3
STARTING FROM MEANING
Contextuality and Its Implications
Kellogg's set up a branch in India and started producing corn flakes to give consumers the real thing. What they didn't realize was that Indians, rather like the Chinese, think that to start the day with something cold, like cold milk on your cereal, is a shock to the system. You start it with warm milk. But you pour warm milk on Mr. Kellogg's corn flakes and they turn to wet paper. You pour warm milk on the sturdier Indian corn flake, it holds up. Does it taste better than Mr. Kellogg's? No. If Mr. Kellogg's is eaten as Mr. Kellogg intended, it is somewhat better than Indian corn flakes. The point is that . . . you have to know something about . . . a place and its cultural rituals.
Having an initial sense of their research question, interpretive researchers begin designing their research project by thinking about the kinds and sources of evidence that would enable them to engage it. Most start off by thinking about the setting(s), archives, event(s), actors, and so forth among which and whom they will conduct their study. Although we would prefer to go directly to those matters, in keeping with the rhythms of an interpretive research project, we have decided to defer that to the next chapter in order to address, first, the reader who is accustomed to thinking from a different starting point. This is the reader who believes that the first step in a research design has to be the identification and definition of concepts, their operationalization in the form of variables, and the stipulation of hypotheses that establish the relationships among them. Underlying this starting point is a particular orientation toward knowledge and its sources. In order to be able to speak more clearly about where interpretive research begins, and why, we need to discuss its contrasting orientation toward knowledge, with its focus on meaning-making and on contextuality.
How does an orientation toward contextuality bear on these matters? Concepts and some form of hypothesizing and conceptions of causality are central to scholarly endeavors from both interpretive and positivist approaches. What is meant by these terms, however, and even the precise terminology used, varies across the two approaches and contributes to significant differences in thinking about designing a research project.
Contrasting Orientations toward Knowledge
All researchers seek to contribute to knowledge. Yet this common goal elides complex questions: Knowledge about what? Knowledge for what purpose? Knowledge for whom? The sine qua non of interpretive research—the sensibility that is its hallmark and which makes it distinctive in comparison with other research approaches—is its focus on meaning-making: it seeks knowledge about how human beings, scholars included, make individual and collective sense of their particular worlds.
In interpretive research, human beings are understood not as objects, but as agents. Such persons are seen as actively and collaboratively constructing (and deconstructing, meaning both critically assessing and changing) their polities, societies, and cultures—along with the institutions, organizations, practices, physical artifacts, and language and concepts that populate these. At the same time, those same political and cultural contexts frame these agents’ possibilities for thought, discourse, and action. Interpretive research understands that the motivation that animates these several activities is meaning—both its expression and its communication to others. As language is at the nexus of meaning, context, and action, interpretive research “tak[es] language seriously” (J. White 1992). For this reason, the interpretive turn in theorizing about ways of knowing joined the linguistic turn (Rorty 1967, Fraser 1995), the metaphoric turn (Lorenz 1998), and other such “turns,” including the practice turn (Schatzki et al. 2001) with its attention to acts and physical artifacts.1
In interpretive methodology, the ambiguity and plasticity of meaning-making and of the systems of symbols (e.g., language, visual images, etc.) used to express and communicate meaning to oneself and to others are understood as creating the possibility for multiple interpretations of acts, events, settings, and so forth. This is, ontologically, the reason for attending to humans’ use of language in constructing their worlds and, epistemologically, in making sense of them. The possibility of the multiplicity of meanings is one of the things that makes connections to context critical for both the conduct of interpretive research and its design: the reasons things take these particular forms and not others has to do with their specific contexts of time and place. It matters that the corn flakes are being eaten in India rather than in the US, in Homi Bhabha's tale that serves as the epigraph to this chapter. Another point about multiple possibilities and the importance of context is the understanding that the artifacts humans create to express and convey meaning—not only language, but also acts and the physical objects engaged in doing and speaking—can take myriad forms, depending on time and/or place. For those studying humans in group form—as neighborhoods, communities, polities, organizations, and so on—a distinctive aspect of research concerns the artifactual meanings that members share and which separate one epistemic–interpretive group from another, perhaps provoking not only disunity but also conflict.
