Language Recognition#

Detect the dominant language of a string or rank candidate languages by confidence.

#Functions

#NaturalLanguage.dominantLanguage(text: string): Language | null

Return the single most likely language for text. Returns null when the text is too short or ambiguous to classify.

NaturalLanguage.dominantLanguage("Bonjour le monde")   // → "fr"
NaturalLanguage.dominantLanguage("今天天气很好")       // → "zh-Hans"
NaturalLanguage.dominantLanguage("")                   // → null

#NaturalLanguage.languageHypotheses(text, options?): LanguageHypothesis[]

Return up to maximumCount candidate languages sorted by confidence (descending).

• options.maximumCount?: number — defaults to 3.
• options.constraints?: Language[] — soft preference; recognizer favors candidates in this set but is not strictly limited to them.
• options.hints?: { [language: string]: number } — prior probabilities (BCP-47 → 0...1).

const ranked = NaturalLanguage.languageHypotheses("Buenos días", {
  maximumCount: 3,
  constraints: ["es", "pt", "it"],
  hints: { es: 0.8 }
})
// → [{ language: "es", confidence: 0.92 }, { language: "pt", confidence: 0.05 }, ...]

#Notes

• Language identification is offline and extremely fast — feel free to call it on every keystroke if needed.
• For very short input (one or two words), the recognizer may return null or low confidence. Provide more context when possible.