The Korean translation market is facing a demand for more diverse media, such as videos, games, webtoons, metaverses, and traditional document-oriented translations, following the recent tendency of cultural content to break away from existing mainstream languages, such as English, Chinese, and Japanese, and spread to Southeast Asian languages. With the high ripple power and potential of the Korean wave, emerging technologies such as artificial intelligence and automatic translation may require a process different from human translation that’s utterly dependent on the human mind.

In such a changing situation, when a company selects a language service provider (LSP)^*, only inquiring about translation experience in a similar field, the translators’ qualifications, or the company’s reputation, the LSP cannot provide all the information. The risk of pulling out becomes slightly higher. Here are three strategies that every localization manager must know when working with LSPs.

^*Language service provider (LSP): an entity that offers varying degrees of translation, localization, interpretation, and language services.

First, secure two or more LSPs.

Currently, more than 99% of domestic LSPs or localization companies are small businesses with annual sales of less than USD 3.4 million. There are many cases where translation is not even the core business of a company or what makes up most of its sales, especially if you’re dealing with a branch of an overseas company.

Considering this weak supplier market, picking just one company would be a precarious choice. It may be challenging to secure stability and scalability to meet your company’s various localization needs and the possibility of changes. But it’s best to initially focus on building a collaborative system with at least two or three companies. It’s recommended that you go through the process of supplier selection, while keeping in mind that the important thing here is that examining a company’s reputation, size, and similar field experience is only the first step in choosing two or three companies to compete with—not, in itself, a factor in the final decision.

Second, check the LSP’s systems and technology.

The current localization demand of enterprises is no longer based on the “knowledge of the language.” Unlike in the past, when no one in the company was well versed in the target language, today’s companies usually have localization managers fluent in both source and target languages^*. However, it should be noted that, in the end, they still choose outsourcing for two reasons: specialization and large-scale content production. From Lexcode’s experience in the field, many localization managers manage localization projects without performing the translation themselves because projects typically require specialized knowledge or are too large to be processed inside the organization.

^*Source language: the language being translated from; target language: the language being translated into

To carry out such a specialized and large-scale project, it’s necessary to have the appropriate tool to unify the quality of dozens or hundreds of workers working simultaneously and conduct quality inspection work through a professional workforce. Saying, “I will do my best with one person’s effort,” is the same as saying, “I will tighten a screw by hand instead of using an electric drill.” It’s an essential choice in many ways to accumulate relevant experience and data while using multiple suppliers and comparing whether such a system exists and what difference it brings to the actual project.

Third, check whether language data are collected, organized, and maximized.

Now, translation is not just a onetime thing but also a progression to the stage of further reducing localization costs and time by accumulating and using high-level language data derived from translation. Here, the localization manager looks at the stage of processing the accumulated language data and using deep learning tools to create an automatic translation solution in the relevant field. To this end, the localization manager must build language data and prepare a strategy to use them step-by-step. It would be ideal if the LSP had such a system, but if not, you must separately request and secure the language data derived from the LSP’s translation. (In reality, even if you do this, there’s a high risk of data loss when a different person starts handling the relevant project, so it’s necessary to check whether a process for gathering language data systematically is established). As the fields of natural language processing and deep learning are actively developing, as long as excellent quality and an appropriate amount of data are secured, it’s possible not only to build an automatic translation model customized for your company but also to derive various types of data from all the information.

Final thoughts

The above are three items that localization managers must consider when selecting an LSP. As the global economy is being reorganized and fragmented, LSP diversification is inevitably a topic of discussion for all companies, not just a few large corporations. Here is a passage from Scott Galloway’s “거대한 가속” (translation: “Great Acceleration”) to wrap everything up:

“테스트를 거쳐 안전하게 처리된 공급망은 더욱 강력하고 안전한 주문 처리 체계를 만들어 이해 관계자들이 실제로 느낄 수 있는 중요한 가치를 제공할 것이다. 공급망의 안전을 보장하기 위해 수십억 달러를 지출하기로 한 베조스의 결정은 미친 사람, 혹은 천재의 눈에만 명확하게 보이는 비전에서 비롯된 것이다.”

Translation: “A tested and secured supply chain will create a stronger and more secure order-processing system, providing significant value that stakeholders can actually feel. Bezos’ decision to spend billions of dollars to ensure the safety of his supply chain was driven by a vision that was clear only to the eyes of a madman or genius.”

If you’re curious about “language data,” this article is for you: From top LSP in Korea to language-data-for-AI services provider

Ham Chul Young, CEO of Lexcode
[email protected]