Understanding classroom lectures

After hearing three lectures as part of a regular university course, students were asked to summarize the lectures. The lectures were graded for accuracy and completeness. Each one was taped, transcribed, and coded, using a coding system based on work by van Dijk and Kintsch (1983) and Meyers (1975a, 1975b). The hypotheses to be tested were: (1) The percentage, but not the number of propositions would correlate positively with the accuracy scores in the summaries. (2) Completeness scores would correlate positively with the total number of propositions, but the number and percentage of propositions coded as macropropositions would correlate even more highly. (3) Signaled propositions would be better remembered than nonsignaled ones. Hypothesis 1 was confirmed in the summaries of all three lectures; Hypotheses 2 and 3 were confirmed only in two of the lectures. Analyses of the errors, lexical substitutions, discrepancies in the grades assigned by the two coders, and differences in the students' scores for the different lectures revealed interesting (and in some cases unexpected) results concerning lexical-semantic networks and grading reliability. Finally, implications for classroom lecturing are discussed.