A graph-oriented computational approach to the grapheme-to-phoneme transcription of Italian written texts is described. The letter-to-sound transcription rule system is built up taking into account the implicit phonotactic constraints of the Italian language. The system follows the mathematical model called Finite States Automata (FSA), which is generalized and augmented to obtain a simple syntax-directed translation schema. The whole set of rules described and implemented is often associated with a particular set of exceptions. These rules are entirely based on the pure orthographic form of the words, and no level of grammatical knowledge of their classes is considered. After describing the rules in their traditional notation and formalism, we consider both the graph-oriented point of view relative to the transcriber implementation and some examples of the transcription procedure. Finally, the advantages and disadvantages of the graph-oriented approach will be shown and discussed, together with future trends and improvements.
