This is an exploration of how 'knowledge of socially organised environments of concerted actions [or] "commonsense knowledge of social structures"' (357 ) is actually developed [the phrase in quotation marks is based on the work of Schutz]. Such knowledge will account for common culture, and it is also a resource for social scientists. Commonsense knowledge offers descriptions of the society 'that its members, sociologists included' [original emphasis, page 357] develop in order to manage and communicate decisions. It is usually taken for granted, but its workings can be revealed by experiment.

The documentary method of interpretation is common to both commonsense and sociology, and refers to a way of joining together past present and future events in order to produce a coherent interpretation. In particular, it can be used to assemble a body of knowledge of social structures --'decisions of meaning, facts, method, and causal texture' (358). Mannheim [sic] offered a description of the documentary method of interpretation as 'the search for "... an identical, homologous pattern underlying a vast variety of totally different realisations of meaning"... treating an actual appearance as "the document of," as "pointing to," as "standing on behalf of" a presupposed underlying pattern' (358). The process works in several ways, and the underlying document is both derived from and used to interpret new events. We all use this method every day to interpret words and actions, and sociologists use it too, as when Goffman detects underlying 'strategies for the management of impressions... [or Merton detects] types of deviance... [or when the US Census constructs] occupational categories' (358). The work of the documentary method can be revealed in an experiment 'designed to exaggerate the features of this method in use and to catch the work of "fact production" in-flight' (359).

[Briefly, 10 undergraduates were told that they would be talking to a student counselor, and were advised they could ask a series of questions which would attract either 'yes' or 'no' answers. The counselor was to provide the answers from an adjoining room. In fact, 'yes' or 'no' answers were provided at random. Subjects heard the answer, and then recorded their comments as they attempted to interpret it. They then recorded their overall impressions, and were subsequently interviewed.

Garfinkel provides unedited transcripts from two actual case studies. It is difficult to summarise these, obviously, and you are urged to read them for yourself. In general, what they show is that the subjects of the experiment made every effort to interpret these random answers in a way that made coherent sense to them, even when the answers provided an initial surprise. They are also extremely forgiving of what appeared to be contradictions in the answers, and seemed to find some genuine meaning in them.]

What the experiment showed is summarised under 11 headings:

(A) All the subjects managed to get through and interpret what they had heard

(B) All the answers were interpreted as answers to questions

(C) The subjects' follow-up questions were responses to the [random] answers, so that a sense of topic was preserved

(D) Subjects were able to interpret answers as making sense of their questions, even when the same answer was given to quite different questions, or when some question was seen to be implied in the answer, even though it had not actually been asked

(E)... 'incomplete, inappropriate and contradictory answers' were managed in various ways -- later answers seem to clarify them, inadequacies were seen as a result of the efficient method, reasons for inappropriate answers were sought, the adviser's motives were interpreted, as therapeutic, for example Garfinkel notes that some students did become suspicious, however, and some apparently 'temporarily withdrew their willingness to continue' (369).

(F) ...'throughout, there was a concern and search for pattern... perceived from the very beginning'

(G) Subjects contributed 'actively and heavily' to the advice they had received, often by assigning 'a scenic source' to random answers [that is seeing in the advice 'the thought formulated in the subject's questions' (369)]. "Upon being told about the deception the subjects were intensely chagrined', but still struggled 'in most cases' to salvage what had happened as a genuine form of advice.

(H) A certain vagueness was maintained, despite the opportunities for clarification in the exchange

(I) Subjects relied on their memberships, referring to the 'features of the collectivity as a scheme of interpretation' -- to establish the good character of the advice, for example. 'These social structures consisted of normative features of the social system seen from within' (370). These assumed social relationships emerged from the activity of interpretation. In particular, they assumed there was a shared a body of commonsense knowledge about such matters as 'family, school, home, occupation': 'These evidences and the collectivity features were referred back and forth to each other' (371)

(J) Subjects constructed a 'reasonable' account of what they had done, by relying on these assumed norms. They assumed that what the adviser had said was an example of a class of events, comparable to past and future events, and agreed with estimates of likelihood and conditions of occurrence, and occupied a possible location in a set of means-ends relationships. Making the advice fit with these norms also made it seem true.

(K) Subjects undertook the work of documenting to manage the responses, but this was not a simple matter of applying rules, rather to do with developing the advice to make it fit commonsense knowledge.

Much sociology uses the documentary method too, in community studies, or Survey Research, or life history, wherever selections are used to describe underlying patterns: these selections might be examples from interviews or numerical tables. Documentary methods may be used informally or far more rigorously. Readers of sociological reports also have to fill in the inevitable gaps in them in the same way. The relationship between actual observations and the underlying occurrence is a 'sign relationship, the reader must consult some set of grammatical rules to decide this correspondence... some theory of the intended events on the basis of which the decisions to code the actual observations as findings are recommended' (373). This is achieved by relying upon the knowledge of social life that the reader and the writer share, as members of 'a community of cobelievers' (373). Therefore, documentary methods are always employed, even in the most rigorous and factual studies.

