Ethnography in ethnomethodology and conversation analysis: Both,neither,or something else altogether?

Abstract

This article focuses on various ethnographic procedures and findings in ethnomethodology (EM) and conversation analysis (CA), addressing the question of how EM and CA relate to ethnography. Given the obvious answer that EM includes ethnography, we also argue that CA does as well, though just how EM and CA do so needs to be qualified and specified. Ethnographic procedures have been used in EM for decades, although often in non-standard ways, and currently with some ambivalence. In CA, it is more common to disavow ethnography in favor of recorded and transcribed interactional exchanges. However, we argue that CA often makes use of ethnographic insights drawn from extended study of recordings, while also identifying “ethnographic” inquiries of a sort that take place within the organizational settings studied. Our aim is to identify the place of ethnography within EMCA by taking an inventory of ways “ethnography” has been used, invoked, produced, and/or disavowed in particular studies and to highlight what is distinctive about those various EMCA uses of ethnography in contrast with more conventional ethnography.

Introduction: Ethnography and ethnomethodology

The question we address in this paper is “How do ethnomethodology and conversation analysis (EMCA) stand with respect to ethnography?” The question is complicated by the fact that conventionally, ethnography tends to be loosely understood as a qualitative research method for describing social, organizational, and/or cultural activities in a way that takes account of the typical member’s point of view. However, just how ethnography is systematic, and just how the “member’s point of view” is taken into account, differs widely across various lines of research in anthropology, linguistics, and sociology. Variations include single-sited versus multi-sited ethnography; participant versus non-participant observation; grounded theory methodology; ethnography of communication; reflexive ethnography; and auto-ethnography, to name just some among many (Clifford and Marcus, 1986; Glaser and Strauss, 1968; Goldschmidt, 1977; Hymes, 1964; Marcus, 1995). ¹ Some approaches emphasize abstract “codes,” “concepts,” “taxonomies,” “sensitizing concepts,” and/or “patterns” that are said to define members’ activities, identities, and cultural objects, while others (including ethnomethodologists) consider codes and coding inherently problematic. The latter object that to elucidate abstract organizational objects for coding, the researcher must reduce large amounts of data to a relatively small number of categories or themes—losing essential information about how the interaction was organized in situ in the process. Approaches to ethnography that seek to elucidate in situ sequential, collaborative, and reflexive features of interaction, including conversation analysis and ethnomethodology, focus on the details of interaction, and/or discursive narratives to reveal local orders of activity.

In this article, we aim to set out, and briefly characterize, the place of ethnography and its various relevancies in ethnomethodology and conversation analysis. Despite the similarity of the words “ethnomethodology” and “ethnography,” they do not mean the same thing and should not be confused with each other. The “ethno” prefix they share is usually translated as “folk” or “people,” but ethnography is often said to be a method (or a family of methods), while ethnomethodology is not a method. It has become the name of a field of research, but more importantly it is a name for the phenomenon the research investigates, rather than the method of investigation. Harold Garfinkel, who coined the term and founded the field, drew an analogy between ethnomethodology and the various ethnosciences (ethnobotany, ethnomedicine, ethnomathematics, etc.) and ethno-arts (e.g., ethnomusicology), which investigate “folk” classifications and practices with plants, medicinal substances, numerical and geometrical reckoning, and artistic practices (Garfinkel, 1974). ² However, unlike these specific ethnosciences, ethnomethodology encompasses a broad array of specialized and commonplace practices, and investigates such practices without presuming that modern academic conceptions of science, mathematics, medicine and the arts provide templates for understanding their “folk,” “native,” “ordinary,” “vulgar,” or “naïve” counterparts. ³

Some ethnomethodological studies are self-described as ethnographies, while others disavow or express ambivalence toward ethnography. As we elaborate in this article, however, even in cases in which ethnography is downplayed in favor of other approaches, ethnographic understanding can often be said to be part of, or relevant to, ethnomethodological and conversation analytic research. Some, especially early, ethnomethodological studies were recognizable ethnographies, in the sense that the researchers who conducted the studies spent extended periods of time in one, or a very few, organizational settings, observing daily activities and to an extent participating in them. Sudnow (1967), for instance, compared two hospitals in different regions of the USA, while Bittner (1967) closely examined the practices of city police in a “skid row” area of a large city, and Wieder (1974) conducted an ethnography of a “half-way house” (a low security carceral institution). In addition to being focused on a place or organizational environment, each of these studies examined a theme: Bittner examined the practices through which the police used their powers to “keep order,” thereby addressing a basic sociological theme of order production; Wieder examined the way inmates deployed the specifications of the “convict code” (consisting of maxims such as “Do your own time”; “Don’t be a snitch”), thereby addressing the theme of how abstract rules or codes are deployed on specific occasions (also see Zimmerman, 1971); and Sudnow examined how staff at two hospitals managed the death and dying of patients. Viewed retrospectively, these studies used what Garfinkel (2002, 2022) later called “analytical ethnography” to elucidate and explicate “perspicuous phenomena” (which we discuss below). Ethnographies of this sort are not limited to the early days of ethnomethodology. More recent studies by Sormani (2014), Vertesi (2015), Eisenmann (2022), and many others continue using ethnography to examine detailed actions that constitute the production of practices and objects in distinctively organized settings.

In addition to performing such ethnographies, Garfinkel and others in EM and CA have also expressed ambivalence about ethnography. This is most evident in CA, where the use of video and audio recordings as data for investigating “naturally occurring” social interaction is often contrasted with ethnographers’ uses of field notes, recollected or imagined examples, interviews, and the interpretation of texts. Nevertheless, as we shall argue, Harvey Sacks, the founder of CA was an avid reader of ethnographies, and conversation analysts sometimes treat “ethnographic” understanding as a background resource and phenomenon of interest in their analysis.

