How WhatsApp Word and Emoji Analysis Works

Find repeated words without uploading your conversation

Chat Analyzer shows frequent eligible words and supported emoji occurrences from your selected WhatsApp records. Both calculations run in your browser. The results describe written content; they do not classify emotions, personalities or the quality of a relationship.

Word analysis uses ordinary nonempty participant text, excluding recognized media and event records. It normalizes case, removes recognized URL and email tokens, then filters a combined Turkish, German and English stopword list. It does not guess the conversation's language or merge word stems.

A tiny fictional word and emoji example

Imagine two invented messages: Lantern lantern trail 😊 and Trail lantern 😊😊. With a browser that supports the required emoji segmentation, the results below follow directly. Repetition inside one message counts more than once.

Why a word may be absent

Displayed top words must contain 2–40 Unicode code points and occur at least twice. The list and word cloud show at most 20 words, ordered by count with deterministic spelling tie-breaks. A word occurring once can contribute to eligible-word totals without appearing in that list.

Numbers, punctuation and emoji are not word tokens. Stopwords from all three supported lists apply together, so a token can be removed because it is common in another language. Some bare domains without a recognized URL prefix can remain ordinary text; this is filtering, not a guarantee that all sensitive terms disappear.

Date and participant filters change the selected text locally. Cross-participant disproportionate-word comparisons require additional evidence and are unavailable for a single selected label. An unavailable result is not a zero or a personality finding.

Review automatically included word cards before sharing

Emoji counts stay unavailable when the browser cannot provide the supported segmentation; InstaScope does not replace them with zero. Neither emoji nor word counts establish sentiment.

Wrapped automatically includes a small qualifying word selection as its first card when the evidence rules are met, so review every card before sharing. The two-message example above is enough for Analyzer word counts, but does not satisfy Wrapped's stronger word-card evidence rules. Chat Vault never stores lexical words or emoji glyphs.

The current Wrapped word gate needs at least 6 eligible text messages, 20 eligible word occurrences and 2 selected words that each occur at least 3 times. It shows at most 6 words after excluding original participant-label tokens from the bounded Analyzer top words. These evidence and filtering rules reduce some disclosures; they are not a sensitive-word detector.