The process of localization is a complex and demanding task, one which includes, but is not limited to, translating, editing, and proofreading work. Automated quality assurance of the translated text occupies a very important role in this process. So what is quality assurance, and why is it needed?
Even when the text being translated is relatively short, human error can lead to the same term being translated differently, a required tag being skipped, or the simple mistake of pressing the space bar button twice. What would be the best way to identify those mistakes and get rid of them to ensure the high quality of the final text? A solution to this problem is provided by a wide variety of automated QA tools like MultiQA, Xbench, Verifika, LTB, and others. Those tools scan the bilingual file for potential issues. What exactly are those “issues”? They can be split into several categories:
- Inconsistent terminology
- Inconsistency in the translation of identical text segments (and vice versa: the same target text derived from different sources)
- The correctness of “variables”: numbers and their format, links, units, punctuation marks, extra/incorrect spaces, etc.
- Inconsistency between the tags in the source and the target
- Spelling mistakes
- Observation of general localization rules (grammatical constructions, text presentation, etc.)
Each automated QA tool has its pluses.
For example, our favorite QA tool, MultiQA, can create several reports that identify the abovementioned issues. It also allows updating files right from the report simply by pressing a button! The types of QA checks that can be accomplished with MultiQA are:
- «Terminology Consistency». This type of check is needed to identify inconsistently and incorrectly translated glossary terms. The system uses a glossary (or several, if needed) that are stored in an online terminology management system. This system, also known as the Terminology Management System, or TMS, stores a database of glossaries sorted by project category. This database is “live” and it is possible to constantly update and edit it during the project’s or term’s life cycle. Another great asset of this tool is the built-in generation of grammatical forms for each term that become automatically entered into the glossary. This allows a dramatic reduction in the amount of false positives at the stage of checking terminology, which in turn reduces the time spent on the process.
- «Inconsistent Segments». This type of check allows identifying inconsistent segments in the translation. The tool scans the source text (and the target) for identical segments, and then checks whether the translation for them matches, or (if we are talking about the target text) whether they come from the same source. All of the mismatched segments are added to the report.
- «Localization Rules». This check ensures that the numbers in the target text match those in the source text, the dates are localized in the correct way, the punctuation is correct, and there are no double/trailing spaces.
- «Tags & Entities». This type of check allows ensuring that all the tags from the source text are present in the target text in the correct way (the same number in the same places).
- «Spelling». This check’s purpose is to make sure that the words in the target text are spelled correctly.
- «Marked Segments». There are situations when either the translator or editor is unsure whether a text segment has been correctly translated. Should this situation arise, they can mark the problematic text segment with a ‘translator doubt’ notification to draw the project manager’s attention to it. This type of check allows filtering all those marked segments into a separate report for an in-depth analysis.
- «Preview» and «All». Those reports cover all translated text segments. The text in «Preview» is presented in two ways: with and without the tags. The second option is useful when there are a lot of tags present, which can make the text hard to read. The «All» report is useful for specialized «final» checks, which include checking words and word forms that are not considered to be terms (for example, the «How to» grammatical structure), special project checks (whether the text presentation meets the customer’s standards), and regular term checks (terms common for the language or individual involved in the project).
- «Updated Segments». A report based on all the checks performed that allows tracking the updates made for bilingual files.
It’s no secret that each project requires a unique approach. That is why we develop tailor-made Quality Management programs that allow us to match customer instructions perfectly. Apart from automated translation quality assurance tools, we also utilize a wide variety of additional instruments. These include various bug trackers, query trackers, online based quality tools (like ContentQuo), etc.
We strive to ensure the quality of our work, and to improve our customers’ experience on a daily basis!
