406 JOURNAL OF LAW AND POLICY
proportions for self-repetition. Self-repetition effects are
significant for Murphy (the defense attorney questioning the
potential juror) and Carnes (the juror being questioned) in that
actual repetition exceeds repetition in randomized dialogues, for
both unigrams and 2+ grams (adjusted p ≤ .01). The effect also
exists for the court with respect to unigrams (adjusted p ≤ .01).
Actual repetition of 2+ grams is not significantly in excess of
randomized repetition for the court, and neither level of n-gram
yields significant actual self-repetition for Wolfe (the
prosecutor).
Figure 6: Allo-repetition in the Cunningham case: shared vs. nonshared
tokens by speaker in actual and randomized dialogue
DialogTypeOtherSharingSpeakerActualWOLFE
OSNOSMURPHYCOURTCARNESRandomizedOS NOS