Comment: We can reconstruct for PL the direct stem *hi (Lezg. hi 'which', Ud. Nidzh. he, Vart. e 'what, which'); the same stem is used in Tsakh. (in conjunction with other interrogatives) hi-šu 'who' and hi-ǯo 'what', and, as oblique, in Tab. hi-ʔan / hi-q̇an 'how many', Ag. erg. hi-na 'who', Bud. he-je 'where'. M. Alekseyev suggested that several interrogative forms in Lezgian languages beginning with n- actually go back to the same stem with reduction of the initial unstressed syllable (thus n- does not represent any independent interrogative morpheme, but is just an old oblique stem marker *-nV). Cf. Ag. Rich. na 'who (erg.)' which is obviously identic with Ag. Tp. hi-na. In this way we can explain Lezg. erg. ni (< *hi-ni), Tab. naʔan 'where' (< *hi-na-ʔan), Tsakh. ne-n 'which' (< *hi-ne-n).
Besides the oblique base *hi-nV- there existed in PL a variant *ha- (sometimes also *ha-nV- with the same oblique stem marker) reflected in Rut. erg. ha-la 'who', Kryz. erg. ha-lɨr id., Arch. ha-n 'what', Bud. ha-nu 'whose'. See Aлексеев 1985,74.