Why does Lithuanian šeima have "ei"? [closed]

Question

Why does Lithuanian šeima have north Samogitian "ei" instead of "ie"?

Answer 1

I’m not a Baltologist (by any means!), but I do know that the split treatment of PIE *eu̯ and *ou̯ (and *au̯) in Baltic, remaining either entirely separate or all merging in *ie, is rather a tangled knot and – as everything else in Balto-Slavic – depends on your views on Balto-Slavic accent and accent mobility.

The least contentious point seems to be that as a general rule of thumb all three diphthongs tend to become Baltic *ie in non-final syllables when they appear in originally stressed syllables and remain distinct when unstressed. (The outcomes in final syllables differ, but they’re not relevant here, so let’s leave them out to avoid muddying the waters more than necessary.)

Christian Stang’s Vergleichende Grammatik der Baltischen Sprachen (p. 52ff.) deals first with some cases where this development fits (with a bit of hand-waving and appeals to analogy and dialect forms here and there). He then deals with some cases where this doesn’t seem to fit, including šeimà. I will quote his argument at length here (p. 62f.):

Anderseits gibt es Fälle, wo sich betontes ai, ei nicht leicht durch konservierende Einwirkung der Ablautreihen erklären lässt, und wo auch eine analogische Erklärung mit Schwierigkeiten verbunden ist. So findet man unter den primären Nomina Fälle wie: véidas (Pl. veidaĩ) , šeivà (Akk. šeĩvą), deivė̃ (deĩvę), veĩslė, kreĩvas, laĩškas, vaĩkas, šeimà (šeĩmą), lett. sàime usw. gegenüber dienà (Akk. diẽną), kiẽmas usw. Man könnte natürlich darauf verweisen, dass in diesen Wörtern einige Kasus Endbetonung, andere Wurzelbetonung haben, dass also ursprünglich einige Kasus den Wurzelvokal ei, ai, andere ie gehabt haben. In einigen Wörtern hätte dann ie, in anderen ei, ai gesiegt. In Wirklichkeit wäre aber dies keine Erklärung. Die ganze Entwicklung kann nicht ein Spiel des Zufalls gewesen sein. Es ist aber eine naheliegende Annahme, dass diẽvas und kreĩvas, dienà und šeimà, giesmė̃ und deivė̃ einmal zu verschiedenen Betonungstyp engehört haben. Man muss nämlich m. E. annehmen, dass das Balt. wie das Slav. ursprünglich drei Betonungstypen aufgewiesen hat: einen wurzelbetonten Typus, einen mobilen Typus, und einen Typus mit fest betontem Suffix- oder Themavokal; vgl. slav. *vőrna, *zimà, *ženà, r. ворóна, зима́ (Akk. зи́му); жена́ (Akk. жену́). In Wörtern des dritten Typus kamen keine wurzelbetonten Kasusformen vor, und der Diphthong ie könnte daher überhaupt nicht auftreten, falls das System bis ins Lit.-Lett. gedauert hat. Im Lit. – und vielleicht auch im Lett. – sind aber später die Wörter des dritten Typus in den mobilen Typus übergegangen (die o-Stämme teilweise in den ersten Typus). Zwischen žiemà – mit Verallgemeinerung des Vokalismus der wurzelbetonten Formen – und šeivà existierte dann kein akzentueller Unterschied mehr. Der Gegensatz ie : ei war willkürlich geworden.

That is, he sets up three different accentuation types in Balto-Slavic:

1. fixed root accent
2. mobile accent
3. fixed accent on the suffix or theme vowel

In the first two types, the stress would fall on the root either always (type 1) or in some paradigmatic forms (type 2), so if the root had a diphthong, this would likely end up as Baltic *ie. In type 3, however, the stress would never fall on the diphthong, which would thus always (barring analogy) remain distinct.

At some point, he argues, type 3 was lost when most of its words were subsumed under type 2 (or in the case of o-stems, partly type 1). At this point, the link between the accent and the shape of the diphthong would have vanished, and we end up with pairs like žiemà and šeivà which have the same accent pattern, but different diphthongs.

This explains things rather neatly, but unfortunately he doesn’t explain (at least not there) why these particular words ended up in type 3 to begin with, which makes it a bit of a free argument: you can always just assume that the words that don’t have *ie originally had type 3 accentuation.

In the case of šeimà, we happen to know that its ancestor, PIE *tk̑ói̯-mo- (as reconstructed by Wiktionary, though tk̑éi̯-mo- would perhaps be a better fit for the Lithuanian form), was stressed on the root (cf. Sanskrit क्षेम kṣéma). We even have the slightly simpler, but otherwise entirely parallel and semantically very close, form *k̑ói̯-mo in Lithuanian káimas ‘village’, which has kept its root accent into Lithuanian.

The attested accent in šeimà clearly shows that the stress patterns of these two words did diverge at some point between PIE and Lithuanian, but when, why and how, I’m afraid I cannot answer.