Volume 4, No. 2, Art. 43 – May 2003

There Is No Objective Subjectivity in the Study of Social Interaction

Tilo Weber

Abstract: The variant of ethnomethodological Conversation Analysis (CA) represented and advocated by Emanuel A. SCHEGLOFF pursues the goal of analyzing discourse in a strictly empiricist manner that excludes the impact of the researcher's subjective "preoccupations" and "presuppositions" from the investigative process.

This paper outlines the SCHEGLOFFian research strategy and characterizes it as representing a methodologist—as opposed to naïve—variant of epistemological realism. It is argued that this approach, while avoiding circularity, fails to make feasible an "account of the object itself" (SCHEGLOFF 1997a, p.174). This line of argument is illustrated by its practical consequences apropos one of CA's classical themes, viz. conversational repair (cf. SCHEGLOFF, JEFFERSON & SACKS 1977).

It is demonstrated that conversation analytic data analysis presupposes decisions concerning the selection, the preparation, and the (re)presentation of the data that influence the analytic results but that cannot be justified empirically. Accordingly, the adequacy of conversation analytic findings hinges on "the practical purposes" of the processes that yield those findings rather than its correspondence with discourse as "an internally grounded reality of its own" (SCHEGLOFF 1997a, p.171). This conclusion applies the ethnomethodological insight in the locally constructed nature of social interaction and reality to social science itself.

Key words: epistemology, methodologism, realism, ethnomethodology, conversation analysis, Emanuel A. Schegloff, conversational repair

Table of Contents

1. Introduction

2. Conversation Analysis. Empiricism Without Circularity

3. The Impact of the Analyst's Standpoint on the Analysis of Discourse

3.1 SCHEGLOFF's methodologist realism

3.2 Presuppositions inherent to the CA research process

4. Analytic Presuppositions. The Case of Conversational Repair

4.1 Preliminaries on other-initiated self-repair

4.2 The history of repair and its epistemic implications for the data analysis

4.3 No empirical analysis of repair without a preconception of its formal properties

5. Conclusion







1. Introduction

When philosophers learnt to distinguish subject from object, modern epistemology emerged in an immediate crisis of subjectivity. Scholars recognized a reality that was of a different kind than the recognizers' consciousness, knowledge, believes, etc. and they declared it the goal of science to gain knowledge of both ontological realms. It was this very distinction, however, that called into question whether human beings' epistemic means, that is, their senses and understanding, were sufficient to transcend individual subjectivity. [1]

A response to this crisis and to objections against realism was the methodologist approach to epistemology and science. Methodologists, most prominently among them René DESCARTES, are aware of the epistemic border. They concede that human perception and knowledge to some extent reflect the individual human's point of view and, thus, are relative thereto. However, they hold on to a realist, non-idealist and non-skeptic stance in believing to reach beyond the subjective and as far as the objects themselves. According to methodologists it all depends on applying an appropriate research method. Any such method admits only premises to reasoning that meticulous scrutiny has proven to be beyond reasonable1) doubt. DESCARTES was the first both to formulate and to execute this program in his Meditationes de Prima Philosophia (1993). [2]

While EINSTEINian physics and HEISENBERG's uncertainty principle are highly sophisticated attempts to cope with similar problems of relativity in the physical world, the difficulties seem to double when social processes are to be analyzed. Whether or not a particular communicative exchange can be labeled "successful," "coherent," or of a particular kind seems to hinge not only on the methods, venture points and criteria accountably applied by the analyst but also the stance that the subjects under observation adopt towards what they are involved in. It seems that when assessing and interpreting their current exchanges, discourse participants are in a position at least as legitimate as that of the outside observer and analyst. [3]

In spite and fully aware of these difficulties, methodologist approaches of various kinds have been pursued by social scientists. Among them, a particular variety of Conversation Analysis (CA) offers itself in several regards as an exemplary case to examine whether methodologism can successfully deal with the problem of subjectivity. First, CA is mainly rooted in a phenomenological tradition that has emerged as an attempt to come to terms with the epistemological problems implied by the subject-object-distinction. Second, unlike other descendants of this tradition (e.g., hermeneutics), CA in the form represented by Emanuel A. SCHEGLOFF2) claims to reconcile its phenomenological premises with what I refer to as a methodologist approach to social interaction. Third, conversation analysts themselves—again SCHEGLOFF is to be mentioned here in the first place—have repeatedly raised the issue of relativity. Doing so (e.g., SCHEGLOFF 1997a, 1997b, 1998, 2000; SACKS, JEFFERSON & SCHEGLOFF 1974), they have explicated their own epistemological position in various debates both by elaborating on their methodology in response to objections from various sides and by criticizing others for having failed to come to terms with the problems conversation analysts claim to avoid. Finally, for a functional linguist it is worth noting that it is CA that has encouraged students of language to learn from the analysis of individual tokens of discourse, to pay attention to seemingly marginal details of linguistic structure and to watch out for the impact these details might have on social interaction. [4]

Following these introductory remarks, the CA methodology, its research strategy and the rationale behind it are revisited in Section 2 mainly on the basis of SCHEGLOFF's programmatic statements. It is suggested that this strategy avoids the problem of circularity by which empiricist approaches are threatened. I characterize the SCHEGLOFFian variety of CA in Section 3 as a methodologist approach to social interaction. It is argued that this approach fails to provide the means necessary to define and to cross the subject-object border. Section 4 substantiates and illustrates this charge apropos the case of conversational repair as one of the most prominent types of discourse phenomena studied by conversation analysts (cf., among many others, SCHEGLOFF et al. 1977; SCHEGLOFF 1992, 1997, 2000; JEFFERSON 1975, 1983; SELTING 1987; EGBERT 2002, to appear). In particular, it is argued, that CA analyses, while mostly refraining from adopting as a starting point an explicit definition of repair, nonetheless are based on an implicit, but unexplicated, and not fully explicable preconception of their intended research object. Accordingly, I conclude in Section 5 that CA cannot possibly succeed in accounting for its data "in their own right." The individual researcher's perspective and venture point leave their irreducible impact on the data analyses because the history that led up to their choice cannot be fully accounted for. To put it differently: it is impossible for an analyst to cross the border from the subject (the researcher) to the object (instances of social interaction) because it is impossible to determine the exact location of this border. Furthermore, this epistemic and epistemological limitation is not contingent on CA but generally pertains to methodologism in the social sciences. [5]

