Language Detector

Paste a sentence and see which language it is in. The detector first looks at the writing system: Hangul means Korean, kana means Japanese, Thai, Greek, Hebrew, Georgian and many Indian scripts each point to one main language. For text in the Latin or Cyrillic alphabets, which dozens of languages share, it compares the words against the most common words of each language and checks the special letters it uses, such as ñ, ß, ø, ł, ı or і. It covers nearly 70 languages.

How to use this tool

1. Paste at least a sentence; two or three work better.
2. Read the language and the confidence level.
3. Check the “most likely languages” table when two related languages are close.
4. The “writing systems” table shows which scripts the text mixes.

Formula

Score = share of words that are common words of the language + 0.35 × (special letters it uses − 2 × special letters it does not use) ÷ special letters

Each Latin- and Cyrillic-script language has a list of its 15–60 most frequent short words (articles, pronouns, prepositions) and the letters beyond the basic alphabet that it uses. A text full of “der”, “und” and ü is German; “och”, “är” and å point to Swedish. Chinese is split into Simplified and Traditional by characters that exist in only one of them (们 or 們), Japanese is recognised by its kana, and Arabic, Persian and Urdu by letters such as پ, گ and ے. The confidence level reflects how far the best language is ahead of the next one.

Example

The sample sentence, “Der schnelle braune Fuchs springt über den faulen Hund, und die Kinder lachen im Garten.”, is detected as German with high confidence: 6 of its 15 words (der, über, den, und, die, im) are among the most common German words, and its one special letter, ü, is used in German but not in English or Dutch.