People struggling with symptoms of depression are less likely to perceive themselves as active initiators of their activities, which is directly reflected in the way they express themselves. Analysing the way people construct sentences can provide an early warning, according to an international study on postpartum depression initiated by psychologists from SWPS University.
[Depression] makes me completely incapable of doing things. (...) Even getting dressed feels like a challenge - these are the words of a patient quoted in a paper published in Depression and Anxiety . Such a profound sense of inability and powerlessness, which often leaves individuals almost entirely unable to act, is a common experience for people suffering from clinical depression. Similar symptoms are associated with postpartum depression; they may appear after the birth of a child.
Symptoms of depression, such as loss of motivation and diminished interest in activities, are often associated with a lower sense of agency. A high sense of agency, on the other hand, is associated with positive emotions and positive mood , emphasises the first author of the study, Marta Witkowska, PhD, a social psychologist from the Center for Research on Social Relations at the Institute of Psychology, SWPS University.
In the latest study, the researchers focused on semantic agency, which reflects the use of terms relating to a sense of control. For example, individuals characterised by a high level of agency write about "achieving goals" or "planning," while those with a reduced level of agency tend to use passive language.
While conventional methods of language analysis in the context of mental health primarily track emotional vocabulary, the method used by the researchers from SWPS University allows to capture of a broader spectrum of linguistic subtleties.
Analysis of posts on social media platforms
In two studies, the researchers investigated whether semantic agency could serve as a new marker for diagnosing symptoms of depression in the postpartum period. The investigators analysed a substantial dataset of over 17,600 posts on Twitter (now X) and over 3,000 posts on Reddit, based on the use of machine learning and expert evaluations, among other things. The platforms were selected deliberately - these are the common outlets where users share authentic, personal experiences. The study focused on postpartum women, a group particularly vulnerable to low mood.
The BERTAgent algorithm calculated the semantic agency score from -100 to 100 for each post. For example, the sentence "I had the overwhelming feeling of sinking deeper and deeper" scored -46.64, representing low agency, while the post "Months into my postpartum journey, I ran my first kilometer, and I couldn’t be prouder!” scored 56.14, representing high agency. The posts were also assessed for the use of specific emotion-related words and first-person singular pronouns. The experts also analysed randomly selected posts.
In both studies, lower semantic agency proved to be a reliable marker of low mood symptoms. It turns out that people with depressive symptoms are less likely to view themselves as active authors of their actions. The analysis shows that posts about personal, painful postpartum experiences are saturated with passive language.
While the link between a lower sense of agency and depression is well documented in the psychological literature, our results show that these patterns can be effectively detected with linguistic analysis , Marta Witkowska points out.
Improved mental health monitoring
As digital communication increasingly emerges as one of the dominant forms of interaction, identifying at-risk individuals based on their social media language presents a promising avenue for prevention and early intervention efforts, the authors of the paper believe.
In the long term, as the field of digital mental health continues to advance, the study results could strengthen our ability to more effectively detect and respond to the mental health needs of persons at risk of depression.
For example, intelligent systems could discreetly monitor the mental health of individuals at risk and suggest consulting a specialist if necessary. This could be a breakthrough in the field of existing mental health monitoring tools, especially in the case of postpartum depression, where prompt intervention is crucial for the health of both mother and child.
The authors caution, however, that before similar algorithms can be used in practice, future clinical research should examine whether analysing language for perceived agency allows to reliably distinguish between individuals suffering from clinical depression and healthy persons, or those experiencing temporary stress.
In addition to researchers from SWPS University, the authors of the paper " Semantic Agency Patterns Signal Depressive Experiences: Evidence From Postpartum Communication on Social Media " include experts from the Polish Academy of Sciences, the Nicolaus Copernicus University in Toruń, the University of Padua and New York University Abu Dhabi.
Depression and Anxiety
Survey
People
Semantic Agency Patterns Signal Depressive Experiences: Evidence From Postpartum Communication on Social Media
8-Feb-2026
The authors declare no conflicts of interest.