Musings on software globalization

The question “is it difficult to translate x?” or “is it more difficult to translate x than y?” is one that pops up regularly when you work with localization, and usually the question is based on a misunderstanding of the basic nature of translation. For instance, I have seen people wonder whether terms such as look and feel, which conveys a single concept, but is comprised of two nouns, would be “difficult” to translate, and whether it would be more easily translated if the term were to be hyphenated (look-and-feel). I have seen similar translatability questions about terms such as headless and green screen.

In all such cases the answer is: if the term is clearly defined (for instance in a glossary), then it is not difficult to translate it. There may be plenty of good reasons not to use a certain term – for instance, I doubt that many native English speakers know what a headless computer is (a computer with no keyboard, mouse or monitor attached) – but translatability is not one of them. The questions seem to come up because people think translators take each word and transfer it literally to the target language, so if the source term is “strange”, then the target term must be equally strange, or it must be difficult to deal with. In reality, of course, translators (at least competent translators) translate meaning, not discrete words, so if they have a good glossary which defines technical terms, then they should be able to find an expression in the target language which conveys that meaning.

So in other words:
Create a well-written source text, which uses well-defined terminology and clear syntax. Avoid jargon or slang which makes the text difficult to understand for most people in the source language. If you stick to those simple rules, you don’t have to worry about “translatability” (if you have reasonably competent translators, of course). ¹

And this advice, by the way, covers the (in my experience) single most common example of “difficult to translate” terms in computer software: long noun phrases like these: “side member bracket assembly medium”, “computer human cognition simulation games”, “hydraulic ground test stand pressure and return line filters”. These are truly difficult to translate, but only because they are very difficult to parse in English. Does “computer human cognition simulation games” refer to “games related to computer and human cognition simulation”, to “simulation games of computer human cognition”, or…? Somebody writing in English can “get away” with these long strings of nouns, since they are an entirely legitimate way of constructing English sentences, but that does not mean that they are intelligible to anybody but the writer (and maybe his close colleagues ). (See this article for a good analysis).

The difficulty of parsing the strings is brought to the surface if they need to be translated, since practically all other languages require that the relationship between the nouns be indicated somehow – thus forcing the translator to find/guess/choose the actual meaning.

Note 1: This does not mean that you can ignore language issues if you are a software developer and you are constructing messages using variables, concatenation, etc. On the contrary – this is where translatability becomes really important, and unfortunately, it is also where it is most often ignored or overlooked. I will cover some of those issues in a sequel to this post.