Slavic *-os (*-om)

The most enigmatic part of (North-)Krivichian (Old Novgorodian) morphology is the o-stem nom.sg.masc. in -e. Whereas the rest of Slavic has merged the nominative and accusative as *-U (as the result of regular sound change *-os > *-uh > *-U; *-om > *-uN > *-U), Krivichian maintains them separate as nom. -e vs. acc. -U.

Zaliznjak's "Drevnenovgorodskij dialekt" reviews some of the attempts that have been made to explain the nom.sg. in -e. They can be divided into purely phonological explanations, and morphological/analogical ones. The latter type of explanation is represented by Sobolevskij's hypothesis that the -e comes from the vocative (this is probably equivalent to V.V. Ivanov's hypothesis that -e stems from a PIE "casus indefinitus"); or Shaxmatov's hypothesis that -e comes from a postfixed copula -(j)e.
Two phonetical interpretations are mentioned, one (by Borkovskij and Zhukovskaja) to the effect that -U became -e in Krivichian, which is obviously not the case (cf. the opposition nom. -e : acc. -U), the other, by Nikolaev, Dybo and Zaliznjak, to the effect that PIE *-os gave Kriv. *-& > -e, instead of -U as in the rest of Slavic.
The hypothesis I had earlier expressed in this forum is basically identical to the one Zaliznjak attributes to Vermeer and Krys'ko: the -e comes from soft-stem *-jos > -(j)e. This means that the ordering of the soundlaws must have been slightly different in North Krivichian as compared to general Slavic, where the raising of /o/ (/a/) before /h/ preceeded the Umlaut of vowels after /j/ [Slavic *-jos > *-jah > *-juh > *-jI vs. Kriv. *-jos > *-jah > *-jeh > *-(j)e]. The main problem with this interpretation is that precisely in the soft stems, the ending -(j)e is hardly ever found, and the nom.sg. is mostly -(j)I (except, obviously, in those documents where /I/ is written with the letter <e>).

The Nikolaev/Dybo/Zaliznjak theory can be vindicated if Old Novgorodian shows other instances of *-os becoming -e. As a matter of fact, there is one such instance, namely the verbal 1pl. -me (jesme "we are", vs. general Slavic *jesmU). However, not a trace of *-os > -e can be found in the the dat/abl. pl., which has -mU everywhere (-amU, -omU, -ImU, -UmU).

Another instance of PIE *-os is obviously the s-stem NA sg. *-os, which in Krivichian as well as in Slavic as a whole is reflected as -o. The s-stem neuter ending has always been the main objection against *-os > -U as a regular phonetic development. However, the evidence that a final *-s (-h) raises a preceding back vowel (ah > uh, a:h > o:h, o:h > u:h) is unmistakable, so the NA sg. -o of the s-stems, like the NA sg. -o of (some) neuter o-stems must have an analogical explanation. I have earlier expressed the opinion that in those forms, the *-s or *-m was replaced by pronominal *-d (as for instance in NA n. *tod). In view of the fact that *-d was apparently lost very early on in Balto-Slavic (otherwise, the vowel would have been lengthened by Winter's law, and we would have had *tod > +ta), I would now revise that in favour of a formulation to the effect that *-s and *-m were simply deleted in the s-stem and *-óm neuters. In the case of the s-stems, the move would appear to be entirely logical: a form *<ta nebas> was transformed to <ta neba> (> to nebo) by deleting the -s, which was felt to be an improper "masculine" morpheme here. The same thing might have happened with the -N of the thematic neuters (*<ta se:dláN> => *<ta se:dlá> > <to sêlo>), although that does not yet explain why barytone neuters became masculines (*<ta dváraN> => *<taN dváraN> > <tU dvorU>), or how Baltic came to lose the neuter gender altogether.