by Koichi Seto, General Manager, International Business Development, Sales & Marketing

 

(The original article in Japanese was posted in October 2020)

 

Current Situation and Future Potential of “Machine Translation”

Language is the biggest barrier when conducting business with companies overseas. Now, machine translation has been expected to break through this obstacle. I am sure that many people have already tried some machine translation tools such as Google Translation. In Japanese-English machine translation, simple sentences are translated correctly in most cases. Long sentences are sometimes translated strangely or with some mistakes, but it is still a handy tool to roughly understand the meaning of the sentences. Although perfect machine translation does not exist at the moment, current translation technology is astonishing progress compared to the past when people had to read English sentences with a heavy dictionary, which took more time.

The origin of machine translation dates back to the 1950s. The first machine translation was “rule-based machine translation (RBMT)” where a computer learns dictionaries and grammar to translate based on fixed rules. After that, it developed to “statistical machine translation (SMT)” where a computer translates by statistically processing a massive amount of data. What is mostly used now is “neural machine translation (NMT)” where deep learning is applied, and some companies such as Google or Microsoft provide services utilizing this technology since 2016.

However, even using such new technology that has been said that the accuracy of translation has been improved even between Japanese and English where linguistic systems totally differ, “What do you want for lunch?” (Japanese) is translated into “What are you going to do for lunch?” (English), and “I have a reservation under the name of Yamada.” (Japanese) is translated into “I have a reservation at Yamada” (English). Further, in the translation app pre-installed on the newest smartphone, “I will miss you.” (Japanese) is translated into “You will miss me.” (English).

These errors are caused by not only the characteristics of Japanese language where sometimes the subject is omitted or the object is unclear, but also the difficulty, with only a single sentence (utterance), in expressing the meaning that changes depending on context.

I am also interested to verify whether or not the current machine translation has an ability to select more appropriate wordings or terminologies when it seems difficult to correctly translate only by the context, but pointing out the problems of machine translation is not the purpose of this article, so I am not going to talk about the details now.

Nevertheless, for the system of translation using artificial intelligence (AI) where deep learning is applied, it is possible to learn the words or terms specific to each company and reflect them in translation. Also, the cost is approximately less than 1/10 of the cost of outsourcing the work to a translator. It is obvious that further progress in machine translation in the future is expected. Also, in web meeting systems that have been necessary in business now, if simultaneous interpreting tools became generally utilizable, it may lead to more business opportunities.

However, when high-level practical translation is needed for business documents or the like, manual proofreading process called “post-editing” has to be involved after machine translation, and the time spent in this process influences the total translation cost. Moreover, manual translation is necessary also for literary works and so forth that require imaginative ideas and cannot be handled with simple machine translation.

As I mentioned above, the conclusion would bet the development of machine translation will not completely eliminate the need for learning foreign languages. Machine translation is useful for processing a large amount of information efficiently, which is similar to spreadsheet software to an extent. (However, results of data processing are not 100% reliable unlike Excel.) Excel is able to perform large-scale complicated calculations, but for simple calculations with several digits, calculators are much easier to use. Although calculators are convenient, memorizing the multiplication table is a bare minimum for living daily life. The same applies to translation. It is better for communication to know basic English words and sentences. Although new convenient tools should be largely utilized, basic skills are still required for utilizing the tools effectively. Above-mentioned “post-editing” also cannot be performed without the knowledge of the object language.

 

International Business Communication after the Achievement of Perfect “Automatic Translation”

If perfect automatic translation is achieved, will communication problems in international businesses be solved? Leaving aside a discussion about the necessity of learning foreign languages for now, I would say, "No."

When being assigned abroad, most of business people are shocked at the differences of cultures, customs and people’s ways of thinking between Japan and foreign countries. To say the least of it, such businesspeople often face difficulties in accurately conveying their intention to others in a foreign country,　and suffer many situations where the local staff do not do what they are told to do for some unknown reason.

Even if technologies enable us to break through the obstacles caused by the barriers of languages, Japanese business people still need to acquire understanding of different cultures and customs, interpersonal skills, and management skills more than ever to cope with various cultures. Although we used to just force those who work in factories overseas such as in other Asian countries to follow “Japanese styles” or “Japanese rules,” now we need to respect the foreign colleagues’ cultures and customs to cultivate a better understanding each other for better management and more effective business negotiations. There are also some countries that do not even have a system of “working at a company” yet, so the first thing to do now is to have a common view in this sense.

Nowadays, several companies have developed testing services that measure the management skills of people when they are placed in cross-cultural environments. This means that language ability is not the only skill needed for business with other countries. Although Japan has been getting ahead of other countries in the manufacturing industry so far, Japanese companies absolutely need to not only gain competitive advantages in new industrial fields but also brush up business skills including communication skills with foreign people.