[prev in list] [next in list] [prev in thread] [next in thread] 

List:       aspell-user
Subject:    [Aspell-user] forbidden words
From:       ge <eleonora46 () gmx ! net>
Date:       2006-06-30 19:56:59
Message-ID: E1FwP7D-0004V1-1A () lists ! gnu ! org
[Download RAW message or body]

Kevin,

I created an aspell dictionary for Hungarian with 870 thousand words.
At building it (using make) however, I get lots of messages like:
Warning: Removing invalid affix 'w' from word Bonaparté.
The flag w means, this word, whenever appears is marked as bad.

This might look a nonsense for you, saying: "Why does then the word appear at all in \
the list, if it is wrong?" However, this makes sense. an example:
There is a Hungarian word: öl (kill)
öl is then conjugated with all the verb prefixes, like 
megöl
elöl
kiöl
átöl
etc....
The creation of all prefixes makes sense, since there are at least 20 of them, and \
adding all verbs with the used prefix would be a lot of work with little result. In \
this case there is only a small possibility, that this will create wrong words. \
However, unfortunately sometimes it does:

There is no such Hungarian word, as elöl, but there is a word elől (ahead),
that needs to be written with ő and not ö. 
Therefore we simply say elöl/w, and that causes to mark the erroneously
spelled elöl word as wrong.

Therefore I consider /w as a useful flag for languages,  that have to generate
lots of word forms.

FORBIDDENWORD w
in the affix file signalizes, that in this affix file /w flag means a forbidden word.

The usage of /w is even broader a little bit. If there are other flags together with \
/w, all word forms, that are created by the other flags are considered as invalid, \
bad words.

There are lots of other examples for the usefulness of the /w flag in a word set. \
Usage of /w saves in a lot of cases additional flags, and flags are a valuable, \
limited resource (max 255) in case of aspell.

What is the behaviour of aspell, and what do you think about the above mechanizm?

Thanks, Eleonora


[prev in list] [next in list] [prev in thread] [next in thread]