>>The real challenge, of course, would be to get a list of words.
>
>Well, done this way there shouldn't be any speed issues:
>Why not feed word a few months of messages, gather the words marked as wrong, remove them from the word list build by all those messages and presto <g>.
>
>Just a special case of preprocessing - and you might implement a "learning behaviour".
>
>regards
>
>thomas
Yes, right - that will have to be done only once. I think that would solve Michel's problem. That is quite an interesting idea.
Difference in opinions hath cost many millions of lives: for instance, whether flesh be bread, or bread be flesh; whether whistling be a vice or a virtue; whether it be better to kiss a post, or throw it into the fire... (from Gulliver's Travels)