"There are various problems as regards language."
In his introduction to Wittgenstein's Tractatus, Russell distinguishes between "the problem as to what is the relation subsisting between thoughts, words, or sentences, and that which they refer to or mean" and "the question: what relation must one fact (such as a sentence) have to another in order to be capable of being a symbol for that other?" He takes the latter to be the basic problem of logic, and the subject of the Tractatus, and he counts the former under the problems of epistemology, which is the subject of this post.
Now, Wittgenstein focused on the logic of sentences. "Only in the context of a sentence does a word have meaning," as Frege originally put it. Well, only in the context of a paragraph, I would argue, does a sentence convey knowledge. The error of logical positivism, we might say, was to reduce the problem of epistemology to a logical problem—they read the Tractatus as a philosophy of science rather than a philosophy of language. Following Wittgenstein, they took the (true) sentence as a the unit of analysis.
When Foucault encouraged us to look not at propositions but statements he was opposing precisely this reduction. The study of "the dispersion of statements" rather than "the interrelation of true propositions" (Heidegger's phrase) improved our understanding of science greatly. But I wonder if it was very helpful to scientists themselves. The virtue of logical positivism was that it got scientists to think seriously about the individual truths they were expressing and the relationship between them. (The narrowness of their epistemology aside, one often hears that positivists are fantastic thesis supervisors. This doesn't surprise me.) I want to propose a unit of epistemic analysis that lies between the sentence and the statement: the claim.
A claim doesn't have to be true in any strict sense. And it doesn't have to be a viable element of discourse. It only has to be supported by the writer's knowledge and this support must be articulated in a prose paragraph around it. The paragraph, on this view, offers an excellent object of study for the epistemologist. We can see what is meant by "knowledge" in a particular field of research by looking at how published paragraphs are composed. What sort of support is offered for what sort of claim? "What," as Russell put it, "is the relation subsisting between thoughts, words, or sentences, and that which they refer to or mean?"
If I am right about paragraphs, then each paragraph is merely the tip of the iceberg of the writer's knowledge (much of which the writer shares with the reader because they are epistemic peers). The claim (expressed in the key sentence) is the apex of the tip of the iceberg. So, knowing where the peak of the tip is, as it were, we can extrapolate beneath the surface; we can take the "dignity of movement" of the paragraph as an indication of the depth of the knowledge that supports it.
But this support will be different in different kinds of text. A research article in the Administrative Science Quarterly is composed in a different way than a feature article in The New Yorker. Both may be thoroughly researched, and both writers may know a great deal about their subject matter. Still, how they know is different, and this difference ought to be apparent in the relative "composure" of the paragraphs that the article is made of. I'm at a conference in Barcelona the rest of this week, but I hope to be able to post an actual comparison of journalistic and academic paragraphs on Friday. Otherwise, it will have to wait until Monday.