The translation process includes several stages: translating a text into another language, editing, and proofreading. Within professional circles, this process is called ТЕР (translate, edit, proof). Translated materials are checked for spelling and punctuation errors, consistency of style and terms, and accuracy in converting abbreviations and acronyms. The translated text must be free from words in the source language and other inaccuracies. The next step is localization testing. This process consists of testing the translation by linguists in the finished product: on web or mobile application pages and in the compiled software. In other words, it is a functional and linguistic analysis of the translated elements displayed on a screen.
Main Aspects of Localization Testing
Correction of errors that cannot be detected during ТЕР
The approach based on translation, editing, and proofreading allows for the very precise adaptation of text from one language to another. At the same time, some common errors cannot be detected when translating materials. Localization testing assesses the visual impact of the translated elements, their internal consistency, and relevance to the whole context.
Translation integrity checks
The analysis of the language version deployment allows the detection of all the components skipped when exporting the source files for translation. These can include interface elements, service messages, user forms, software dialogue boxes, meta tags, and other items that comprise the finished product.
Final analysis of translation quality from the user perspective
Managing inaccuracies before the project launches in another region makes the language versions of the product as seamless and natural as possible for a foreign audience. With localization testing, you can be 100% certain of the translation quality. This is particularly important for companies aiming to emphasize their status and gain foreign users’ trust.
Where Localization Testing is Used
Typically, projects that need further testing by translators are characterized by large text and visual blocks, the use of professional terms, and nonstandard and specialized vocabulary. It is often extremely difficult to interpret content and simultaneously control the quality of translation in text format alone. Examples include:
- Software
- Mobile and desktop applications
- Websites
- Industrial and medical equipment manuals
- Electrical diagrams of complex equipment
- Construction plans
Localization Testing Types
Language testing. This confirms the linguistic and cultural nuances in created language versions and checks the naturalness of the text; that is, the correct interpretation of set expressions, idioms, humor, and other attributes that convey the message and mood of the source text.
Cosmetic testing. This includes product visual analyses. For instance, localization testing of a website or a mobile application on a particular device helps reveal problems of font display and size, diacritical mark support, and the correctness of line breaks.
Pseudo localization testing. This includes assessing the readiness of the product for analysis and editing, as well as general trends in the language pair. It is used to identify segments with issues and inconsistencies, such as important files that are missing or essential differences in the translated text, before full testing, which can save an expert’s time later.
Materials Required for Linguistic Localization Testing
The typical set of documents provided by a client for testing includes the following:
Project build
This term comes from the verb “to build.” A build is a set of sources comprising a whole product. A customer typically provides a build as a downloadable folder. Localization testing includes processing the text and graphical components of the build.
Test plan
This is a localization testing checklist with all the items that testers should consider. These items may include interface elements, blocks with unusual vocabulary, audio content for voice acting, slang, and set expressions.
Localization file (or localization kit)
This document must contain text for all language pairs in the project. During testing and error identification, all amendments will be made to this file. The localization kit helps identify irrelevant data and clarify ambiguous points in the texts.
Screenshots
As an alternative to a build, a customer can provide screenshots. This is an option for small projects that allows the client to spend much less time preparing for testing. A screenshot folder is compact and easy to download.
Localization Testing Stages
The localization process typically includes several steps:
1. Receiving customer files and specifications
If your project has already been translated by Kenaz, simply use the test plan to define the main aspects that require final verification. Otherwise, additional clarification of the requirements and expectations may be required.
2. Testing and reporting
Experts start to check the materials. Every localization test aims to identify errors or bugs. An error list (bug report, issue report, issue tracker, or issues log) is a list of detailed recommendations on improving the source files. Often, a partial or total lack of context for the translation, rather than translator or editor errors, is the source of linguistic bugs.
3. Amendments to the localization kit and transferring updated materials to a customer
Before making any corrections, translators evaluate the validity of the amendments, i.e., whether the changes are appropriate or not. Some items in a bug report may be well-founded decisions rather than translation or localization errors. At this stage, linguists communicate with each other and review any uncertainty with the client. The approved corrections will be included in the localization kit.
