/kusa-⁽ⁱ⁾/

Puhvel (HED:K 289f.) is skeptical of any connection between Luw. kūšainta still badly understood at the time and kušattar, on the one hand, and kušala(i)- as attested in the Hittite texts, on the other. In his opinion, the underlying verb is *kušā- reflected in a deradical *-lo- stem in Hittite, and in the (according to Puhvel, improbable) case that Luw. kūš(a)i- (or kūša(i)-) and kušattar are etymologically connected, they would be based on *kušā- as well.

In contrast, Starke 1990a:327-329 and 515 (followed by CLL:113) is positive about deriving the three words from the same Luwian root. If this is correct, the starting point is Pre-Luw. *kus- or *k^u̯s- ‘to scrape’, reflected in the original root verb of the ḫi-conjugation HLuw. verb /kusa-⁽ⁱ⁾/ ‘to erase, efface’ (< PAnat. *k^u̯os-/k^u̯s-⁽ⁱ⁾).For the full Luwian thematization of root verbs of the ḫi-conjugation, see Sasseville 2021a:387-400. From the root an instrument noun with the Luwian suffix -āl- n. (< PAnat. *-ó-lo- n.), i.e. *k^u̯s-āl- or *kus-āl- ‘curry, comb’ was derived. This, in turn, is the base of a secondary denominal kušal(a)i- ‘to curry’ (cf. also Ehelolf 1938a:vii fn. 3 and HEG:A-K 669f. with references), which is attested as a loan word in Hittite.Outdated Neu 1968a:142 n. 4 (following Kammenhuber 1961b:155 n. e) with a segmentation as noun *^GIŠkuša- + verb la- ‘to make loose’; cf. already Neumann 1963a:41. Furthermore, the root *kus- or *k^u̯s- gave rise perhaps to a verb in -a(i)- (< *-eh₂-i̯e/o-) in kūšainta, probably via an abstract noun *k^u̯s-eh₂- or *kus-eh₂-, which may be based on an adjective *k^u̯s-o- or *kus-o- ‘scraping’. The same adjective also underlies the abstract noun kušattar.

[E.R.]

On semantic grounds, Luw. /kusa‑⁽ⁱ⁾/ ‘to erase, efface’ should not be connected with Gk. ψάω ‘to rub’ (contra HED K:290), which rather belongs to PIE *pes- ‘to rub’ (cf. LIV²⁺ s.v. Neu: *pes- ‘reiben’, following Oettinger 1979a:327). This root is also underlying the PIE word for ‘penis’ (cf. Lat. pēnis) continued in Anatolian as Hitt. pešn- ‘man’, cf. EDHIL:670, with doubts on the semantic connection with ‘to rub’.
A semantically more promising but phonologically difficult (cf. below) derivation is cautiously proposed here. HLuw. /kusa‑⁽ⁱ⁾/ ‘to erase, efface’ can be traced back to ‘to scratch, scrape’. Furthermore, the above-mentioned derivational base of both CLuw. kūšainta and kušattar may have meant ‘scraping, scratching’. Thus, the semantic connection with a root meaning ‘to scratch, scrape (vel sim.)’ is preferable to the proposal made by HED K:290 (cf. above). One of the roots with such a meaning is reconstructed by LIV²:372 as PIE *kseu̯- ‘to scrape, grind’.

According to HED K:290, HLuw. /kusa‑⁽ⁱ⁾/ ‘to erase, efface’ and its inner-Anatolian cognates should be compared to Gk. ψάω ‘to rub’. Morphologically, the derivational bases of both Gk. ψάω ‘to rub’ and CLuw. kuš(a)i- ‘to erase, efface’ could go back to a common eh₂‑stem PIE *k^u̯s‑éh₂- ‘scraping’. Under this proposal, the underlying root would have to be reconstructed as PIE *k^u̯es- ‘to scratch, rub’. However, since the Greek verb is by now connected with PIE *pes- ‘to rub’ (cf. LIV²⁺ s.v. Neu: *pes- ‘reiben’) and PIE *pés‑os- ‘penis’, this equation is not tenable anymore.
Alternatively, it can be tentatively proposed to affiliate the Luwian formations to the root PIE *kseu̯- ‘to scrape, grind’. A regular formation of, e.g., a zero‑grade o‑stem would result in PIE *ksu̯‑ó‑, albeit being not in line with pre‑Luw. *k^u̯s‑a- or *kus‑a‑. Note, however, that nearly each Luwian formation mentioned above ultimately contains a vocalic suffix. If they were derived from the root PIE *kseu̯‑, the anlaut /ksuV-/ would be realized as *ksu̯V‑. This triphonemic complex consonant cluster might have been dealt with either by splitting up *ksu̯V- to *ksuu̯V-  according to Lindeman’s Law in monosyllabic forms or by metathesizing *ksu̯V- to *kusV‑.However, no such metathesis is attested in Anatolian. Hitt. šumaš ‘you (pl.)’ is most likely not a parallel for such a metathesis (probably not < *uš‑maš), cf. Katz 1998b:138–141, EDHIL:116, LIPP II:859.The result would be an ordinary CVC‑root, which additionally is reminiscent of the semantically close (and etymologically related) root PIE *kes- ‘to scratch’. This root is continued in Anatolian as well, cf. PAnat. *kés-/ks-´^(mi).

(Gk. ψάω ‘to rub’)

cf. PAnat. *kés-/ks-´^(mi)