It is shown that grammatical inference is applicable to natural language processing. Given the wide and complex range of structures appearing in an unrestricted natural language like English, full grammatical inference, yielding a comprehensive syntactic and semantic definition of English, is too much to hope for at present. Instead, the authors focus on techniques for dealing with ambiguity resolution by probabilistic ranking; this does not require a full formal Chomskyan grammar. They give a short overview of the different levels and methods being investigated at CCALAS for probabilistic ranking of candidates in ambiguous English input.