Using natural language processing to identify the rhetoric of deception in business and competitive intelligence email communications.

摘要

This study extended previous work that identified linguistic-based cues (e.g. an increased use of modifiers), through the use of natural language processing techniques, as effective discriminators between truthful and deceptive messages in negotiative email communications. The extension of the previous research included linguistic-based cues as they related to management informative tasks comprised of business and competitive intelligence information, rather than negotiation-based information sharing. A follow-up survey captured demographic and socioeconomic and sociocultural information about participants. Such types of information have previously been shown as having the potential to influence participants' language use. The results of the study indicated that many of the linguistic-based cues that effectively discriminate between deceptive and truthful message senders differ in both type, and in many cases, direction, by genre discourse (negotiative versus informative), though some commonalities were found. Participants in this study also showed differences in affective characteristics associated with their levels of comfort in composing either truthful or deceptive messages. The confidence levels of participants also varied by management information types. Finally, when considering demographic (sex) and socioeconomic (income level) variables across informing conditions (truthful, deceptive), differences were found within several of the linguistic-based cues both independently and as a result of the moderating effects of sociocultural and/or demographic variables. These findings provide evidence of (1) the need to consider the genre of the writing when trying to detect deception; (2) the potential affective influences of deceptive writing tasks; (3) an area in higher education, forecasting, which may need to be remediated; and (4) the need to consider demographic and sociocultural information in the selection of cues to apply in detecting deception in business and competitive intelligence emailed communications.