Walking the Technical Side of Localization: Orchestrating QA, OCR and DTP to Really Work for You
On the execution level, every localization project has two major distinct dimensions – linguistic and technical. While the linguistic dimension encompasses language adaptation issues, such as language pairs, translation, and review, technical dimension includes physical aspects of the project, such as file formats, required software, digital resources and various additional services that need to be performed before, during and after the project, all with the aim of ensuring the quality and usability of the final localized materials. Both dimensions are of equal importance and we may say that the difference between translation and localization happens at the junction point of these two dimensions.
Fixing the Quality Issues in the Source
We have touched upon the direct link between the quality of source materials and the quality of translation in our previous article. This theme also extends to the technical realm – poorly formatted, uneditable or poorly scanned source documents will definitely not improve the quality of the localized materials.
The main reason for this is that, regardless of whether we use an AI-powered machine translation engine and post-editing or traditional professional human translation, all the major phases of localization process are performed in a CAT (Computer-Assisted Translation) software that can only be used with editable files.
CAT infrastructure is what enables boosted productivity, cuts costs and shortens the turnaround time, so it is easy to see why the properties of the source files are one of the essential elements of the localization process.
In cases when source files cannot be easily imported into a CAT software, there are ways to process these files to enable smooth execution of the localization project, such as:
- OCR or Optical Character Recognition
In certain cases, the materials for localization exist only in the form of scanned documents, or they include images or scanned elements such as seals, tables, or even written text. To make these files editable, technical localization teams perform OCR processing to generate a searchable and editable textual file that will be used in the further phases of the process.
- Alignment
In cases when there are (editable or uneditable) translated versions of documents that were not generated in a CAT tool, it is possible to perform the OCR and subsequently the alignment process to develop translation memories that will enable faster, cheaper and leaner localization with improved lexical consistency.
- Development of Terminology Bases
It is always recommended to extract specific terms and phrases – from taglines to any project-related terms that require special attention – before the localization process begins, and to develop terminology bases that will contain client-approved translations of these terms and phrases. This approach will save time and resources and ensure consistency and quality of the localized materials.
Fixing the Quality Issues in the Target
Quality Assurance (QA) process is performed on translated and reviewed files to check and resolve any errors that may have slipped through. The QA process is also computer-assisted and can be performed within the CAT tool or with independent QA applications.
What is usually checked are issues that involve spelling, formatting, tags, numbers and consistency. Depending on the project requirements, there are various types and levels of quality assurance, and dedicated QA programs are developed for each project.
It should be noted that the QA process is performed on purely textual files before they are sent to further processing, depending on the file format. After this stage, further quality control steps may include:
- DTP or Desktop Publishing + Proofreading
Even the best laid translations may go awry once the translated text is imported into a page layout software. Different languages (and different scripts) have different lengths, there may be issues with text breaks, fonts, layouts and everything in between.
This is why DTP should be performed with proofreading that will enable the linguist team to check their translation in context and address any issues before the localized materials are published.
- Software testing
In cases when the subject of localization is software, testing of localized apps (or, alternatively, screenshots) is performed for the same reasons as proofreading – to enable the linguists and QA teams to check their work in context and address any issues.
Under One Roof
While each of these processes can be performed by different teams and by various organizations, the record shows that centralizing the entire localization process and the supporting processes provides higher quality and faster turnaround times – simply because the linguist and technical teams within the same organization who work in accordance with harmonized rules and procedures are essentially all working on the same task with the same goal.
There is little room for “not my job” mentality, misunderstandings of technical and language aspects of the source material and similar weak spots that may lead to errors, delayed product launches or suboptimal localization.
Lest we forget, we localize content to localize the experience, and one of the fastest and safest ways to do it is by bringing various talents whose services are required for localization under the same roof. This enables linguists to know how their translations will be used and they will have an opportunity to adapt their work accordingly and to check it in context, while the tech people will be aware of the importance of linguistic subtleties in their line of work, and the result can only be a high quality localization.