Researchers in positivist modes also seek to understand humans and their behaviors; but what language and other artifacts mean for these researchers, in terms of their uses in research, is quite different. In positivist methodology, ambiguity is the enemy of measurability, where each instance of a phenomenon must be classifiable in one and only one category. As concepts need to be operationalized to produce variables for measuring phenomena and for assembling data sets (on the same unit of analysis) for subsequent statistical assessments, initial attention to their specification is imperative. In this process, their definition is, and must be, both precise and abstracted from context. Short of a pilot study, conceptual definitions are “locked in” to the stipulated variables and their definitions at the point of measurement (e.g., when a survey is administered). If, in the course of the research, it is judged that the measure of a concept is poor, rendering it an inadequate indicator of subjects’ meanings, the concept cannot be redefined and re-measured in that same study.2
Just as interpretive research is distinctive in its focus on meaning-making (knowledge about what?), it also has different sensibilities about the uses of knowledge: knowledge for what purpose, for whom? Interpretive methodologists dispute the usefulness (and desirability) of knowledge that claims to “rise above” its context. This does not mean that interpretive researchers do not theorize. Instead, they seek to theorize on the basis of knowledge that makes clear its connections to specific (kinds of) human beings in specific, historically and culturally understood settings.
This presents a strong contrast with positivist methodology, for which “generalizability” is a widespread concern. Researchers in that tradition are often asking, “Are the results generalizable?” This orientation implicitly places responsibility for the applicability of “findings” from one research setting to potential others on the shoulders of the researchers; i.e., it is they who must demonstrate that their results hold for all other settings or for those specified in the “scope” conditions of mid-level theorizing (which delimit that theorizing from a-historical, a-cultural theorizing). In this research approach, the overarching goal is building general theory for the purposes of prediction (and explanation).
By contrast, members of interpretive research communities ask: “Is the research sufficiently contextualized so that the interpretations are embedded in, rather than abstracted from, the settings of the actors studied?” Seen from an interpretive perspective, the positivist orientation toward the general obscures the intimacy (and inseparability) of the link between research purpose (anticipated learning, in light of broader theoretical concerns) and the context(s) that sparked or drove the research. Interpretive methodology shifts responsibility for the applicability of research learning to other research settings from the researchers to the readers of the research. As Lincoln and Guba (1985) long ago argued, scholars should describe their research contexts in sufficiently “thick” ways that readers of their work (including researchers engaged with other research questions) can assess the relevance of the research to their own settings.3 Whereas positivist methodologists understand “findings” as useful for building general theory solely for the purposes of prediction and explanation, interpretive methodologists observe that research can also be useful for a variety of other purposes—not only explanatory ones, but emancipatory and critical ones, as well.4 The quality or value of contextualized knowledge (theory) is to be assessed by users, whether academic or other, who decide themselves the extent to which that knowledge fits their circumstances and purposes, i.e., whether it works in context (Avenier 2010, Tsoukas 2009). The centrality of context to interpretive methods lends weight to treating contextuality as a more appropriate indicator of the achievements of interpretive research than “generalizability,” its equivalent in positivist methodology: it is a better fit with interpretive methodological orientations to knowledge creation and use.5
The interpretive orientation toward knowledge, with its focus on meaning-making (instead of a priori model specification) and contextuality (rather than generalizability), ripples through the entire research design process.6 Contextuality provides a direct methodological rationale for the “thick description” (Geertz 1973) that has been widely taken up by methodologists as a key characteristic of qualitative–interpretive writing, which embeds meaning in context. But this desideratum requires “thickly written” fieldnotes, themselves resting on “thickly crafted and experienced” observations, interviews, and documentary reading. To take as an example the field research narrative that Geertz uses in developing this concept, a “thin” description might be rendered this way: Cohen, a Jewish trader, stole some sheep from a Berber tribe, and so the French authorities imprisoned him (1973: 7–9).
