Word Class Assignment (WCLA) in a Text-To-Speech (TTS) system is necessary for correct grapheme-to-phoneme conversion, and as a precursor for syntactic analysis (necessary for the calculation of intonation contours). Since the words of a language do not form a closed set (especially in languages like German and Dutch that can freely form new compounds), a lexicon will always be incomplete and thus must be supplemented by modules for morphological decomposition, and rule-based WCLA. Moreover, most words can have several word classes; thus, WCLA requires statistical and/or syntactical evaluation of a word class lattice.