Proto-Finnic declension

If *s : *h gradation is suffixal rather than radical, then that means the rule is that it weakens at the beginning of every non-initial odd-numbered syllable, right? But then, as you say, it was analogically extended to trisyllabic forms, and it apparently was extended to *mees as well. On the other hand, when I applied the older radical gradation rule for *s, out rolled the illatives *kuninkasehen and *kuninkasihen, which explain -s- of the Finnish forms kuninkaaseen and kuninkaisiin very well indeed. So in this case, the *-s- cannot have been gradated and must have remained. If that's true, then it begs the question why *s was analogically gradated to *h in all the other cases but not in this one. I really wonder what rules, if any, applied to *s : *h gradation in Proto-Finnic times.

I've now added alternative forms of the passives with both *t and *tt in the ending, and applied syncope to the former one as usual. But the present passive connegative is a bit tricky. If the regular present passive was *-ksen, then the expected equivalent form of the connegative would be *-ktAk. Is this right?

As for gradation of tt < *kt in South Estonian, that could very easily be analogical. After all, *ht : *hd in Finnish is analogical as well, isn't it? I don't think you can put too much weight on that.

I don't know a lot about the inflections of other Finnic languages than Finnish and Estonian. I know that Finnish has a potential, but I thought it was a typically northern feature, so is it attested in South Estonian at all? Also, was contraction of *-sen- > *-sn- > *-ss- regular in the inessive? Given that it was regular in the potential, I assume so.

When I created the conditional forms, I assumed that the sibilant in the mood marker was *c and that it may have formed through assibilation. But I did that because otherwise, with the radical gradation of *s, it would end up as *h more often than not. If *s was not subject to radical gradation after all, then *s could of course be the real phoneme here, rather than *c. But Sami has what seems to be a precursor to *c, so is *c right after all?

I've now adjusted the non-present 1pl and 2pl ending to have a single consonant instead of a geminate. But of course, the single *t in the 2pl would have triggered syncope of a preceding *e. This would include at least the optative *-tedek > *-t'ek, and the potential *-nedek > *-ndek. But of course if the *-n- of the potential marker also contracts with the preceding consonant in some cases, so what would happen in this case? *-lndek, *-nndek, *-sndek don't seem like allowable combinations, so they must have either simplified, or contraction of one syllable must have blocked the other from doing the same. What do you think is most likely here?

How regular was contraction of *-het- (< *-šet-) and *-ket-? In Finnish it rarely appears at all, only clearly affecting tehdä and nähdä. Is it safe to assume that contraction of at least *-het- > *-ht- was regular?

The illatives of *s-stems are actually from forms like †kuninkaasen < *kuninkahe-sen with the stem in the weak grade as expected, and hence the ending in the strong grade. The long-vocalic endings -seen, -siin are by analogy to diminutives like punaiseen < *punaise-hen.

The sibilant in the conditionals has certainly been *-c-, yes. The mood marker ,has been explained to have developed from the continuative verbal suffix *-ice-, as in ilo 'joy' : iloitse- 'to revel'.

On syncope:

• I'm not sure what you mean by "contraction in the inessive". We have no evidence of any former vowel between *s and *n in the ending.
• Loss of two subsequent vowels is not attested anywhere, no. I guess the analogy of the other personal endings would have kept the 2PS potentials from contracting.
• IIUC there is no original *-te- element in the optatives: the long vowels are shaped after the relatively frequently used 3PS forms. Hence *-gotek, etc.
• I think *-ht- < *-šet- is pretty much regular, yes. Though of course this is not found in the inflection of words like hanhi or karhi where another consonant precedes. CCC contracted forms like *kante-ta- > *kantta- > *katta- 'to cover' were fossilized already by Proto-Finnic.
• Syncope before *k is exceptional and indeed does not seem to have occurred anywhere else than in nähdä, tehdä and their passives nähnyt, tehnyt. I wonder if the fact that all other -ke-stem words are back-vocalic (e.g. lukea, pukea) or nominals (e.g. mäki, väki) has something to do with this. It may also be relevant that these verbs have an exceptional inflection in Livonian, based largely on the monosyllabic stems nǟ-, tīe- (which look like as if they were from *näxə-, *texə- and not *näkə-, *tekə-).

Thank you for your answers. I meant essive not inessive, sorry about that.

For *s : *h gradation, should I apply the following rules?

• If the stem ends in -Vse- and the nominative ends in -s, then it always weakens when a vowel follows it. The following illative -s- will appear in the strong grade.
• In all other cases, the illative -s- weakens to -h-.

Would these rules give the correct inflections, or are there other considerations? What about verbs with -Vse- stems?