Why a long ē in rēx, rēgis but not in regere or regiō?

Question

I'm assuming it's not a phonological thing—like, if the ē in rēx was compensatorily lengthened because of g—>x, then it would be regis, not rēgis, right?

Answer 1

My first thought was that this looks like an ablaut variant. After looking into it a bit, this seems possible, but far from certain. I'm not familar with the field of IE literature, but using Google, I found a few sources that mention the matter, although I don't know how reliable they are. (So please don't assume this post contains the correct answer!)

• Wiktionary at least indicates the vowel length difference existed already in Proto-Indo-European:

  *h₃reǵ-

  *h₃rḗǵ-s (“king”)

• Some pages from A Grammar of Modern Indo-European by Carlos Quiles and Fernando López-Menchero (p. 188) that I found in Google Books seem to indicate that the reconstruction of the complete paradigm of the "king" word in PIE is a bit difficult, and there are alternative reconstructions of the nominative form. Apparently, the length in the Latin nominative could be ultimately derived from another form of the PIE word via analogy. I don't think this is likely to be a high-quality/well-credentialed source as it is part of a questionable movement to revive PIE, but there are some citations that you may want to check out. I may edit this post to replace this with a better source if anyone has a suggestion.

To me, ablaut is a bit of a "black box", but I am no Indo-European scholar—it's possible that someone will be able to explain the reason why lengthened-e grade is used in the "king" noun in Latin while short-e grade is used in the (present-tense forms of) the verb. There are theories that explain ablaut grades as arising from earlier differences in things like stress placement. But the Wikipedia article about ablaut that I linked to above says that clear cases of lengthened-grade forms that have to be reconstructed back to PIE are relatively rare.

• "The alleged “lengthened” grades in the roots of some Latin nouns" (2013), by Lucie Pultrová, actually explains the long vowel in rex and some other words as the result of stress shift at some point, but as the title indicates, Pultrová is of the opinion that this occured some time after the split of PIE, so that it belongs to the history of Latin specifically rather than being inherited from PIE ablaut patterns. This was interesting to me, but I don't quite understand what the phonetic motivation is supposed to be for going from zero-grade (when unstressed) to long vowels (when stressed). Pultrová seems to agree that the phonetics of the proposed sound change are questionable and requests that phonetic specialists review the data presented in the paper. Pultrová has written a book, The Latin Deverbative Nouns and Adjectives, of which you can see a review here that describes some of her other thoughts about the development of Latin stress and vowels.

I guess my main thought at this point is: good question! I hope someone can provide better insight into the answer, but if not, you've identified an interesting problem in Latin etymology.

Answer 2

de Vaan posits an IE *h₃reǵ- alternating with *h₃rēǵ-, but he does not offer an explanation for the alternating quantity of the vowel. Sihler 271,4 offers IE *reh¹ǵ-, implying (if I understand him) that rēx is from the cited full-grade form *reh¹ǵ-, and that reg- is from the zero-grade *rh¹ǵ-. The “classic” theory (e.g. in Walde) is that rēx is a long grade nominative.