But, how that ethnographic work is performed matters. Ethnomethodological studies of work differ from more conventional ethnographies both in their insistence on what Garfinkel (2002) called “the unique adequacy requirement of methods”: an injunction that traditional ethnographers—even those engaged in participant observation research—would likely associate with the extreme of “going native” that they warn their students to avoid; they also differ in their focus on sequential/organizational detail rather than concepts/meaning. Unique adequacy is not the same as going native, because EMCA research does not take an individual point of view, thus retaining its analytic edge, while nevertheless requiring mastery of the culture/practice in question. Additionally, we argue that because EMCA locates (or even dissolves) meaning in sequential organization, the focus on the production and performance of detail rather than on the presumed efficacy of abstracted organizational concepts is, ironically, more intensely focused on the production of what the word “meaning” glosses than on an hypostatized “making” of meaning. ⁴

Unlike many (perhaps all) other approaches to ethnography, EMCA ethnographies involve closely detailed analysis of orderly properties of embodied action and social interaction, including the timing and sequencing of continuous streams of (often recorded) “naturally occurring” and “naturally organized” conduct in medical, legal, artistic, navigational, and ordinary settings. Especially crucial for such analysis is to recover the “point of view” or “orientation” of the “actors” (or “members”) to the actions they produce as they are produced. Interaction is treated as an assembly process in which how the pieces are assembled and in what order and by whom determines the recognizable/meaningful social character of interaction.

Thus, while ethnography continues to have an important place in both CA and EM, it is a varied and unusual place that merits close examination. A substantial amount of work in ethnomethodology ⁵ is ethnographic in the familiar sense of being based on personal observation of activities, requiring extended time in relevant social-organizational circumstances that involve formal and informal interaction with particular informants. Other uses of ethnography in ethnomethodology are quite different in how they are done, why they are done, and in the outcomes produced.

The aim of the present paper is to point out some of the things that distinguish ethnomethodological ethnography from other forms of ethnography, and then sketch out how ethnography continues to be featured in EM and CA in their particular ways. We open with a broad discussion of features that distinguish ethnomethodology and ethnography in Section II, followed in Section II by more detailed discussions of five ways in which ethnography remains relevant in EM and CA. In doing so, we focus on the distinctive ways that ethnomethodologists and conversation analysts can, and do, respecify the place of ethnography in, of, and as, the production of social activities.

Distinguishing features of ethnomethodological ethnography

One general feature of traditional ethnography that distinguishes it from ethnomethodology is that ethnographers typically begin with the idea that a culturally coherent group is organized around significant symbols and conceptual oppositions that are somewhat durable and can be abstracted from the particular actions they represent, or serve to “govern.” Accordingly, ethnographers look for patterns in the way symbols are represented and used, the way “rituals” are conducted, and data are often organized into taxonomies and “maps” of a “culture.” ⁶

Another difference is that many ethnographers treat “social facts” as deep patterns of normative behavior that are laid down over time and hence somewhat durable. Some ethnographers who do this look for such patterns in aggregated (coded) data sets, sometimes using software tools to exhibit and define normative patterns, while others focus on the details of ritual situations, attempting to parse out their symbolic significance and cultural functions. Such ritual situations tend to be restricted to particular occasions that highlight durable symbolic systems and time-worn normative patterns. Consequently, this approach treats a slice of social life as standing for the organization of the whole.

Ethnomethodology by contrast treats cultural categories/symbols as fragile achievements and looks for how they are practically and interactionally accomplished and used on actual occasions. The difference between the approaches of Garfinkel and Geertz to describing and analyzing social practices illustrates the point (see Rawls and Turowetz, 2021). Geertz and others who take the durable approach, treat ritual practices as symbolic of established cultural meanings. For EMCA, on the other hand, interactional practices are used to achieve meaning on the spot—and can do so independently of durable cultural/symbolic meanings—which is the significance of Garfinkel’s focus on indexicality and reflexivity. What things are/mean is a constantly changing cooperatively achieved affair.

One of the distinguishing features of ethnography as done in EMCA is that the researcher searches for relentlessly performed varieties of language/culture games that are only glossed by the rules, standards, and symbolic meanings that other researchers tend to focus on. The ethnomethodologist treats these as making relevant what, after being achieved in interaction, come to be seen as the normative patterns in interaction that other researchers focus on. In other words, what the conventional ethnographer takes as the raw data, the EM ethnographer treats as achieved social objects, whose assembly is evident and accountable to members—and can therefore be made evident and accountable to research. Because of this, ethnomethodologists who undertake ethnographies tend to look for different phenomena than other ethnographers.

First, they often look for settings of action that make problematic or “perspicuous” (Garfinkel, 2002: 181–182) how members locally constitute assemblages of categories, distinctions, and identities. Such settings are important because the work of assembling social objects is taken-for-granted, that is, it is a hidden background to focal relevancies, until contingencies (often associated with problems or troubles) elucidate their relevance.

This ethnomethodological approach contrasts with a more typical ethnographic investigative procedure that treats durable symbolic meanings as transcendental templates people can “somehow” use for the reproduction of patterns of behavior without having to achieve their meaning in situ. The presumption in ethnomethodology is that members collectively produce order and meaning on the spot, moment-to-moment, by orienting the relevancies of rules and expectancies in ways that can be empirically documented. The objective is not to find an abstract norm indicated by behavioral signs or proxies, but rather to locate the order in the scenes that people are actually producing (and orienting toward) in structured interactions for “each next first time.” In other words, the ethnomethodological approach treats actual interaction as displaying an ordered character that is constitutive of meaning that abstract norms can only approximate (and typically fail to even approximate), a treatment that is the reverse of more conventional approaches.

Second, because the presumption is that people are actually using something like language games, or culture games (Garfinkel, 2019[1960]) to assemble social objects, it follows that troubles can occur—and that when they do occur, noticeably and accountably as troubles—the researcher will be able to discern what the problem was by looking at how participants orient to that trouble explicitly in efforts to repair or circumvent it on the spot. While other ethnographers address problems, they tend to treat them as indications/violations of general norms and ritual taboos, rather than as evidence of the situated devices being used by members to organize and repair sequences of action. For ethnomethodology, the inquiry is more focused on the situation: What rule were people invoking that got violated? Why was the action in question treated as a problem? What counts as a violation? How is such a violation evident? Strange as it may seem, was the relevance of a rule a consequence of its violation (Garfinkel, 1967: 75)? What “work” was being done by treating it as a violation? And, finally, why does it matter?

The ethnomethodological interest in trouble—and the marginalized

This orientation toward the in situ accomplishment of meaning/order means that ethnomethodologists are often interested in practical trouble: things that go wrong and how members deal with such trouble, as ways of making taken-for-granted organizations of interaction perspicuous (i.e., accountable, relevant to practicing members, and demonstrable in descriptions of practices). How members handle and/or repair trouble is the key to how order and meaning are being achieved in the absence of trouble.

