Inductive learning of pronunciation rules with the Depes system

The development of grapheme-to-phoneme rules by means of a highly flexible software environment called Depes is discussed. The first version of the Depes system proved useful for manual rule development. The current version provides tools to facilitate and automate the inductive learning of pronunciation rules. As input for the automatic inductive learning of rules, the system only requires a list of words (each word in its orthographic and phonemic form). The result of the iterative induction process is an ordered list of pronunciation rules. This list is an ideal starting point for further manual rule development and refinement.<>