--- title: "Language-Based Mental Health Insights: What Word Choices Reveal About Depression and Anxiety in High-Pressure Jobs" slug: language-based-mental-health-insights-2026-utsa-plos-one-911-dispatchers category: health category_label: "Health" author: "BrainWavePost Staff" date: 2026-07-10 tags: ["mental health", "linguistics", "natural language processing", "depression", "anxiety", "workplace wellbeing", "first responders", "911 dispatchers", "PLOS One", "UT San Antonio"] read_time_minutes: 8 canonical_url: https://brainwavepost.com/article/language-based-mental-health-insights-2026-utsa-plos-one-911-dispatchers source: BrainWavePost --- # Language-Based Mental Health Insights: What Word Choices Reveal About Depression and Anxiety in High-Pressure Jobs *Health · 2026-07-10 · BrainWavePost Staff · 8 min read* > A new peer-reviewed PLOS One study from UT San Antonio finds that the proportion of negatively valenced words 911 dispatchers use to describe stressful calls reliably predicts depression and anxiety symptoms — pointing to a new class of opt-in, language-based wellness tools for high-pressure workforces. > **How this article is sourced** _(info)_ > > Every claim below is drawn from peer-reviewed research and official university communications: PLOS One, UT San Antonio Today, npj Mental Health Research (Nature Portfolio), Cognition & Emotion (Taylor & Francis), and PubMed Central. Each statement is tagged with a numbered citation linking to the primary source. [1][2][3][4][5] A peer-reviewed study published in PLOS One and announced on 9 July 2026 by UT San Antonio has quantified something clinicians have long suspected: the specific words people use to describe stressful events can be an early, hidden warning sign of depression and anxiety — especially in high-pressure professions. [1][2] ## What the study actually measured The research team, led by psychology professors Vivian Ta-Johnson and Sandra B. Morissette, analyzed written narratives from 106 emergency call takers and dispatchers (ECDs) in collaboration with the San Antonio Police Department and Bexar Metro 911. [1][2] Participants first completed a standard self-report assessment for depression, anxiety, and stress, then typed detailed descriptions of a high-stress workplace event. [2] Their narratives were scored using natural language processing tools built on a validated database of roughly 14,000 English words, each rated on two dimensions of emotional reactivity: valence (emotional tone, positive to negative) and arousal (emotional intensity, calm to panicked). [1][2] - **106** — Emergency call takers and dispatchers analyzed in the PLOS One study [1][2] - **~14,000** — Validated English words in the emotional-rating database used for scoring [2] - **2 dimensions** — Emotional reactivity was measured as valence (tone) and arousal (intensity) [1][2] ## Negatively valenced words tracked with symptoms — intensity did not Narratives containing a higher proportion of negatively valenced words — the study cites examples such as 'blood,' 'danger,' 'hate,' and 'shoot' — reliably predicted more acute symptoms of depression and anxiety. Follow-up tests confirmed that individuals meeting clinical thresholds for moderate or higher depressive and anxiety symptoms used significantly more negative language than those below the threshold. [1][2] Surprisingly, arousal — the intensity of the language — did not predict anxiety, depression, or stress symptoms. As Ta-Johnson explained, the calls dispatchers describe are almost uniformly intense (suicides, medical emergencies, violence), so language reflecting intensity is 'to some extent, an unavoidable part of the profession.' [2] What varied — and what predicted symptoms — was tone, not volume. The findings also suggest positive and negative word choices operate on independent tracks rather than as opposites: highly negative language flagged mental health risk, but a lack of positive words did not, likely because chronically high-stress jobs suppress positive expression for contextual reasons rather than clinical ones. [2] ## Why this matters for a workforce that often falls through the cracks Emergency call takers and dispatchers face a well-documented occupational gap. They are routinely exposed to indirect, continuous trauma, yet in many jurisdictions they are not classified as first responders — limiting their access to dedicated wellness resources and funding. They also lack closure, rarely learning what happened to the callers they helped. [2] That combination — chronic exposure, structural under-support, and unresolved emotional processing — is exactly the environment in which a low-friction, opt-in screening signal could matter most. The study's authors envision brief periodic written reflections that language software could