This means that, as in Garfinkel’s (1967, Ch. 5) study of Agnes (who at the time was called an “intersexed person” undergoing gender-reassignment surgery), ethnomethodologists are often interested in persons who contend with the possibility of marginal, stigmatized, and/or “deviant” identities/categorizations, because their experiences of trying to fit into the world and achieve “normal ordinary” identities give them a heightened awareness of taken-for-granted ways of living. Troubled and marginalized situations of action thus provide especially pertinent ethnographic sites, both in Garfinkel’s early research on “deviant” and “racial” identities (1940, 1942, 2012[1947], 1949) and “degradation ceremonies” (1956), and in later research on “perspicuous settings” (2002). Because “trouble” and persons who experience trouble frequently, as a result of how they are categorized in societies, are the key to how ordinary society works, ethnomethodologists have produced significant research that bears on questions of how racism, sexism (Rawls and Duck, 2020) and other forms of exclusion and marginalization are accomplished in social interaction. ⁷

Because ethnomethodologists focus on the situated order properties of interaction and the rules that gloss them, when they study a site ethnographically, they look for different things than what most ethnographers look for. For instance, Duck (2015), in studying a drug-dealing neighborhood, describes the way people make order in that space; focusing on how and what people do there is different from places that are not so severely marginalized; locating order making activities that are missed by ethnographers who pursue what their presuppositions tell them orderly cultural patterns should look like. Duck inverts the usual practice by looking for order where others find disorder. EM studies of basketball (Jimerson, 1999; Macbeth, 2012), policing (Bittner, 1970; Meehan, 1986; Meehan and Ponder, 2002), and extreme camping (Burke et al., 2008), to name a few, take a similar interest in how the activities studied are organized moment-by-moment by their practitioners. Such studies focus on situated sequential order, rather than on symbols, rituals, or taboos—often using trouble as an entry point.

Making social facts is not a morally neutral enterprise

There is a further point of difference between ethnomethodology and conventional ethnography. Garfinkel’s early studies of the social categories “Red” “Negro” “Criminal” and “Jew” in 1947–1948, positioned his research against the popular argument that social science should be neutral with regard to morality. Garfinkel maintained that the making of social facts—social categories—is not a morally neutral enterprise and that as a consequence the “objectivity” of such facts resides in moral and emotional features of their achievement. ⁸ While in social science, moral issues are typically assumed to complicate “the objective clarity of the symbol’s meaning,” Garfinkel (2012 [1947]: 21) argued that in actual use, the “Red” (i.e., Communist) symbol, like other ideal symbols, gets its objectivity from “the maximum heaping up of these emotional factors” (ibid.). These emotional and moral factors are interactional achievements. To attempt to remove their moral character in the interest of “objectivity” would in its own way produce an “unscientific” divestment of the social phenomenon’s evident, actual, and consequential features.

As Whitehead (2021) notes, Sacks’ analyses of membership categories in the 1960s also demonstrated moral characteristics of categories and how they can be mobilized to tacitly account for a referred-to person’s actions (as provided-for by their membership in the category in question). This, he says, constitutes a mechanism through which the common-sense knowledge associated with racial categories can be reproduced as a “by-product” of whatever the participants are doing, rather than being the “design target” of their activities—thus burying the moral implications of category use. Sacks thus offers an early example of how studies of category use can provide insights into such features as implicit, tacit, and taken-for-granted Race-based privilege, racialized ways of seeing, and associated obstacles to the accomplishment of empathy across racial lines—that can be grounded in the details of participants’ orientations and conduct in everyday interactions.

To note another relationship between ethnomethodology and ethnography, research on the moral loading of certain categories is being taken up by ethnographers who are not ethnomethodologists. These researchers have been documenting how particular categories of person are being oriented toward interactionally in ways that produce forms of exclusion that deny them membership in situations. Garcia (2017), for instance, argues that the category “illegal” has been “racialized” in its use, and others have made similar arguments with regard to religion (Husain, 2017; Mohammadpour, 2022) and immigration (Castañeda, 2018). These ethnographers are examining the ways in which certain categories are being used and felt—their moral loading as Garfinkel might say—in an effort to document how racism and racializing processes have become embedded in categories other than Race in ways that enable people to engage in racial exclusion while denying that racism is occurring. Beaman (2017) refers to this as a racial grammar in the Wittgensteinian sense.

While such studies look at categorization in ways that are similar to Garfinkel’s early arguments about the moral loading involved in the use of social categories, an ethnomethodological study of “racialization” would focus more on how the process works in an actual interaction—which is important for revealing tacit aspects of how such category use excludes members from participation—but the overall point would be similar. ⁹

Ethnographers and ethnomethodologists have also both studied accounts and narratives. Here again, the difference is in the focus (or lack thereof) on the order cooperatively produced in an accountable situation. When Garfinkel did his ethnographic study of courtrooms from 1940 to 1942, he focused on “accounts” involving Race categories that were produced in court (see Garfinkel, 1949). Garfinkel’s question was: What would the people in that place accept as an account involving Race? He found that the accounts for Black and White behavior that were expected and acceptable in court were quite different as was their legal relevance. In what he called the “White court,” it was acceptable to say things about Black defendants that it was not acceptable to say about White defendants and vice versa.

In his early short story of segregated seating on a bus in Virginia, Garfinkel (1940) was interested in which presuppositions about Race were being used to make order and sense of the Civil Rights protest taking place on that bus. He noted that conflicting sets of presuppositions (characteristic of the North versus the South) were interfering with sensemaking in real time. He also pointed out various assembly practices that were particular to the bus itself as a situation—some of which related to how the driver and passengers were accountable for their behavior. In that early work, Garfinkel had not yet figured out how to make the presuppositions he was interested in visible, so he represented them as mental dialogue. By 1948, he had figured out that he could get access to social and sequential presuppositions by studying interactions that go wrong (as that one did) and then study recordings of those interactions in detail and talk to the participants about the trouble they experienced. By 1957, Garfinkel (1957) was referring to categories of people (like Agnes) who experienced frequent troubles of this kind (and as a consequence could talk about them) as “natural experiments” in their own persons.

