Eric is quite right about SOUNDX. It is Anglo-Amnerican and is taken from Don Knuth's <Art of Computer Programming>. It will probably mangle a lot of non-English-like sounmds, but it is still probably better than having *no* sound-matching at all. The received wisdom is that a 7-char domain works best (actually not the first 7 chars, nor the first 7 phonemes, but the first 7 voiced consonants). This may be a bit too much if we want to match 22 languages. If anyone has a better routine... Peter Flynn