The concept of discourse analysis was originally developed by linguists seeking to take linguistic analysis beyond the groundlevel building blocks of language: phonemes, morphemes, words and syntax. Hence, the continued centrality of one definition of discourse analysis the study of language beyond the sentence. However, research in this tradition has still focused on language as a self-contained system. An alternative approach to discourse analysis views discourse as language in the world language as it functions in potentially all aspects of human life. This approach emanates from the various disciplines studying humans scientifically linguistics, sociology, anthropology, psychology, communication and so on. Sociological studies of conversational interaction, for example, seek to discover principles of social organization therein, while psychologists examine (among other things) how language is processed by the brain. It must be added that this approach to discourse analysis frequently blends elements from different disciplines, because complex human realities like discourse do not divide neatly along disciplinary lines. Two more recent conceptualizations of discourse and its analysis are critical discourse analysis and what might be termed Foucauldian discourse analysis. First, critical discourse analysis emanates from neo-Marxist understandings of social inequality and how language functions to maintain and foster such inequality. Second, the postmodernist concept of discourses (Foucault 1972) examines how language works together with other social practices to naturalize perspectives on human beings which have the effect of defining and controlling them for example, conceptualizations of the body in Western medicine. These two approaches to discourse analysis have become popular in recent years in applied linguistics and are sometimes combined (e.g. Fairclough 2003). Discourse Analysis and Ethnography: Complementary or in Opposition? Ethnographers and discourse analysts have sometimes debated whether their respective approaches are complementary or
oppositional. Here we focus on the single most significant
disagreement in this area between ethnographers and conversation analysts, the latter representing a highly influential approach to analysing spoken interaction. Historically, ethnographers did not collect verbatim spoken interaction data from the people they studied; doing so was virtually impossible before the advent of the portable tape recorder. Where ethnographers focused on language at all, they almost always focused on ritualized and monologic forms. By the 1970s, however, ethnographic studies were being conducted which included spoken interaction; these were first undertaken in the ethnography of communication framework (Hymes 1964), which sought to study the particular linguistic practices of sociocultural groups. A related innovation was microethnography, first developed as a methodological option in classroom ethnography (Erickson 1992). Applying these approaches, ethnographers carried out major studies, including some which traced at-risk students educational difficulties to their culturally based verbal interaction styles vis-a-vis tacit middle-class norms (e.g. Heath 1983). Starting in the 1960s, a group of sociologists led by Harvey Sacks (e.g. Sacks et al. 1974) developed conversation analysis (CA) (see Wilkinson and Kitzinger this volume), seeking to discover principles of social organization within momentto- moment social interaction rather than via externally imposed, top-down concepts like culture and social class. They based their findings on the painstaking analysis of detailed transcripts of conversations, and later other kinds of interaction. A central tenet of CA is that the emic structure of talk can only be determined from within the linguistic context of interaction, as reflected in interlocutors own responses to talk. That is, recourse to transcript- extrinsic (Nelson 1994) information of the sort traditional ethnography gathers for example, information not demonstrably relevant to participants in particular interactions is ruled analytically out of court. More specifically, conversation analysts and their allies critiqued ethnographic studies because they: (1) depended on a priori categories and assumed contextual influences such as cultural norms, socio-institutional identities (e.g. doctor, female, working
class) and local factors (e.g. past relationships among individuals)
to explain social behaviour, instead of basing their explanations directly on interactional data. In this view, CA portrayed social behaviour as dynamic, emergent and situated vis-a-vis the interactional contingencies of the moment, versus static ethnographic accounts; and (2) were based on questionable evidence, such as unsystematic, retrospective accounts of ethnographic observations and interviews of research participants regarding social practices which, albeit their own, they could not adequately explain because such practices were tacit and unreflective that is, just the way things are (e.g. Maynard 1989; Schegloff 1992). Ethnographers responded in various ways. First, they countered that conversational transcripts provide only partial information regarding the identities, social relationships, and contextual background needed to understand social behaviour exactly the kind of information ethnography excels in collecting. Second, they argued that CAs emphasis on interactional structure led to arid accounts of social behaviour, wherein form was privileged at the expense of meaning. Third, they suggested that the long-term nature of ethnographic studies yielded knowledge of regularities in social behaviour which conversation analysts, who tended to focus on single, momentary interactions, had no special access to (e.g. Cicourel 1992; Duranti 1997; Moerman 1988). This debate has been partly resolved by the fact that there is now a substantial history of combining these approaches in highly effective ways (e.g. Goodwin 1990; Moerman 1988).1 In many senses the two approaches are highly complementary: Each is strong where the other is weak. First, regarding what CA can contribute to ethnography, fine interactional detail provides valuable material for sociocultural analysis, material which can complement data gathered through, for example, observations and interviews because interaction is central to the organization of culture as well as social organization (Goodwin 1990: 1). Likewise, the ethnographic problem of attaining emicity is partly addressed by CAs commitment to studying participants own orientations to the interactive behaviours of their interlocutors. Such evidence can be used to test ethnographic interpretations of
what is going on in the social lives of those being studied, since
social life fundamentally involves interactive coordination. Second, regarding what ethnography can contribute to CA, rich, longitudinal descriptions of social life and language use among particular groups can flesh out fine-grained analysis of momentby-moment verbal interaction. The same is true for more immediate contextual details, such as the pre-existing personal and social relationships between interlocutors, or the larger activities engaged in while talk is proceeding. Theoretical concepts such as social class, power and culture, properly used, can also help analysts understand the complex sociocultural realities being studied. To sum up the convergent possibilities of ethnography and CA in particular, and ethnographic and discourse analysis in general, close description of the moment-by-moment constitution of social life in talk-ininteraction can both fundamentally enrich and be fundamentally enriched by broad descriptions of social behaviours, norms and values. From this perspective, incorporating discourse analysis and ethnography. Critical Ethnography Critical ethnography is a form of ethnography with antecedents in neo-Marxist critical theory (May 1997). Its primary objective is to unveil the unequal distribution of power in society, and to change it for the better (Kincheloe and McLaren 2000; Talmy 2010). Critical ethnography thus differs from the previous two approaches in its broader focus and direct critical and emancipatory intent, although, as already noted, all three approaches have historically concerned themselves with social inequality. As with ethnography in general, critical ethnography depends on long-term, intensive, emically oriented analysis of particular social situations. However, it differs somewhat in its immediate focus: A critically-located ethnographic methodology highlights the interplay between social structure, material relations, and agency; addresses the ways that social structure is (or is not) instantiated, accommodated, resisted, and/or transformed in the micropolitics of everyday life; contends with issues of ideology, hegemony, and culture; critically addresses its own historically-, materially-, and culturally-specific interpretations; works toward change; and does
so with the collaboration of research participants. (Talmy 2010:
130) As with other ethnographic approaches, discourse analysis also adds substance and rigor to critical ethnography. The most common approach to discourse analysis within a critical ethnographic framework is critical discourse analysis (CDA) (see Wodak, this volume). Like critical ethnography, CDA is directly concerned with exposing inequality and injustice, in this case through analysing language as a means of naturalizing unequal social structures and relations (Bhatia et al. 2008). CDA enables researchers to generate, warrant, and elaborate (critical) claims in demonstrable and data-near terms (Talmy 2010: 131). In the following section, we provide an extended example of critical ethnography combined not with CDA per se but rather CAoriented discourse analysis. This example concretely illustrates: (1) the specific nature of discourse oriented critical ethnography; (2) the combined use of discourse analysis and ethnography in general the main topic of this chapter; and (3) the basically open-ended nature of ethnographic practice.