Eric Liu: Translation World Shifting from English-Centric to Chinese-Centric

I remember once having a meeting with a person working in an international business who didn't know the difference in quality between machine translation and human translation. He assumed everything had to be good because the translation was in letters and not Chinese characters! So many people have this very simplistic idea of language and communication.

Due to this, at Alibaba we spend a lot of time on internal communication and training, with seminars and presentations telling people why we need to translate high-quality content into different languages, and then explain the process so they can get comfortable with what we do.

There is also a flipside to this fact, and that is about translation as an industry. I have been surprised by how much translation has become a commodity. It serves an absolutely fundamental need, rather like water, but it has been fundamentally commoditized. The question for me is: how can you add value to translation if you are a company?

TAUS: So how do you address the commodity problem?

Eric Liu: One reason translation is a commodity is because it is an intermediate rather than a final product. People use it to develop a further product – rather like a microchip that you can’t sell as a retail item. So we have kept wondering how we can add more value.

One of the answers is that if you can improve products by using effective metrics, then you can derive plenty of value. At Alibaba for example, one of the advantages we have is that there is a very clear business use case – e-commerce. So, we are moving language technology and infrastructure deeply into our business processes and transactions, so we get feedback on how well they are doing in terms of satisfaction or page views and so on. These are not our ultimate goals, of course, but they are very useful metrics for us to calculate how can we improve our service to Alibaba’s business.

TAUS: Do you use your host company’s KPIs for your metrics, or your own system?

Eric Liu: We look at all the important factors. Top executives don’t need to know about the intricacies of our business. The best way is to show that language localization has added major value to the entire group.

In fact, we’re juggling with two things and it is a major balancing act. On the one hand we are trying to find ways of breaking through the shackles of the commodity issue, of being an invisible process like a call center or an administrative back-office – just something that needs to work. We try to do this as I have explained by identifying the value-add we can contribute and having the right metrics and goals.

On the other, we have to be competitive in building our technology and infrastructure. This means integrating machine translation and other automated processes, and also more innovative things we are doing in speech and post-editing, into the business. So we have to balance these two forces.

TAUS: How is Alibaba's technology strategy being supported?

Eric Liu: We have a large team here at the Alibaba Language group, including engineers and product people, all working in a tight-knit, focused unit under one roof. This covers all localization, a machine translation group and smaller working groups focused on novel services such as voice. We have our own e-commerce data, so we don’t need to collaborate to a dramatic degree on that particular front outside the company. The overall aim is to develop scalable solutions with a highly-focused team.

TAUS: Do you see any specific features in the Chinese market that gives it an advantage in terms of advances in translation business practices?

Eric Liu: I don’t feel there is any inherent advantage in the Chinese situation in terms of technology or processes, such as the fact that the Chinese didn’t invest so much in what today is legacy technology such as translation memories.

In terms of data, the current laxer privacy situation around translation data in China is likely to become less significant in the next two years. I think that the major issue then will be a lack of rich data for lots of “smaller” language pairs. But we have developed and maintained a large-scale crowdsourcing platform as part of our offering, so we feel fairly positive about being able to source translations for a broad range of pairs as a basis for building data resources.

However, the big shift in my opinion is that Chinese as a source language as well as a target language is going to play an increasingly key role in the overall volume of future translation. Chinese to other languages and vice versa could become a major feature of future business for us. Today, already, there is a similar number of Chinese speakers as there is of English speakers on the web. Scales are tipped towards Chinese in mobile, which means even more engagement and time spent. Not to mention that the Chinese are more internationally-minded when they are online. The web, in the mentality of the Chinese, has always been a key tool to discover the world outside. Over the next five to ten years, significantly more Chinese speakers will be using the web to buy products and services, book travels, engage in business, and enjoy content. All of that will often in some ways involve translations and the need for linguistic services will multiply as a result. One day more content will be translated from Chinese to German or Turkish than from English. By then, I hope that everyone in business will know what translation is and why it’s so important to us!