Saturday, October 1, 2005

Banal Features Analysis

Alt. title: "Dull Feature Analysis."  Today I'm working on a small-time application of discourse analysis for Method~ologies.  We're looking at a corpus of eight student essays.  Initially, I considered how I would graph Bazerman's concept of "intertextual reach," which he defines as "how far a text travels for its intertextual relations" (89).  How far is that?  How do we account for the span of these traces--meters, leagues, years, decibels, lumens?  Maybe referential density could draw on network studies.  How?  We could establish a near intertextual reach as reference-gestures that share another source.  This would involve a triangulation of citations: Bazerman--let's say--cites Porter and Prior.  But Porter also cites Prior.  Porter is intertextually nearer than Prior (who does not cite any other source in common with Bazerman).  I'm making this up.  The far reach would describe the solitary reference--the singular text-trace that is not shared by any other source cited in the primary text (the text whose traces and reaches we are surveying).  But I wanted to think about intertextual reach as a quality that could be determined by triangulating citations.  Applied to a batch of student essays where works-to-cite are predefined, intertextual reach seems wobbly--a stretch, as in...look at how they reach alike.

I'll need something else. 

And so I got out all of my fingers and toes and went about counting commonplace features--dull features.  In her work on awk sentences and evidentials, Barton applies a method of linguistic analysis she refers to as rich feature analysis.  Rich feature analysis can lead to inductive (data-first) or deductive (theory-first) claim-making.

Rich features have both linguistic integrity (i.e., they are structural features of language, so they can be defined in linguistic terms and then categorized, coded, counted, and otherwise analyzed empirically) and contextual value (i.e., they can be conventionally connected to matters of function, meaning, interpretation, and significance).  The connection between a feature and its contextual value is a convention of language use.  In this method, then, the connection between structure and function is the primary focus of analysis. (66)

That's where I'm at for now--thinking through this stuff.  I'm tempted to complement the terms Barton emphasizes, but I'm just as inclined to make the case that dull features also have linguistic integrity and contextual value.  One distinction, perhaps, is that banal/dull features don't connect to "matters of function, meaning, interpretation and significance" in quite the same way as rich features.  Banal features are, perhaps, second-class features, in this sense; in conventionalized reading-for-meaning, they are there and yet not there--these features.  Yes, of course...I'm going to need a truckload of caveats to clear this up.

Posted by Derek Mueller at October 1, 2005 2:30 PM to Methods