screen to flag individuals who might benefit from extra mental-health resources, without imposing on staff who prefer not to participate. [2] ## The wider evidence base for linguistic markers The UT San Antonio finding does not stand alone. A longitudinal mixed-methods study published in Cognition & Emotion tested how positive- and negative-valenced words in natural language, and their change over time, relate to depression and anxiety symptoms across 40 participants and 1,440 narratives — reinforcing that valence-based linguistic features carry clinically meaningful signal beyond a single occupational sample. [3] In parallel, a randomised cross-over trial published in npj Mental Health Research (Nature Portfolio) evaluated linguistic markers of depression and anxiety across eight different text-data types and frequencies in 218 symptomatic adults — an important step toward understanding which kinds of text (diaries, prompted reflections, transcripts) yield the most reliable linguistic markers for real-world deployment. [4] And a broader NLP study published in the Journal of Affective Disorders (available via PubMed Central) used natural language processing to identify patterns associated with depression, anxiety, and stress symptoms during the COVID-19 pandemic, showing that linguistic screening approaches generalize beyond a single crisis or profession. [5] ## What language-based tools are — and are not > **Educational summary — not medical advice** _(note)_ > > This article summarizes peer-reviewed research. Language-based screening tools are supportive indicators, not diagnostic instruments. They are not designed to replace clinicians, monitor employees punitively, or make decisions about care on their own. If you are struggling with your mental health, please contact a qualified professional or a local crisis line. [1][2] The UT San Antonio researchers are explicit that these tools are not meant to diagnose mental health conditions, monitor employees punitively, or replace human therapists. [2] Their intended use is proactive wellness: opt-in written reflections that a language model screens for elevated risk markers, so that additional support can be offered — an especially important design choice for first responders, who frequently avoid formal mental-health services because of workload demands and stigma. [2] Looking forward, the team plans to test whether changing how telecommunicators narrate their trauma — for example, writing in a more positive or past-tense frame — can influence how they process it over time, including as part of narrative writing treatment interventions. Partnerships with additional police departments and telecommunication divisions are in the early stages of being established to replicate and expand the work. [2] ## The bottom line Word choice — specifically the proportion of negatively valenced words in first-person accounts of stressful events — carries a measurable, replicable signal about depression and anxiety, at least in the high-pressure environments studied so far. [1][2][3][4][5] The immediate opportunity is not diagnosis by algorithm, but a new class of unobtrusive, opt-in wellness indicators that can help under-supported workforces — starting with 911 dispatchers — get to human help sooner. ## Sources (clickable) - [1] PLOS One — 'Linguistic markers of emotional reactivity and their association with anxiety, depression, and stress among emergency call takers and dispatchers' (Ta-Johnson, Morissette, et al.): https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0350551 - [2] UT San Antonio Today — 'Study finds word choice is linked to depression and anxiety symptoms in 911 dispatchers' (9 July 2026): https://news.utsa.edu/2026/07/how-word-choice-is-linked-to-depression-and-anxiety-symptoms-in-911-dispatchers/ - [3] Cognition & Emotion (Taylor & Francis) — 'Natural language sentiment as an indicator of depression and anxiety symptoms: a longitudinal mixed methods study' (2024): https://doi.org/10.1080/02699931.2024.2351952 - [4] npj Mental Health Research (Nature Portfolio) — 'A randomised cross over trial examining the linguistic markers of depression and anxiety in symptomatic adults': https://www.nature.com/articles/s44184-025-00140-y - [5] Journal of Affective Disorders / PMC — 'Using Natural Language Processing to Identify Patterns Associated with Depression, Anxiety, and Stress Symptoms During the COVID-19 Pandemic': https://pmc.ncbi.nlm.nih.gov/articles/PMC11927753/ --- _Canonical article: [https://brainwavepost.com/article/language-based-mental-health-insights-2026-utsa-plos-one-911-dispatchers](https://brainwavepost.com/article/language-based-mental-health-insights-2026-utsa-plos-one-911-dispatchers) — © BrainWavePost. Educational content; see the article page for full disclaimers._