A Computational Narrative Analysis of Children-Parent Attachment Relationships

Abstract

Children narratives implicitly represent their experiences and emotions. The relationships infants establish with their environment will shape their relationships with others and the concept of themselves. In this context, the Attachment Story Completion Task (ASCT) contains a series of unfinished stories to project the self in relation to attachment. Unfinished story procedures present a dilemma which needs to be solved and a codification of the secure, secure/insecure or insecure attachment categories. This paper analyses a story-corpus to explain 3 to 6 year old children-parent attachment relationships. It is a computational approach to exploring attachment representational models in two unfinished story-lines: "The stolen bike" and "The present". The resulting corpora contains 184 stories in one corpus and 170 stories in the other. The Latent Semantic Analysis (LSA) and Linguistic Inquiry and Word Count (LIWC) computational frameworks observe the emotions which children project. As a result, the computational analysis of the children mental representational model, in both corpora, have shown to be comparable to expert judgements in attachment categorization.