Prosodic and multimodal markers
of humor in conversation 45
3.2
Pauses
We find that the mean length of pauses in the data was 0.70 seconds. Only one of
the four instances of humor that occurred with a within-turn-pause has a length
that was above the average pause and that by a mere .07 seconds. Three instances
of humor occurred as overlapping turns, and hence had no pause at all preced-
ing them. Hence it follows that at least half of the instances of humor were not
preceded by significant pauses. However, it would be misleading to think that the
remaining instances of humor were examples that support the theory that an in-
stance of humor is preceded by a pause that marks it, since they were all instances
of pauses between turns. Two considerations arise: 1) pauses between turns are
significantly longer that within-turn pauses, and 2) pauses between turns cannot
be interpreted as “marking” the following turn, unless the next speaker has been
selected in the previous turn. Finally, we should point out that none of the pauses
occur before the punch or jab line in the strict definition. In the broad definition
(i.e., the pause-based unit that contains the punch or jab line), all pause-based
units containing instances of humor will be preceded by a pause; therefore, it
would be meaningless to argue that our data support the pause-marking hypoth-
esis. Only pauses that precede immediately the phrase containing the punch/jab
line, above average length within-turn pauses, or exceptionally long pauses be-
tween turns could count as evidence of marking of the punch or jab line. Under
these restrictions, we found that none of the instances of humor was preceded by a
signaling/marking pause. It should be noted that Wennerstrom and Siegel (2003)
found an average pause duration slightly under 0.75 seconds between syntactically
complete units. Furthermore, Wennerstrom and Siegel (2003) report that pauses
of 1.5 seconds are not unusual between turns. Hence, the pauses in our conversa-
tion are not exceptional.
3.3
Pitch
We found that our two speakers tended to produce humorous turns at a very
slightly higher pitch than serious turns, on average. The difference between hu-
morous and non-humorous turns was not significant.
The differences between
serious turns and punch lines, jab lines, and irony taken individually were like-
wise not significant. The differences between speakers were also insignificant. See
Table 3/4, Hertz values. The zero figures in Carmen’s support reflect the fact that
she did not produce any.