A thicker description would add the “whys” and “hows” that underlie this statement's “whats,” contributing all manner of historical, demographic, economic, geographic, and cultural nuances, as relevant to the research question and what the researcher is seeking to theorize, that contextualize the seemingly simple event. Such is, in fact, the research tale that Geertz narrates, over three pages of text. Neither the events nor their description need to be out of the ordinary: what matters is the detail, as this bit from Liebow's examination of Tally's corner's fathers shows:
Together with Calvin, a frail and ailing forty-year-old alcoholic and homosexual who looked after the children in exchange for a place to live, Leroy bathed the children, braided the girls’ hair, washed their clothes at ‘the Bendix’ (laundromat), played with them, and on their birthdays went shoplifting to get them gifts.
(Liebow 1967: 52)
The information is presented without judgment; and what is significant is what it will enable Liebow to argue, later, with the support of additional evidence in terms of the data's implications for the theoretical issues that concern him in this study. In order to be able to produce a research text of such detail, a researcher has to have observed, talked, and/or read enough, and noted all of that, to generate material for those layers of contextualization.7
Good interpretive research designs think through the crafting of situations—the selection of research-relevant settings, actors, events, documents, and so on— that can lead to those sorts of “thick experience,” as well as of ways to “thicken” encounters in the field (see discussion of mapping for exposure and intertextuality in Chapter 5). In thinking through the specifics of a research project, investigators seek to design research in ways that respect and preserve the meaning–context link such that readers are enabled to understand local preferences for warm milk over cold, even if it means breakfasting on less tasty corn flakes, as Bhabha puts it.
Contextuality and the Character of Concepts and Causality
Putting contextuality front and center in a logic of research bears significantly on how one thinks about concepts and on the understandings of causality that are ensconced in hypotheses. We further defer a discussion of where and how interpretive researchers begin to design research in order to discuss these two matters, as they are commonly ones on whose use interpretive researchers are challenged and on whose grounds interpretive research is seen as deficient. We seek here to show researchers not familiar with interpretive logics of inquiry why this is not the case, by contrasting interpretive understandings of “concepts” and “causality” with positivist ones.
Concepts: Bottom-up In Situ Development
All researchers use concepts. We conceptualize as we build theories from our empirical research, and we use the concepts developed in that process to communicate with one another. Following Geertz (1983), we characterize these sorts of concepts as “experience-distant”—part of the scholarly world, but not commonly part of the worlds studied by social scientists (unless they are studying researchers, as science studies scholars do). Interpretive researchers seek to understand the worlds of those they study from the latter's perspectives. They seek, to complete Geertz’ taxonomy, the concepts that are “experience-near” (1983: 57)—used by those on the kill-floor of a slaughterhouse (Pachirat 2009a), in government discussions of national security (Cohn 2006), or on the receiving end of US government welfare processing requirements (Soss 2000), to take but three examples. In part, this practice of searching for experience-near concepts derives from the conviction that participants possess valuable “local knowledge,” concepts and their situated definitions that have grown out of their own daily practices and interactions, reflecting their own lived experiences of the setting, events in it, interactions, and so forth: that is what researchers want to understand.8 Learning experience-near concepts may provide entrée to such knowledge, which for those using the concepts in everyday ways is often tacit, in Polanyi's (1966: 4) sense that we “can know more than we can tell.”
