Is there a built-in way to detect the language of a text to automatically choose the correct stemmer for the Bayesian classifier?