Validity of LIWC in measuring personality expression via spoken language

• Main Authors: Goh, Bei Jun, Chen, Jiayu
• Other Authors: Lin Qiu
• Format: Student Research Paper
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Social sciences::Psychology
• Online Access: https://hdl.handle.net/10356/165905
• Institution: Nanyang Technological University

Past research on personality using the Linguistic Inquiry Word Count (LIWC) shows that in written language contexts, extroverts use more words from the positive emotion and social process categories. However, when applying LIWC to spoken language contexts, there were mixed or insignificant results. This suggests that the two LIWC categories may not be as valid in capturing extraversion in spoken language contexts. We propose that the weakened findings could be attributed to participants’ usage of voice cues as alternative channels for expressing extraversion i.e. mean pitch, pitch variability, mean intensity, intensity variability and formant frequency. We conducted two studies with two speaking tasks each, a confederate was added to the second study to prompt the expression of extraversion. Our results found self-reported extraversion to be insignificant with LIWC’s positive emotion and social process word use, mean pitch, pitch variability and formant frequency in both studies. Variability of intensity was significant with extraversion in study one while mean intensity found significance in study two. LIWC’s word count variable found positive significance in both studies. Interestingly, when independent judges coded for positive and sociable states, a significant positive correlation with participants’ extraversion was found in both transcripts and audio recordings for both studies. We conclude that while traditional methods using human judgement of personality might still be necessary, LIWC and machine audio softwares can be supplementary tools that provide quick and cost-efficient analyses. We also highlight the importance of establishing language-based measures of personality in spoken language.