in Finnic is by the presence/absence of the final -i in the nominative. Syncope and epenthesis of -e- would have removed any other differences. —CodeCat 00:30, 9 November 2014 (UTC)
 * I mainly mean: whose reconstruction is it to have *-c and not *-t, and to assume some sort of later levelling? There was definitely a contrast between these (e.g. *ohudet > Fi. ohuet plural of 'thin' vs. *ohudec > Fi. ohuus 'thinness'). --Tropylium (talk) 01:50, 9 November 2014 (UTC)
 * It's my own, based on the assumption that there was no distinction between consonant stems and e/i-stems. If there was, then the inflection module needs to be changed to handle it. —CodeCat 01:53, 9 November 2014 (UTC)
 * Ah, I was wondering if you had found some recent reference on this topic.
 * I'm not sure if I follow what the declension problem is though. This word has no *-i, and hence there should be no reason to assume assibilation. It should be perfectly possible to continue treating consonant-stems and vowel-stems the same, as long as you allow for a distinction between stems ending in *-c(i) ( : *-t-), and those like this that seem to have had plain *-t (which definitely existed).--Tropylium (talk) 02:14, 9 November 2014 (UTC)
 * Well that's the problem. The word has no -i because if it had one, it would have been apocopated anyway. So it's more a matter of assuming it had once been there, or that it never was. Right now the inflection module changes final -t to -c in such nouns because it assumes the -i had always been there. But that apparently needs to be changed so that it knows when nouns had no -i and when they did. How would this apply to verbs, though? Verbs also have an endingless form, the third-person singular present. The same code that produces the -c in nouns also produces this form, so any change would affect such verbs too. —CodeCat 02:21, 9 November 2014 (UTC)
 * If you're asking what is the correct underlying phonological analysis, that is an entire research problem of its own. But as long as we mainly care about getting the surface forms right, I could imagine three options:
 * The vowel-stem/consonant-stem distinction.
 * Phonologizing the *t/*c contrast in words like this, + a rule that sorts stem-final *-(C)c- and *-(C)t- into the same declension class everywhere else.
 * Phonologizing the *t/*c contrast everywhere (and not just in the geminates).
 * I think there is still no single *t/*c contrast in verb stems. Though things should be amendable if new issues turn up.--Tropylium (talk) 23:43, 9 November 2014 (UTC)