Properties of a coding system for traumatic memories

Introduction Narrative studies have focused on the language used by the individuals to describe stressful or traumatic experiences. Hence, linguistic procedures have been applied aiming to obtain information about autobiographical memories and trauma processing. However, there is a general lack of agreement about how to measure narrative aspects. Software programs for this purpose are limited, since they don’t capture the language context, and systems based on judge's rates are not free of subjective biases. Objectives This study presents a coding system developed to analyze several language categories related to traumatic memories and psychological processes. Structural aspects (e.g., coherence) and content dimensions of traumatic narratives (e.g., emotional or cognitive processes) are measured. Each narrative aspect is coded by raters using both dichotomous (presence/absence) and numerical values (Likert scale). Aims To propose a structured coding system for traumatic narratives that considers the language context and maximizes consensus among different raters. Methods Traumatic narratives from 50 traumatized women and stressful narratives from 50 non-traumatized women have been evaluated according the system developed. Three blind raters coded each narrative. Results Inter-rater reliability data are provided for the different narrative categories. The agreement between raters is discussed for both structural and content language domains. Conclusions The analysis of the inter-rater reliability allows exploring subjective biases in assessing different structural and content language dimensions. This study advances in the development of a procedure to analyze autobiographical narratives in a valid and reliable way, with a special focus on traumatic and other unpleasant memories.