Abstract: The present invention provides a facility for identifying the unknown language of text represented by a series of data values in accordance with a character set that associates character glyphs with particular data values. The facility first generates a characterization that characterizes the series of data values in terms of the occurrence of particular data values on the series of data values. For each of a plurality of languages, the facility then retrieves a model that models the language in terms of the statistical occurrence of particular data values in representative samples of text in that language. The facility then compares the retrieved models to the generated characterization of the series of data values, and identifies as the distinguished language the language whose model compares most favorably to the generated characterization of the series of data values.