Language Detector

How it works

First the writing system is identified from the characters (Latin, Arabic, Cyrillic, Greek, Hebrew, Devanagari, Chinese, Japanese, Korean, Thai…). Scripts used by a single language identify it directly. For shared scripts, the detector counts each language's most frequent function words and characteristic letter combinations, and normalizes them into confidence scores. It needs a sentence or two: single words are often ambiguous.