*h₁ḗrH-/h₁érH-

Comparing HLuw. /irh(i)-/ c. ‘border’ with Hitt. e/irḫā- and arḫā- c. ‘line, boundary’, we may reconstruct either an ablauting root noun, which would have been secondarily thematized in Hittite, or an original thematic noun. Either option would yield an i-mutated stem in Luwian. Kimball 1994b:88f. suggests a vr̥ddhi-formation *h₁ērh₂-ó-, which would explain the i-vocalism and the plene spelling of the ending, whereas the root vowel -a- in the variant arḫā- remains unaccounted for. Melchert 1994a:84 prefers an accented full grade of the root, which would account for the a-vocalism due to the sound change *-éRC- > -áRC-. The vocalism in e/irḫā- is taken to be analogical with the denominative verb irḫā(i)-^(mi) ‘to make the rounds’, in which unaccented *e would yield Hitt. i. However, this fails to account for the Luwian root vocalism, which requires a lengthened grade (thus already Rieken 1999a:69); the putative sound change of pretonic *e to i in Luwian adduced in EDHIL:247 and Mouton & Yakubovich 2019a:218 n. 14 does not exist (see Melchert 2019b:365f. and Sasseville 2021a:294f. with n. 49). Kloekhorst (EDHIL:247) suggests a hysterokinetic ablaut of a *-eh₂- stem, which would have been secondarily thematized, postulating *h₁ér-h₂-, *h₁r-éh₂-m, *h₁rh₂-ó-, which leaves the Luwian evidence unexplained.No concrete explanation is offered in HEG A-K:365-367 and HED A:134f.

Taking all the currently available evidence seriously, we must reckon with an acrostatic common gender root noun *h₁ḗrH-/h₁érH- (+ *h₁r̥H-é/ó-), which would have yielded Hitt. erḫ-/arḫ- and likewise Luw. /irh-/arh-/. The thematization seen in Hittite would be comparable to the one found in the word for ‘foot’ padā- c. (Rieken 1999a:19). The vacillation in the root vocalism in Hittite is seen in the denominative verbs as well, i.e. irḫā(i)-^(mi) beside arḫā(i)-^(mi) ‘to make the rounds’. In Luwian, the lengthened grade was generalized in the noun and therefore it was kept in the formation in *-u̯o- found as the base of the following derivatives, e.g. ēḫḫui-^(ti), ēr(ḫū)wa- and ēr(ḫu)i(t)-. On the other hand, the full grade of the weak cases is continued grammaticalized in the adverb CLuw. āḫḫa. Concerning the following derivatives, i.e. the verb āḫḫuwāḫḫuwa- and its verbal noun aḫḫūwattar/-t(ta)n-, if they truly belong here, it looks like another derivation in *-u̯o- (see under lemma) built this time on the weak stem (unless directly built on the preverb āḫḫa in pre-Luwian). On the other hand, the two derivatives in question could be analyzed differently, i.e. extended root verb in -u-, i.e. *h₁ErH-u-, which would mean that āḫḫuwāḫḫuwa- is not a factitive but a u-extended formation like e.g. CLuw. malḫu-. However, this hypothesis cannot be verified yet, and thus more evidence is needed.

[D.S.]