All ethnographic observation is selective. No one can record everything. Even having comprehensive video does not solve this problem; selections still need to be made. What ethnomethodology has the ethnographer doing that is different from most ethnography is not merely collecting more detailed data, but rather, searching the data for the rules/expectancies—of what Garfinkel called language/culture games (2019[1960])—the assembly practices of interaction—instead of looking at symbols, beliefs, rituals, or for normative patterns.

It makes a big difference.

The ethnomethodologist as an ethnographer typically comes up with descriptions of rules/expectancies to which participants orient when making sense, in a more detailed way (and with a different sense of detail) than the more traditional ethnographer. Another big difference is that most ethnographers treat symbol systems and social objects as durable—as “epistemic objects.” For the ethnomethodologist, by contrast, meanings, social objects, and identities do not exist unless and until they are mutually and cooperatively constituted in and through interaction on a particular occasion. Participants may anticipate them, but until they actually achieve these objects together, they might as well not exist. Ethnomethodology focuses on how members do this, rather than trying to set the achievements into durable symbolic, ritual, or structural schemas as most other ethnographers do.

Because of this focus on the detailed cooperative sequential production of social objects and identities, ethnomethodologists also take into account a strong time dimension (Rawls, 2005). The meaning of anything at Time 1 may be very different from the meaning of the “same” thing at Time 2. This is one of the reasons why the idea of durable objects and durable knowledge that holds across sequences and situations is treated as a problem by EM and CA. Objects and meanings may sometimes be the “same” across time—but this is not something that either the participants in interaction or EMCA researchers can take for granted. ¹⁰

This presumption of the fragility and changeability of objects and identities is one of the principal things distinguishing the kind of analysis the ethnomethodologist does from other approaches to ethnography. That objects can change in a second is the heart of the matter and it makes a big difference. Fragility and changeability relocate the emphasis on conformity, generalizability, and analytics that is associated with approaches to durable epistemic objects, symbols, and/or cultural knowledge, to the mutual cooperation and conditions necessary for cooperation (Trust Conditions in Garfinkel, 1963) that the fragile achievement of social objects in situ requires. That objects can remain substantially the same much of the time does not change this requirement—the point is that stability is not inherent in the objects—therefore, it cannot be the basic mechanism people are using to assemble shared objects together.

Displaced and ambivalent uses of ethnography in EM and CA

Leading proponents of both EM and CA have, at times, distanced themselves from ethnography, as indicated by the title Deconstructing Ethnography (Button et al., 2015) and by some of Garfinkel’s remarks suggesting that ethnography provides only limited access to the phenomenal fields of a practice, the details of which “in any actual case” are not “available to introspection to ethnographic reportage to the analysis of ethnographic documentation or to documented arguments except, and at best, as documented conjectures” (Garfinkel, 2002: 189). In CA, it has become almost obligatory to use audio and video recordings as materials, because field notes and other reconstructed descriptions are not sufficiently detailed to permit the required sort of repeated inspection and analysis of relationships between turns and sequences of turns (Atkinson and Heritage, 1984: 3). However, as early as 1948, Garfinkel was himself relying on recordings whenever he could make them, and in a 1964 lecture, Sacks mentions that the tradition of Chicago School ethnography, which he acknowledged was out of fashion in sociology, was in his estimation “the only work worth criticizing in sociology, where criticizing is giving some dignity to something” (1992a: 27). Sacks added that, despite his fundamental differences with their way of working, he would “very much recommend” more recent work by anthropologists such as Hymes (1962) and Goodenough (1964). Ethnography also features in more recent work in EM and CA, though in ways that differ from more familiar variants in sociology and anthropology. Four ways in which ethnography features in EMCA are what we will call (1) endogenous ethnography, (2) gleaned ethnography, (3) analytic ethnography, and (4) praxiological ethnography. A final section (5) categories and categorization, considers the ethnomethodological approach to the problem of categories. The first two variants are occasional, thematic and methodological resources in CA, the third is explicitly, albeit ambivalently, produced in Garfinkel’s studies of work, and the fourth is a program that has been realized in a limited way in EM. The fifth takes up a preoccupation with categories that is endemic to ethnography and examines the ways EM and CA have approached the documentation of categories and category use.

Endogenous ethnography

“Understanding” is a key term in ethnography. It applies to the ethnographer’s efforts to grasp the meaning and practical implications of actions and linguistic expressions that are particular to a society, work group, or other organization of activities studied, but it also applies within the production and recognition (accountability in ethnomethodological terms) of activities themselves. Understanding is thus fundamental to both the ordinary production and the professional sociological explication of action/interaction. Moerman and Sacks (1987) observe that “understanding” also is endogenous to the production of sequentially organized conversation, particularly in reference to the enactment of transitions between speakers in accordance with the “one speaker speaks at a time” basic rule in ordinary conversation. ¹¹ Achieving transitions between speakers with minimal gap and overlap requires:

… the collaborative location of transition points, and the collaborative use of means for arriving at who speaks after any current speaker. These are tasks which, on the situated occasions of their solution, are tasks of understanding. And participants so interpret them. They take failing to talk when one has been selected to and another stops as evidence of failing to understand what has been said. (Moerman and Sacks, 1987: 183)

In both the more traditional form of ethnography and in the analysis of natural conversation, the analytical aim is to understand the native understandings (in conversation analysis, the participants’ “orientations”) of the concerted actions they produce and coordinate.

In addition to describing how “understanding” is endogenous to conversation, conversation analysts also have identified “ethnography” as itself endogenous to interaction as an occasional way of soliciting understanding, explicating understanding, and expressing understanding of action and interaction. For example, in his discussion of testimony at the 1992 trial of four Los Angeles police officers accused of using excessive force when apprehending and arresting Rodney King, Charles Goodwin (1994: 616) points out that an “ethnography of police practices” was among the resources the defense lawyers used to “analyze the events” on the famous videotape of police officers repeatedly beating King while he lay face-down on the pavement. Such an “ethnography” was conveyed, for example, by a police sergeant testifying as an expert witness for the defense of the accused officers, who was granted status as an expert witness by virtue of his experience with training police on the use of force. The witness explained to the court how police “perceive” and react to the actions of potentially dangerous suspects. What Goodwin called “professional vision” was presumably acquired through such training, and the witness then presented a frame-by-frame explication of the video in accordance with such “vision.” Goodwin used the term “ethnography” to characterize the way the witness presented the “point of view” of the police officers shown on the tape during the arrest of a suspect. ¹² This account deployed a contrast between how the tape appeared to millions of television viewers as an unequivocal display of disturbingly and unjustifiably violent actions by the police, versus how it was made out in expert testimony as a “reasonable” exercise of police power (a contrast that is disturbing in its own right). ¹³

Another, perhaps more technical, instance in which the word “ethnography” appears as an endogenous feature of the accountability of conversational actions is in Terasaki’s (2004) analysis of pre-announcement sequences (sequences initiated by a speaker that indicate that they have news to announce, but without saying in so many words what the news is). Terasaki (2004: 191) describes instances in which recipients of pre-announcements deploy “ethnographic” understandings. A pre-announcement, such as “Hey, we got good news!” taken in isolation does not specify the “good news” it adumbrates, but as Terasaki points out the substantive news may very well be evident to a recipient with relevant “ethnographic” knowledge of the speaker and their circumstances. What goes for participants also goes (though in a differently organized way) for overhearing analysts.

