Shortly after the release of Adobe InDesign CS for Windows and Mac OS in 2003, it quickly became regarded as the new desktop publishing (DTP) standard in document production. Companies that publish InDesign documents in multiple languages still face challenges, however, in developing appropriate procedures and automation tools for efficient and reliable localization processes. This article reviews four relevant conversion tools currently available and highlights common problems in the localization process as well as how graphic design agencies, localization agencies and their clients can best avoid them.

Assumptions

This article is based on the assumption that after the InDesign documents are created and approved by graphic design agencies or internal client DTP departments in one single language, companies normally use external vendors for translation (while document preparation is carried out in-house) or outsource the whole process to their localization service providers (LSPs). Processes described are those followed at our company, milengo Inc. Please note that milengo Inc. does not exclusively support or endorse any of the programs mentioned in this article.

The actual localization process typically includes preparation of source documents for export/import procedures (preprocessing); export of translatable text from an InDesign file to a format that can be accessed by translation memory (TM) applications; translation of content by local milengo members; import of translated text into InDesign; adjustment of the document's layout and features (post-processing), and QA and approval.

Three main types of features are relevant for export/import before/after translation. First, text paragraphs should still be in their original position (frame, page) after translation, and the original formatting (paragraph and character styles) should be retained. Second, image frames are used in two ways in InDesign documents: at fixed positions related to page borders or within the text (floating). Both kinds of image frames should be retained and should stay in their original positions relative to the respective reference elements. The third type of feature is index marks and hyperlinks within the text, which are used to automatically create document indexes and (when exporting PDF documents) for hyperlinks to internal references, external documents or websites. Marks and hyperlinks should be exported in a way that enables translators to translate the associated entry text along with the respective text phrases. After translation, all types of marks should be imported in a way that does not impair the internal processing of InDesign.

An ideal conversion tool would not require any pre- or post-processing of InDesign documents; would export/import all relevant InDesign document features correctly; would export to a file format that is compatible with different TM tools; and would present translatable text to the translator in a layout that is as close as possible to the original InDesign document layout, especially regarding font formatting and tables.

As the ideal conversion tool (still) does not exist, the localization agency will need to perform the pre- and post-processing. The actual work required for pre- and post-processing depends heavily on the respective conversion tool s ability to export and import document features. Therefore, we will first talk about the available conversion tools and then consider some aspects of how to make the best possible use of them.

Tool Overview

For the past decade, TRADOS S-Tagger has been the tool of choice for localizing FrameMaker documents. For InDesign files, no single conversion tool has been established yet as a comparable standard. The best known tools currently available for converting InDesign CS files into translatable file formats are TRADOS Story Collector for InDesign CS (released in 8/2004 with TRADOS LSP 6.5.5.438); SDLX InDesign CS Filter (released in 10/2004 with SDLX 2004 Bonus Release); STAR Transit InDesign CS Filter (first released in 5/2004 with Transit XV Service Pack 11); and ECM SysFilter for InDesign (first public version released in 3/2004). The table "Filter Tools for InDesign CS — Feature Overview" compares the relevant features of these tools.

TRADOS Story Collector. When using Story Collector, the hyperlinks, table of contents and index must be removed manually from the InDesign document before the translatable text is exported. If this step is neglected, InDesign will crash during the import procedure. Story Collector places inline frames at the beginning of stories or into the first frame during translation import. They must be pasted back into their original places manually. Index marks can be translated within the text but are sometimes splattered by tags. While tables are difficult to recognize in the tagged format, tags are doubly protected. They are locked by default, and TagEditor alerts translators about missing or wrong tags in fuzzy-matched translations.

SDLX. SDLX exports translatable text into its own ITD file format. By default, the table of contents and index are ignored during export since they must be regenerated after translation anyhow. Inline images are handled well. They are placed correctly after importing the translated text into InDesign. Index marks can be translated in SDLX, and translators can easily recognize them by noticing the square brackets. Their relation to text terms, however, is not always obvious. Hyperlinks are stripped from the InDesign document during translation import and must be restored manually. SDLX recently released a small tool for TRADOS users designed to convert between ITD and TTX formats.

STAR Transit. The Transit Filter handles all relevant features correctly and is therefore quite close to an "ideal" tool. The Transit user interface helps translators to understand the main text structure by presenting bold text in bold face and table text in tables. In our tests, however, the formatting from InDesign paragraph styles was not shown in the translator's interface. Unfortunately, the exported TTC format is proprietary, and files can only be translated with Transit.

ECM SysFilter. ECM SysFilter exports translatable text into RTF formatted files. These files are especially appealing to translators because the original text layout, including tables, is transferred to the translatable file. Index marks can be edited directly in the text. But if the InDesign document has inline images, a DTP engineer has to copy each image manually onto the InDesign pasteboard before exporting with the SysFilter. After translation, all images must be manually pasted back into the text. This task can be time-consuming when multiple target languages are involved. To make things worse, hyperlinks are stripped from the document during translation import and must be restored manually. On the other hand, the well-preserved document formatting and tables in the exported RTF file provide a valuable "emergency option." If the automatic import process fails, it is possible to copy and paste the translation manually from Microsoft Word to InDesign. The use of the standard RTF format gives language service providers and translators the freedom to choose their preferred CAT tools.

