Originally only accusative, from Proto-Tocharian *ñäś, ultimately from Proto-Indo-European *éǵh₂, compare Tocharian A näṣ m and Tocharian B ñaś. The first element ñ- probably came from a reduplicated Proto-Indo-European *(h₁)méme (the first laryngeal can be reconstructed by Hittite evidence, compare Hittite 𒄠𒈬𒊌 (ammuk)), dissimilated as *méne, a zero-grade *mne > *mñä > *ñä; the second element -ś is comparable to Ancient Greek ἐμέγε (emége)), Old Norse mik and Gothic 𐌼𐌹𐌺 (mik) (< Proto-Indo-European *mne ǵe), but Tocharian A pronoun origins remain opaque to Adams. To Petersen, the unpalatalised n- is probably related to first person plural pronouns, such as Sanskrit नः (naḥ) and Latin nōs, explained on the basis of an earlier plurale majestatis, but as Jasanoff explains “ there is no trace of the PIE oblique stem *nē̆s / *nō̆s / *n̥s- neither *nos, *nōs, *nes or *nēs could have yielded näṣ directly ”
Later on, Pedersen (1941: 135) suggest that the *ñ- in pre-Tocharian A was dissimilated to n- under the influence of the following palatalised *-ṣ, which is approved by Winter (1965: 203) and Adams (1988: 153), although no satisfactory account is given for the final sibilant. The first person singular pronoun is a Tocharian A innovation, the final -k is believed to represent a particle, Van Windekens sets up *mne-kʷām parallel to *mne-kʷim (> ToB. ñaś), Krause and Thomas adopt a similar formula, *me-kw(e) from which both ToA. pron. derive from, Adams opts for an apocopated *-kw(e), but does not link it to the ToB. pron. The vocalism remains problematic, Schmidt assumes a contamination with the 2 sg. pron. nom. tu obl. cu. Jasanoff however explains “ by starting from a nom. sg. *eǵō final *-ō seems to have first given *-u in Common Tocharian therefore, to assume a Common Tocharian development of *eǵō to *ñäku with subsequent rounding of *ñäk- to *ñuk- and regular loss of the final vowel in Tocharian A.” For the source of the gender contrast in Tocharian A, Jasanoff explains “Common Tocharian seems to have inherited two nom. / obl. forms from Proto-Indo-European — *ñäś, whence A näṣ and B ñäś, continuing the old accusative *me-ge; and *ñäku, whence A ñuk, continuing the old nominative *eǵō ”
He further explains “For an illustration, purely typological, of the kind of sociolinguistic effects that might have been induced by the weakening of the ego : me contrast in Common Tocharian, it will be useful to reflect briefly on the consequences of the corresponding weakening in modern English. As is well-known, the “correct” distinction between I and me tends to be lost in spoken English in several syntactic environments, including 1) after the copula, as in It’s me (him, her, us, them) for it is I (he, she, we, they); and 2) in compound noun phrases of the type X and I, X and me, which appear in substandard speech as invariant X and me or me and X (cf. Me and Fred are going to the movies, etc.), and in hypercorrect usage as invariant X and I (cf. the very common between you and I). The It’s me construction is absolutely normal; it is used in informal speech, at least in America, by speakers of both sexes and almost every educational and socioeconomic background. But It is I, and especially It is he and It is she, have their proper sphere as well. They are high-register variants, appropriate to a formal lecturing style or to polite conversation between educated speakers who are not on informal terms. In some speech situations the choice between It’s me (him, etc.) and It is (he, etc.) may be a matter of conversational strategy: the former is self-consciously colloquial and invites greater intimacy; the latter is aggressively "correct" and asserts social or educational distance. As such, It is I is sometimes associated with an overly careful, priggish. even schoolmarmish way of talking a style that tends to be stereotyped in the popular mind as a female characteristic.”
ñuk f