2. Conversation Analysis. Empiricism Without Circularity

Since the publication of A Simplest Systematics for the Organization of Turn-Taking for Conversation (SACKS, JEFFERSON & SCHEGLOFF 1974) in Language, functionally oriented linguists have been influenced by CA mainly in the variety represented by the authors of this classical paper and, most recently, by Emanuel A. SCHEGLOFF. The central reason for this cross-disciplinary interest seems to derive from a common focus on what SCHEGLOFF calls "talk-in-interaction." Functional linguists have both felt convinced by the CA way of minute discourse analysis as well as applied their own methods and concepts to CA themes like turn-taking, conversational repair, and projection (to point out just a few recent ones, cf. the papers in SELTING & COUPER-KUHLEN 2001 and FORD, FOX & THOMPSON 2002). [6]

For the present purpose of methodological debate, SCHEGLOFF's contributions are rich sources because they extensively explicate, discuss and account for a specific variety of CA and contextualize it by relating it to other approaches including speech act theory (cf. SCHEGLOFF 1992) and critical discourse analysis (cf. SCHEGLOFF 1997, 1999a, 1999b). This section presents the essence of the SCHEGLOFFian research strategy (cf. 1997b, 2000) and points out how this strategy avoids certain problems that emerge from the empiricist way of constituting and looking at objects of inquiry. [7]

As a descendant of ethnomethodology, CA has a strong interest in the (ethno-) means and methods that the members of a given discourse community deploy in making sense to each other and of the interaction they are currently engaged in (cf. TEN HAVE 2002). It is characteristic for SCHEGLOFF's understanding of CA that it aims at the discourse participants' own displays and accounts of these means and methods as the prime kind of evidence to analyze them: "[T]he events of conversation have a sense and import to participants which are at least partially displayed in each successive contribution, and which are thereby put to some degree under interactional control." (SCHEGLOFF 1997a, p.163) [8]

According to this view, each contribution to a discourse makes perceptible, i.e., shows as opposed to describes or implies, the contributor's understanding of his/her own or another speaker's previous activity to co-participants as well as to observers.3) The analyst's task is to identify the means and manners by which this is done. [9]

If one suspends, for a moment, the issue of whether a concern for participants' accounts and displays allows the analyst to neutralize the contingencies implied by his/her own individual perspective and to gain access to discourse as the "objects in their own right" (SCHEGLOFF 1997a, p.179), another methodological question will immediately arise: accounts and displays of what objects should the researcher be attentive to? Subscribing to a radical empiricist program, the conversation analyst has to dispense with the universe of entities implied by his/her own ontology to let "the data themselves" suggest what their structure is and what—from the participants' perspective—relevant discourse phenomena worthy of closer analysis are. If, however, the conversation analyst, at the very beginning, does not know what s/he is looking for in the data, it seems vain to expect him/her to find anything. The empiricist is faced with a dilemma that may well be couched in the form of a paradoxical question: how can I know what I'm looking for before I have seen what I have found? [10]

SCHEGLOFF's way of coping with this dilemma is to define the first stage of the analytic process as an inquirer's passive and receptive self-exposure to data that "tell" the observer what to be attentive to rather than an active and intentional search. This making oneself receptive to the self-explicative data is part of a research strategy that can be summarized—mainly by quoting SCHEGLOFF himself—as a sequence of four consecutive steps:

  • "In naturalistic inquiry of the sort I am committed to, it happens that, while examining a naturally occurring event (or, rather, a record of it), one notices something [...] that presents itself as 'Oh, I've seen something like that before!'"

  • One finds earlier "cases" of this something that one thinks one has encountered before and sees "whether they hold up as relevantly similar."

  • A "[...] common next step is to assemble a collection of candidate other instances," thereby including candidates "generously."

  • One excludes "accountably" the bad examples from the collection. In this process one is forced "to make explicit just what it is which makes them different from our targets, and thereby potentially [is forced] to specify progressively just what (if anything) is distinctively going on in the fragments which set us off" (for all quotes above, s. SCHEGLOFF 1997b, pp.501-502). [11]

The way SCHEGLOFF introduces this investigative strategy emphasizes the receptive role of the inquirer at the initial stages of the research process. Accordingly, there is an unspecified discourse phenomenon that "happens" to "present itself" as noteworthy to the analyst who may be—note the "change-of-state-token" in SCHEGLOFF's formulation—even surprised by the data's claiming his/her specific attention. It is obvious, however, that SCHEGLOFF's analyst has to be well prepared prior to adopting the receptive mode by having chosen a particular event to examine and by having been ready for making audio- and/or videotapes of that event. [12]

Exposing him/herself to the data (I) the analyst notices "something," an event token of a yet undetermined type. The researcher (II) then refers back to his/her memory or, better yet, to databases looking for fragments that are somehow similar to the original one. The collection (III) is generously enlarged by adding novel fragments as candidate members of a supposed category. Collecting "generously" means that all tokens are included of which it appears to the least possible that they may turn out to be category members. Up to this stage of the inquiry, no explicit definition of the type of phenomena looked for has been used as a criterion for identifying new candidate tokens. Nor has a definition been formulated as a result of the preliminary analyses. What ties together the generous collection of candidate cases of a yet undetermined type is their unspecified likeness in the eye of the analyst. [13]