Interpretive research design plans for concept “development” to take place during fieldwork, not before it. In one sense, this might be considered development to the extent that a researcher needs to develop her own understanding of what for situational actors is common sense. Depending on the research question, design possibilities for acquiring local concepts range from the choice of quite formal methods (e.g., some versions of grounded theory, Charmaz 2006; ordinary language analysis, Schaffer 2006) to more open-ended attention to how participants talk, their use of special terms, jargon, vernacular, modes of reasoning, metaphors, etc. Hanging out with medical students, for instance, Becker (1998) was puzzled by their use of the word “crock.” Finding out and exploring what this word meant to them—a patient who did not advance students’ learning—and how they used it became central to his understanding and analysis of medical education. Looking at shop-floor workers’ interactions on the job, Roy (1959) explored what they termed “banana time,” jokingly named after the ritual “theft” of one colleague's mid-morning snack, to generate insights into job satisfaction in routinized work settings. Soss (2000) found that welfare recipients enrolled in two government programs, Social Security Disability Insurance (SSDI) and Assistance to Families with Dependent Children (ADFC), used the same phrase—being treated “like a number”—but with contrary meanings in each program due to their quite different contexts. SSDI recipients used the phrase to express their feeling that they were treated impersonally, but with respect, whereas AFDC recipients used the same words to convey their feeling of being dehumanized by their treatment. In all three examples, the concepts were already “developed” in the field by those who used them in everyday sorts of ways; what researchers needed to do was to develop their own knowledge of those concepts in their situated usages—to learn how to use them in the “local language” of each research setting.
By contrast, the a priori concept formation characteristic of positivist research design, essential to the operationalization of variables, fixes the meaning of experience-distant concepts in ways that preclude or make exceedingly difficult the interpretive research goal of understanding participants’ views. Recognizing this limitation, some positivist pilot studies look for ways to ameliorate this difficulty for some topics and to some extent. For example, rather than relying solely on their own terminology and on questions they craft themselves, some survey researchers hold focus groups before finalizing their questionnaires to try to learn the language that is meaningful to potential respondents; they then use that knowledge to frame better survey questions. From an interpretive perspective, however, such preparatory work does not foreclose the possibility that participants responding to the reframed survey questions will interpret the same words or phrases in different ways, whether from one another or from the survey designer's intended meanings, due to their own distinctive contexts.
It bears mention that an interpretive research focus on bottom-up concept development does not assume blind acceptance by researchers of what they are told. We discuss in Chapter 5 how researchers check their sense-making across multiple sources, seeking a sort of thickness by “mapping” the research setting to gain exposure to multiple perspectives on the research focus, thereby achieving a kind of “intertextuality” across sources of evidence. For instance, a researcher might map participants across the neighborhoods that make up the community that is the study's setting, or departments or hierarchical levels across a bureaucratic organization, seeking out multiple possible views on the subject of analysis, which may include contradictory narratives. This means taking participants’ views seriously—as authors of their own lives—even if the researcher, in analyzing them, also offers critical perspectives or insights into the tacit understandings and assumptions that underlie those views, including discussing conflicting views heard or read in various corners of the research terrain.
A focus on local knowledge and concepts, then, does not mean that interpretive researchers foreswear the use of experience-distant concepts. Researchers might coin their own experience-distant concepts, as called for by a specific analysis, or draw on those common within a particular research community in order to join the theoretical and analytic conversation taking place there. Still, bottom-up concept articulation and use is a key marker of interpretive research's commitment to embedding human participants’ meaning-making in its social, political, cultural, and historical contexts. The research ideal is to be able to use local concepts in everyday, adult ways, rather than tripping over them as a child or neophyte would. This extends to the particularities of everyday speech, as when Liebow quotes one of his corner men as saying, “. . . he stone took care of her and her children” (1967: 84) and then explains in a note that “stone” is “An intensive, in this case meaning ‘Really took care’” (1967: 84, n. 17).
But What of Hypothesizing? Constitutive Causality
Because of this commitment to meaning-making in context and the attendant engagement with participants’ concepts in situ, many key positivist design concerns are not relevant to interpretive research design. These include three that figure there centrally: (1) the a priori definition of concepts (just discussed); (2) assessment of the validity and reliability of the variables that operationalize them; and (3) construction of relationships among variables—whether in the form of null and research hypotheses (for simple bivariate relations) or as multivariate relationships (for a regression or other statistical or formal model).9 But if these concerns are not engaged, what happens to causality, and to hypothesis-testing that seeks to establish causality?
