34 Introduction
the study. In addition to performing in-class coding and observation, the other
coders carried out group interviews with students in several of the schools.^10 To
avoid possible variation by subject matter, we studied only Contracts classes. I chose
to study first-semester, first-year classes to catch the socialization process at a critical
moment; it is during this first semester that students receive their primary initia-
tion into distinctively legal language and thought. The Contracts course in these
eight schools was a required class, and students were assigned to sections, so that
their own preferences and choices were not an influence on their selection for a
particular class. Although there was one exception, we generally included only
professors with significant amounts of teaching experience at their law schools, so
as to maximize any influence the institution might have on teaching norms and
styles.
As noted, the classes were first taped and coded by in-class coders. The in-class
coders noted the identity of each speaker in terms of race and gender, whether a
turn was called on or volunteered, and any other particularly salient aspects of the
interaction that might not be captured on tape. For example, in one case a student
responded to a professor’s question by putting her head down on her desk; this
nonverbal response would be noted on the coding sheet. Coders also tracked any
notes the professor put on the board.
The tapes were subsequently transcribed at the American Bar Foundation in
Chicago. Using tapes, transcripts, and in-class coding sheets, each turn in each class
was then coded for the following information:
- Speaker identity (students were assigned numbers in order to track repeat
players across classes). - Speaker gender.
- Speaker racial identification.
- Order of appearance number (for each class, students were assigned a number
based on the order in which they spoke in that class). - Linguistic type (turns were coded in terms of the kind of speech occurring:
monologue, various kinds of dialogue).^11 - Kind of turn (called on without volunteering, volunteered and called on,
spoke without being called on, etc.; spoken, nonverbal, or silent turn). - Length of turn (timed to half of a second).
- Evaluation (whether the turn contained a positive or negative evaluation of the
preceding turn). - Speaker transition type (for turns that overlapped but did not interrupt
previous speech, we coded whether they were facilitative backchannels [as
when someone comments “right,” or “mmm hmm” in the background while a
primary speaker is talking], or other kinds of background comments).
As with the earlier, in-class phase of the project, the transcript coding for each
school was overseen by a single coder.^12 Extensive cross-checking of the coding was
performed by proofreaders who checked the coding sheets, data enterers who en-
tered the coding sheets into the computer, quantitative analysts who performed
the data cleanup and statistical analyses, and the two project managers.^13 We could
and did, on multiple occasions, go back to the tapes and transcripts to double-check