Abstract: A system and method for generating native-sounding speech in a system prompt or in a text-to-speech processor or application is disclosed. A number of locales are defined representing particular languages, regions and dialects, and a set of locale-specific rules are defined for each locale. The locale-specific rules define how certain types of words, symbols, characters, punctuation and other elements should be interpreted. The locale-specific rules are used by a messaging system to generate native-sounding prompts that are played for callers. The locale-specific rules may also be used by a text-to-speech process to generate native-sounding speech for the user. The locale-specific rules may be implemented in a software class, program or application. The locale-specific rules incorporate a hierarchical structure in which rules that are common to a number of languages are grouped at one level and more specific rules are organized at lower levels in the hierarchy.