Our engineering process is based on a model comprising extensive Localization Quality Assurance procedures. Each localization project is completed within a consistent, quality-centric framework. Quality and workflow processes can be adapted to integrate with existing client environments. During all phases of the localization process, our management and production teams working from offices in Taipei and Beijing will control every aspect of the process through our quality assurance procedures.

Quality Assurance (QA) comprises pre-translation, translation and post-translation processes; without proper pre-translation document preparation and the use of appropriate tools during translation, any post-translation QA efforts will be futile, endless and costly. Allow us to elaborate further:

Pre-translation document preparation: Firstly, we must understand that a translator, regardless of his/her technical background, should only work on the text; a translator should never be asked to handle program codes in the documents on top of translation. Hence, before handing over the documents to our translation team, our project managers and engineers will make sure that all contents or tags (e.g. Framemaker tag) that are unrelated to the text are “protected”, such that they cannot be changed by the translators. This protection is carried out using selected macro programs. In this manner, we can ensure the accuracy of the non-textual components of the translated documents.

Next, our in-house linguists will go through the textual contents of the documents in the perspective of a translator. In this step known as Pre-QA (PQA), the original documents are examined in detail before translation, and any questions arising will be highlighted to the client; at the same time, we will construct a glossary in the original language.

The use of tools during translation: The use of Translation Memory (TM) tools during translation is of vital importance. These tools ensure consistency throughout a document, and can help our clients save cost. However, the use of TM alone is insufficient; it has to be coupled with the construction of a Glossary. When the translator is translating a sentence containing a key term, the existing translation(s) of that term will pop-up in real-time; the translator can then decide if the suggested translation is applicable in the current sentence. Why is building a Glossary important? When a project requires multiple translators working simultaneously, the Glossary is crucial in ensuring consistency (as explained above).

Most problems can be avoided with the three-tier approach described above; what remains is human error. How then do we detect human error?

Post-translation QA: This QA process comprises a two-step procedure: the first step of which is the preliminary review by the proofreader to uncover mistranslations, punctuation errors and other obvious problems; the second step involves the use of a QA software (TQC) developed internally by our company for error detection. During TQC, we will first import all glossaries and interface translations to form the bases for error detection. Next, we will import the translated document (bilingual) provided by the translator. An error report will then be generated, covering a total of 10 review items, including inconsistency in key terms (term inconsistency), multiple translations of identical sentences in the document (multi-translation), parts that have not been translated (missing translation), inconsistency in punctuation marks (punctuation error), inconsistency in numbers between the original and translated documents (wrong number), and inconsistency in tags and program codes between the original and translated documents, etc. This error report will be provided to the translator for verification and amendment where necessary. Through the above approach, human error is minimized.

The key to this comprehensive QA process is “time”. This means we require time in addition to that determined by daily translation output to implement these procedures. We seek our client’s understanding on this additional time requirement.

Preparation and application of the Glossary

1. Differentiation based on product types:

The existing Glossaries are differentiated based on the type of products, e.g. mobile phones, servers, WiMax, etc.

2. Based on this differentiation, the English strings to be retained will be filtered out.

3. The existing TM will be referenced for translations of technical terms (excluding terms in English to be retained).

4. A database will be created from the translated Glossary and the English strings to be retained.

5. Using relevant QA tools, the translated document will then be checked again for errors based on the Glossary.

A good Glossary should offer sufficient quantity (of terms) and comprehensive coverage. On top of that, it must be suitably differentiated for translators to reference it efficiently. Imagine a translator having to manually sieve through a Glossary with over 30, 000 entries while translating. Not only is this impractical, the entire translation process will be consequently slowed down. The translator will be put off by the Glossary, and the existence of the Glossary will be meaningless. Hence, all Glossaries should be reasonably differentiated based on the type of product, such that only the Glossary relevant to the product (or project) will be provided to the translator when a project comes up (irrelevant ones will be left out). Only then is the Glossary valuable, and speed and quality are ensured through concise workflow and procedures. We should minimize the amount of work that a translator needs to do in addition to translation; only then can he/she focus better on translation itself. If the Glossary takes the form of an Excel or Word file, it will slow down the translator’s speed, and disrupt the flow of the translation. Hence, the Glossary should be constructed as a dialog box that pops up automatically, at an appropriate time, to provide translations of relevant words to the translator without requiring him/her to open other files. This increases the accuracy and efficiency of the translation work, and will not disrupt the translator’s workflow. In addition to constructing a Glossary in the form of a Database, we will also integrate all Glossaries into the word bank of our QA tool. Upon receiving the translated document, we will perform a check using the tool, and then generate an Inconsistency Error Report in a blink of an eye. This report will be given to the translator for his/her verification and amendment.