[liblouis-liblouisxml] Re: Backtranslation problem
- From: lars@xxxxxxxxxxxx (Lars Bjørndal)
- To: liblouis-liblouisxml@xxxxxxxxxxxxx
- Date: Tue, 20 Jan 2009 15:09:21 +0100
Hello, John!
Thank you very much for your information about the allround/debug
tools. Thanks to these tools and to your explanation, I found the problem
with my personal table regarding the "ei"/elleri" problems. It was due
to litdigit be defined too early in the table.
There are another problem, however, that I'm not able to slove. This
is also reproducable with the english en-us-g1.ctb: If you load this
table and run "50-ring", you get problems with backtransatlin:
"50-r9n7". I would like to get "50-ring", of course. Is this a problem
with the engine or with the tables?
Thanks!
Lars
"John J. Boyer" <johnjboyer@xxxxxxxxxxxxx> writes:
> Lars,
>
> Here are the results I got from lou_debug and lou_allround. I have
> gone through the steps in some detail for those who may be unfamiliar
> with these two tools.
>
> To see what is happening in back-translation in lou_debug you must use
> the dot pattern of what you are interested in. The word eller translates
> to dots 15. I called lou_debug with No-No.g2.ctb. At the command: prompt
> I typed p (enter). This took me to the particular: prompt. Here I typed
> d (enter) for dots. At the -> prompt I typed 15 (enter). This showed the
> definition of the dot pattern 15 together with the rules associated with
> it. The rule for eller occurred twice, but there was nothing else
> unusual. Trying the dot pattern 24 for the letter i also produced
> nothing unusual.
>
> Then I tried lou_allround. At the command: prompt typing t (enter)
> brought up a request for a table name. I typed No-No-g2.ctb (enter) and
> got back to the command: prompt. Here I typed r (enter) for run. I then
> tried both eller and ei. Both were translated and back-translated
> correctly. I then pressed enter with nothing on the line to get back to
> the command: prompt. Here I chose b for back-translation only. After
> confirming this, I again tyred r (enter) at the command: prompt. I then
> tried just the letter e and it was back-translated to eller. I then
> tried ei. It was back-translated to ei.
>
> I next looked at the table in a text editor. There are rules for eller
> at lines 18 and 146. I also noticed that forskjeller is defined as both
> begword and endword at lines 206 and 207.
>
> Maybe this will give you some clues.
>
> John
>
> On Mon, Dec 29, 2008 at 11:57:46AM +0100, Lars Bj�rndal wrote:
>> Hello!
>>
>> John, I think this is food for you.
>>
>> I have a really strange issue regarding backtranslation, that I cannot
>> solve by myself.
>>
>> I'm using liblouis-1.4.0, and liblouisxml-1.7.0 at the moment. One
>> table that can be used to reproduce this, is No-No-g2.ctb.
>>
>> If I backtranslate the word "ei", which is a Norwegian word by itself,
>> it's backtranslated into "elleri", e.g. the letter "e" is
>> backtranslated into "eller" as it should have been if the "e" was
>> alone, and the letter "i" is added. The word "eller" is defined as
>> word in the table, not begword. If I try to backtranslate "ef", this
>> does not happen.
>>
>> The same happens with "bli". "bl" is backtranslated into "blant", and
>> the "i" is added. "blant" is defined as word too.
>>
>> I've read the tables up and down, but I cannot understand why this
>> should happen. Hope you can help me!
>>
>> Lars
For a description of the software and to download it go to
http://www.jjb-software.com
Other related posts:
