A Lexical Analysis of the Child Attachment Interview in an Adolescent Sample: Markers and Correlates of Borderline Personality Disorder

Lexical analysis is a simple tool in understanding how one’s language reflects their inner world. The present study uses this methodology to investigate how adolescents with Borderline Personality Disorder (BPD), or BPD characteristics, speak about their attachment relationships. Given the theoretical and empirical evidence of borderline pathology being a developmental concept, with its roots in the attachment relationship, this methodology, which has not been widely used in investigations of BPD, and not at all in studies of adolescent BPD, has the potential to shed some light not only on the subjective experience of family and attachment in adolescents with borderline pathology, but also add support to previous findings regarding the concordance of BPD phenomenology between adolescents and adults. Transcripts of Child Attachment Interviews (CAI) conducted with N=301 adolescents, split into three groups (BPD, Healthy Controls, Clinical Controls), were analyzed using Linguistic Inquiry and Word Count software (LIWC). An ANCOVA framework was used to test for group differences, and Pearson’s correlations to test for dimensional relationships between lexical categories of interest and score on a continuous measure of Borderline Symptomology (BPFSC). Results yielded several significant differences between the healthy control group and both clinical groups (words related to anxiety, sadness, death), some significant differences only between the BPD group and the healthy controls (we), and two significant differences specific to the BPD group (words related to positive emotions, anger). There were no significant differences in total word count, first-person singular pronouns, or words related to insight. Dimensional analyses revealed similar patterns, with positive relationships between BPFSC total score and first-person singular pronouns and words related to anxiety, anger, and death and negative relationships between BPFSC and first-person plural pronouns and positive emotion words. Again, there was no significant relationship with word count or words related to insight. Implications of these results are discussed and placed within the context of the larger Borderline and lexical analysis literatures.