This task prepares existing translations for uploading to our translation memories. Then, they are reused in new projects. This process is called alignment.
Translation memory or TM is a database of original source text and its respective translated segments. It works by automatically reusing any segments that are the same or similar on new projects.
- The more you translate, the more it grows, and the higher the chance is of translation memory matches.
- Translation memory is effective when source texts use identical or similar phrases.
- It’s also useful for updating documents or making small changes, since the majority of the translated text can be reused from the translation memory.
In translation tools, segmentation involves dividing a source text into units for translation. These are called segments and they are a key component of alignment projects.
Depending on the configuration of the system, the source text is divided in different ways. Some of the most common options for segmentation are:
- Segments coincide with sentences and end in a full stop, a question or exclamation mark, a quotation mark, etc.
- A new segment is created after a paragraph break
- In bulleted lists (such as this one), individual segments are created for each item
When you work on an alignment project, keep this in mind and try to ensure segments match these characteristics.
A way to recycle translated material created outside of our translation tool. It requires the source and target versions of the documents to be aligned. The process creates a bilingual TMX file that feeds to a Translation Memory and reused in new projects.
- The system takes two files, source and target, and segments them into translation units.
- Then, it attempts to match and connect the source and target units, with various degrees of success.
- After that, a translator goes through the alignment and makes the necessary adjustments to match source and target text correctly
- Finally, the Project Officer imports the resulting TMX into the translation memory.
If there are any questions about the process, please do not hesitate to contact your Project Officer. They will help you out!
You will be provided with:
- A link to the alignment project.
- The source and target files as reference material
Here is how you approach this task:
Familiarize yourself with the general structure and content of the source and target files. This helps you make informed decisions when aligning.
Open the project link and take a look at the interface. It looks similar to the screenshot below.
The platform is simple and easy to use:
a. The original and translated segments are displayed on the left and right, respectively.
b. Match the translations with the original segment by either:
- Drag and drop the segments next to each other.
- Hover your mouse pointer over one segment and click on the check mark that appears on the top left to select it. Then repeat the process with the corresponding translation and click on the Align button in the tool ribbon at the top of the screen.
Merging, Splitting, and Hiding: As seen earlier, segments can be created in different ways, and you may need to modify the segmentation by using the Merge or Split buttons if the process did not work correctly. You can also hide a segment if you think it should not be part of the alignment.
a. Merging: combine two segments in either source or target by selecting and clicking on the merge button.
b. Splitting: divide a segment into two or more parts by selecting it and clicking on “split”. In the pop-up window, place the dividers in the correct position and then confirm.
c. Hiding: select the segment you wish to hide and click the button.
Work your way through the document until you have aligned all units.
a. Spot checks: you may speed through large sections of the document which are properly aligned. In such cases, please:
- Start from a pair of properly aligned segments.
- Scroll down approximately 30-50 segments and check if they continue to be properly aligned.
- If they do, repeat step 2. If they don’t, return to the segment in step 1.
- BE CAREFUL: the tool attempts to align segments automatically but it may fail. Be extremely careful when you proceed with spot checks.
b. Empty segments: if you are unable to find a match for a specific segment. The corresponding text may be:
- Within a neighboring segment (you will need to use split or merge to correct this)
- Located in an earlier or subsequent section of the project. Look for the content in segments close-by.
- Simply missing. In some rare cases, you may be unable to find the unit. After reviewing the document carefully, hide the unmatched unit.
Perform one final review of the document.
Handover notes: you may have issues in the source file, problems you were unable to resolve, or areas in which you think improvements are needed. Submit them to the PO while you do the review. Otherwise, compile any pending points or comments in an email and share them with your PO.
Check that all source segments are aligned with the corresponding translated unit.
Verify that there are no miss-alignments by spot-checking the project every 100 segments.
Check at all flagged issues are resolved.
Note: The screenshots are used for illustration purposes. The process can be done in another aligner tools like SDL Studio or a simple spreadsheet.