Novel computational algorithms to index lexical markers of psychological distancing and their relationship to emotion regulation efficacy over time.

The degree to which one employs an objective or spatially/temporally distant perspective via language, i.e., linguistic distancing, has previously been shown to be positively associated with well-being. We sought to further elucidate relationships among language and emotion over time as a function of the implementation of sub-tactics of psychological distancing. In Study 1, we developed novel deep machine learning algorithms to identify the degree to which linguistic patterns reflect two types of psychological distancing, namely objective (OBJ) and spatial/temporal (FAR) distancing. In Study 2, in an expressive writing-based longitudinal emotion regulation training task, participants transcribed their thoughts while viewing negative or neutral stimuli over 5 sessions in one of three ways: by implementing objective language (Objective group), by implementing spatially/temporally distant language (Far group), or by responding naturally. We found that the OBJ and FAR algorithms significantly predicted changes in task-based self-reported negative affect in the Objective group and found no significant associations in the Far group. The relationship between the algorithm scores and self-reported negative affect was stronger in the Objective group compared to the Far group. These findings describe sensitive linguistic distancing algorithms that are capable of tracking changes in self-reported negative affect. These results may be useful in developing novel, unobtrusive emotion regulation assessments and interventions that utilize natural language processing.