It is fundamental in CA to consider “analysis” as endogenous to the production of talk-in-interaction, and the challenge for practitioners of CA is to identify and describe how analysis is performed in situ by participants. The attempt to recover participants’ “orientations” to the conversations they produce is in line with ethnographic efforts to provide “thick descriptions” that correctly grasp the orientations internal to the production and recognition of the action (Geertz, 1973). Following Goodwin and Terasaki, we can say that “ethnography” can also be treated as an endogenous methodological practice that participants use as a resource for producing intelligible actions—and thus subjected to analysis.

In Goodwin’s case, the “ethnography” is an expert witness’s account of how a trained police officer would see the videotaped scene. The witness’s account respecifies what millions of viewers who repeatedly watched the video on their television screens took to be apparent without need of a tutorial (and, for many, the tutorial was a tendentious exercise in obfuscation rather than a lesson to be accepted at face value). The expert witness’s account not only provided details that an uninstructed viewer would fail to see, it reframed the scene in an entirely different way that, not incidentally, tended to exonerate the police officers at the scene.

In Terasaki’s case, the “ethnography” has to do with what Garfinkel (1967: 37ff.) glossed as “background expectancies,” “background understandings,” or “common understandings” that are endogenous to ordinary conversation. Garfinkel (1967: 38) demonstrated the salience of such understandings with a simple exercise that required the students who performed it to record and transcribe a short sequence of an ordinary conversation, and then in two columns present what was “actually said” (the transcript) in the left-hand column, and in the right-hand column what the parties understood was being talked about (what they “meant”). As Garfinkel elaborates, the students found it impossible in the right-hand column to write an exhaustive account for a non-participant of just what the participants meant and understood. Their frustration furnished material for Garfinkel’s demonstration that conversational organization relies upon what is not said in so many words, and in a sense cannot be said without creating trouble and confusion.

Terasaki (2004: 189) presents an excerpt from a conversation involving four parties, identified, somewhat confusingly, with the letters D, D2, R, and R2 (double-brackets mark simultaneous starts of utterances in lines above and below them, and double slash marks denote the start in overlap of the line below the current utterance):

D: How are you all?

R: Oh very- very well.

D: Good.

D: Hey we got good news.

R: I know.

  [[

R2: What's the good news.

D: Oh ya do?

  [[

D2: Ya heard it?

D: Oh good.

R: Except I don' know what a giant fullicullar lympho-blastoma is.

D: Who the hell does, exc//ept a doctor.

R: Well

R2: Mm

R: (I d'n)

D2: This is nice did you make this? ((accomplishes shift in topic))

Terasaki (2004: 189) observes that in this and other instances,

… the features of the Pre-Announcement First [‘Hey we got good news’] appear to provide few materials in themselves for locating what among the possible pieces of news they have heard is here being referred to, e.g., ‘Have you heard?’ It appears that Recipients utilize ethnographic/biographic aspects of the scene in conjunction with Pre-Announcement components ... to come to that determination.

The “pre” is incomplete in itself and its designed incompleteness requires the listener to do some work—work that in many cases is accomplished immediately and effortlessly. Other “aspects of the scene” are not scenic in the sense that a recording device would reveal them as materially evident, but neither are they “private” matters hidden in the heads of knowing participants.

Although, as Garfinkel’s exercise demonstrates, an obstinate recipient can always demand further specification, a fairly “thin” gloss can often be sufficient for revealing just what constitutes the “good news” that “we” received, and just who is included under “we.” An overhearer (including the professional conversation analyst listening to the recording or reading the transcript) also can make out what D alludes to, and R immediately recognizes, as the “good news,” but only by inferring retrospectively from R’s later mention of “a giant fullicullar lympho-blastoma” that the “good news” can be taken as “good,” perhaps in light of a successful medical treatment or a diagnosis of a type of lymphoma that is less life-threatening than initially feared. An overhearer with a more extended transcript at hand very likely would be able to make out further details, but Terasaki’s reference to ethnographic/biographic aspects is not a reference to revelations made through her own “method”; instead it is a way of speaking of the background understandings that enable R to recognize the as yet unspecified “good news,” which enables both D and R to proceed to discuss “it” without ever saying just what D was ready to announce. (The presence of R2, who apparently has not heard the news, also enables D and R to display to R2 what the news was.)

Gleaned ethnography

Although conversation analysts quite appropriately credit their treatments of recorded interaction with being more empirically precise and materially grounded than ethnographic field notes, they also rely upon a variant of ethnography derived from recordings that we might call “gleaned ethnography.” ¹⁴ By this, we mean details about the occasion, the parties, and their relationships, and many other features that are apprehended through repeated playback of a recording, and in some cases, other recordings involving one or more of the same participants. As features of a gestalt contexture, such background details may not feature directly in the analysis of thematic phenomenon such as a recurrent type of sequential organization, but they can be important for establishing a preliminary understanding of “what is going on.”

A lecture by Harvey Sacks (1992) clearly shows how “gleaned ethnography” can enter into the analysis of particular interactional sequences. Sacks discusses a series of recorded phone calls among a group of friends, and from the several calls, he derives details of their biographical situations and relationships that inform his analysis of a fragment taken from one of the calls. ¹⁵ These details come into play in Sacks’ analysis of how information can be “buried” in an announcement:

A: Hello.

B: Happy Thanksgiving from Balboa.

A: Why thank you dear.

B: Did you just get up?

A: Yeah.