At this point, the conceptual and explicative work sets in. "Bad" cases, i.e. sequences that are found to lie outside the presumed category, are excluded from the collection on rational grounds (IV). That is, the exclusion of each single fragment has to be accounted for with reference to features that make them distinct from the members of the category. In turn, the defining traits of a category gradually emerge in the course of this process. [14]

The progressivity of the research strategy as sketched by SCHEGLOFF and a lack of a natural point for closing the generous collection of candidate cases imply an inherent openness and recursivity of conversation analytic conceptualization. Novel data may justify the exclusion of candidates thus far considered in the category and the reconsideration of fragments excluded earlier on grounds that do not hold up any more in the light of the new material. Accordingly, the first explicit but preliminary account of the emerging category at stage IV is followed by a stage III'4) at which novel candidate tokens are considered and others, potentially, are reconsidered that were excluded previously. In the light of these novel data, a reformulation of the first account may turn out to be warranted (IV'), then new material is brought to the inquirer's attention (III''), and so forth. [15]

SCHEGLOFF's research strategy was introduced in this section as an answer to the empiricist's dilemma. This answer is an effective one insofar as it avoids circularity without resulting in disorientation in the face of seemingly chaotic and unstructured data. Progressivity and recursivity of the conversation analytic research process allow the researcher to arrive at concepts as outcomes rather than inputs to empirical data analyses. Conceptualization, in this view, is based on an open collection of candidate cases whose inclusion or exclusion is to be accounted for in explicit terms. This, in turn, sheds new light on all other candidates that previously were classified in or out of a supposed category. The question, however, of whether an inquirer can possibly be as neutral and unprejudiced as SCHEGLOFF assumes it remains to be discussed. [16]

3. The Impact of the Analyst's Standpoint on the Analysis of Discourse

Does the SCHEGLOFFian research strategy achieve its proclaimed goal of "grasping" its "objects in their own right" and to account for discourse as "an internally grounded reality of its own that we can aspire to get at analytically" (SCHEGLOFF 1997a, pp.170, 171)? In order to answer this question, I will first characterize SCHEGLOFF's (and other's) variety of CA as a kind of methodologist realism in a Cartesian tradition. In this regard, it differs fundamentally from other descendants of phenomenology in general and ethnomethodology in particular. I then argue that such a research strategy forces the analyst to view his/her data in the light of preconceptions that cannot possibly be accounted for and that, therefore, leave their indeterminable mark on the results obtained by the analysis. [17]

3.1 SCHEGLOFF's methodologist realism

Within the phenomenological paradigm, epistemological realism is an uncommon position. The reason for a specifically ethnomethodological antirealism is put by TEN HAVE in the following way:

"For ethnomethodology common sense practices are the topic of study, but those practices are also, unavoidably, used as a resource for any study one may try to undertake. [...] So the problem for ethnomethodology is how common sense practices and common sense knowledge can lose their status as an unexamined 'resource,' in order to be a 'topic' for analysis. [...] This [...] problem seems to be in principle unsolvable, one is bound to lose either the resource or the topic. So what one has to do is to find practical solutions, which are unavoidably compromises." (TEN HAVE 2002, p.18) [18]

SCHEGLOFF is aware of arguments along those lines and against the concept of an "object of inquiry in its own terms'." He acknowledges these concerns, if in an ironic way:

"The very idea hints at a methodological and epistemological naivety that is unbecoming in our better universities, its seems to betray a touching belief in a 'reality'—and one which is accessible, furthermore—that seems unaffected by a properly sophisticated skepticism." (SCHEGLOFF 1997a, p.171). [19]

He continues proclaiming his own position:

Let me then make a clean breast of it. In my view, if ever there was an object of inquiry furnished internally with its own constitutive sense, with 'its own term,' with a defensible sense of its own reality, it is talk-in-interaction, and most centrally ordinary conversation." (ibid.) [20]

SCHEGLOFF's (1997a) talking of "objects in their own right," the possibility of "a partial account of the object itself," as well as of "an internally grounded reality of its own" does not leave much room for doubt in his realism. Still, the specifics of his view deserve some closer examination. Judging by SCHEGLOFF's own quote, it is obvious that the realism advocated here is not naïve, if epistemological naivety is to be attributed to an analyst who takes as real what senses and understanding disclose to her/him on first and unprepared exposure. On the contrary, the first stages of the conversation analytic process aim at diligently neutralizing the individual inquirer's unnoticed preoccupations, prejudices and background assumptions. SCHEGLOFF's analytic method represents an epistemological tradition that reaches back to DESCARTES' (1993, meditatio I) methodological doubt and includes HUSSERL's (2002a) phenomenological epoché. [21]

SCHEGLOFF's methodologist realism and approach to establishing the uninvolved spectator (cf. HUSSERL 2002a)5) aims at realizing an objective subjectivity in the following way: Since observation is unavoidably relative to the observer's subjective perspective, the study of social interaction, insofar as it involves observation, has to adopt a subjective viewpoint. The participants' perspectives on their own and on their co-interactants' efforts at jointly making sense, even though being relative to particular standpoints, are privileged in two ways. First, they are "endogenously grounded" (SCHEGLOFF 1997a, p.163) in discourse; that is, social interaction hinges fundamentally on the subjective sense that the participants make of it whereas the analyst's impact—ideally6) —is nil. [22]