4. Regression testing
This is follow-up testing. It can be performed to control the changes made when a customer updates their build. At this stage, we can determine if the recommended corrections have been implemented and the overall result of applying them. For instance, we can assess how efficiently the text size on buttons and in interface menu items has been modified and determine whether a further reduction is required. Regression testing allows you to ensure that your localization is perfect and improves the initial testing results.
Most Common Localization Errors
that can be Exposed by Testing
High-quality testing includes data analysis over a wide range of parameters. A professional approach inclusive comprehensive material processing to detect errors that are difficult or even impossible to exclude at the translation stage. Let’s take a closer look at the most common errors.
The lack of significant components
The files for translation provided by a customer are often not fully uploaded and lack important segments. This problem can only be detected at language version deployment. For example, there may be untranslated pop-up messages, error notifications, feedback forms, or program dialogues.
Translated blocks do not relate to the general context
The majority of the translation process is the work on the text. When translating a website or an application, the source file contents can be in the form of a series of text blocks, individual phrases, and words. Often, at the launch of a language version, we see errors in display, format, the internal consistency of page elements, etc.
Localization testing provides a chance to analyze all the translated segments on the page along with their layout and relevance to the overall context. Finding these issues can refine the product and make it more user-friendly.
Mismatch of fonts and encodings
Frequently, texts translated into different languages must be converted from one format to another (e.g., text export to XML or TXT format from the programming environment). When reimporting the translation to the initial format, the encodings can be mismatched, and such mismatches can result in corrupted localized content display.
The length of the translated text doesn’t match the image
When text is translated, it’s often longer than the source text. If this difference is only a few characters, it’s hardly noticeable. However, if the permissible length of the text in a block is restricted, the translated segment must be adapted to the given space. Such errors are difficult to foresee at the translation stage because they depend on design features and, hence, are commonly found at the testing stage.
Incorrect hyphenation
Translations are usually inserted into a code by developers who may not be familiar with the localization languages and may not be able to identify incorrect hyphenation or word breaks. This is another reason the localization testing should be performed by professionals who are proficient in the target languages.
Unlocalized meta tags
The translation of meta tags not only allows your web project to rank higher in search results but is also an important localization attribute. For instance, if the meta tags of your website haven’t been translated, a user can see text in an unfamiliar language in their browser.
Inappropriate wordings
While localizing text, it is important to ensure that the translation triggers the right associations and emotions in users. In addition to correctly conveying the content and emotional weight, we must avoid meaningless constructions and the incorrect use of gender and case.
Data format mismatch
During testing, number, time, length, and weight formats, as well as currency specific to the target region, should be checked. These elements can be missed during translation.
Inappropriate interpretation of graphical elements
High-quality localization includes the use of the same fonts on visual elements in new language versions. Button and menu item text should be visible on all devices. Color combinations should not have any associations with a national flag or other cultural features of a region that are completely different from the author’s concept.
Drawbacks of audio files
The adaptation of training materials, such as lecture and seminar notes, often includes the translation of voiceover texts. Transcription, accentuation, overall sonority, and the preservation of the speakers’ semantic and emotional weight are of particular importance in such materials.
How the Depth of Localization Testing Affects the Cost
The depth of the localization testing indicates how well a project is developed. More thorough localization testing incurs additional costs and requires more time, but it will help to eliminate any flaws that may damage native speakers’ perception of the final localized product. Customers with large amounts of data and limited budgets can use alternatives when only part of the project needs to be tested.
Clients with a limited budget willing to order localization linguistic testing can specify the number of hours they can pay for and specialists will test the most important components and usage scenarios of an application, website, or other materials.
Localization Testing Results
Localization testing allows source materials to be precisely adapted to the regional and cultural features of other countries. The lists of errors drafted by professionals are detailed and consistent with customer requirements. A typical bug report contains detailed information on each issue:
- The issue type and description. The report includes information about the error and its nature—whether it is linguistic, functional, visual, etc. This allows for immediate identification of the subject area and a specialist to resolve the issue.
- Bug illustrations enable you to see the current problem and show how the correct form should appear.
- Error path. This may be a series of steps to the issue location or a link to the corresponding page.
Amendments based on testing results will help to eliminate all localization issues and provide foreign users with a flawless and natural-looking translation. Localization testing can be performed through a translation agency with all the necessary technical and human resources.