| Pages in topic: [1 2] > |
How to make the source texts more reusable? Thread poster: Jia Deng
|
|---|
There are 10 weather forecasts to be translated. I put the first five in one Word document and the next five in the other. Before translation, the TM is empty. As we know, the lanaguage of weather forecasts is different. There are a lot clauses like: "Cloudy this evening and tonight with rain, heavy in places. " I edited the source texts. I made all this knid of clauses into complete sentences, for example, "The weather will be cloudy this evening. The weather will be cloudy tonight. There will be rain. The rain are heavy in places." I used many rules of conttrolled language in order to make the source texts more TM-friendly. Then I translated the first five weather forecasts. When I translated the next weather forecasts, there were not as many matches as I expected. Why? What can I do besides lowering the fuzzy match threshold? Are weather forecasts suitable to be translated using translation memory?
| | | |
John Fossey 
Canada
Local time: 12:23
Member (2008)
French to English
| More info needed | Dec 9, 2010 |
Perhaps you could say a little more about what you are trying to do. Your profile says nothing except that your account type is student. Are you trying to learn about translation memories?
Are you using the TM for a small job? TMs are more useful once a significant volume has been accumulated. What is the purpose for which you are using translation memory?
What language pairs are you working in? Some languages segment better than others.
With a little more information about what you are trying to accomplish you might get a better response.
| | | |
| More information | Dec 9, 2010 |
Thank you very much for your reply. I am a student. This is my assignment, which requires us to translate weather forecasts with Wordfast. I have tio think about how to make the source text more TM--friendly and reusable. The aim of this assignment is to see whether students understand well how TM works. Students are also required to give some suggestions on how to write translation memory-friendly weather forecasts.
| | | |
| language pair | Dec 9, 2010 |
My language pair is English---Chinese
| | |
|
|
|
I wonder what "how to make the source text more TM friendly and reusable" means.
Why do you have to edit the original source text?
| | | |
Jia Deng wrote:
I used many rules of controlled language in order to make the source texts more TM-friendly.
I don't think there is much more you can do. Use short sentences with a standard structure, always use the same words for the same concept. Not much else I can think of. Of course, one would need to see your files and procedures to make sure that all your segments are entered into the TM and the correct TM is used appropriately on the new document (segmentation etc.), but you're probably doing it right.
I don't have any experience in this area, but I do know weather forecasts are the standard example for TM-friendly (and machine translation friendly) texts.
See http://en.wikipedia.org/wiki/METEO_System
| | | |
Erik Freitag
Germany
Local time: 18:23
Member (2006)
Dutch to German
+ ...
| Don't arbitrarily change the source text. | Dec 9, 2010 |
One more thing that comes to mind:
It is probably not a good idea to arbitrarily change the source text. If the question is: What should the source text look like in order to improve TM leverage, my answer would be: It should look like the texts the TM was actually built from. If you build a TM from actual weather forecasts, it will very likely contain segments like "Cloudy this evening and tonight with rain, heavy in places". Chances are that it will not contain many segments of the "The rain are heavy in places" kind.
Kind regards,
Erik
| | | |
| changing source texts | Dec 9, 2010 |
efreitag wrote:
It is probably not a good idea to arbitrarily change the source text.
We all seem to be stuck in the translator mindset, when this question/assignment is not from the translator's perspective. Of course this is a translation website so that's not too surprising. (Although I remember somebody on here mentioned that he is involved in controlled language, I think he even teaches people how to write controlled langage. Perhaps he will drop in.)
It seems quite clear to me that this school assignment is designed to give a bit more perspective: to see in practice how source texts and the use and usefulness of CAT tools in the translation process are interrelated. This is specifically related to scenarios where you (well, the client) can control the source language (some technical manuals and indeed weather forecasts are written in controlled language to make translation/processing/interpretation easier and more reliable).
