A growing literature on the relationship between the ethnographer and his or her subject of study goes beyond a confessional first-person stance that focuses primarily on the author, and directs attention to the construction of ethnography in its double spheres of the research process and creating the ethnographic text. This literature contributes to a general movement toward a more hermeneutical approach to ethnography that involves a radical shift of ethnographic knowledge from observation of reality (as transparent, empirical, and objectified) to understanding arrived at through mutual experience in interaction with the ethnographic other. Such knowledge is characterized by reflexivity and acknowledgement of prejudice (Gadamer 1975), so that explicit examination of one's bias and preunderstandings is a basic, positive step in a dialectical movement between interpreter and interpreted (Marcus and Cushman 1982: 38). In this view, ethnographic knowledge is constructed (and reconstructed) in an ongoing, reiterative manner (Dwyer 1979), so that knowledge of both the other and the self changes markedly over time (Crapanzano 1977). A significant shift in the construction of ethnographic texts incorporates these dimensions of ethnographic knowledge. In a number of experimental works, the ethnographer is introduced into the text that now is viewed as a by-product of relationships established with a series of people who actually do more than inform. Their personal interaction creates a ground, or a tie, within which cultural knowledge is defined and understood (e.g., Bohannan [Bowen] 1954; Briggs 1965; Read 1965; Rabinow 1977; Riesman 1977; and Dumont 1978). I refer here not to confessional narratives about the experience of fieldwork, which Dumont (1978) points out are strikingly similar, but to narratives which take the relationship between first-person research and the organization of ethnographic reporting. Yet problems remain with these experiments. Marcus (1980: 508) questions their relation to conventional ethnographies, and the relation between the first-person narrative (still regarded as subjective, soft and unscientific (Clifford 1983: 132) and the third-person narratives by the same authors (objective, hard, and scientific) is unclear, even contradictory. Ethnographic knowledge depicted in first-person narratives calls into question the third-person construction of certain and objective knowledge about others.2 New perspectives on ethnographic knowledge also lead to questioning the authority of the ethnographer to construct a discrete text from an experience permeated with ambiguity, ambivalence, and interpretation (Clifford 1983: 120). Finally, experiments in opening up the text to include more native voices to disperse authority are criticized by Clifford (1983), Tyler (1982) and Marcus and Cushman (1982). Most problematic for them is the fact that dialogues remain representations of dialogue, since they are texts, not dialogues, and retain the authority of the ethnographer as author (Clifford 1983: 134). Dialogue rendered as text...is no longer dialogue, but a text masquerading as a dialogue, a mere monologue about a dialogue since the informant's appearances in the dialogue are at best mediated through the ethnographer's dominant authorial role (Tyler 1982: 3, in