Grammatical Inference a Magical Mystery Tour

How do children learn language? Is it already in their heads when they are born, or do they acquire language through experience? Is it possible to program a computer to learn a language using techniques which are possibly available to the human infant? Grammatical Inference strives to automatically infer a grammar for a language from a subset of the strings in the language, and possibly also a set of strings not in the language. There are many useful applications which will beneet from a general Grammatical Inference Engine, such as speech and image recognition, natural language processing systems, and database query engines. In this paper I give a general overview of the relatively new eld of Grammatical Inference, starting from its application in a speech recognition system. I cover most of the commonly used techniques, and discuss some of the objections raised by linguists to these techniques. In a subsequent paper I will describe my project in detail and will cover the more philosophical aspects of this eld, such as the human language acquisition debate. These papers will hopefully serve to illustrate the power of proba-bilistic approaches to Linguists and Engineers alike.