B: Did you get your newspaper this morning? Mac saw it out in front of your house and put it up on your porch.

A: Oh, why thank you.

Sacks (1992, Vol II: 177) observes: “The basic thing here is that we have a technique, not for telling A how her newspaper got on her porch, but for saying, ‘My husband, whom you know left me, came home.’ And telling that in such a way as to not require that it be responded to right now.” The announcement that Sacks finds B delivering to A is not the announcement that an overhearer of the isolated sequence would be likely to apprehend. An overhearer would, perhaps, understand that B announces that “Mac” (presumably, B’s partner) performed a small favor for A, and that A responds by acknowledging the favor. The point Sacks makes, however, is that the announcement artfully conveys “buried information,” which is recognizable on just this occasion, by just this speaker, and by just that recipient. As Sacks explains it:

So the newspaper and the husband have come to be affiliated in such a way that this announcement, burying the information that B’s husband came home, but allowing for it to be picked up on, can be seen as really an elaborately designed event. So, a way was found to design a presentation of the information that the husband is back, without requiring that it be responded to, since responding could be kind of sticky and you may be willing to have the other person respond if they wish, but you may not want to require them to respond, i.e., to make it then and there a topic […]. (1992, V.I:178)

Sacks adds that by deploying such a buried announcement one can convey news in a way that gives the recipient something else to respond to, thus making the topic available without requiring the recipient to speak to it.

Consider the “ethnographic” details that Sacks mobilizes in his analysis: details about the parties’ lives that he, as an analyst, is able to use for understanding what they are talking about, and how they are talking. Presumably, from repeatedly listening to the entire phone call, as well as to other recorded phone calls among A and B, Sacks gleaned an understanding about the relevant identities of the parties, about their relationships to one another as friends and neighbors, and particularly about B’s current situation with her spouse. These details compose a background saga about B’s relationship to her husband Mac: about the fact that Mac had left her, and about Mac’s gesture with the newspaper, together with the collateral inferences that A can be relied upon to draw from B’s focal mention of Mac’s neighborly gesture.

The gleaned ethnography that Sacks deploys in this case is a distinct variant of analytic ethnography. However, it is being used to support a sequential analysis, and not to develop a normative or symbolic system that holds irrespective of sequential details. Whether or not his gleaning is uniquely adequate for discerning the action accomplished in and through the announcement in this case, the ethnography helps elucidate what would be opaque to a reading of the transcribed fragment in isolation. The “work” (ethnographic work) in this instance is not an occupational practice, but a members’ practice, involving intertwined actions, identities, and relations among a few parties, a project that establishes what the parties are doing.

Analytic ethnography

As noted earlier, analytic ethnographies take up recurrent themes and problems and locate them with specific settings of activity. For example, Lynch (1993) outlines an ethnomethodological approach to classical epistemological topics, such as observation, representation, measurement, and the fact-artifact distinction, and treating them as “epistopics”: topics to be explicated with detailed on-site investigations in laboratories and other settings of inquiry. This approach is derived from Garfinkel’s (1991, 2002, 2022) proposals to “respecify” classic methodological themes and distinctions as concrete work practices performed in “perspicuous settings.”

Early in his career, even before beginning his PhD studies, while he was in the Army Airforce from 1942 to 1946, Garfinkel developed an abiding interest in the production of work, and the use of models (or “mock-ups”) as part of a training program for aircraft mechanics (Garfinkel, 1942). Learning from his wife Arlene about her work as a lipid chemist, as well as from their network of prominent colleagues and friends who worked in such fields as computer science and neuro-psychology, and from his own, as well as his colleagues’ and students’ studies of engineers, lawyers, police officers, musicians, and other practitioners, Garfinkel cultivated a long-term interest not only in ordinary activities, but also in the shop practices and embodied work through which specialized activities are produced. Through these interests and experiences, he developed a distinctive conception of locally and organizationally adequate methods as social phenomena, rather than treating methods as the exclusive possession of a social science.

Starting in the 1970s, when he formulated an explicit program for studies of work in the sciences and professions, the central aim and theme of the studies of the natural sciences was to respecify each natural science as itself a discovering science of practical actions (Garfinkel, 2022). The idea was that natural science practices necessarily must investigate and devise singular practical actions that are adequate to the phenomena they investigate. To say that such locally organized practices are discovered is to acknowledge that they are not simply applications of textbook methods, because they involve situated improvisations with complexes of phenomena, instruments, and embodied activities. Nor is the configuration and adequacy of such methods accessible to standard sociological methods.

Garfinkel (1992[in press], 2002: 181ff.) credits Sacks with devising a procedure for investigating a general theme by finding a particular situation of practice in which it is used as a routine task. Garfinkel dubs the procedure “Sacks’ gloss” and traces it to a conversation in 1963, when Sacks told him of a procedure he devised for exploring a distinction by literally taking it to the streets. The distinction evidently was familiar to Sacks when he was a law student, and it is related to property law. It had to do with the difference between things that can be appropriated by “anybody” without legal liability (“possessitives”) versus things that are owned by somebody and cannot be legally appropriated by others without permission by the owner (“possessables”). According to the story Garfinkel tells, Sacks went on to say that he could easily investigate this distinction through research at the law library, but he did not want to do that; instead, he aimed to find a “work group” whose routine practice was to use the distinction. Speaking in Sacks’ voice, Garfinkel recounts the following:

What I want to do, I want to find a group, a work group, a group whose work somewhere in the city is of this sort: that what they’re doing is such that when I tell them what I’m looking for, they will tell me what I’m really talking about, and they will give it to me out of the ethnography that they’re capable of reciting from their own work experience. So they’ll know better than I what I’m really talking about, and I want to treat the distinction as their knowing better than I. (Garfinkel, 1992 [in press]: 15, emphasis in original)

Note that the name “ethnography” was attributed not to Sacks’ own investigation, but to the informants who would teach him what they understood his distinction to mean as a practical matter. The work group that Sacks located was a police detail whose task was to identify abandoned cars parked on streets alongside other cars that were presumed to be legally owned and registered to their owners. The distinction in such a case was highly specific to the practical work of locating such vehicles, inspecting them, and deciding if they were abandoned or not.

