The correspondences between PEC and PWC, in what concerns the timbre features, are regular in any case (note the palatalisation + labialisation in PWC, which can correspond both to *ü and to *qw + *e in PEC). In PWC, however, we would expect a form like *q:́ʷa (with tense *q:) corresponding to a long vowel in PEC. This discrepancy is still to be explained, but of course it does not make the comparison less valid.

    See Trubetzkoy 1930, 276, Abdokov 1983, 105.

    There is also a possible (but very problematic) Kartvelian parallel: Georg. (m)sχali, Megr. sχuli, Chan. cχuli, Svan. icχi 'pear' (see Charaya 1912, 26; Gudava 1954, 706-707).

    Pharyngealisation must certainly be reconstructed for PL; however, it was lost (neutralised before a front vowel) in the Tab.-Ag. area (otherwise in Ag. dialects we would have reflexes X̀/ħ), reappearing later in the same position in a number of dialects (Tab. Düb., Ag. Rich.).

    The most archaic root structure is preserved in Tsakh. (Gelm. ǯɨχIajä), pointing to the PL *-a-Auslaut (with the obl. base *χIerä-). The Arch. form has a suffixed -t, either inherited from the PEC variant *qHǖr-di, or, more probably, under the influence of the Lak. form quIrt.

    As for the strange first part *ǯu- (usually yielding regularly ǯV- or žV-, but devoiced in the first weak syllable before a voiceless *χI in Lezg., Ag. Rich. and Bud.), it is most probable that we deal with a compound *ǯum-χIera "quince"+"pear" with the first component distorted.

    See Trubetzkoy 1930, 84; Лексика 1971, 162; Гигинейшвили 1977, 92.

    See Mészáros 1934, 378; Balkarov 1969, 177; Shagirov 1, 238-239.