Comments:EAS 155, Martin 234, Miller 1976, 370-371, 1979b, 17-31, Martin 1996, 35. The parallel seems quite plausible (on a similar interchange between "island" and "forest" see, e.g., under *ā́rV), but vocalism raises some doubts. The reconstruction *ši̯àmi is possible if one assumes PT *simek < *semek (which is quite possible, because the word is only attested in OUygh. and may have actually been semek), and PTM *šumi < *šimu (cf. the Udehe (An.) form simb́o - perhaps archaic).