Lexicon Development for Speech and Language Processing

Frank Van Eynde's book examines how computational dictionaries are built, filled and kept up to date for systems that process spoken and written language. Its opening chapter maps the field of computational lexicography and introduces the core concepts and debates that the later chapters explore in depth.

The next two chapters look at how hand-built lexicons are organised, first covering grammatical information such as word classes and morphosyntax, then sound-based information such as morphophonology. Both use a declarative, constraint-based approach and show how general patterns can be captured formally to make a lexicon more consistent and less repetitive.

Two further chapters take the opposite route, extracting lexical data automatically from text corpora with inductive, data-driven methods. They also cover tokenization, lemmatization and shallow parsing. The final three chapters turn to speech synthesis and speech recognition.