Even those who write for a living may not keep creating better translations top of mind. Because we work with people from a of variety cultures—at least in a cosmopolitan area like Silicon Valley—we tend to assume that everyone speaks and reads English pretty proficiently. We don’t think about how something we’ve written will sound when it’s translated into French or Bengali.
Val Swisher, the CEO of Content Rules and an expert in global content strategy, content development, and terminology management, thinks about better translations all the time. She finds that clients often blame translators when they learn that their translated content is poor. She comments, “Blaming the people doing the work might make sense. However, in my experience the cause of poor translation is often not the person trying their best to translate the content.”
There are three primary reasons for poor quality translations.
Poor source content
Source content is likely the first place to look. Long sentences are a big stumbling block. Ms. Swisher has seen 30-, 40-, and 50-word sentences. (Her all-time high is 95 words!) Long sentences are difficult to understand in English and impossible to translate. So limit sentences to 26 words or fewer. Grammatical errors are problematic for translators, so polish your grammar. Finally, be mindful of tone of voice. Your company style may be warm and friendly, with “sentence fragments, made up words, and all sorts of punctuation,” but that can create a translation challenge.
Ever worked on a project that was nearly finished, only to have additional input appear at the last minute? (Most writers have.) The same thing can happen during translation. Hiccups occur, when new and revised content appears late in the translation process. Software development typically requires feature freeze, and translation should require content freeze.
Another workflow issue is lack of competent in-country content review (ICR). This second set of eyes is critical. Otherwise, why go through the time, effort, and expense involved in translation?
Translation memory problems
Wikipedia defines translation memory as a database that stores previously translated content. The goal is to help human translators do their jobs faster and easier. Ms. Swisher has looked her clients’ TMs and found 1) multiple translations for the same content, 2) the same translation for the same content multiple times, and 3) mismatched source content and translations. A bloated or corrupted translation memory impacts translation quality and raises costs. The solution? Scrub your TM after every translation or at least at scheduled intervals.
The need to write for translation is not something most of us think of very often, but we should. I believe that when we consider potential translations, our own source content will be cleaner, clearer, and easier to read.
Aside from her role as CEO of Content Rules, Val Swisher is deeply involved with Translators without Borders, a humanitarian medical translation project. I think you’ll find this YouTube video inspiring.