Pre- and Post-processing

Depending on the export tool used, hyperlinks, auto-generated text and inline images may need to be removed from text frames before the translatable text is exported, as described earlier. It might also be useful to adjust the document to some of the layout guidelines described later in this article, which will in many cases simplify post-translation processing. Before beginning the translation, a test import of the still untranslated text should be performed for each converted document to confirm the accuracy of the export/import procedure. Since the TRADOS-exported and SDLX-exported text does not provide a reasonable overview of the original text structure, DTP engineers should create a PDF file of the source document for the translators to use as a reference.

After the translated text is imported, the target language attribute should be applied to all styles to ensure correct hyphenation. Next, all text elements need to be checked and adjusted for the complete text display, as well as text positioning and formatting. Hyperlinks and inline images must be restored. Finally, the table of contents and index need to be generated to finalize the process.

The manual work required to complete the full pre- and post-processing is time intensive, error-prone and more costly compared with the conversion of FrameMaker files for example. To better streamline the process, milengo decided to embark on the development of a proprietary InDesign plug-in that automates the pre- and post-processing tasks associated with InDesign files. Without any further manual editing, inline frames, hyperlinks and bookmark sections can either be fully exported into RTF format, or the InDesign files can be preprocessed to use with the preferred export CAT tool.

Three Typical Issues

Whatever tool you decide to use, InDesign itself might prove to be an issue. InDesign files from versions 2.x, 3.0 and 3.0.1 (August 2004 update) are not compatible with each other and cannot be converted downward. Therefore, when exchanging documents, it is crucial that you ensure that all parties working on the documents use exactly the same InDesign version. Adobe has announced downward compatibility via the InDesign Interchange Format for the recently released 4.0 version.

Missing fonts and image files are a second frequent problem. It is necessary not only to keep track of all fonts and EPS files used in the document, but the files used to create the EPS files as well. It is advisable that the DTP provider/department deliver the complete file set together with the final document. Collecting them later may turn out to be a time-consuming and rather costly effort.

The third issue can also be the most frightening. After importing the translation into InDesign documents, parts of the text might be misplaced; there may seem to be disappeared graphics overlapping text paragraphs; the formatting becomes corrupt; and table-of-contents entries are missing. Normally, these problems are not caused by the export/ import process itself, but rather stem from the original document layout. Everyday experience has shown us that InDesign documents sent to us for translation may use insufficient or inconsistent layout techniques. An example would be table-of-contents entries that have been added manually to the automatically generated table-of-contents without applying appropriate paragraph styles to the respective headlines in the text body. Although doing so makes the text visible in the printed version or PDF document, the automatic table-of-contents generation would not compile correctly after translation and would therefore need to be manually corrected again. Displaced or invisible text typically occurs if the frames in the original InDesign document do not leave any space for increased text length in other languages.

To avoid any potential predicaments that may stem from the layout, the authors highly recommend that users have localization in mind as early as possible when creating InDesign documents. If the documents are to be translated into multiple target languages, it might even be useful to adjust source documents to meet these guidelines prior to exporting the translatable text. Doing so can significantly reduce additional adjustments and alterations of each of the target documents after translation. See "Layout Guidelines for Preparing InDesign Documents" for some suggestions.

Workflow Aspects and Quality Assurance

Localization can be greatly simplified if potential pitfalls are considered during the initial document creation. First of all, the graphic design agency or department, the LSP and the client should agree as to which InDesign version will be used. If you use layout specifications, take localization issues into account. All fonts and image files should be stored safely.

Before starting the actual localization of large documents, perform a pilot test by running a small portion through the entire localization process for each target language. This is to make sure that all document-specific and language-specific technical issues are identified and addressed as early as possible. While the translation will usually be performed in the target-language countries, it is highly recommended that you perform all conversions and layout adjustments at a central location. The party who exported the translatable text should be responsible for importing it back again.

Localized documents should be approved by both the LSP and the client. The DTP staff in charge of QA should be provided with a PDF export of the localized document so that they can record and track their changes by adding comments. Requested changes should be implemented at the same site responsible for all other document conversions and adjustments.

Conclusion

All available tools are capable of exporting and importing the main text threads, but only the Transit filter can fully convert all relevant InDesign features. Unfortunately, Transit only works with a proprietary file type that cannot be employed with any other translation tool. TRADOS or SDLX users may likely choose their filter based on the relevant features required for the particular document or project. For example, if there are several inline images to process, SDLX would be the preferred filter. If translators have to fiddle with complex tables or if compatibility is a priority, ECM SysFilter would be the best choice.

As within any aspect of the entire localization process, careful preprocessing and post-processing play a crucial role in the successful localization of InDesign documents. Whatever tool you decide to use, the document layout should ideally be developed with localization in mind, and a well-defined workflow with your localization provider should be defined to significantly reduce any unnecessary manual work on behalf of the DTP engineer. Therefore, when outsourcing InDesign localization projects to localization agencies, make sure they not only prove their localization competence, but that they also have in-depth expertise in InDesign and DTP. We therefore recommend that agencies and clients engage in pilot testing phases in the early stages of a relationship to help avoid any pitfalls in the localization process.