Kui :ēja (ēji-) "to be separate from, detached, move aside, make way for"
Khuttia Kui :ēsp- "to winnow"
Kuwi (Fitzgerald) :ērḷali "to be separated"
Kuwi (Schulze) :ērsinai "to sort"
Sunkarametta Kuwi :ērs- "to winnow"
Parja Kuwi :ēhp- "to winnow"
Tekriya Kuwi :ērsp- "to winnow with a sidewise motion"
Kuwi (Israel) :erl- (-it-) "to go by a different path"
Additional forms :Also Kui ēspa (ēst-) to separate; Kuwi_F ēṛsali to separate; Kuwi_Isr ērs- (-it-) to separate, keep separate
Notes :An interesting and unique case. The original causative form seems to be preserved in Kui_T; in other languages either -r- or -p- have fallen out. As for the original auslaut cluster, on the base of the correspondence "Kui -j- - Kuwi -rl-" (-ṛ- in Kuwi_F is misspelled as usual) we reconstruct a PK *-Rl- which is indirectly indicated by external data (cf. Konda er_lis- to separate). The vowel is long in the causative form; in the simple form the short vowel, preserved in Kuwi_Isr, is primary; in other dialects vowel length is the result either of compensatory lengthening or analogy.
Kui :ēra (ēri-) "to spy, scout; n. spying, scouting"
Kuwi (Schulze) :hēnai "to scout"
Sunkarametta Kuwi :hēnḍ- (hēṭ-) "to see"
Parja Kuwi :hēnḍ- (hēṭ-) "to see"
Additional forms :Also Kui pl. action ērka (ērki-); Kuwi_S hēri kiyali to see
Notes :Besides the fact that conjugations do not match in Kui and Kuwi, the phonetics here is dreadful: irregular -r- in Kui, irregular -nḍ- in Kuwi_Su and Kuwi_P, irregular aspiration, irregular almost everything. Perhaps it would be wise to reconstruct some kind of cluster here, but first of all, it is not clear what exactly should be reconstructed, and, on the other hand, this is the only such case. Maybe it would be wiser to separate the two roots; but in this case we have to postulate at least three roots: *ēr- (Kui and Kuwi_S hēri, although the aspiration is still unclear); *ē- (*sē-?) in Kuwi_S, *ēnḍ-p- (*sēnḍ-p-?) in Kuwi_Su and Kuwi_P. External data give us only *ēr-; probably it would be justified to trace hēnai back to *hēr-nai? The obvious morphological irregularity here is that the verb in Kuwi belongs to t-conjugation without any obvious reason (the form hēnḍ- is absolutely unique for this conjugation, since hēṭ- < *hēnḍ-p- should be expected). It is probably this irregularity that leads to such irregularities in phonetics.
