Using Machine Translation to Your Advantage

Let’s pretend for a minute that you’re the content manager for a successful American company looking to expand internationally. Your boss comes to you one Monday morning with an ambitious plan — she wants you to manage the translation of your 10,000+ database of technical documents for French and Spanish-speaking markets. The clincher? You’ve only got three months and a lackluster budget to make this translation miracle happen.

Modern technology to the rescue! If you’re in this position of preparing huge volumes of content for new language markets on a tight schedule, you’ll find an ally in what localization experts call “machine translation”.

What is machine translation?

Broadly speaking, machine translation (MT) is a computer-generated attempt to match the language output that human translators perform. While the human brain has the capacity for understanding cultural context and nuances of expression within a relatively small sample, machines clearly do not (IBM’s Watson notwithstanding, perhaps). For example, if you use MT for a technical software document that includes the words “shift” and “step” throughout, the machine will have no way of knowing on its own if you’re talking about dancing (“shift” to the left; take two polka “steps” to the right) or computer programming (and by the way, this is a real-world example). Machines rely on volumes of data, processing power, and human linguists for this much-needed contextual information.

Is machine translation the right solution for your project?

Here are seven quick questions concerning your content that will help you assess if machine translation may be right for your project.

If you’re in the blue with most of your responses, keep on reading. MT may be a good translation solution for your content.

How do you stand to benefit from using machine translation?

To put it bluntly, you’ll save on the two of most precious assets in the eyes of your executives: time and money. With machine translation, you’ll be able to achieve the impossible, working within those tight budgets and turnaround times. While a human translator may successfully translate 2,000 to 3,000 words-per-day with a high accuracy rate the first time around, a machine-translation process with human post-editing can more than double productivity while achieving close to the same quality — all within a significantly reduced budget.

What’s the downside?

As is often the case, with speed comes some compromise. Machine translation can introduce inaccurate meanings, redundancies, awkward sentence structures and, at times, unintentionally funny mistakes. For example, if the translation engine is not properly set up, your document could end up talking more about the polka than you’d like. The good news is that the machine eliminates some uniquely human errors like typos.

Some mistakes are not so funny — especially in mission-critical healthcare and consumer safety scenarios — when a mistranslated sentence could mean life or death. The translation engine’s capacities are determined by humans that program it. MT can only perform aptly if the proper prep work is done to generate contextually accurate drafts, and post-translation editing is still a must to address some of the mistranslations and stylistic issues that can occur.

So how can you balance the speed and cost benefits of machine translation with potential pitfalls?

A human/machine compromise (or hybrid solution) is the answer.

Machine translation with what we call pre- and post-editing is a methodology in which a linguist “trains” or programs the machine-translation engine to correctly translate context-specific terminology, phrases with double meanings and case-based client exceptions to rules where the MT platform may have otherwise made a mistake. In the case of your polka steps, the linguist would program the word “step” to always be translated in the context of computer programming. The content is then processed by the machine translation software and then after translation, a professional human translator reviews the output and edits it for technical accuracy, style and comprehensibility.

A vital part of this hybrid approach relies on translation memory technology (a.k.a. TM, as opposed to MT). Translation memory stores previously translated and approved phrases so they can be reused rather than retranslated from scratch.

Well-managed translation memory prevents repeated mistakes with machine translation. This improves the quality of subsequent machine translations (and saves you more time and money).

How can you prepare your content so it’s machine-friendly?

As we’ve seen, machine translation is ideal for high-volume projects with relatively straight-forward communication goals. Because MT works best on short, simple, grammatically-accurate sentences, one way to optimize your content is to pare it down. Removing synonyms and flowery language will make the translation process smoother. Over the long term, train your technical writers to keep their sentences short and direct. You can also work with your translation agency’s linguists to create a glossary of terms with their approved translations.

The moral of the story

The human touch is still essential in translation when it comes to your brand voice and marketing copy. But for a lot of high-volume technical content, including procedural text, technical manuals and internal documentation, being accurate and understandable is good enough. And this can be achieved through a hybrid MT process. “Good-enough” quality is what the situation calls for when you only have three months to get to market with three million words of technical support documentation, or when anything more pulls valuable budget dollars from other places that need the extra human effort.

The one caveat is that you should make sure to ask your translation partner about source content optimization, setting up the machine translation engine and human post-editing. These additional interventions in the MT process will guarantee that your content is comprehensible and accurate in your target languages.  

Contact us today for information on how our machine translation services can make your company’s international ambitions realizable and affordable.