Because some speech recognition and text-to-speech technologies are better at some languages than others, you can use different technologies for processing different languages. You specify one technology for “all locales” and then override this for the exceptional languages. This can be specified on a per node or per application basis. Specifying the technology for a new language or switching between technologies for any language is done without altering the application itself.
If you are using language-only locales, make sure you have speech plug-ins configured for all locales or the single-part locale.
The current application locale is assumed for all speech recognition attempts, and is used for text-to-speech unless specified within the application. The current application locale for VoiceXML 2.1 applications is either the operating system locale or the application default locale, if one has been specified. The current application locale for Java applications is the node default locale or the application default locale, if one has been specified.