Methodological positivists seek to predict phenomena, ideally on the basis of a causal law or mechanism connecting specified independent variable(s) to the phenomenon under investigation (the dependent variable). Positivist methodology treats causality mechanistically: in what might be called a “billiard ball” understanding, investigation looks to see how one thing—in the analogy, a moving cue stick—leads to another—a ball that then moves on impact. Positivist-informed research seeks to identify that first thing in order to be able to predict ensuing reaction and, thereby, to control its movement or development or, if that is not possible, to move other things out of its way or prepare for its effects. This orientation is especially clear in Campbell and Stanley's (1963) influential book that extends the logic of experimental design from the laboratory to field settings.
There is considerable debate about “causality” within interpretive research communities (see Bernstein 1978, Taylor 1985) and, perhaps, no consensus. P. Jackson (2011), for example, articulates three versions that are distinctive from what we have termed “billiard ball” causality. L. Hansen (2006), by contrast, eschews all discussion of “causality.” In our view, methodological interpretivists seek understanding within specific settings: how the actors in them understand their contexts, explicitly and/or tacitly, and why they conduct themselves in particular ways. This “why” takes the form of “constitutive” causality,10 which engages how humans conceive of their worlds, the language they use to describe them, and other elements constituting that social world, which make possible or impossible the interactions they pursue. It is an effort at explanation that does not insist on producing abstract accounts of events. What it is after is explanation that rests on “descriptions of the . . . interweaving of codes [of meaning] in particular situations,” processes whose parts are contingent upon one another, rather than being “logically derivable from the codes themselves” (Hammersley 2008: 55).
Anderson's (2006) treatment of the power of nationalism, for instance, can be understood in this constitutive fashion—as the conception of an “imagined political community” for which millions have been willing to die. Or consider the rechristening of the estate or inheritance tax in recent US debates over tax policy as the “death tax” and the ways in which this reframed conceptualization has been a mobilizing idea for anti-tax activists (Luntz 2007; cf. Lakoff 2008). In an extended historical analysis, P. Jackson (2006) details how certain post–WWII German politicians drew on the rhetorical commonplace of the Abendland—the broad notion of Western civilization—to solidify their political position and alliances, while marginalizing their competitors. These are examples of a constitutive causality that seeks to explain events in terms of actors’ understandings of their own contexts, rather than in terms of a more mechanistic causality.11
These contrasting aims and understandings of causality, themselves embedded in distinctive conceptions of human agency, mean that hypotheses and hypothesizing are also understood quite differently. Because interpretive methodologies rest on local knowledge, interpretive research designs commonly do not specify formal hypotheses that a study is expected to falsify or support in a single, definitive test. Instead, the researcher's understanding of the relationships that in positivist-informed research would be articulated in the form of hypotheses is allowed—and expected—to develop over the course of the research project, as it unfolds. Whereas many forms of positivist research, say a laboratory experiment, require a completed design prior to generating the data that will test the hypothesis, interpretive research requires an iterative process of researcher sense-making which cannot be fully specified a priori because of its unfolding, processual character. Initial research expectations are treated as educated provisional inferences that will be considered and explored, rather than as formal hypotheses that will be “tested” in the narrower sense implied by the standard usage of “hypothesis testing.”
The Centrality of Context
Research logics and purposes matter. The logics of interpretive and positivist methodologies are distinctive, as are their intended purposes, and this means that their respective approaches to conceptualization, hypothesizing, and causality are, likewise, distinctive. As we have just discussed, the logic of interpretive inquiry— focused on meaning-making in context—requires researchers’ central attention to the concepts used by the human beings they study. This logic also means, consistent with the iterative character of hermeneutic sense-making, that researchers’ initial conjectures are assessed and reassessed in the field. Perhaps most striking, it entails a conceptualization of causality that is at odds with the conventional wisdom offered in the vast majority of methods textbooks.
Having laid this groundwork explaining why interpretive research does not begin with the same orientation toward knowledge as positivist research, we can now pick up the thread of interpretive research design and its own rhythms: establishing the setting(s), event(s), actors, and so forth among which and whom researchers will conduct their studies and, following that, the evidentiary character of that material.