Prototypicality and social influence

In this project, we aim to understand how people express the norms and values of the group – the group prototype – in the style in which they write.

Self-categorisation theory suggests that we enact the group prototype of a valued group identity when the social identity is salient. Research also suggests that group members that are seen as prototypical for the group are particularly popular and influential (e.g., as leaders). However, it is currently unclear whether and how members of online communities enact a shared identity through the use of a particular group prototypical style, and whether using such a style makes posts more influential. Our research fills this gap by:

(1) assessing whether linguistic style is related to the prototype of social groups

(2) examining whether posts written in a group prototypical style are more influential in online communities

 

 

Does linguistic style reflect distinct group prototypes?

Here we examine whether differences in linguistic style between online communities reflect the group prototype. We are using natural language processing techniques, combined with machine learning (e.g., extra random tress) and multidimensional scaling (MDS), to test whether groups that share a similar prototype according to the classification by Deaux et al. (1995) also share a similar linguistic style.

Pre-registration: https://osf.io/dac5p

Are posts written in a group prototypical style more influential?

In this part of the project, we examine whether posts written in a group prototypical linguistic style are more influential in online forums than those written in a less prototypical style. For this analyses, we use naturally occurring data from clearnet and darknet online forums, and natural language processing and social network analysis techniques.

Cork A, Everson R, Levine M, Koschate M. (2020). Using computational techniques to study social influence online. Group Processes & Intergroup Relations, 23(6), 808-826. doi:10.1177/1368430220937354 (open access)

---

These questions are examined by Alicia Cork as part of her PhD project, supervised by Miriam Koschate-Reis, Richard Everson and Mark Levine.

Funding: EPSRC-NPIF studentship “Detecting and preventing criminal network involvement from digital footprints” (1929614; industry partner: NCA)

Alicia is affiliated with the Alan Turing Institute as part of an Enrichment Scheme placement.