Automatically identifying humorous and persuasive language produced during a creative problem-solving task
Despite being an important component of problem-solving ability, relatively little is known about the linguistic features of creativity and the related constructs of elaboration, humor, and persuasion. In order to better understand the linguistic features of these constructs, two analyses were performed to examine relationships between linguistic features and human judgments of humor, persuasion, creativity, and elaboration. First, linguistic indices derived from automatic text analysis tools were used to predict human categorizations of utterances for humor and persuasion in a corpus of natural dialogue produced during a creative problem-solving and divergent thinking task. Four linguistic indices related to the use of function words, word age-of-acquisition, and spoken word frequency distinguished humor and persuasion with approximately 50% accuracy. These linguistic features, along with incidence scores for humorous and persuasive categories based on human ratings per dialogue were then used to predict human ratings of creativity and elaboration from the same corpus. Less variation in use of function words significantly predicted both creativity and elaboration scores, and lower spoken word frequency and higher incidences of humorous utterances significantly predicted creativity scores. These results demonstrate the potential for linguistic features to explain creativity and elaboration and highlight connections between humor, persuasion, and creativity.
|Conference||30th International Florida Artificial Intelligence Research Society Conference, FLAIRS 2017|
Skalicky, S. (Stephen), Crossley, S.A. (Scott A.), McNamara, D.S. (Danielle S.), & Muldner, K. (2017). Automatically identifying humorous and persuasive language produced during a creative problem-solving task. In FLAIRS 2017 - Proceedings of the 30th International Florida Artificial Intelligence Research Society Conference (pp. 282–287).