Second, the participants' perspectives are—according to SCHEGLOFF—the same for and objectively accessible to everyone who is attentive to them. To refer to the analyst's activity as observation, therefore, appears to be misleading insofar as observation presupposes a particular point of view. In this sense, the conversation analyst does not observe the sense makings of the participating individuals. S/he rather aims at recording receptively the "displays" of the participants' "demonstrable" understandings. The "prism of disciplined and molecular observation" invoked by SCHEGLOFF is designed to be a tool indispensable for the social scientist in order to "refracture" the world "at the level of the lived reality of the events which compose it" (SCHEGLOFF 1997a, p.180). Furthermore, it is supposed to guarantee, very much in line with HERITAGE's (1984) microscope metaphor, that everybody who looks through this prism is exposed to and presented with the same reality. [23]

3.2 Presuppositions inherent to the CA research process

In order for CA to comply with SCHEGLOFF's standards, it has to be both non-circular and "about" a reality that is objectively accessible and independent of the inquiry by an analyst. It has been argued with respect to the first of these goals that SCHEGLOFF's research strategy does indeed allow for the analyst's being accountably and non-predictably led away from his/her original intuition (the "noticing of something") by novel data. What in the following is questioned is the contention that this starting point, the initial stage of the conversation analytic research process is or can be as unprejudiced and epistemically neutral as SCHEGLOFF suggests. It is argued that the analyst's presuppositions cannot be methodically neutralized and, what is consequential with regard to the subject-object distinction, they cannot be fully accounted for either. [24]

To substantiate these claims it is advisable to have another look at what SCHEGLOFF describes as the initial stage of CA analysis:

"In naturalistic inquiry of the sort I am committed to, it happens that, while examining a naturally occurring event (or, rather, a record of it), one notices something [...] that presents itself as 'Oh, I've seen something like that before!'" (SCHEGLOFF 1997b, p.501) [25]

At this point, the analyst does not know what s/he is looking for or even that s/he is looking for something. The "something" happens to present itself in an unspecified way and further analysis may eventually suggest that there was not really anything at all to begin with (cf. above, stage IV). While the receptive exposure to some discourse fragment thus described does not seem to be contingent on a particular researcher's preoccupations and preconceptions, this data examination does not take place without preparation on the analyst's part. A list of preceding activities includes the following:

  • deciding to make audio and/or videotapes of a piece or several pieces of talk-in-interaction

  • deciding on what exchanges to make a record of

  • asking certain individuals to agree with being recorded

  • having the technical equipment ready

  • putting up the camera(s)/audio recorder(s) at particular places in the room

  • making the record or having it made

  • putting up the equipment for watching the tapes. [26]

All these activities and decisions bring with them consequences for the nature of the data that the SCHEGLOFFian analyst examines at stage I. And all of them are based on choices, either made in an accountable manner or rather accidentally. These choices are consequential on earlier deliberate or accidental choices, and so on. [27]

To consider briefly one of the above mentioned factors, the focus on telephone conversations in the early CA studies of the 1960s seems grounded in the technological contingencies of those days rather than the nature of social interaction. In general, everyday conversation has offered itself as a favorite data source because it has been readily available without much administrational, or even legal, preparation. Apart from these practicalities, there are good, if both contestable and contested7), reasons to prefer so-called natural conversations to interview data or literary fiction for example. [28]

Whether or not those arguments hold up against objections from, for example, structuralist or generativist linguists, must and can be left undecided here. It is important to note, however, that the use of particular discourse fragments as data is necessarily preceded by a particular history. This includes, in the better cases, the analyst's accountable reasons for using this very material. Even if the pre-history of the first exposure to the data, to some extent, can be accounted for in descriptive and theoretical terms, these cannot be the terms of the data themselves. In addition to being thus "exogenously grounded," each conversation analytic data choice is to an undeterminable extent the result of processes that themselves are consequential on previous processes and so on. This history must remain withdrawn from full reconstruction. [29]

A concept that was referred to as objective subjectivity, that is, a participant's subjective point of view that is objectively accessible to analysts regardless of their respective preconditions and backgrounds, was previously identified to be a fundamental component of SCHEGLOFF's methodologist realism. This foundation or "bedrock" (HERITAGE & ATKINSON 1984, p.15), however, was found not to support the argument resting on it. Its alleged objectivity turned out to be flawed because SCHEGLOFF's "naturally occurring event (or, rather, a record of it)" is not a neutral venture point for an ensuing analysis, but is contingent on a number of individual decisions and choices by the analyst. [30]

4. Analytic Presuppositions. The Case of Conversational Repair

The study of conversational repair, for several reasons, is a suitable example for the purpose of illustrating the conversation analytic research strategy outlined above, the non-circularity of this strategy, (some of) its presuppositions and its limits with regard to the discourse reality it makes accessible.8) Repair is one of the classical themes of CA and one that has drawn some methodological attention by the protagonists of CA themselves. To begin with, the concept of repair is introduced by quoting its authors. Subsequently, a particular structural subtype, viz. other-initiated self-repair, is presented and some of the features typical of CA conceptualization are pointed out. This is followed by remarks on the history of research in repair that illustrate the non-circularity of the project. Next, I argue that this very research history and, particularly, those contributions to it that have succeeded the original work by SCHEGLOFF, JEFFERSON, SACKS and others imply preconceptions not intrinsically grounded in the data analyzed by the respective researchers. Finally, consequences from these findings for the study of conversational repair are suggested. [31]

4.1 Preliminaries on other-initiated self-repair

In the abstract of SCHEGLOFF, JEFFERSON & SACKS' (1977) classical paper on The Preference for Self-Correction in the Organization of Repair in Conversation, the authors present their understanding of the concept repair: "An 'organization of repair' operates in conversation, addressed to recurrent problems in speaking, hearing, and understanding." (ibid., p.361; emphasis mine, TW) [32]

In line with and elaborating on this first explication, SCHEGLOFF writes at a number of occasions:

