"Gender" is an obligatory grammatical category. It is a linguistic term and has no connection with biological sex or social personae. Indo-European languages, with their proclivity for tripartite constructions, label gender "masculine," "feminine," and "neuter." The labels only very roughly correlate with biological sex, are applied to innumerable nouns with inorganic or even intangible referents, and may override the actual biological sex of a referent.
Gender in Algonkian languages is designated "animate" or "inanimate" by linguists and correlates roughly with organic and inorganic referents. Ethnographers--among them A. I. Hallowell (1992) and, most recently, R. Brightman (1993)--report that animate nouns refer to beings perceived as imbued with power, and inanimate nouns to objects considered lacking power. Human perception, not physical science criteria, assigns gender, resulting in some inorganic phenomena taking the animate case (Corbett 1991:12, 20-24); personal experience of a transforming manifestation can idiosyncratically determine gender choice for particular referents.
Inuit and many other northern First Nations see infants as reincarnations of other people and sometimes of spirit beings (Mills and Slobodin 1994). The incarnated soul must be recognized by giving the infant its name, and an infant may incarnate several souls, requiring it to carry several names. Rasmussen mentioned a woman with 16 names, "as a bag round her" (quoted by Saladin d'Anglure 1994:90). A soul may incarnate in a body with genitals different from those of its former body, and a child may simultaneously incarnate souls that formerly took male and female bodies, inducing shifting social personae and dress in the living person. Social personae were normatively differentiated into men and women, but to survive in the North, all adults needed familiarity with the range of skills only ideally categorized as "men's work" or "women's work" (Kehoe 1991:431-433).
Quite recently, the label "two-spirit" caught on with many American Indians who are openly not, or not exclusively, heterosexual. The label is not "traditional," and even if it were for some nations, it could not possibly be traditional for all the hundreds of American First Nations. Furthermore, why "two-"? Rasmussen's Inuit acquaintance had 16 souls. At the conferences that produced the book, Two-Spirited People, I heard several First Nations people describe themselves as very much unitary, neither "male" nor "female," much less a pair in one body. Nor did they report an assumption of duality within one body as a common concept within reservation communities; rather, people confided dismay at the Western proclivity for dichotomies. Outside Indo-European-speaking societies, "gender" would not be relevant to the social personae glosses "men" and "women," and "third gender" likely would be meaningless. The unsavory word "berdache" certainly ought to be ditched (Jacobs et al. 1997:3-5), but the urban American neologism "two-spirit" can be misleading.
Where does that leave the honestly concerned archaeologist? If one is working with historic or protohistoric material for which an ethnic affiliation can be well supported, the simple answer is to ask native speakers, or look into dictionaries, for the people's own words for the social personae "man," "woman," and whatever additional personae may be recognized. Terms without easy Western equivalents can be glossed with descriptions. For archaeological material too old for reasonably strong affiliation, adult males and adult females can be identified from skeletons but not by bald assumptions about associated artifact significance. In fortunate projects where artifacts lie with, and co-vary with, biological evidence of sex, the inference may be drawn that the social personae we gloss as "men" and "women" are thus marked. Without skeletal or DNA coefficients, patterned assemblages may be postulated to signify "man," "woman," or another social category.
Although we are probably stuck with the ethnocentric Indo-European conflation of gender and "sex" (marking social personae), let us not compound the confusion by inscribing non-Western social personae under the Indo-European tag "third gender." Thaumaturges, spiritual adepts, clowns, individuals substituting for the lack of a more appropriate kinsperson in a prescribed role (such as the Zuni La'mana We'wha), and individuals uncomfortable in the persona usually ascribed to an adult with their genitalia, may (or may not!) be marked by exceptional artifact associations. One catchword obfuscates the diversity created in human societies and the fluidity conceptualized in many American Indian cultures. Thomas (1997:171, n. 8) explains that the Navajo word nádleeh "means constant state of change...and nádleehí means one who is in a constant state of change."
And while we're at it, although it is a separate issue, those hundreds of sovereignties that Europeans invaded, fought, made treaties with, and settled among in the Americas were America's First Nations. Hundreds of First Nations remain on the continent, encapsulated by the Anglo and Hispanic dominant states. The word "nation" means "born to." The First Nations of Canada advocate this usage. It is respectful and unambiguous.