SOFTWARE LOCALISATION CALLS FOR SKILLS AND TECHNOLOGY
When you want to take software or an application to a new market area, you usually need to conduct software localisation. This means translating the software’s user interface and other texts into the language of the target area. This is necessary because usually people prefer to use software products in their mother tongue.
In this article we shed more light on what types of issues are related to the software localisation. Find out, what needs to be considered when having software translated.
Where are texts located in an application?
The translation of software texts (also referred to as software localisation) is a translation type of its own. It differs from the translation of more fluent text types quite significantly. When localising software, each character set or string needs to be processed as a separate unit. The unit does not necessarily have a connection with the surrounding character strings. Will the translator be able to see the environment in which each text is located in the actual programme? That depends on the delivered material and the utilised translation tool.
Some tools especially designed for software translations enable the viewing of the text in a so-called WYSIWYG view. This enables the translator to see the actual environment in which the character string to be translated is located. Often, it makes a substantial difference, whether the string is, e.g. a dialogue title or a menu command. Let us take as an example a source language verb, such as “Print”. Thanks to the WYSIWYG view, the translator will determine whether “Print” should be translated in a designating title format (Printing) or a more imperative form (Print).
Again, the context can also be determined on the basis of other types of information. If the text to be translated is an XML file, then in some cases the file’s elements provide information on the position in which the string is to appear. (More information on translating XML files can be found here.)
Sometimes a specific string to be translated can contain very little information, e.g. a single preposition. In such cases, it is practically impossible to translate the expression without more definitive information. Additional information often proves indispensable to translate such strings correctly.
How are variables taken into account?
Typically, texts will also contain placeholders (without any text to be translated) referring to the exchangeable variables. First step is to recognise these references as such variables. Second is to formulate the surrounding text so that it accounts for the location of various variables in the sentence.
In Finnish, for example, one must consider the need for inflecting words. In English, you can say ‘save %1 to %2’. In Finnish, however, as there are no prepositions, and the direction and focus, for example, are expressed using grammatical cases, the inflection requires the use of an auxiliary word. Without any specific information, the word often used is ‘target’: save %1 to target %2.
Can texts fit in the space provided?
A practical challenge is also posed by the difference in the length of various languages. In the programme, the space reserved for the text is usually restricted. However, there are languages that have a natural tendency for longer expressions. Therefore, limiting the length of all languages strictly to the length of the English text is not necessarily ideal.
In any case, the length restrictions are a fact that needs to be considered. At its simplest, the verification of string lengths and shortening of the texts can be implemented in Excel as a separate manual work phase. If the material is sent for translation in a format which can be processed with a dedicated tool for software localisation, then the string lengths can possibly also be verified using the same tool.
Will terminology remain consistent?
If at all possible, a translation memory should be used for translating software texts. Terminology and style will remain more consistent when the translator is able to see how the previous strings have been translated. It is especially important in a situation where there is more than one translator working on the same project.
Use of a translation memory offers other benefits as well. If the text is abundant in repetitions, this will allow a more favourable price for the commissioner of the text. The translation of subsequent updates to the software is also going to be easier and more affordable, as the previous translations can already be found in the translation memory.
In addition to using an actual translation memory, creating a term list using a terminology tool may provide significant help for future translation needs.
Many programmes are used in the Windows environment. Thus, it is good to use consistent terminology with other Windows software. The software company Microsoft has facilitated this work by providing the translations for all of their most important software in all languages on their Language Portal website. The same site can also be used for downloading style guides and term databases (in TBX format).
There are also dedicated applications for the verification of internal consistency within a text, such as the Apsic quality assurance tool called Xbench.
Tips for commissioners of software localisation
- The translation of software strings is more demanding than translating conventional running text. Therefore, it is good to reserve a sufficient amount of time for completion of the translation.
- It is advisable to consult the person in charge of project management at a translation agency. They will be able to determine the most suitable file format for translation work.
- The version of the software to be sent for translation should be the most recent one. Changes made during or after the translation phase will slow down the process.
- Any restrictions on the length of character strings should be determined in advance. This decreases the amount of work needed during the software testing phase, as localised texts will fit in the space provided for them.
- All relevant reference material, such as possible previous translations or term lists, should be delivered together with the texts to be translated. Captured screenshots may also provide assistance, as long as the images can be connected with the correct strings to be translated.
- The best translation result is achieved when possible queries by the translator can be answered preferably by a person who is well acquainted with the relevant software.