Giving Computers Personality? Personality in Computers is in the Eye of the User - Coverbal Synchrony in Human-Machine Interaction - page 57

et al., 2005) and can therefore be considered a valid interpretive feature.
Our research focused on a psychotherapeutically oriented intervention,
and its influence on the language behavior described above.

￿ What effect does a system-initiated psychotherapeutically oriented

intervention have on the emotional state of a user?

￿ Can ego-involvement be increased by the implementation of an

intervention?

6.2 Methods and results

The transcribed interactions before and after the intervention phase
were examined, along with the dialectal speech and the overlaps
induced by the subjects. First, the transcripts were analyzed for
the verbosities of the individual subjects. The verbosities of the
subjects within those phases before and after the barrier did not differ
signifi cantly ( F = 1.48, p = .226) between the experimental ( M pre = 127.4;
M post = 141.9), and the control group ( M pre = 136.8; M post = 147.6). This
was advantageous for further analysis, because the intervention had
no effect on the verbosities.

Dialectal speech behavior

Code-switching between formal and dialectal speaking styles was only
expected for those people using dialectal language. Therefore, only
those subjects that produced at least one example of dialectal linguistic
behavior were included in the analysis ( N = 100). First, the number of
dialectal forms of language was divided by the total number of each
subjects' verbosity to calculate a percentage value. This enables a
between subjects comparison of the dialectal speech behavior. During
the corrective phase (stage 5, see Figure 2), immediately after the
subjects received either the empathic intervention or no intervention,
statistically significant differences were determined between the
experimental ( M = 1.4) and control groups ( M = 0.7). The subjects of
the experimental group produced significantly less dialectal speech,
whereas before the intervention phase the groups did not differ from
each other.

Overlaps

Again, the number of overlaps was scaled according to verbosity,
in order to create a percentage value. A distinct variance was found
between the determined overlap of the experimental and control
groups. The subjects of the experimental and control group not