Sociologists report that what they are doing is seeing behind surface appearances or actual occurrences. In the case of the subjects of the experiment it is more a matter of coming to terms with the situation, making the unknown known. For sociologists specifically, documentary methods involve the following qualities:

(1) An interest in connecting future events to present ones. How to bring this future about is 'characteristically vague or unknown' (374). Thus researchers applying coding rules to a set of replies have in mind a likely future distribution, although it is common to retain a certain vagueness permitting future decisions [weird example!].

(2) There are alternative paths to a future state, and these are 'characteristically sketchy, incoherent, and unelaborated' (374). For example, the task of managing rapport between interviewers and interviewees is commonly managed by 'a set of ad hoc tactics for adjusting to present opportunity... [rather] than of acting in the deliberate and calculated realisation of a plan' (375).

(3) Official methodological sequences are often sketched in and rationalised retrospectively, especially at the stage of writing an article for publication [including Garfinkel too with this article?].

(4) Investigators are often unable to predict the effects of their actions until they actually get involved.

(5) Researchers can decide post hoc that a particular outcome was the main goal of the activity after all.

(6) Investigators often find their goals becoming clarified as they proceed.

(7) There is a characteristic lack of perfect information which forbids precise calculation of effects or consequences.

(8) Researchers' actual procedural know-how is rarely codified. Nevertheless, researchers must produce something that looks reasonable. Inviting colleagues to comment upon the reasonableness of findings implies some some of check against the above features which govern professional documentary work.

Sociologists therefore rely upon commonsense at crucial moments of interpretation, where some correspondence between observations and intended events is being constructed. This is concealed by certain features such as:

(1) Some rational statement of a goal to be achieved which contains criteria which suggest that the investigation has been adequate

(2) There is some [post hoc?] story showing how the research has progressed through a series of states in order to reach the anticipated state, the goal.

Ideally, the goal itself under procedures to achieve it should be specified in advance, so that some logical precedence over actual observation is being claimed. This then permits a 'prediction', or an 'operational definition' to be used to describe the actual observation. Against this ideal, the documentary method looks erroneous and distorting, and sociologists sometimes claim to be making progress away from it. A 'flood of textbooks on methods was written to provide remedies for such situations' (377). They claim to offer advice on developing calculable procedures, mathematical models, statistical schemes of inference and the like. 'Immense sums of foundation money, criteria defining adequate research designs, and many careers rest on the conviction that this is so' (377). [Some interest in the workings of the academic community and the outside world after all then? How does all this happen though?]

'Yet it is common knowledge that in the overwhelming number of researches that are methodological acceptable, and, paradoxically, precisely to the extent that rigorous methods are used, dramatic discrepancies are visible between the theoretical properties of the intended sociological findings... and the mathematical assumptions there must be satisfied if the statistical measures are to be used...'(378). It is common to see statistical measures as somehow indicating findings rather than literally describing them -- which reintroduces the documentary method and the norms of reasonable argument.

Are there not some procedures to reduce the worst aspects of documentary methods? Restrictions on ex post facto analysis, procedures to perform follow-up studies? In other words, can the documentary method be modified or even eventually replaced for sociology? One alternative is simply to stick with literal observation and description, but this reduces the sociological importance of the study. Another option is to try and describe the workings of the documentary method as it affects sociological evidence, to describe the 'rules that actually govern the use of... methods... and making... intelligible [the production of sociological facts]' (378).

Commentary

Reading this piece again after many years reminds me of how important this sort of criticism actually is for sociology. In some ways, Garfinkel's criticisms have had a considerable impact, seen best of all, perhaps, in the ways in which some ethnographers in particular have attempted to disclose their actual rules of procedure. For more statistical work, it still seems common for crucial matters of interpretation to be left implicit, or hidden away in some of the technical appendices.

My re-reading also prompted me to think of more critical remarks about Garfinkel's work. Apart from the irritating pseudo-technical style of much of it, which I have protected you from, dear reader, it strikes me as being pretty unreflexive itself. Is ethnomethodology exempt from a reliance on the documentary method and its implicit commonsense decisions and interpretations, for example? Did Garfinkel's list of features of sociological and commonsense understandings really arise at the end of a series of procedures of investigation, or was it the other way around?

There are also the other obvious problems: the huge generalisations about sociological studies, the curious reliance upon 'common knowledge' about the flaws in these studies, and the considerable amount of 'soft quantification' when attempting to lend significance to the results of the experiment (using terms such as 'most', 'in most cases', and so on). You might also be able to see something in Gouldner's criticisms: the sadism of tricking students ending in their 'intense chagrin'; the solemn elaborations of a couple of observations into 11 'rules' (Why 11? Can we reduce these, say, to 5 main ones? Why are some examples put under existing rules and not made into rules of their own?).