Why we complain : a two-factor model of complaining in language use
Access full-text files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
When people experience dissatisfaction or frustration, they often express their feelings through complaining. However, very little is known about everyday complaints. In order to understand how people make complaints, this project proposed a two-factor model of complaint expression, with the factors of certainty and emotional involvement. There were two primary goals for this project. First, this project examined how people make complaints with different expectations, particularly in language use. Second, this project explored whether listeners understand individuals' expectations behind complaints. Four major complaint expectations were identified by content analysis in the pilot study (N = 276). Computerized text analysis was used to examine the relationships between language markers and the four complaint expectations. The factor of certainty was assessed by personal pronoun use and certainty words, whereas the factor of emotional involvement was assessed by the use of negative emotion words. Study 1 (N = 272) used multiple-choice questions to measure complaint expectations and replicated the language findings from the pilot study. Study 2 (N = 247) manipulated complaint expectations by experimental instructions to investigate language usage. The results suggested weak associations between manipulated expectations and language use. Study 3 (N = 204) focused on listeners and examined if they could identify the accurate expectations behind complaints. The results confirmed previous findings about the overconfidence effect in social behavior. An additional analysis examined the accuracy rate of computerized detection methods and then compared the computer's performance to human judges' accuracy. The results showed that the accuracy rate from the computerized text analysis was around 25% to 30%. Human judges performed slightly better than computerized text analysis with a 30% to 35% of accuracy rate. This is one of the first research projects that has attempted to detect and recognize human intentions surrounding complaining using language modeling.
Department
Description
text