XML and Electronic Commerce: But What About Documents?   Table of contents   Indexes   How STEP/SGML and XML/EDI Initiatives Will Impact Future Development of IETMs.
 
 

Automating Language Translation Requires Document Management, Workflow, and Application Tools

 
Jean-Charles   d'Harcourt
  European Sales Director
  Xyvision France
69, rue d'Aguesseau
Boulogne   France  92100
Phone: +331 46 10 50 40
Fax: +331 46 10 50 54
Email: harcourt@xyvision.com Web: www.xyvision.com
 
Biographical notice:
 
Jean-Charles d'Harcourt
 
Jean-Charles d'Harcourt is the European Sales and Marketing Director for Xyvision. Prior to joining Xyvision a few months ago Jean-Charles was co-founder and VP Marketing and Sales at GRIF S.A. the European leading company in SGML and HTML publishing Tools. Jean-Charles has more than 7 years experience in the document management business and was graduated in 1983 from the European Business School.
 
ABSTRACT:
description
 
Effective language translation requires more than just document management.
 
Workflow and application-level software are needed to perform the following functions:
 
  • establish links between source and target language versions,
  • permit some components to not be used in all target languages,
  • permit target languages to contain components not found in the source language information,
  • automate determination of which components need translation,
  • provide automated tools for notifying users which data needs to be translated,
  • create translation packages, if sending the information to a service bureau for translation,
  • provide a locking mechanism to guard against sending out updates of source material while previous version is still being translated (synchronization between component versions),
  • provide facilities for word counts, if sending the information to a service bureau for translation,
  • check for completeness of information upon return from service bureau.
 
Most companies today that need to produce documents in multiple languages lack a standard process for localizing information. Often, translation can only occur when the source language version is complete and approved for publication, creating a long distribution time between the source language and other language versions. This causes problems for such industries as international product manufacturers, who may need to ship new models to many countries simultaneously. Companies that attempt to translate documents at the same time they are written have difficulty coordinating the numerous revisions and versions of both text and graphics, often resulting in redundant or inconsistent translations, unnecessary expenses, and erroneous information.
 
The combination of SGML/XML with document management enables corporations to manage their information as components. By managing the data as modules or objects, changes to source material can be tracked at a finer level. As a result, only changed information needs to be retranslated. However, document management alone does not solve the problem. Workflow and application-level software are needed to automate the translation process and to insure that changed information has all been translated prior to publishing.
 
The application-level software must include provisions for:
 
  • Linking updated source language components to applicable target market document structures, keeping in mind that one target market may consist of different language versions. For example, the Belgium market may need both a French and Dutch version of the information.
  • Creating translation projects which lock information components to prevent modification while being localized.
  • Integrating software translation tools if needed.
  • Creating translation packages for service bureau or freelance translators who do not have access to the repository.
  • Counting translated words for billing purposes.
  • Validating and checking-in translated files.
  • Removing locks and publishing localized document.
 
Workflow plays a significant role in the management of the translation project. A typical workflow would would consist of the following steps. Writers create and edit information as reusable components in a source language document structure. A separate document structure is created for each target language that requires translation. When information changes, the application links the updated source component to the applicable target version. While a component is being localized, the corresponding source version component in the project is locked so that no further modification can take place until the translationis complete. This does not prohibit users from editing the source documents for additional updates, but instead prohibits users from creating a new translation project until the previous project has been checked-in or manually voided (i.e., locks removed).
 
Once the project is created, it is either forwarded to a machine translation application for electronic translation or a translation package is created to send to a servive bureau or freelance translator.
 
The translation package should comprise:
  • The source language components, current and previous version.
  • The last translated version.
  • A "diffed" version of the data which contains both the previous and current versions of the source language.
  • An Acrobat document showing how the composed pages look with all graphics and other page elements in place, possibly with change bars to show which information has changed.
 
Using these components the translator localizes the data and passes the translated data back to the repository where it is stored with the appropriate market document structure. On import, the application must validate the incoming files. When all updates and translations are complete, the project is forwarded with the correct components for each language to a publishing engine for final composition and pagination, or to the Web for electronic access to the information.
XML and Electronic Commerce: But What About Documents?   Table of contents   Indexes   How STEP/SGML and XML/EDI Initiatives Will Impact Future Development of IETMs.