In Google Translate, Dokdo was translated as the Japanese-style name Takeshima, and kimchi was translated as paocai, which means Chinese-style pickled vegetables, so the controversy grew. Professor Seo Kyoung-duk of Sungshin Women's University pointed out that these results are clear errors. He said notation errors on a platform used a lot around the world can wrongly tell Korea's history and culture. Professor Seo explained that the Japanese translation of Dokdo shows the name used in Japan's territorial claim as it is, and changing kimchi to paocai also makes the two foods look like the same thing. In particular, the Korean government set the Chinese notation of kimchi as xinqi in 2021, so he pointed out that the current translation result does not match the government standard either. The article also said this is not the first time this kind of problem has happened. Professor Seo said he recently had notation errors for Gyeokryeolbiyeoldo on Google Maps and a search error for Dokdo Airport corrected. He stressed that a global platform like Google should handle Korea-related history and culture notations more carefully.
원문 보기
This is not just a simple typo, it is a problem with how the translation engine makes basic choices
At first, you may think, 'Why does it translate this like that?' But Google Translate is not a dictionary that memorizes one right answer and gives it back like a person. It is a neural machine translation system that chooses the most likely expression from large-scale translation data. If you understand this, it becomes easier to see why controversial names can get results different from what people expect.
Simply put, a translation engine is closer to a probability calculator than a test grader. If one word is called differently in different language areas, and the notations are mixed in international articles, web documents, and user sentences, the model may choose the one that appears more often or looks more likely. So place names tied to territorial disputes like Dokdo, and food names tied to cultural identity like kimchi, are areas that can easily become unstable technically too.
What matters is that this result does not mean exactly the same thing as 'the platform officially adopted one country's position' right away. But it is clear that it may be the result of a general-purpose model being pulled by usage frequency and common practice. If you understand this much, you get a standard for judgment: you cannot see this controversy as only a simple bug, but you also should not immediately decide it was intentional.
When you look at a translation result, before asking 'Did someone change it with bad intent?' you can first ask 'What data did this model see to choose this default?'
Even when the same notation controversy repeats in translation, maps, and search, you will not quickly tie them all together as having the same cause.
It looks like a bug, but what is really working
On the surface, it looks like one line of 'mistranslation,' but inside, completely different principles are moving. If you know this difference, it becomes easier to understand why it is not fixed right away either.
| Comparison item | What people expect | How a real general-purpose translation engine works |
|---|---|---|
| Translation standard | Keep one correct answer fixed | Choose the expression with the higher probability among several candidates |
| Input processing | Match words one by one like a dictionary | Look at the sentence context and nearby signals together |
| Controversial proper nouns | Official spelling should come first | The frequency and customary spelling in training data can work strongly |
| How to fix | Fix only the wrong parts right away | A model update, policy review, and a separate glossary update may be needed |
| User feedback | If you report it, it changes right away | It is used as input for quality improvement, but an immediate change is not guaranteed |
Why does the name Dokdo connect right away to the territory issue?
The spelling of Dokdo is not just a simple Korean-Japanese translation issue. That is because changes in the name are directly tied to changes in records of administrative control and sovereignty claims.
Step 1: Records of Usando and Ulleungdo are called up as the origin
The Korean side sees records where Usando and Ulleungdo appear together in documents like the 1454 Geography Section of the Annals of Sejong and the 1531 Revised and Augmented Survey of the Geography of Korea as the starting point of the historical story of Dokdo. The key point here is which island the names in old documents refer to today.
Step 2: In the 1900 Korean Empire Imperial Ordinance, 'Seokdo' appears
Imperial Ordinance No. 41 of the Korean Empire included Seokdo in the jurisdiction of Uldo County. Korea explains this by interpreting Seokdo as today's Dokdo, saying it was already under control within the modern state administrative system.
Step 3: Japan's fixing of 'Takeshima' in 1905 became a turning point
In 1905, Japan incorporated the island into Shimane Prefecture and confirmed Takeshima as the official name used today. Because the act of naming and the incorporation measure went together, the spelling itself came to be read like a sovereignty claim sentence.
Step 4: The 1906 report saying 'Dokdo belonging to this county' became evidence for the Korean side
In the report by Uldo County head Sim Heung-taek, the expression 'Dokdo belonging to this county' appears. Korea sees this as evidence that the name Dokdo was actually used in the field of administration.
Step 5: Today, the choice of name itself has become a diplomatic message
In the international community, Dokdo, Takeshima, and Liancourt Rocks carry different meanings. So if automatic translation shows Takeshima like a default, in Korea it is understood not as a simple mistranslation but as a spelling that affects awareness of the territory issue.
Dokdo, Takeshima, Liancourt Rocks — why are there three names for the same island?
