Sentiment Analyzer 🔒 Runs locally in your browser — nothing is uploaded.

Lexicon-based text sentiment scorer

⚠ English-only, lexicon-based heuristic — NOT AI. This tool looks up each word against a small, hand-built list of ~200 English sentiment words and applies a few simple rules for negation ("not good") and intensifiers ("very good"). It runs entirely in your browser: no model, no API call, no training data. It cannot understand sarcasm, irony, long-distance negation, mixed tone, or context the way a human (or a real NLP/AI model) would — treat the result as a rough, directional signal, not a verdict.

Paste some English text above to see its sentiment.

About this tool

This sentiment analyzer uses the same broad approach as classic lexicon-based tools like AFINN and VADER — but with its own original, hand-built word list rather than any published dataset. Every word in the input is looked up against roughly 200 common English sentiment words, each carrying a valence score from -5 (very negative) to +5 (very positive). Two simple context rules refine the raw score: a negator ("not", "never", "isn't") within three words flips and softens the sign of the word that follows, and an intensifier ("very", "extremely") directly before a word scales its magnitude up.

All matched scores are averaged and normalized into a single "compound" score between -1 and +1, then classified as Positive, Negative, or Neutral using fixed thresholds. Because this runs as plain word lookup plus a couple of rules — not a trained AI model — it is fast, fully private (nothing leaves your browser), and fully transparent: the breakdown table below shows exactly which words were matched and why their score changed. It also has real limits worth keeping in mind: it only understands English, it cannot detect sarcasm or irony, and negation more than three words away from its target word will be missed entirely.

Frequently asked questions

Does this work for languages other than English?

No — the lexicon behind this tool is a hand-built list of English words with sentiment scores, so it only recognizes English vocabulary. Text in other languages will show mostly "unmatched" words and an unreliable near-zero/neutral result, not because the sentiment is actually neutral but because none of the words are in the lexicon. There is no plan to add other languages' lexicons in this tool as-is; each language would need its own carefully hand-built word list, not just a translation of the English one, since word connotation and intensity don't translate directly.

Can it detect sarcasm?

No — this is a fundamental, well-known limitation of ANY lexicon-based sentiment approach, not just this tool. A sentence like "Oh great, my flight got cancelled again" is dripping with negative sentiment to a human reader, but scores positive here because "great" is a strongly positive word and the tool has no concept of tone, context, or irony. Detecting sarcasm reliably genuinely requires a contextual language model, which is explicitly out of scope for this lightweight, transparent, lexicon-based tool.