Garfinkel (2022; Garfinkel et al., 1988) developed “Sacks’ gloss” into a procedure for opening discussions with laboratory scientists. He used the metaphor of “coat hangers” to suggest how recurrent themes that had proven effective for eliciting stories from practitioners in previous instances would be introduced into the current discussion. He compiled a long list of these themes (“slogans” or “contingencies” as he also called them), including the following ones as well as many others (Garfinkel, 2022: 23):

“Losing the phenomenon”

“Wasting time”

“Making an experiment work”

“An issue can get settled”

“Dread of, and provisions for, demonically wild contingencies”

“Custom fitting imported methods and equipment to local, vernacular details

of shop work and shop talk”

“The local availability to ‘our shop’ of improvisational and bricolage expertise”

When presenting any of these themes for an informant’s commentary, Garfinkel would mention it and say, “Tell us what we’re talking about, and of course tell us whether we are all wet” (2022: 36). Consequently, when the stories would pour forth, as they often did, the analytical theme (“what we’re talking about”) becomes reflexively enriched, often in unanticipated ways.

A point that Garfinkel highlights, both in his account of Sacks’ gloss and in his “coat hangers” procedure, is that when requesting elaboration on the theme, he is not presenting a robust generalization but is instead using words in a way that is ignorant of what they can possibly mean to the practitioner. To state it as an oxymoron, it is an informed agnotological procedure that effectively elicits stories. ¹⁶

Having set out this procedure of analytic ethnography, Garfinkel offers a list of “dissatisfactions” which “are essentially unavoidable and essentially without remedy” (2022: 50). The dissatisfactions arise from the limits imposed by the analyst’s substantive ignorance of what the informants provide in verbal accounts: the ethnographer is unable to reproduce, or even describe in an informed way, the work done by the informants. While the account may be of interest in the social sciences or science studies, it cannot be taken seriously as a contribution to the science described. This takes us to “praxiological ethnography.”

Praxeological ethnography

In connection with studies of work, Garfinkel formulated what he called the unique adequacy requirement of methods. Superficially understood, this is not unlike more familiar requirements for anthropologists and historians to master the vernacular language and practical understandings in the cultural, organizational, or historical situations they study. However, the agenda of describing the real-time, in-course, concerted production of members’ practices that Garfinkel focused on required a deep immersion in and through a mastery of practical actions that most ethnographers do not even attempt to accomplish. Unique adequacy was more than a methodological requirement, however, as it was made relevant by a conception of social actions as being locally performed by members, moment-to-moment, in concert with other competent members. This contrasted with the conventional ethnographic conception of practical actions as manifestations of underlying social structures that could (indeed should) be described abstractly. The aim, therefore, was not to apply specialized methods, concepts, and models developed in a social science field in order to attain adequate knowledge of other fields and their practices, but instead to conduct studies that met conditions of adequacy that are endogenous to the fields studied. This requirement necessitated a kind of “hybrid” study of specialized language and practical understandings.

Garfinkel held out David Sudnow’s (1978) study of his progressive development of competent playing of improvisational jazz at the piano keyboard, as an example of such a study, and especially praised Eric Livingston’s (1986) explication of the work of following mathematical proofs, and Stacey Burns’ (1997) legally informed studies of lawyers’ work. Because of the difficulty and time involved in attaining personal competency in technically demanding fields, such studies are rare, and the attempt to produce hybrids of ethnomethodology-music, ethnomethodology-mathematics, or ethnomethodology-law have not gained much of a foothold in the academic social sciences, although workplace ethnographies have attained a tenuous place in high-tech organizations. Lucy Suchman’s (1987) analytical ethnography performed at the Xerox Corporation Palo Alto Research Center was in part inspired by Garfinkel’s account of instructed actions and became a landmark study for a development in which ethnographic research was used in conjunction with the design and implementation of computerized artifacts; a development that was later dubbed “technomethodology” (Dourish and Button, 1998). Often, such studies are performed by teams involving computer scientists, ethnomethodologists, anthropologists, and managers, so that hybrids of ethnomethodology/computer science are achieved through collaboration, rather than by any single member of such teams. The extent to which such research has realized Garfinkel’s demanding specifications of unique adequacy and hybrid studies is an open question, however (Button et al., 2015; Button, Lynch and Sharrock, 2022, Ch. 7).

One of the authors of the present paper gained insight into the benefits of such collaborative ethnography, in a study of systems engineers. Rawls listened to how the information designers formulated their problem. They talked about their databases in terms of Chomskyan models of universal grammar into which they could insert durable objects. The problems they talked about reproduced the problems with Chomsky’s model—there is no interactivity. It was a case of observing engineers trying to design a system for assembling meaning without acknowledging the situated assembly work that people ordinarily do, and then watching them fail as a consequence—and observing them talk about that failure. The analysis was facilitated by including an information designer with a grasp of EM as a research partner. In contrast with the team of design engineers they studied, ordinary people clearly can make sense in interaction. The knowledge of how they do so could be useful in solving practical problems in Information Systems Design.

The study was ethnographic in trying to show how the assumptions that systems engineers were making sometimes made it impossible for them even to talk about their objectives. Because they were making assumptions we might call “epistemic”—about the durability of objects and meaning and allowing those assumptions to shape the way they both worked and talked—they were unable to make their objects work across knowledge boundaries (Rawls and Mann, 2015).

For the observer, troubles in their interactions could be used to locate places where intelligibility broke down, and then—like Garfinkel with Agnes—to focus on what the team of information designers said was going on at those places. What they thought was going wrong turned out to be associated with which theory of categorization/knowledge they were using—which made it impossible for them to talk across this boundary. Using an EMCA approach, by contrast, the information designers working on the research team were able to show what the problems they were having looked like and why they were having them.

Finally, Garfinkel also performed and celebrated studies that were neither analytic ethnographies, nor efforts to investigate practices that are endogenous to contemporaneous sciences, professions, and other skilled activities. This set of studies involved efforts to bring under examination the embodied and equipmental work of practically building and operating Galileo’s inclined plane demonstration and pendulum “experiments” (Garfinkel, 2002, Ch. 9); and a demonstration with a prism in hand of Goethe’s visual argument for refuting Newton’s theory of color (Bjelic and Lynch, 1992). Unlike historical efforts to reproduce the apparatus used in centuries-old scientific investigations, these efforts were often casual in their “replication” efforts, aiming instead to make perspicuous (to bring under examination) aspects of the embodied and interactional work of handling materials, tools, and measurement devices to “make experiments work.” The point was not to recover what a natural philosopher like Galileo might have encountered when conducting his demonstrations (if, indeed, he actually performed them), but to elucidate how the phenomenon (e.g., the law of free fall) could be “found,” “reproduced,” and “lost” in the course of embodied work aiming to demonstrate it (Garfinkel, 2002: 264). Such studies take advantage of the relative simplicity of the materials and relations involved in classic scientific demonstrations, as compared with studies in contemporary sciences that require large budgets and years of training. Livingston (2008) provides an array of variously skilled exercises that readers can readily follow and in many cases undertake to perform.

