My own strategy has been to create a method of analyzing textual data qualitatively and with replicability Relying in the main on techniques developed in linguistic pragmatics, the method simulates a reading of a text, producing a model of the practical reasoning of its author and providing insight into the processes of worldmaking. The method forces analysts to identify explicitly every inference in the simulated reading. I do not propose to describe the method in great depth here. I lack the space. Readers unfamiliar with linguistic pragmatics may balk at some of the terminology in what follows. They should simply skip over the “jargon” or consult other work in which I present fuller descriptions of the method that do not presume familiarity with the terminology (Duffy 2008; 2012). For simplicity of exposition, I will assume the text under study to be a negotiation dialogue.
Briefly then, pragmatic analysis simulates a reading by creating an inventory of the explicit and implicit contents of the text. The implicit contents are those not stated by the speaker, but intended by the speaker to be inferred by the hearer. Here, my collaborators and I rely upon linguistic pragmatics, including speech act theory, which animates much of rule-oriented Constructivism.
We cast each proposition in the dialogue as a formal locution, recognizing that some speech acts are indirect. We inventory each proposition’s illocutionary and perlocutionary intent (which, following Bach and Harnish (1979), we term “reflexive intentions”). We also inventory any conversational implicatures. These are propositions that must be true in order for speakers to remain in conformity with Grice’s (1989) principle of cooperation and its associated maxims of conversation. Following contemporary practice, we treat what Searle (1969) termed the “felicity conditions” of speech acts as conversational implicatures. We also inventory any presuppositions. These are statements that must be true in order for a proposition to have a truth value. They do not follow deductively from the proposition because they hold both for the proposition and its negation. In dialogical contexts only, we also inventory any undischarged argumentative concessions and commitments of the conversants (Rescher 1979).
At this juncture, we posit an “action theorem,” or a proposition that describes a participant’s position at the conclusion of conversation. We then derive the action theorem from the propositions in our inventory If we cannot, we look for errors of commission or omission in our analysis. Once we are able to infer the action theorem from these propositions, we perform a sensitivity analysis, systematically removing individual proposition to see whether the derivation still holds without it. If not, that proposition is deemed crucial to the analysis. The sensitivity analysis typically is not needed when working with relatively small text, as the propositions crucial to the derivation are obvious to the analyst. From this crucial subset of propositions, we construct a syllogism that shows the derivation of the action theorem.
We consider this syllogism a model of the practical reasoning of the conversation participant. It might then count as part of a causal explanation. Of course, we may well have made an error. After all, it is only a simulated reading. We may well have introduced error because of the influence of our personal experiences on our inferences. Fortunately, the method is replicable, as it has forced us to write down every inference. Anyone who disagrees with our analysis should be able to point to the inference they believe is faulty. Thereby, the method focuses criticism on the actual point of disagreement.
When we first presented this work to the Peace Studies Society (International), game theorist Steven Brams asked how we knew whether the speech of the American or Soviet diplomats was not “cheap talk.” That is, how do we know whether their verbal productions were not disingenuous, designed to advantage the speaker strategically. We agreed that this was possible, but pointed out that speakers cannot always speak strategically because no one would wish to converse with them. But we did impose on ourselves the restriction that, in strategic contexts, inferences grounded in the Gricean maxim of quality (to be truthful) must be backed by evidence (e.g., consistency with statements in other contexts) of the truthfulness of the utterance. Subsequently, we showed that this quest to verify a speaker’s sincerity pertains to the participants as well as to the analysts (Duffy and Goh 2008).