RE: Note about resolving conref.

["JoAnn Hackos" <joann.hackos@comtech-serv.com>]

Hi Bruce,

Unfortunately, translation memory can't handle this problem. The conref'ed words are generally stored in a separate collection topic which just has a list of words. The words are translated separately from the topics. Then, when the output is run through the OT, the words are embedded in the text. However, since the words are translated separately, they will often be in the wrong case in the translated text (subject, object, and so on). That means the resulting text is clearly ungrammatical, especially in heavily inflected languages. That's what the best practice document says.

 

By the way, the same happens in EN when the noun changes from an initial consonant to an initial vowel or if it changes from singular to plural.

 

Translation memory does nothing to alleviate the grammar problems that can occur when single words are conref'ed into various grammatical structures.

 

JoAnn

 

JoAnn T. Hackos, PhD
President
Comtech Services, Inc.
710 Kipling Street, Suite 400
Denver CO 80215
303-232-7586
joann.hackos@comtech-serv.com

 

 

---

From: bruce.esrig@gmail.com [mailto:bruce.esrig@gmail.com] On Behalf Of Bruce Esrig
Sent: Sunday, June 29, 2008 10:46 PM
To: JoAnn Hackos
Cc: dita-translation@lists.oasis-open.org; Bryan Schnabel; christian.lieske@sap.com; dpooley@sdl.com; fsasaki@w3.org; rfletcher@sdl.com; Howard.Schwartz; ishida@w3.org; rmraya@maxprograms.com; tony.jewtushenko@productinnovator.com; ysavourel@translate.com; Michael Priestley
Subject: Re: Note about resolving conref.

Aha ... the repetition that results from conref can be handled by translation memory software. Distributing translation jobs to multiple translators would also work as long as you're using a hosted, instantly updated translation memory database that is shared across translators.

 

But if the job is to be distributed and the translation memory won't be shared in real time, then the translation team might benefit from getting a separate copy of the material-to-be-conrefed so that the material-to-be-conrefed can be translated once. The common material will then be in the translation memory database when the material-after-conrefs-are-resolved is distributed for translation.
 

Best wishes,

 

Bruce Esrig
 

On 6/29/08, JoAnn Hackos <joann.hackos@comtech-serv.com> wrote:

Here is an idea about resolving conrefs that I received from Michael Priestly. Please consider it for our committee meeting tomorrow. I am looking for a process that we can recommend in an amendment to the conref best practices document.

 

___

It would require custom XSLT. Probably not much, using one of the existing conref XSLTs as a base. Prior to sending files to translation, the author would run a packaging process that runs the XSLT against each of the files they intend to send. The packaging process would require some custom batch files to run the XSLT, probably ant-based if they're using the DITA open toolkit since they could borrow from existing processes again.

___

 

JoAnn T. Hackos, PhD
President
Comtech Services, Inc.
710 Kipling Street, Suite 400
Denver CO 80215
303-232-7586
joann.hackos@comtech-serv.com