Blog Archives

Localization and Language Technologies

I wrote a brief overview of localization and the use of language technologies. You can view it here at my Seneca College web space, or read the full text below.

Localization is an important concept for technical communicators to understand. Localization is not just theory, it’s big business! This webpage provides an overview of the practice of localization and the use of language technologies.

What is Localization?

According to LISA (Localization Industry Standards Association), localization is “the process of modifying products or services to account for differences in distinct markets.” In order to accomplish this process, three main categories of issues need to be addressed:

Linguistic Issues: Lingustic issues are the focus for technical communicators concerned with localization. Primarily, linguistic issues pertain to the translation of a product’s user interface and documentation. Linguistic issues also extend to the translation and re-engineering of linguistic-dependent applications such as applications programming interfaces (APIs), search engines, or wizards.

Content and Cultural Issues: A key content and cultural issue is the presentation of information (icons, graphics, colours, forms of address, etc). For example, Mexican audiences prefer more graphics in their documentation. It is also important to consider the localization of functionality. As in, a car sold in Britain should not be left-hand drive, and accounting software must be modified to conform to locally accepted accounting principles.

Technical Issues: Supporting local languages, especially those that don’t use roman script, may require re-design and re-engineering. For example, Arabic script is bi-directional (numbers and foreign-language words run from left to right, but everything else is right to left). Also, some east Asian languages require twice the disk space per character compared to English. Therefore, adapting products to local scripts requires changes to the code (for software), product design, packaging, etc.

Designing for Flexibility and Translatability

In order to achieve successful localization of a product, flexibility and translatability must be considered from the beginning.

Designing for flexibility ensures that the basic product can be easily adapted to local variations (making sure bi-directional characters and Asian scripts are supported from the beginning).

Designing for translatability ensures that language components of a product are clearly written, clearly identified, and easily accessible, and that they are kept physically separate from other aspects of the product. For example, a software product designed for translatability will ensure that items containing text (buttons and menus) are large enough to accommodate for the fact that words are of different lengths in different languages.

Factors Influencing Localization

Factors influencing the extent of localization include:

  1. The nature and scope of the product.
  2. The size of the target market and audience.
  3. The length of the product lifecycle.
  4. Anticipated frequency of updates.
  5. Competitor behavior.
  6. Market acceptance (as in “is there a sustainable market for this?”).
  7. National or international legislation.

Canada provides a great example of localization due to legislation, as by law all products must be fully translated into French. For more information, consult the English FAQ at the Office québécois de la langue française.

Language Technologies

Language technologies (also called “linguistic tools”) are designed to improve the speed and accuracy of translation, therefore making the translator’s job easier. Rather than replace humans as the masters of language, language technologies enable people to better deal with one of humanity’s most complex inventions—language—and improve human-computer interaction. As such, language technologies are the key tools used by localization experts.

      Terminology management systems (TMS)

Terminology refers to the specialist vocabularies associated with specific sectors and applications—they are the foundation on which good translation is built. Terminology management systems are databases of terms that are translated into different languages. A good TMS is concept-based rather than word-based, allows effective synonym management, and can be integrated with word processors to enhance translator productivity. Popular TMSpackages (according to Google) include AnyLexicLingo, and TermWeb.

      Translation Memory (TM)

TM refers to a database of previous translations in which the source and the target language texts have been broken down into segments that are aligned with each other. When a subsequent version of the source text is compared with the original, translation memory identifies the equivalent translated segments and inserts them into the new target text. Efficient use of translation memory depends on the quality of the original translation (since errors will be consistently replicated), the accuracy of alignment, and how updates to the original source text are handled. Translation memory is not perfect: editing is required to identify subtle changes of context, and to adapt texts to local markets. For example, in a car’s user manual you would write “open the hood” for a US audience, and “open the bonnet” for a UK audience.

      Globalization Management Systems (GMS)

GMS are an extension of Content Management Systems, and allow for the translation of large, constantly-changing websites (such as that of an international technology corporation like Hewlett-Packard or Apple). GMS typically have an engine that monitors site content for change, and a component that passes content to translators or other linguistic tools for further processing. GMS also manage the workflow and synchronization of translated content with the source-language website.

      Machine Translation (MT)

Machine translation is similar to translation memory, but differs in that it also performs linguistic analysis, using rules and statistics, on the texts submitted to it. MT systems break down the source text into its constituent parts before translating them and reassembling them in the relevant target language. The results do not compare to high-quality human translation, and require substantial post-editing. Even so, MT can still offer substantial productivity gains. A good, publicly available example of a MT system is Google translate.

Guidelines for the Use of Language Technologies

Avoid false expectations: Language technologies enable people, but do not replace them. Human intervention in the form of post-editing will always be required.

Match technology to your needs: The use of inappropriate tools will lead to substandard and/or unusable results. For example, machine translation is generally unsuitable for advertising or marketing texts, since these texts demand a high degree or originality and style.

Garbage in, garbage out: The quality of results produced by language technologies is crucially dependent on the quality of the original text.

Invest for the long term: Localization requires careful planning and implementation to produce the best results. Remember: language technologies are continually-expanding repositories of knowledge and are improved by building up over time.

For more information on localization and the localization industry, consult the Localization Industry Standards Association Localization Industry Primer.