Though focused on historical experiments and demonstrations, such studies were part of an effort to highlight a distinctive mode of praxeological investigation. Other investigations involved disruptions of the taken-for-granted embodiment of activities such as carrying on a conversation, writing on a blackboard, or playing a board game. Some disruptions were induced by wearing inverting lenses or speaking through an audio feedback side-tone delay machine that would delay a speaker’s audio feedback of their own voice as they spoke. Others involved investigations of disabilities (Goode, 1995; Garfinkel, 2002; Ch. 6; Robillard, 1999). When discussing insights drawn from such studies, Garfinkel often used gestalt themes, referencing Gurwitsch’s (1964) phenomenological treatment of them in his account of phenomenal fields (Eisenmann and Lynch, 2021; Garfinkel, 2021). Themes such as figure-ground are usually discussed in terms of visual perception and demonstrated with reversable figures such as Jastrow’s “duck-rabbit,” but Garfinkel’s disruptive exercises and accounts of troubles and provisional solutions enacted in contexts of disabled action and interaction were designed to draw attention to the spatial and temporal arrangements and embodied movements of tools and implements in labs and kitchens—in situations of work.

Categories and categorization

A standing problem for sociology has to do with assigning social categories to data. For EMCA, however, the more important issue is the prior question of how categories are actually achieved on particular occasions of use. Following Durkheim (Rawls, 2009), the idea is unless and until they are achieved in interaction, there are no categories either in the data, or that can be assigned to the data.

Sacks (1972) in his work on membership categories took up the issue as a members’ problem. Sacks repeatedly pointed out that numerous categories are available to “correctly” describe a person and that person’s action, and that the selection of relevant categories is both a routine matter in conversation/interaction and an unsolved methodological problem in sociology. Sacks also showed how the routine selection of categories enacts social and racial inequalities in social interaction. By studying how categories of person and action are co-selected in conversation, Sacks did not “solve” the methodological problem for sociology, but instead shifted it from its conventional status as a general problem to an ordinary phenomenon handled in ordinary talk, which is also amenable to description.

Garfinkel began focusing on the problem of how to characterize categories of social objects in 1947. At that point, his focus was on how ordinary people were formulating and using categories—in actual situations—and he designed his initial dissertation research to study how people oriented a specific category: “The Jew as a Social Object.” This “category” research continued with studies of the “pre-medical candidate” and later studies of Agnes—a transsexual said to be “passing” as a woman (where “passing” was a term used at the time that presupposed a difference between a person’s underlying identity and the identity they were presenting publicly).

This distinction has become increasingly problematic in recent decades in ways that Garfinkel’s pioneering research foreshadows. In all cases, Garfinkel focused on the interactional achievement of the category. What we learn from Agnes is that every one of us is performing our sex/gender identity every day. The matter, he argued, could never—for anyone—be settled by reference biologically to what they “really” were. What we need to understand is how participants in interaction orient toward such categories and achieve their relevancy, the rules they use in doing so, and how the resulting achieved (or assigned) categories figure into an overall interactional structure of categories or identities in the situation and in society.

It is also important to distinguish categories like Black and Jew, which are assigned by Others, whether or not members are interactionally aligning with those categories or not, from categories like Gender that can at least sometimes be achieved by competent members as they choose. These “assigned by Others” categories are not achieved in the same sense—or rather, they are achieved by majority (or otherwise dominant) members of the overall society (and also by majority members of the minority group) at the expense of a particular member. In other words, Others impose these categories on members who are then identified in ways they have no say over, creating further barriers to participation—which is one meaning of “racialized.” ¹⁷

The treatment of categories as relentlessly achieved in interaction—rather than as durable symbols or identities—has a particular application to contemporary studies of Information Systems. In research at MITRE on Information Systems Design (Rawls and Mann, 2015), with regard to an attempt to design a “universal” database (a few years before the initially disastrous roll out of “Obamacare” highlighted the problem), the designers kept saying that the problem was going to be “What is the What?” They were stating the EM categorization problem as a practical problem for their work (Garfinkel, 2008). They needed a “What” that would hold constant across situations and databases—an “Information Object”—that they could load their information onto. Their problem was that there is no such thing. Objects do not hold constant across different databases. What they eventually did was to synchronize the databases—which did not at all solve their initial problem—although ironically it preserved the false conception of transferrable or “universal” data objects that had led to their initial problem.

Conclusion

Ethnomethodology takes up the challenge of ethnography in order to give prominent attention to the “member’s point of view,” and the ethno-methods available to members, as a constitutive feature of social actions and more broadly a generative condition of social order, and to do so without privileging individual agency or mentality. The EMCA focus on the moment-to-moment, oriented-to, production of practical and interactional order necessitates a very particular locus of ethnographic attention. We argued that, while professional conversation analysts sometimes eschew ethnography, Sacks read and admired ethnographies and also used them with some qualifications. For Sacks and others in CA, ethnography is made relevant both as a background gleaning of contextual details that set up the analysis of focal interactional structures, and as a way to describe what parties to an occasion themselves perform, invoke, or display as grounds for their actions.

Garfinkel mentions “Sacks’ gloss” as a procedure for locating resonant theoretical themes and distinctions in mundane settings of practical work in EM. In the program he developed in studies of work, Garfinkel ambivalently used what he called “analytic ethnography” to develop such resonant themes in unanticipated ways by inviting practitioners to expound upon their circumstantial relevance. He also recommended a demanding way to conduct “uniquely adequate” explications of practices through descriptions that are grounded in a through mastery of the practices described, rather than through the detached mediation of interviews and fly-on-the-wall observation. The characterization of socially organized actions in EMCA is thus transformed from a classical “problem” for the investigation of social order to a phenomenon that members achieve through their concerted production of local orders of activity.