How Far Can Machine Translation and Crowdsourcing Go?

Our project manager and I exchanged ideas about the potential of (MT) machine translation and crowdsourcing, which we’d like to feature on LIMPID blog.

I came across an article, which featured a startup company that offers machine translation and crowd translation as the core business model with cheap rates and fast service. It looks promising.

Technically, they are one of our competitors. Instead of following my impulse to blog with a subjective perspective, I consulted with my colleague to confirm some language and translation concepts.

Conversations, Remarks, and Insights

“It’s not an innovative business model, it has existed for quite a few years, if not, decades,” my colleague replied after reading the article. “To simplify, it’s “machine translation” and “crowdsourcing.”

Machine translations (MT), including Google Translate (GT), has been helpful despite the imperfect translation and funny phrases. GT, for example, has improved in the recent years, of which became useful for end-users and corporate end-users.

Technology has broken language barriers. It has bridged the gap between a “Ni hao ma,” and “Ich liebe dich.”

However, here’s the premise I’d like to expand with our project manager since I find our discussion fascinating and thought-provoking.

“How far can machine translation and crowdsourcing go? Is this model, known as translation-as-a-service, accurate enough to kick us out of translation and localization job?”

The premise points us to business correspondence and how we should use these tools.

These language translation startups offer low prices for the translation services. Moreover, the price is tempting for startups and small businesses that have limited budgets. However, in business, there’s a timeless statement often quoted in sales and web copies: “You get what you pay for.”

As you read between the lines, you know where we are heading.

Real ‘Deal’ for Business Deals

And the truth is…

“Technically speaking,” according to our project manager, “for tier 1 (category 1) language such as French and German, it’s possible to offer a low-cost translation.”

Why? It’s because they are classified as easy languages to learn and translate according to my research. You can check the category of languages on About World Language.

“But for CCJK (Asian language) Chinese Simplified, Chinese Traditional, and Korean, it’s not likely to replace humans.” Of course, with the technology and innovative tools available, MT can translate CCJK, but not as accurate as to how a linguist can work on it, especially in legal documents and localization of marketing collaterals.

If you’re dealing with these type of documents, you can’t gamble on the cheaper alternatives because you need the—‘real deal’—professional translators, editors, and linguists to manually translate the content from the source and contextualize every word on those documents.

When MT + Crowdsourcing Isn’t Relevant

You can’t adopt a ‘fast food’ business model in this area. Professional translators will have to spend the time to work on the project, and while the cost is higher than MT and crowdsourcing services offer, you also get what you pay for. Would you sacrifice the price over quality when a business deal is at stake? Plus, you need professional translators who are expert in the industry/niche of the business. Knowledge in both Chinese language and marketing for professional translations is a requisite.

Let’s say MT + crowdsourcing delivers the translated document in Simplified Chinese. Well, after the MT translates it, human editors will correct it. It’s not bad. However, our project manager drops the question: “It boils down to, what are the qualifications of these editors working on general translations?”

Moreover, I’ll add more: “Are these human editors expert in the niche of that document they’ll translate?” As what our project manager said, “With the budget they are tooting, they cannot hire a professional translator. Guess whom they will attract? —Amateurs, looking for quick money on the internet.”

When Machine Translation and Crowdsourcing Meet?

But before wrapping this blog, let’s take a look on the business model of MT + crowdsourcing. I agree with our project manager said that these services have an edge in translating short texts and emails. These kinds of services do have a market share in our industry because they are cheap and fast. You can use it to translate documents for personal use.

I tried one of the services and together with the project manager, checked the results of a translated document in Simplified Chinese. Here’s what we found out:

1. It’s not careful enough to distinguish Simplified Chinese and Traditional Chinese.

2. MT isn’t consistent with terminologies.

Final thoughts

If the project requires translation of advertising campaigns, relevant business and legal documents, and brainpower to think of localized concepts for the Chinese audience, go for the ‘real deal.’ One of the perks of hiring professional translators is that they will spend time checking the document thoroughly and use the right terminologies. They will consider the niche, industry, and wording fit for the Chinese audience.

Again, for quick translation of tweets, updates, email correspondence, an overview of documents, MT + crowdsourcing is one of the cheapest and fast alternative aside from Google Translate.