"By 'repair' we refer to practices for dealing with problems or troubles in speaking, hearing, and understanding the talk in conversation (and in other forms of talk-in-interaction, for that matter). I want to underscore the phrase 'the talk' in my reference to 'problems in understanding the talk'; for we did not mean to include within the scope of 'repair' all practices addressed to problems of understanding (like understanding exactly how the Internet works), only the narrower domain of 'understanding what someone has just said'—though there can on occasion be only a fuzzy boundary between these. [...]" (SCHEGLOFF 2000, p.207, for almost identical quotes, cf. 1987, p.210, and 1997a, p.503) [33]

Within the class of repair sequences, SCHEGLOFF et al. (1977) distinguish four structural types along the lines of which participant initiates the repair and thereby makes it a relevant next move, and of who performs the repair, i.e. makes an attempt at overcoming or solving a problem "of speaking, hearing, and understanding." The participant roles involved are self, i.e. the one who has produced the problematic utterance (or trouble source or repairable), and other, i.e. the recipient of the trouble source. Table 1 presents the repair typology in a schematic form:

Table 1: Structural types of conversational repair (cf. SCHEGLOFF et al. 1977) [34]

The following sequence, extracted from a corpus of dinner table conversations (cf. WEBER 1998, Chapter 4), instantiates the type of other-initiated self-repair that will be in focus below:

Sequence 1: Südöstlich [35]

In the current episode, Theo has been describing to the dinner host, Curt, where he and his wife are staying during their visit in town. Curt initiates repair upon Theo's problematic turn (03) by, first, expressing surprise (06) and, then, making manifest exactly what element causes him trouble (07). In response, Theo, after what may be described as a change-of-state token in HERITAGE's (1984b) sense, completes a self-repair by correcting his earlier information. [36]

4.2 The history of repair and its epistemic implications for the data analysis

Following these preliminary explications of what (other-initiated self-)repair is, a few aspects of the manner in which the concept is introduced by its authors are worth highlighting. This pre-history of the repair concept includes Harvey SACKS' (1995a,b) lectures in which many of the phenomena eventually referred to as repairs are discussed. Most of the time, these sequences are not subsumed in a single category or addressed by a unifying term (cf., e.g., 1995a, pp.2-3, 1995b, p.447).9) When SACKS, borrowing a concept from Erving GOFFMAN and modifying it for his own purposes, introduces remedial exchange as a summarizing label, this is indicative of a process of conceptualization that is still going on rather than brought to a conclusion.10) A particular (sub-)class that seemed particularly salient to the early conversation analysts was addressed as corrections (cf. SACKS 1995b, p.115, pp.120-121, p.146; JEFFERSON 1975 and the title of SCHEGLOFF et al. 1977). As SCHEGLOFF et al. note: "The term 'correction' is commonly understood to refer to the replacement of an 'error' or 'mistake' by what is 'correct.' The phenomena we are addressing, however, are neither contingent upon error, nor limited to replacement" (SCHEGLOFF et al. 1977, p.363). [37]

The eventual introduction of the concept of repair and the subsumption of a variety of types of conversational phenomena in it is one more step in a process of conceptualization that is inherently non-circular. The case of repair, thus, can be taken as an exemplary one to illustrate that the CA strategy may lead analysts to modify their initial intuitions and conceptualizations in the light of newly encountered data (cf. above, stages III, IV, III', IV' ...). [38]

Given these remarks on the early history of research in repair, it is obvious that SCHEGLOFF et al.'s explication of repair right at the onset of their 1977 paper does not mark the beginning of the authors' concern for phenomena of this kind. While the report's retrospective and summarizing nature is a necessary feature of conversation analytic prose of scientific reports in general, for CA this implies important consequences. In the context of this approach to social interaction, terms like repair are considered legitimate only insofar as they are "intrinsically grounded" in the data. It is, however, impossible to represent in scientific writing more than a small portion (of the transcripts of video or audio records) of all the original events the inquirers exposed themselves to in the course of an open and recursive research process. [39]

Eventually, the persuasive power of the conversation analysts' coining and using a term like repair seems to rest on two factors: the reader's trust in the conceptualization process's reflecting the terms of "the objects in their own right" and the value the proposed term has for the reader's own practical purposes (e.g., for making sense of a collection of discourse data). Going back to the data themselves (the only legitimate criterion of categorization) is, under normal circumstances, impossible. This seems true of the interested, but rather uninvolved readers of the 1977 paper as well as of conversation analysts who, for decades, have been working with the repair concept while being mostly detached from the original data with reference to which the concept was introduced. [40]

Considered in this light, an analyst determined to pursue a strictly SCHEGLOFFian research strategy and to choose a venture point that is epistemically unprejudiced would have to be unaware of all pertinent findings obtained by others. For many reasons, this seems as futile as unrealistic. Catching on, contributing to, and even intentionally ignoring a scholarly tradition implies a loss of epistemic innocence. This is true also for an author building up on his own previous work and, hence, for SCHEGLOFF (2000, p.206) when he reproduces his explication of the term repair from an earlier paper (cf. 1997b, p.503) that, in turn, was based on the original Preference for Self-Correction study (SCHEGLOFF et al. 1977, p.361). [41]

4.3 No empirical analysis of repair without a preconception of its formal properties

In the previous section, I have referred to a tradition of studying conversational repair. But what is repair or, rather, what do tokens of repair look like, and what makes them "observably and accountably" different from phenomena of other types? These questions are crucial for empirically minded analysts interested in repair. According to SCHEGLOFF's outline of the analytic process (cf. above, stages III and IV), these inquirers have to scan their databases for candidate repair tokens and subject them to detailed structural scrutiny. Even if the criteria necessary in this process are applied in a generous fashion, they have to be defined in terms of surface features that are suitable to draw the analyst's attention and that allow him/her to identify discourse fragments accountably as candidates for membership in the supposed category of repairs. [42]

