ATTRIBUTION OF MUTUAL UNDERSTANDING 395
In a BBC television interview in 1997, Jeremy Paxman
interviewed Michael Howard, the former Home Secretary in the
UK government.^38 The text, prior to treatment as described in
Part I, is included in Appendix A. The transcript of the Howard
interview is a relatively famous example of an interview in
which questions were asked directly, but did not yield
responsive answers. If the interlocutors were not both native
speakers of English, one might reach the generous conclusion
that Howard did not understand the question asked by Paxman.
Abstracting over the context, with knowledge of the roles
involved—news presenter and politician—one is more likely to
infer evasiveness on the part of the politician, since the language
used is not manifestly complex. In any case one can conclude
from the text of the discussion that the conversation did not
instantiate a collaborative flow of information. This conversation
is a useful one to show how the proposed measures fare in
assessing the level of mutual engagement exemplified. This is
demonstrated below.
The mean counts of shared tokens by levels of n are
provided in Table 3.
Table 3: Shared token counts in the Paxman-Howard interview
OtherShared SelfShared
n-bar 1 2+ 1 2+
Actual 4.71 1.94 3.04 .06
Randomized 4.24 1.58 2.88 .01
The statistical significance of repetitions is assessed by
comparing the proportions in token counts that were shared
between each utterance and its immediately preceding utterance
of the totals that could have been shared in each case. For
neither speaker, for neither level of n in the comparisons here is
(^38) See Interview by Jeremy Paxman with Michael Howard, former U.K.
Home Sec’y, Newsnight (BBC television broadcast May 13, 1997), available
at http://www.youtube.com/watch?v=Uwlsd8RAoqI.