Speech content analysis in feigned depression

The present study investigated the utility of employing computerized speech content analysis software to detect feigned depression in psychologically healthy adults. Twenty-two subjects between the ages of 19 and 54 years old, who scored lower than 10 on the Beck depression inventory (BDI. 1996), were asked to speak freely in response to a neutral stimulus. Participants performed this task in pseudo-randomized order for two experimental conditions: a control state and feigned depression. Computerized transcript analysis data were then subjected to paired t-tests to detect differences based on participants' lexical choices under the two conditions. Results indicated that one semantic category, tool, was significantly different between the feigned depression and control conditions. Another semantic category of interest, arousal, demonstrated a strong trend toward significance. Computerized lexical choice categories previously found to accurately identify persons with depression were not found to be significantly different in the discourse of our participants in the feigning condition. These results are discussed in terms of the use of computerized speech content analysis as a method for possibly identifying persons who may be intentionally feigning depression.