[basex-talk] Can the full text tokenizer be instructed not to tokenize at hyphens?

A customer wants to include composite terms such as 'third-generation' as single tokens so that they may be offered in a completion list. I don’t think this is configurable, or is it? Gerrit