RE: [TFUI] Re: Locale independence and grammatical structure

>-----Original Message-----
>From: yahoogroups-1e+8sjoZAN9Wk0Htik3J/[email protected] [mailto:yahoogroups-1e+8sjoZAN9Wk0Htik3J/[email protected]]
>Sent: Monday, 28 February 2005 12:02 PM
>To: [email protected]
>Subject: Re: [TFUI] Re: Locale independence and grammatical structure
>
>
>
>From: "Phlip" <phlip2005.at.gmail.com-hHKSG33Tihg+1LFaA5+Ji38/[email protected]>
>To: "[email protected]" 
><TestFirstUserInterfaces.at.yahoogroups.com-hHKSG33Tihg+1LFaA5+Ji+zKKks/[email protected]
>thjr.com>
>Sent: Sunday, February 27, 2005 1:16 PM
>Subject: Re: [TFUI] Re: Locale independence and grammatical structure
>
>
>>
>> Marc Brooks wrote:
>>
>>> > What you're describing - in English - is generally 
>considered a bad
>>> > practice by those with experience in localization. So rather than
>>> > adding complexity in order to allow pluralizing, changing 
>case, etc.
>>> > in other languages, I suggest you remove complexity by storing
>>> > every phrase separately.
>>>
>>> +1 or I8 ;)
>>
>> The OP (who is not here) described localizing a game. That may have 
>> included
>> the equivalent of RPG conversations. So the number of 
>individual strings
>> could undergo combinatoric explosion.
>
>That makes it really hard. Good natural language translation
>is still a long way off, and good translation for something like
>a game is even farther off.

Have you looked at John Sowa's book on Conceptual Graphs - http://www.jfsowa.com/cg/cgbib.htm ?
He described using a graph to represent the meaning of a sentence, which could then be processed by a language output module.  It's a while since I read the book, but I recall that he looked at the word order in a number of languages and showed that each one corresponded to a different algorithm for traversing the graph.

If you used his approach you wouldn't have to translate natural language, "only" synthesize it. Since his book is over 20 years old, I expect there will have been advances since then.

Simon
>However. If I had to do something like this, I'd investigate
>having the output be some kind of linguistic structure that
>would then go to a language (and cultural!) dependent
>module that would convert it to the target language. That
>same layer would handle input as well.
>
>I suspect this would still be in the area of a PhD research
>project, but if you had a relatively small specified set of languages
>and only the interactions in a game, it might be doable.
>
>>
>>> > efficiency is not that big a deal. Program complexity is. And the
>>> > time and effort of the localizers is definitely a big deal.
>>>
>>> +1 really, and localization is just as hard as internationalization.
>>> Trust us, we've done it.
>>>
>>> > Suggestion: Use comments in a text file - or an extra column in
>>> > a database - to tell the localizers how each string is used.
>>>
>>> Exactly right.  My ORM has explicit support for localized 
>strings for
>>> lookup tables and such, for precisely this reason.  It's so hard to
>>> graft on this stuff and make it SIMPLE, so "get ugly early"[1] and
>>> just do it right the first time.
>>>
>>> [1] http://www.softwareforyourhead.com/
>>
>> Oh, another thing: The game might be finished, in English.
>
>Game Over.
>
>John Roth
>>
>> -- 
>>  Phlip
>>  
>http://industrialxp.org/community/bin/view/Main>/TestFirstUserInterfaces
>>
>>
>>
>>
>> To unsubscribe, email:
>> TestFirstUserInterfaces-unsubscribe-hHKSG33TihhbjbujkaE4pw@public.gmane.org
>>
>> Yahoo! Groups Links
>>
>>
>>
>>
>>
>>
>>
>> 
>
>
>
>To unsubscribe, email:
>TestFirstUserInterfaces-unsubscribe-hHKSG33TihhbjbujkaE4pw@public.gmane.org
> 
>Yahoo! Groups Links
>
>
>
> 
>
>
>
>


To unsubscribe, email:
TestFirstUserInterfaces-unsubscribe-hHKSG33TihhbjbujkaE4pw@public.gmane.org
 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/TestFirstUserInterfaces/

<*> To unsubscribe from this group, send an email to:
    TestFirstUserInterfaces-unsubscribe-hHKSG33TihhbjbujkaE4pw@public.gmane.org

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/