Passive Linguistic Patterns in Social Media Posts Serve as Early Warning Signs for Postpartum Depression

Passive language on social media is a reliable sign of postpartum depression, according to new research using machine learning algorithms.

By: AXL Media

Published: May 1, 2026, 8:26 AM EDT

Source: Information for this report was sourced from EurekAlert!

The Correlation Between Sentence Structure and Mental State

The way individuals construct their self-narratives provides a significant window into their psychological well-being, particularly regarding their perceived ability to influence their own lives. Researchers from SWPS University have found that individuals experiencing symptoms of depression are far less likely to describe themselves as active initiators of their own activities. While traditional linguistic analysis often focuses on emotional vocabulary, this new study shifts the focus toward semantic agency, which tracks the use of terms related to control and intent. According to Marta Witkowska, PhD, a high sense of agency is strongly associated with positive mood, whereas its absence is a hallmark of the helplessness felt during clinical depression.

Leveraging Machine Learning for Digital Mental Health

To test the reliability of language as a diagnostic tool, investigators analyzed a massive dataset comprising over 20,600 posts from platforms such as X and Reddit. The researchers employed a specialized algorithm known as BERTAgent to calculate agency scores for each post, ranging from -100 to 100. This machine learning approach allowed the team to distinguish between active language, such as "achieving goals" or "planning," and passive language that reflects a feeling of being overwhelmed. By utilizing platforms where users share authentic and personal experiences, the study was able to capture the raw linguistic subtleties of women navigating the postpartum period.

Semantic Agency as a Precise Diagnostic Marker

The analysis revealed that lower semantic agency is a consistent and reliable marker for low mood symptoms. Posts characterized by low agency frequently included phrases describing feelings of "sinking" or an inability to perform basic tasks like getting dressed. In contrast, high-agency posts focused on accomplishments and milestones, showing a clear connection between active verbs and emotional resilience. This finding suggests that even when a user does not explicitly use words like "sad" or "depressed," the passive structure of their sentences can signal a deep-seated loss of motivation and a diminished interest in daily activities.