The CA literature on repair provides for extensional accounts in the form of open lists of structural phenomena to be classified as repairs11) but it does not offer, with a noticeable exception to be acknowledged below, an intensional structural definition. SCHEGLOFF himself, in the texts quoted above, specifies what repair is by explicating what repairs do in talk-in-interaction, what repairs are deployed for by the participants, and what the discourse function of repairs is. Accordingly, repairs or, more specifically, repair sequences allow the participants to deal with "problems or troubles in speaking, hearing and understanding the talk in conversation [...]" (SCHEGLOFF 2000, p.207). [43]

Functions, however, are not perceptible. A participating recipient will be able to identify the function of a co-participant's contribution only if this function is expressed and signaled to him/her by some surface structural property of the speaker's utterance. The same limitation applies to the conversation analyst insofar as s/he has to account for the data from the subjective but privileged point of view of the participants' own understanding (cf. SCHEGLOFF 1997a, p.180). A common function, thus, is not a property with reference to which a class of phenomena could be established by way of empirical data analysis. In the case of conversational repairs, it is impossible to observe directly interactants' being in and overcoming certain mental states like the experience of trouble. This suggests not only that a definition of repair, preliminary as it may be, can be given in structural terms, but that an implicit understanding of what tokens of this presumed category look like must have guided their study from its very beginning. [44]

If this is so, it is fair to propose a structural definition of repair that accounts for those sequences of discourse presented in the literature as instantiating the category. In accordance with FOX and JASPERSON (1995, p.80), the term conversational repair is applied to sequences that are characterized by sequential discontinuity and retrospectivity.12) Sequential discontinuity refers to the property of repairs to interrupt other ongoing activities in a prosodically, syntactically and/or otherwise manifest manner. Repairs are retrospective in that they relate back to and recycle some previous piece of talk. [45]

I do not claim here that this explication definitively formulates the necessary and sufficient conditions for membership in the category of repairs. Nonetheless, the two criteria mentioned, on the one hand, seem compatible with the repair sequences presented in the literature and, on the other hand, can be used as part of a heuristic for compiling new collections of candidate repairs.13) Finally, the definition is in line with SCHEGLOFF's reasoning in at least two regards. Its relative broadness guarantees that the collection assembled will be a generous one. Furthermore, SCHEGLOFF himself (cf. stage IV above) requires of the analyst to exclude tokens form the generous collection in an accountable manner thus being forced "to make explicit just what it is which makes them different from our targets [...]" (SCHEGLOFF 1997, p.502; emphasis mine, TW). [46]

"[W]hile examining a naturally occurring event (or, rather, a record of it), one notices something [...] that presents itself as 'Oh, I've seen something like that before'" (SCHEGLOFF 1997b, p.501). I have argued that what SCHEGLOFF describes in this quote is not without preconditions on the analyst's part. For the inquirer to notice something during the examination of an event means for him/her to become alert to a certain structural feature of that event. This feature is capable of drawing the observer's attention only because s/he has encountered a similar event before. Another analyst, one with a different background of experiences, might not have noticed anything when exposed to the same piece of data or s/he might have noticed something different. If this is so, the objects "present themselves" to an individual whose starting point, rather than being neutral, is the outcome of a particular experiential biography. This applies to the pioneer researcher, who is the first one to concern him/herself with a certain class of phenomena, and it applies, a forteriori, to everyone willing to continue and contribute to conversation analytic work initiated by others. [47]

5. Conclusion

In this paper, I have reconstructed a variety of Conversation Analysis that has been pursued and advocated by Emanuel A. SCHEGLOFF as its most prominent, but by no means only (cf., e.g., HERITAGE 1984a and HERITAGE & ATKINSON 1984), representative. With regard to its epistemological stance, this research strategy was found to instantiate methodologist realism in the Cartesian tradition. In this respect, the SCHEGLOFFian approach is exceptional as well as an integrated part of the phenomenological paradigm (cf. TEN HAVE 2002). [48]

The approach was examined along the lines of two criteria: its ability to avoid circularity and its claim to account for the nature of discourse "in its own terms" as "an internally grounded reality of its own" (SCHEGLOFF 1997, p.171). The recursive conversation analytic process of conceptualization is not circular because, with every novel data fragment, it potentially confronts the analyst with the necessity to modify or to give up his/her previous understanding of an emerging category. Hence, the starting point of the analysis does not determine its outcome. [49]

I have rejected, however, the assumption that the starting point can be epistemically neutral and, thereby, objected to a fundamental premise of SCHEGLOFF's methodologist realism. In particular, I have argued that the viewpoint of the participant, whose understanding of his/her activities is privileged in several ways, is not to be interpreted as an objective subjectivity that provides the analyst with direct access to discourse as a "reality of its own." On the contrary, I have proposed that the analysis of conversation is to a significant extent relative to the point from which it proceeds; this venture point is the result of choices based on preconceptions on the part of the analyst that are as unavoidable as they are impossible to account for. [50]

The case of repair was introduced here as an example to illustrate both the non-circularity of the CA research process as well as its failure to account for discourse on an epistemically neutral basis. I have outlined the (pre-) history of the repair concept and suggested that it represents an open and non-circular process of conceptualization. The category of repair is a—preliminary—outcome of this process without being presupposed by it, for example, by way of having implicitly determined the early conversation analysts' research intuitions. [51]

At the same time, this conceptual history indicates that repair cannot be a category endogenously grounded only in the respective data that are found to justify it. Since even ignoring something is a way of acknowledging it, it is impossible for conversation analysts to look at their data without being in some way "preoccupied" by the classical CA concepts such as repair. Furthermore, preconditions do not only apply to those who carry on previous work started by themselves or by others. The very first observer who noticed "something" that later turned out to be a case of repair must have been alerted to that noticeable token by some perceptible property of it. Being noticeable, however, is not a property of an object in its own right, but an effect that may begin when this object is observed by an individual with a particular experiential background. [52]

