Re: (Case FC2450947) Stemming Special Cases
Martin Porter <[email protected]> Sat, 24 May 2014 09:54:18 +0100
| Newsgroups | gmane.comp.search.snowball |
|---|---|
| Message-ID | <CALc6duzFb0eXDd2vXFTyXwbMeoutjOEnwemQdLiz1YEP3ksVTQ@mail.gmail.com> |
(Sorry for the terrible delay in replying which was caused by technical difficulties with snowball-discuss.) The paste / past error is interesting; I had never noticed it before. The -e is of course removed so that paste conflates with pasting. The alternative would be to restore the -e after removing the -ing from pasting and to leave paste invariant. That would not work if "past" was a verb. For example there is a problem with rout route which are distinct verbs in English. "routing" is the -ing form of both verbs, and therefore ambiguous. One could extend the definition of shortv to include "st" preceded by "a" and possibly preceded by consonants. Verbs affected are taste paste waste haste and no others I think. Again we're in luck, in that tast, past, are wast are not verbs in English. On the other hand tast, hast, wast, are hardly words of contemporary English, so it comes down to just paste / past again. Despite the idea of a general rule, it should be fixed by an addition to the exception lists. Martin