PERFECTING THE ART

A good technical translator can translate about 2,000 words per day. An editor can proofread about 4,000 words per day. Desktop Publishing (formatting) can take a few hours – or many days. Every translation project is unique and timeframe will vary based on many factors.

Translation is a contextual translation whereby the original foreign text is maintained with any language idiosyncrasies intact, whereas transliteration is translating the text verbatim, during which any context may not be lost.

Back translation can be defined as the procedure according to which a translator or team of professional translators interpret a document previously translated into another language back to the original language. Usually this process is made by a translator or translators who had not been previously involved in the project and who have no prior knowledge of the objectives or its specific context.

Transcreation stands for translation + creation. When translating marketing communication content, translation alone is often not enough to meet the quality of requirement of localized materials. Translated texts must be rewritten or modified to meet the target language and culture requirements of different international markets. However, transcreated content is generally not suitable for TM (translation memory) leverage as the repurposed segments may not be 100% accurate without context.

Pseudo-translation is an exercise whereby each translatable text string in a software resource file is automatically replaced with a dummy string with a similar length. The dummy string usually contains accented characters or random characters to represent text in another language. The pseudo-translated resource file is then compiled for localization quality assurance. Pseudo-translation is a quick way to identify issues such as compiling errors due to double-byte or multi-byte characters or missing translations caused by hardcoded strings.

A translation style guide is set of instructions that your translators abide by when they translate your content. Typically they include information like if product names should be translated or left in English, preferred international currency, number and date formats, and ways to handle acronyms/abbreviations, and capital/lowercase letter conventions, etc.

Machine translation (MT) has increasingly gained acceptance in today’s localization world. Depending on the MT engine used, the quality of the machine translated text must be reviewed and edited by a trained human linguist to correct grammar and accuracy errors. The amount of post-editing efforts depends on the language pair as well if the MT engine is adequately trained with existing translation memories as well as industry specific terminology glossaries.

A software product must go through internationalization (I18N) testing, localization (L10N) testing (cosmetic and linguistic), and functionality testing before international release. Internationalization testing should be performed ahead of its localization counterpart to make sure it’s L10N ready and that it’s able to run on foreign language platforms. Localization testing is the step that validates the translated software for truncation as well as layout issues and translation errors. Functionality testing is similar to source application testing and is usually performed as part of localization testing.

Terminology management is the single most important step you can take to improve translation quality while reducing localization costs. One way to get started is to begin a translation glossary. You can either select terms manually or use a tool such as Term Extract to assemble the terms automatically. Each term should be precisely defined so as to not leave any ambiguity. Next, have your translators translate the terms into your target languages.

In our opinion, translation memory (TM) is an intellectual property of the client. This is because the client has already paid a language service provider to translate the content. TM is created automatically during the process of translation with the use of CAT tools. However, some LSPs refuse to turn in project TMs and use them to lock in a client. One way to get the TMs back is to perform manual translation alignment with previously translated files. Unfortunately, doing so can result in fuzzy match penalties and the aligned segments must be reviewed to ensure match-up and quality.