[Edited at 2010-12-09 10:35 GMT]
| | |
|
|
|
Fernando Larrazabal wrote:
I wonder what "how to make the source text more TM friendly and reusable" means.
Why do you have to edit the original source text?
To make the source the more TM friendly and reusable means to edit the source texts in order to see more repetitions and matches when I analyse them in PM perspective and and tranlate them. I used the rules of controlled langauge.
| | | |
| Thank you very much for your replies. | Dec 9, 2010 |
Thank you very much for your replies. I will go on my assignment and tell you the result.
| | | |
Samuel Murray
Netherlands
Local time: 18:23
Member (2006)
English to Afrikaans
+ ...
| On using controlled language | Dec 9, 2010 |
Jia Deng wrote:
I used many rules of conttrolled language in order to make the source texts more TM-friendly.
Controlled language would be good for glossary matching, but not for TM matching. TM matching is based on the entire sentence. For example, if you translate these two sentences:
There will be rain in the morning.
It will be cloudy today.
...and then translate this sentence:
It will be cloudy today, with rain in the morning, but it will clear up later.
...you will probably not get a fuzzy match at all because each of the previous two sentences make up less than half of the current sentence, and very few CAT tools offer matches lower than 50%.
However, if you had added "rain in the morning" and "cloudy today" to your glossary while you translate the previous sentences, then you could have use the glossary insert function when you arrived at the third sentence (depending on your target language).
Also, there is no need to edit the source text -- if you want the translation to read more naturally, you can make the translation more natural even if the source text is unnatural. No-one is doing to see the source text, so you don't have to change it before you start translating.
| | | |
Erik Freitag
Germany
Local time: 18:23
Member (2006)
Dutch to German
+ ...
| misunderstanding | Dec 9, 2010 |
FarkasAndras wrote: efreitag wrote:
It is probably not a good idea to arbitrarily change the source text.
We all seem to be stuck in the translator mindset, when this question/assignment is not from the translator's perspective. Of course this is a translation website so that's not too surprising. (Although I remember somebody on here mentioned that he is involved in controlled language, I think he even teaches people how to write controlled langage. Perhaps he will drop in.) It seems quite clear to me that this school assignment is designed to give a bit more perspective: to see in practice how source texts and the use and usefulness of CAT tools in the translation process are interrelated. This is specifically related to scenarios where you (well, the client) can control the source language (some technical manuals and indeed weather forecasts are written in controlled language to make translation/processing/interpretation easier and more reliable). [Edited at 2010-12-09 10:35 GMT]
I may have not expressed myself quite clear. I was indeed trying to answer from the OP's point of view, not a translator's one.
What I was trying to say was that in order to improve TM leverage, the source text should be written in such a manner that it matches the language used to construct the TM. I doubt that "The rain are heavy" will be heard often in weather forecasts written by people with some knowledge of the English language.
In other words: at least some of the "many rules of conttrolled language in order to make the source texts more TM-friendly" that the OP used seem to be somewhat questionable. It is no surprise that this doesn't improve TM leverage.
| | |
|
|
|
Heinrich Pesch
Finland
Local time: 19:23
Member (2003)
Finnish to German
+ ...
| Restrict choice | Dec 9, 2010 |
First you should go a step backwards and create a set of statements which may be used in those forecasts:
Rain at night.
Rain in the morning.
Rain all day.
etc.
Then you can translate these and will get only 100 % matches as long as you stick to the set of statements.
But with Computer Aided Translation tools there is always the danger of false matches. The shorter a segment the more it is bound to deliver false matches or no match at all.
| | | |
NMR (X)
France
Local time: 18:23
French to Dutch
+ ...
|
| my assignment | Dec 9, 2010 |
Thank you very much ofr your replies. My assignment won't be judged on the quality of the TTs. The assignment is about editing the STs in order to them more TM-friendly. Then give suggestions on how to write translation memory-friendly weather forecasts.
| | | |
| Pages in topic: [1 2] > |
Login to reply/comment 
Translation news related to Wordfast