In conclusion, I would like to suggest that the arguments presented on the epistemological conditions of empirical discourse analysis are not contingent on particularities of the research strategy discussed above. Methodologist realism in the social sciences is generally ill-designed because there is no conceivable method that would allow an analyst to establish clearly and distinctly the aspects of an observation that are due to his/her way of looking at the data and the terms of these data in their own right. Accordingly, (the inquirer's) subject, (his/her) object, and the border between them are indiscernible and, thus, the concepts are irrelevant in the realm of empirical social analysis. [53]


I would like to thank Matthias BRUNK for a lot of helpful comments on a draft of this paper.


1) As it has turned out, one of the main problems of modern epistemology is to determine what "reasonable" means in this context. <back>

2) As we can learn from TEN HAVE (2002), it is necessary to differentiate between epistemological positions within CA. The methodologist form of SCHEGLOFF's and others' (cf. HERITAGE 1984) realism is not common ground among those who consider themselves conversation analysts and certainly not among all scholars rooted in the tradition of GARFINKEL's ethnomethodology. <back>

3) It seems to be neither coincidence nor a matter of ideological prejudice—as suggested by BILLIG's (1998, pp.549-551) critique of SCHEGLOFF—that CA has emerged as an approach that primarily but not exclusively was applied to conversation. Rather, the motivation seems to have been at least partially of a methodological nature since conversation defined as a sequence of more than one participants' alternating turns is the kind of discourse most open to this kind of analysis. <back>

4) I use III', IV', III'', IV'' ... rather than V, VI, VII, VIII ... to number the consecutive stages of the CA research process in order to underscore the recursivity of this process. At stages III, III', III'' ... the collection of candidate tokens is increased, at stages IV, IV', IV'' tokens are excluded accountably. That is, identity of Roman numbers represents identity of functional stage (i.e., increasing the collection or excluding tokens from the collection) in the research process; the apostrophes are meant to indicate that the "material", the collection of candidates, change in the course of the analysis. <back>

5) HUSSERL entitles one of the texts published in this volume Transzendentale Epoché und Reduktion. Die Etablierung des unbeteiligten Zuschauers (HUSSERL 2002b). <back>

6) Cf., however, the methodological problems related to the so called observer's paradox (cf. LABOV 1972, p.209). <back>

7) Cf., e.g., Michael BILLIG's (1999a, pp.549-550; cf. also 1999b) critique of "the rhetoric of 'ordinary conversation'" in the context of his illuminating debate with SCHEGLOFF (1997a, 1999a, 1999b). <back>

8) The following remarks are partially based on considerations presented in WEBER 1998. <back>

9) The inquiries on which these remarks are based have made use of Gail JEFFERSON's index of SACKS' lectures. Even though SACKS himself, it seems, does not use the term, JEFFERSON has included into her index an entry repair. This entry refers to various sections of the lectures that, in the light of the 1977 paper and pertinent subsequent work, can be said to deal with cases and types of repair. <back>

10) This seems particularly obvious when looked upon in the light of a parenthetical remark by which SCHEGLOFF continues his definition of repair as quoted above. After having specified what repairs are not, he goes on: "(Nor, I might add, did we mean to refer to efforts to deal with tension or breakdown in the interaction, or violations of its so-called ritual order—what Goffman (1971) termed 'remedial interchanges'.)" (SCHEGLOFF 2000, p.207). <back>

11) SCHEGLOFF mentions the following open list of phenomena: "'Trouble' includes such occurrences as misarticulations, malaproprisms, use of a 'wrong' word, unavailability of a word when needed, failure to hear or to be heard, trouble on the part of the recipient in understanding, incorrect understandings by recipients, and various others. Because anything in talk can be a source of trouble, everything in conversation is in principle, 'repairable'" (SCHEGLOFF 1987, p.210, cf. also 1997b, p.503 and 2000, p.209). <back>

12) The two aspects, retrospectivity and discontinuity, are also implied by FOX and Robert JASPERSON's (1995, p.80) definition of (self-initiated self-) repair: "We define repair here, then, as any instance in which an emerging utterance is stopped in some way, and is then aborted, recast, or redone." <back>

13) For a more elaborate discussion of this definition and its heuristic value for distinguishing candidate repairs from discourse phenomena of other types, cf. WEBER 1998 and 2002. <back>


Descartes, René (1993). Meditations on First Philosophy in focus. Edited and with an introduction by Stanley Tweyman. London, New York: Routledge. (Meditationes de Prima Philosophiae. Paris. 1641).

Billig, Michael (1999a). Whose Terms? Whose Ordinariness? Rhetoric and Ideology in Conversation Analysis. Discourse & Society, 9, 543-582.

Billig, Michael (1999b). Conversation Analysis and the Claims of Naivety. Discourse & Society, 9, 572-576.

Egbert, Maria (2002). Der Reparatur-Mechanismus in deutschen und interkulturellen Gesprächen. Habilitationsschrift. Oldenburg: Carl von Ossietzky Universität Oldenburg.

Egbert, Maria (to appear). Other-Initiated Repair and Membership Categorization—Some Conversational Events that Trigger Linguistic and Regional Membership Categorization. Journal of Pragmatics.

Ford, Cecilia E.; Fox, Barbara A. & Thompson, Sandra A. (Eds.) (2002). The Language of Turn and Sequence. New York: Oxford University Press.

Fox, Barbara A. & Jasperson, Robert (1995). A Syntactic Exploration of Repair in English Conversation. In Philip W. Davis (Ed.), Alternative Linguistics. Descriptive and Theoretical Modes (pp.77-134). Amsterdam, Philadelphia: John Benjamins.

