# The relationship between language features and PTSD symptoms: a systematic review and meta-analysis

## TL;DR

This study finds that certain language features, like death-related words and anger, are linked to PTSD symptoms, suggesting they could help screen for the disorder.

## Contribution

The paper provides the first meta-analysis showing specific language features consistently correlate with PTSD symptoms across multiple studies.

## Key findings

- Death-related words and negative emotion words are positively correlated with PTSD symptoms.
- Body-related words are linked to hyperarousal, intrusive, and avoidance symptoms of PTSD.
- Word count and anger-related words also show significant positive correlations with PTSD symptoms.

## Abstract

The aim of this study is to investigate the relationship between language features and symptoms of Post-Traumatic Stress Disorder (PTSD) to determine if language features can serve as a reliable index for rapid screening and assessing PTSD.

A comprehensive literature search was performed using Pubmed, Embase, Cochrane Central Register of Controlled Trials, Web of Science, and Ovid databases, augmented by backward reference tracking, to gather pertinent literature concerning language features and traumatic stress disorders published until August 2024.

Twelve observational studies were included, comprising a cumulative sample size of 5,706 cases. Various language analysis tools, such as Linguistic Inquiry and Word Count (LIWC), manual coding, and machine learning techniques, were employed in the studies. Meta-analysis findings revealed a positive correlation between death-related words and PTSD symptoms (OR 1.32, 95%CI 1.10 to 1.59, I² 79.4%, p = 0.004), as well as significant positive correlations between negative emotion words and PTSD symptoms (OR 1.21, 95%CI 1.11 to 1.32, I² 30.5%, p < 0.001), anger-related words and PTSD symptoms (OR 1.14, 95%CI 1.11 to 1.17, I² 0.0%, p < 0.001), word count and PTSD symptoms (OR 1.20, 95%CI 1.09 to 1.31, I² 11.2%, p < 0.001). Additionally, a positive correlation was observed between body-related words and hyperarousal symptoms of PTSD (OR 1.26, 95%CI 1.15 to 1.37, I² 0.0%, p < 0.001), intrusive symptoms (OR 1.40, 95%CI 1.16 to 1.68, I² 0.0%, p < 0.001), and avoidance symptoms (OR1.29, 95%CI 1.21 to 1.37, I² 0.0%, p < 0.001). Death-related words (OR 1.16, 95% CI 1.08 to 1.25, I² 0.0%, p < 0.001) and word count (OR 1.18, 95% CI 1.10 to 1.27, I² 0.0%, p < 0.001) were observed positive correlations between intrusive symptoms of PTSD. Conversely, no correlation was found between the use of words related to sadness, anxiety, positive emotions, first-person pronouns, sensory, cognitive-related words and PTSD symptoms.

Death-related words, anger-related words, negative emotion words, body-related words and word count in Language features hold promise as a reliable indicator for rapid screening and assessing PTSD; however, further research is warranted to investigate their relationship with PTSD symptoms across various cultural contexts, genders, and types of trauma.

https://www.crd.york.ac.uk/PROSPERO, identifier CRD42024528621.