Even when pointing to the same landform, if the names are different, the message they carry also becomes different. This table makes it clearer why people in Korea are sensitive even to a single spelling.
| Spelling | Main side that uses it | |
|---|---|---|
| Dokdo / Dokdo | Korean government · Korean media | |
Meaning it carries The official name used by Korea, highlighting effective control and the historical narrative | ||
| Takeshima / Takeshima | Japanese government · Japanese official documents | |
| Liancourt Rocks / Liancourt Rocks | Some international documents · attempt at neutral wording | |
| Dokdo/Takeshima together | Overseas articles introducing the dispute | |
Why it becomes a cultural debate when kimchi is translated as paocai
The kimchi issue is not just about one food name. International standards, cultural symbols, and government standards have built up one by one, so now it is starting to be treated almost like a proper noun.
Step 1: Kimchi was treated as a separate item in international standards
In 2001, the Codex Alimentarius Commission (CODEX) started managing kimchi under a separate food standard. This means kimchi is closer to an independent food category, not just 'general Asian pickled vegetables.'
Step 2: In 2013, UNESCO listed not kimchi itself but 'kimjang culture'
What UNESCO listed was not just one dish, but the community culture of making and sharing kimchi. So in Korea, kimchi became more strongly established not just as a side dish, but as a symbol of everyday culture and identity.
Step 3: The 2020 paocai standard controversy made the translation issue bigger
After Chinese media reported a 'paocai international standard,' in Korea the habit of calling kimchi and paocai the same thing became a big problem. A translation habit quickly turned into an origin debate.
Step 4: In 2021, the Korean government set the Chinese notation as 'xinqi'
The Ministry of Culture, Sports and Tourism changed the public terminology guideline and stated that the Chinese translation and notation of kimchi is xinqi (辛奇). You can think of this measure as an attempt to make it clear at the labeling stage that it is 'a food different from paocai.'
Step 5: The standard changed, but the field and platforms follow slowly
Even in surveys of translation tools in 2023 and restaurant menus in 2024, the paocai wording was still there. It shows that systems change first, but data and habits change later.
Where are kimchi and paocai actually different?
If you look at this difference, it becomes easier to understand why Korea does not agree with the explanation, 'Isn't it just a similar word in Chinese?'
| Comparison item | Kimchi | |
|---|---|---|
| Basic nature | Korea's fermented vegetable food | |
Paocai A general name for Chinese pickled vegetables or pickled vegetables from a specific region | ||
| How it is made | After salting, it is seasoned again with chili powder, garlic, salted seafood, and more, then fermented | |
| Flavor structure | Spiciness, savory taste, and fermented aroma build up together | |
| Cultural context | Connected to a community culture of making and sharing together, like kimjang | |
| Misunderstanding in translation | Its unique food culture can become blurred | |
Government standards and global platform standards have different goals from the beginning
A lot of people get confused here and ask, 'The government decided it, so why does the platform not follow it?' The answer is that the target and the goal were different from the start.
| Item | Government standard writing | |
|---|---|---|
| Target | Public sector such as the state and local governments | |
Global platform translation standard General users around the world and the whole service | ||
| Main goal | Unifying administrative terms and securing formality | |
| How the writing is decided | Directives, notices, and writing guidelines | |
| Speed of change | If revised, it can be reflected in public documents relatively quickly | |
| How the private sector reflects it | Recommendations, consultations, correction requests, public opinion pressure | |
Even with the same Google, maps, search, and translation make errors in different ways
On the outside, they all look like 'Google notation errors,' but the inside structure is quite different. So you need to look at the correction path differently for each one.
| Service | Main data unit | |
|---|---|---|
| Translation | Sentence, segment, model output | |
Where errors are likely to happen Training data bias, context interpretation, proper noun handling | ||
| Search | Query interpretation, snippet, automatic translation result | |
| Maps | Place entity, address, merged data from external sources | |
| Difference in correction method | Many model and policy updates / search logic adjustment or display improvement / place data review and user suggestion reflection | |
Why one line of automatic translation is not a small problem
The numbers are different kinds, but the shared message is clear. Machine translation is already used very widely, and especially in cultural context translation, errors can build up.
So how should we read this news?
Now you can see the flow, right? Results like Dokdo→Takeshima and Kimchi→Paochai come technically from the probabilistic choice problem of a general translation model, but socially they are read much more seriously because of the context of history, culture, and sovereignty. So this news is not a story that ends with 'one translator got it wrong.'
One more important point is that even if there is a government standard, global platforms do not automatically follow it. The official spelling used in the public sector and the product operation standards of private platforms have different structures. So when you look at this kind of issue, you need to separate these questions: was a standard made, did the platform reflect it, and with what default was it shown to users.
In the end, when you read the next news, you can organize it like this. First, check which service this spelling came from. Second, check whether that name is connected to issues of history, culture, or territory. Third, check how far apart the government standard and the platform default are. If you keep just these three points, when a similar controversy comes up, you will feel much less lost and you will be able to read more accurately what the real issue is.
Do not immediately decide that one translation result shows intention. Look at technical structure + historical context + whether the standard was reflected together.
If you learn this frame, then next time a map, search, or translation controversy appears, you will quickly know what to check first.
We will tell you how to live in Korea
Please give lots of love to gltr life