What Is Emotional Tagging in a Journal App?
"Just name what you're feeling" is common advice, and it turns out to be backed by more than intuition. Emotional tagging in a journal app takes that idea and automates the first step — reading what you wrote and suggesting a label for the feeling behind it — so the benefit of naming an emotion doesn't depend on you remembering to do it yourself.
Gratitude Bestie uses an on-device Core ML model to suggest an emotion tag for each journal entry, drawn from a 15-category taxonomy — Gratitude, Joy, Peace/Calm, Pride, Hope, and more — entirely on your device, with no journal text ever sent to a server.
The research behind "name it to tame it"
In 2007, UCLA psychologist Matthew Lieberman and colleagues published a brain-imaging study in Psychological Science examining what happens when people put an emotional reaction into words. Viewing emotionally charged images activated the amygdala, the brain region central to processing fear and emotional intensity. But when participants labeled the emotion in words — simply choosing a word that described how the image made them feel — amygdala activity decreased, while activity increased in the right ventrolateral prefrontal cortex, a region tied to cognitive control. Putting a feeling into words appeared to measurably dampen the emotional reaction itself, not just describe it after the fact.
From labeling to tagging
Journaling already does half of this naturally — writing about an experience is itself a form of labeling. Emotion tagging adds the second half: a specific, named category attached to the entry, rather than a vague sense of "that was a good day" or "that was rough." Over many entries, those tags become data you can actually look back on, which plain free-form writing doesn't offer on its own.
Why on-device matters here
Journal entries are often the most personal thing someone writes. Running the classification model directly on the device — rather than sending entry text to a cloud service for analysis — means the tagging step never requires your writing to leave your phone. The tradeoff is that the model is more limited than a large cloud-based language model would be; the upside is that privacy isn't a design afterthought.
Where Gratitude Bestie fits
Every entry in Gratitude Bestie gets a suggested tag from a 15-category set — Love, Affection, Happiness, Excitement, Pride, Hope, Relief, Peace/Calm, Gratitude, Trust, Compassion, Inspiration, Amusement, Curiosity, and a neutral fallback — computed on-device as you write. The Trends screen then turns those tags into a donut chart of your emotional mix over time, and Memory Lane can resurface entries by emotion, not just by date. See the full feature breakdown for how tagging, Trends, and Memory Lane work together.
FAQ
Does putting feelings into words actually change anything, or is it just a saying?
It's measurable, not just a saying. A 2007 brain-imaging study by Matthew Lieberman and colleagues, published in Psychological Science, found that labeling an emotion in words reduced activity in the amygdala — the brain region driving the emotional reaction — while increasing activity in a prefrontal region associated with cognitive control.
Is on-device emotion tagging the same as sentiment analysis?
They're related but not identical. Basic sentiment analysis usually just scores text as positive, negative, or neutral. Emotion tagging goes further, classifying text into specific named emotions — like gratitude, pride, or peace/calm — which maps more directly onto what affect-labeling research actually studied.
Can an on-device model read my journal accurately every time?
No system is perfect at inferring emotion from text alone, and an automatic tag is a suggestion, not a diagnosis. Its main value is turning something you'd otherwise have to label yourself into a low-effort default — you can always write past what the tag captures.
Free, on-device, and private by design.