How language detection works
Every language has characteristic letter patterns: English is full of “th” and “ing”, German of “sch” and “ein”, Spanish of “ión” and “que”. The detector uses franc, an open-source library that compares the three-letter sequences (trigrams) in your text with profiles of 82 languages built from translations of the Universal Declaration of Human Rights. It runs entirely in your browser.
You get the most likely language, a confidence indication, the closest alternatives, and which writing systems appear in the text, useful for mixed-language documents.
Getting reliable results
- Length matters. A few words can match several languages; a sentence or two is usually enough.
- Related languages are hard to tell apart: Spanish and Galician, Danish and Norwegian Bokmål, Malay and Indonesian, Bosnian, Croatian and Serbian (in Latin script). Check the alternatives list.
- Names, numbers and code carry no language signal; remove them if results look odd.
For scripts used by a single language (Greek, Georgian, Armenian, Thai, Korean Hangul) the script itself is a strong clue and is shown separately.
Worked example
Here is what the Language Detector returns for its built-in example. Click Try an example in the tool above to load it, then change it as you like.
La lecture est à l’esprit ce que l’exercice est au corps. Elle nourrit la curiosité et ouvre des portes vers d’autres mondes.| Result | Value |
|---|---|
| Language | French |
| Confidence | High |
| Words | 22 |
| Main script | Latin |
Questions
How many languages can it detect?
82 languages, from the most widely spoken to many regional ones, using the franc library.
Why was my text detected as the wrong language?
Very short texts and closely related languages are the usual causes. Add more text or check the list of close matches.
Last reviewed · How Textkiln tools work