Garfinkel, Harold (1967). Studies in Ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall.

Have, Paul ten (2002, September). The Notion of Member is the Heart of the Matter: On the Role of Membership Knowledge in Ethnomethodological Inquiry [53 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research [On-line Journal], 3(3). Available at: http://www.qualitative-research.net/fqs-texte/3-02/3-02tenhave-e.htm.

Heritage, John (1984a). A Change-of-State Token and Aspects of its Sequential Placement. In J. Maxwell Atkinson & John Heritage (Eds.), Structures of Social Action. Studies in Conversation Analysis (pp.299-345). Cambridge: Cambridge University Press.

Heritage, John (1984b). Garfinkel and Ethnomethodology. Cambridge: Polity Press.

Heritage, John & Atkinson, J. Maxwell (1984). Introduction. In J. Maxwell Atkinson & John Heritage (Eds.), Structures of Social Action. Studies in Conversation Analysis (pp.1-15). Cambridge: Cambridge University Press.

Husserl, Edmund (1992). Gesammelte Schriften Band 5. Ideen zu einer reinen Phänomenologie und phänomenologieschen Philosophie. Buch 1. Allgemeine Einführung in die reine Phänomenologie. Hamburg: Meiner.

Husserl, Edmund (2002a). Zur phänomenologischen Reduktion. Texte aus dem Nachlass (1926-1935). Herausgegeben von Sebastian Luft. Dordrecht u.a.: Kluwer Acad. Publ.

Husserl, Edmund (2002b). 4. Transzendentale Epoché und Reduktion. Die Etablierung des unbeteiligten Zuschauers (Herbst 1926). In Edmund Husserl, Zur phänomenologischen Reduktion. Texte aus dem Nachlass (1926-1935). Herausgegeben von Sebastian Luft (pp.86-99). Dordrecht u.a.: Kluwer Acad. Publ.

Jefferson, Gail (1975). Error Correction as an Interactional Resource. Language in Society, 3, 181-199.

Jefferson, Gail (1983). On Exposed and Embedded Correction in Conversation. Studium Linguistik, 14, 58-68.

Labov, William (1972). Sociolinguistic Patterns. Philadelphia: Univ. of Pennsylvania Press.

Sacks, Harvey (1995a). Lectures on Conversation, vol. 1. Edited by Gail Jefferson, with an introduction by Emanuel A. Schegloff. Oxford: Blackwell.

Sacks, Harvey (1995b). Lectures on Conversation, vol. 2. Edited by Gail Jefferson, with an introduction by Emanuel A. Schegloff. Oxford: Blackwell.

Sacks, Harvey; Jefferson, Gail & Schegloff, Emanuel A. (1974). A Simplest Systematics for the Organization of Turn-Taking for Conversation. Language, 50, 696-735.

Schegloff, Emanuel A. (1987). Between Micro and Macro. Contexts and Other Connections. In Jeffrey C. Alexander, Bernhard Giesen, Richard Münch & Neil J. Smelser (Eds.), The Micro-Macro Link (pp.207-237). Berkeley: University of California Press.

Schegloff, Emanuel A. (1992). To Searle on Conversation. In John R. Searle, Herman Parret & Jef Verschueren (1992). (On) Searle on Conversation. Compiled and Introduced by Herman Parret & Jef Verschueren (pp.113-128). Amsterdam & Philadelphia: John Benjamins.

Schegloff, Emanuel A. (1997a). Whose Text? Whose Context? Discourse and Society, 8, 165-187.

Schegloff, Emanuel A. (1997b). Practices and Actions. Boundary Cases of Other-Initiated Repair. Discourse Processes, 23, 499-547.

Schegloff, Emanuel A. (1999a). "Schegloff's Texts" as Billig's Data: a Critical Reply. Discourse & Society, 9, 558-572.

Schegloff, Emanuel A. (1999b). Naivete vs Sophistication or Discipline vs. Self-Indulgence: a Rejoinder to Billig. Discourse & Society, 9, 577-582.

Schegloff, Emanuel A. (2000). When "Others" Initiate Repair. Applied Linguistics, 21, 205-243.

Schegloff, Emanuel A.; Sacks, Harvey & Jefferson, Gail (1977). The Preference for Self-Correction in the Organization of Repair in Conversation. Language, 53, 361-382.

Selting, Margret (1987). Verständigungsprobleme. Eine empirische Analyse am Beispiel der Bürger-Verwaltungs-Kommunikation. Tübingen: Niemeyer.

Selting, Margret & Couper-Kuhlen, Elizabeth (Eds.) (2001). Studies in Interactional Linguistics. Amsterdam, Philadelphia: John Benjamins.

Weber, Tilo (1998). Shared Background and Repair in German Conversation. Boulder, CO: University of Colorado, Boulder.

Weber, Tilo (2002). Reparaturen – Routinen die Gespräche zur Routine machen. Linguistische Berichte, 192, 419-454.


Tilo WEBER, linguist, is currently working on a book on Shared Background and Repair in German Conversation. His main research interests lie in the areas of discourse, linguistic categorization and language philosophy. He is Hochschulassistent at the Martin-Luther-Universität Halle-Wittenberg.


Tilo Weber, Ph.D.

Martin-Luther-Universität Halle-Wittenberg
Germanistisches Institut
06099 Halle

E-mail: weber@germanistik.uni-halle.de


Weber, Tilo (2003). There is no Objective Subjectivity in the Study of Social Interaction [53 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 4(2), Art. 43, http://nbn-resolving.de/urn:nbn:de:0114-fqs0302432.

Revised 6/2008

Copyright (c) 2003 Tilo Weber

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.