Investigating the impact of linguistic features of text on readers’ phenomenological experiences.

Research suggests that individuals have a wide range of phenomenological experiences while reading; however, less is known about the factors that give rise to these varied experiences. The present study examined the relations between readers’ phenomenological experiences and the linguistic features of the texts they read. We combined data sets from seven studies that all employed a similar methodology, wherein individuals reported the characteristics of their thoughts at several points during reading (i.e., task-unrelated thought, visual imagery, verbal thinking, valence, personal connections). Natural language processing (NLP) techniques were then used to calculate theoretically driven linguistic features of the texts used in each study along multiple dimensions. Specifically, we examined how word concreteness, word frequency, valence, and cohesion were predictive of readers’ thought reports. Results indicated that (a) more frequent words were associated with less personal connections, (b) more concrete language was related to more visual imagery, task-unrelated thought, and positive valence but less verbal thinking and personal connections, (c) more cohesive language was related to more verbal thinking but less visual imagery, (d) positively valenced language was associated with less personal connections and visual imagery, and (e) more negatively valenced language was associated with more negative valence and verbal thinking but less personal connections. Overall, this study suggests that multiple textual dimensions can potentially influence and shape readers’ phenomenological experiences.