TIP # 38 - This tip was sent on the week: 1st - 7th May, 2005
This tip applies to the following programs:
ESI Professional and the Split Function - Part V
Last week we translated the example:
“I am using the new word processor”
If some of you actually tried translating this sentence in ESI Professional, you probably wondered why the results that you obtained were so different.
ESI automatically translates this sentence as
Estoy usando el nuevo procesador de texto.
Big surprise! What happened with “palabra nueva”? What happened with “neologismo”? Where did the word “texto “ come from?
What actually happens in this particular sentence is that ESI “sees” the MWE “new word” and recognizes that it is in its database, but at the same time “sees” that keeping this MWE would interfere with “processor”, that it would not be compatible. Therefore it decides to SPLIT the expression even before you have a chance to read it. Almost simultaneously, it detects that – after splitting the MWE “new word”--- a second WME “word processor” can be formed and does not interfere. Thus, it ”decides” to take on this second cluster instead of the former one.
Here is a sequence of what ESI “sees”:
i. “I am using the new word + processor”
ii. “I am using the new + word + processor”
iii. “I am using the new + word processor”
ESI does this series of processes following a very complex set of rules and making use of our proprietary semantic markup in the database. In order to handle MWEs in this way, Word Magic Software had to invent a new grammar system specifically put together to handle word clusters instead of single words, or we should say, besides handling single words. Thus, ESI Professional uses standard grammatical rules plus Word Magic’s own Word Cluster Grammar. It is like working with “word molecules” instead of single “word atoms” in the construction of each phrase.
The advantages of doing this are quite evident: many a time ESI Professional is capable of rendering translations only expected from a human professional. It does have its disadvantages though: Word Cluster Grammar is still “under construction” and probably will require years upon years of constant development, a task which we continuously do here at Word Magic on an everyday basis.
That’s why ESI needs User intervention once in while, to SPLIT a MWE when its rules fail to do so, to select a different word more in line with the context of the sentence… or so many other details and nuances that a good translation job requires. This is why Word Magic has incorporated the Interactive System into ESI Professional: to facilitate this interaction and allow the User to be always in control.
As you might expect, not always is ESI capable of splitting a MWE correctly. When it fails to do, or when it splits the “wrong MWE”, you would have to use the SPLIT button in order to mend the output translation.
In the following weeks we well see a few more examples of correct automatic splitting and other cases where ESI fails to identify the correct word cluster and you have to use the SPLIT button to repair this yourself.