What emotions do the words of a text carry? Find out in seconds.
Paste any text into the NRC Emotion Viz and see its emotion word usage come alive: how positive or negative it is, how it rises and falls from sentence to sentence, and how it compares with everyday language on social media or with a control text of your own. Every comparison comes with a statistical significance test, so you can tell a real difference from noise. Choose from six NRC lexicons and 17 affect and emotion dimensions, from valence, arousal and dominance to anxiety, warmth, trust, joy and fear. It is free, needs no installation or coding, and runs entirely in your browser.
Who can use it, and for what
- Psychology and affective scienceExamine emotion word use in diaries, interviews, open-ended survey answers or therapy transcripts. Compare groups or time points and study emotional variability across a text.
- Mental health and well-being researchTrack anxiety and calmness words, valence and their variability in personal narratives, as rough indicators at the level of groups, not diagnoses of individuals.
- Digital humanities and literary studiesTrace the emotion arc of a novel, play or poem. Compare authors, genres, periods or translations, and see which words drive each rise and fall.
- NLP and AIGet an interpretable, lexicon-based baseline. Inspect datasets before training, compare human-written and machine-generated text, and sanity-check model outputs.
- Corpus linguisticsCompare a target corpus against a reference corpus on emotion word frequencies and mean scores, with significance tests and effect sizes in place of raw counts alone.
- Computational social scienceStudy public discourse on social media, in forums or in news. Upload dated posts and follow emotion by hour, day, week, month or year.
- Political science and communicationCompare the emotional tone of speeches, manifestos, debates and press releases across parties, speakers or years.
- Journalism and media studiesCheck how headlines or coverage of an event are framed emotionally across outlets, and back a story with tested numbers.
- Social psychology and person perceptionMeasure warmth, sociability, trustworthiness and competence language in how people, groups or brands are described.
- Marketing and user researchSummarize the emotional character of reviews, feedback or brand messaging, and compare products or campaigns.
- Education and writingGive students a hands-on way to see tone and emotional pacing in their own writing and in the texts they study.
- Art and creative practiceShape the emotional pacing of a story, script or song lyrics, or use emotion arcs as material for data art and visual pieces.
- Health communication and public policyCompare the tone of public health messages, patient forums or policy documents over time.
- History and archival researchFollow emotion word use in letters, newspapers or parliamentary records across decades.
- Data scienceAdd interpretable emotion features to an exploratory analysis. Profile a text dataset quickly, compare segments with tested differences, and export the scores as CSV for your own models and dashboards.
- Amateur data enthusiastsExplore your favourite book, your group chat, your blog or the lyrics of an album, and download the scores as CSV for your own charts.
Commercial use requires a licence from the National Research Council Canada (NRC). Email the contact below and your request will be passed on to the relevant authorities at NRC.
What you can explore
- Six lexiconsNRC VAD Lexicon (v2.1 and v1), NRC WorryWords, NRC WCST (Words of Warmth), NRC Emotion Intensity Lexicon and NRC Emotion Lexicon.
- Affect and emotion dimensionsvalence, arousal, dominance, anxiety and calmness, warmth, sociability, trustworthiness, competence, and anger, anticipation, disgust, fear, joy, sadness, surprise and trust.
- Statistical significance testingevery target figure is tested against the control, with an optional correction for the number of tests and a summary table per family of dimensions.
- Controlsmillions of posts from X and Reddit, or your own text, CSV or JSON file.
- Viewswhole-text scores, scores by sentence, line or paragraph, emotion arcs, highlighted emotion words, date-based analysis and emotion words used with pronouns.
- Adjustmentsthresholds for which words count, and optional handling of negation and intensifiers.
Use with care
When using an off-the-shelf emotion word analysis tool, treat its scores as rough indicators of word usage, not definitive measures of what an author or reader feels. Results depend on the tool’s vocabulary, design, and the choice of target and control texts, so comparisons can mislead if those texts differ in length, genre, context, or language. Avoid using the output to label individuals or make high-stakes decisions, and explain the tool’s limits when sharing findings. The tool page lists the papers on ethical considerations and its terms of use.
Contact
Dr. Saif M. Mohammad, National Research Council Canada
saif.mohammad@nrc-cnrc.gc.ca