Aligning Existing Translations

To make translations which are not created in Across available to Across and use them, you can use the alignment function. This function loads document pairs consisting of the source document and the corresponding translation to Across. For this purpose, Across analyzes the source document and the target document and matches the corresponding segments in the source and target documents. Upon completion of the alignment, the matched segments are imported to crossTank as translation units and can subsequently be used, e.g. for pre-translations.

The Alignment Wizard is available for performing alignments.

After Across has analyzed the source and target documents and aligned the individual segments, we recommend checking whether the alignment is correct.

If the alignment of two Word documents in Across does not yield the desired result, this may be due to the different formatting or document structure of the two documents. In this case, open the two documents alongside each other in Word (e.g. using the menu item Window > Compare side by side with <document name> or View > View Side by Side) and activate the document map or the navigation pane for each document (via View > Document Map or Navigation Pane). If the documents have different structures, you should correct these, e.g. by assigning the selected style anew. Subsequently, you can align the two documents once more. The results of this alignment should be much better than before.

Requirements

To be able to perform an alignment, the documents containing the source and target texts must exist in an electronic format that is supported by Across.

Please note that the alignment of XLIFF documents is currently not possible.

Alignment algorithms

For the alignment of the matching segments in the source and target documents, Across uses an algorithm that you can select in the Alignment Wizard.

  • The following algorithms are available for selection:
  • Comparison by structure only: is the fastest algorithm; however, it only takes the structure of the document (e.g., the number of paragraphs in the source and target document) into consideration.
  • Comparison By structure and content: takes both the text structure and the text content (numbers, abbreviation, formatting, etc.) into consideration in the comparison. Therefore, it is smarter than the comparison by structure only and usually delivers better results, but is also somewhat slower.
  • Comparison by identifiers: Takes IDs used in certain document formats only (e.g., resource files) into consideration. This algorithm is best suited for text pairs with identical IDs.