A new study by Canada-based interdisciplinary laboratory, Citizen Lab, has found that popular translation services and software in mainland China automatically censor information deemed sensitive by the Chinese communist regime, choosing to skip certain sentences based on the content.
The Citizen Lab, which is based at the Munk School of Global Affairs at the University of Toronto, in its July 30 report titled “Lost in Translation: Characterizing Automated Censorship in Online Translation Services” tested five major Chinese online translation services—run by four Chinese companies and one American company—and found that they all automatically censor words, phrases, or sentences related to content deemed sensitive in China.
According to the study, the four Chinese companies are Alibaba, Baidu, Tencent, and Youdao, while Microsoft’s Bing Translator is the only foreign company allowed to be used in China.
Citizen Lab said after testing the translation services of these five companies, the researchers found 11,634 censorship rules targeting sensitive content.
“The translation services’ censorship primarily targets political and religious expression that runs counter to the Chinese Communist Party’s (CCP) agenda,” the report highlighted.
“Notably, we found a surprising absence of censorship relating to pornography, eroticism, or other more popular targets of censorship, suggesting that the censors either did not expect their censorship rules to be studied or are no longer concerned with hiding the censorship’s true political agenda,” the report noted.
The interdisciplinary laboratory, which studies information controls that impact the openness and security of the Internet and that pose threats to human rights, said these translation review rules are targeted and automatically applied and will partially or completely omit the content that users want to translate.
The Citizen Lab study pointed out that except for Alibaba, several other companies “performed censorship silently and therefore possibly without the user’s knowledge.”
Once sensitive words, lines, or sentences are triggered, the translation services will automatically and “silently omit triggering sentences or lines without any notification,” the study said.
The researchers found that among the services, Alibaba had the strictest censorship, followed by Youdao and then Tencent, while Baidu and Bing have relatively fewer censorship rules.
“Our work reveals the unfortunate reality that, even if users in China have uncensored access to news or communications platforms, what they read or write may still be subject to automated censorship if they must translate between languages,” the report read.
Almost all censorship rules apply to simplified Chinese, traditional Chinese, English, or a mixture of these languages, the study found.
However, the censorship applies more to translations from foreign languages to Chinese rather than from Chinese to other languages, it added.
The Citizen Lab study identified that most religious content censored was referred to Falun Gong, such as 法轮大法 (Falun Dafa) or ‘f a l u n d a f a’ (Falun Dafa in fullwidth characters).
“Microsoft’s Bing heavily censored Falun Gong, including many coded references to it such as 功轮法 (Gong Lunfa [Falun Gong backward]) and 发伦功 [a homonym of Falun Gong in Chinese],” the report said.
“Many references to Falun Gong-associated news media outlets were censored, including 大纪元 (The Epoch Times) and NTDTV + 新唐人电视台 (NTDTV + NTDTV),” it added.
The research further found that the censorship rules targeted content and words related to “dissidents,” “party-state leaders,” criticism of the Chinese government or the CCP, and Tiananmen, which refers to the democracy movement and the Tiananmen massacre on June 4 in 1989.
Citizen Lab said content related to the Covid-19 pandemic censored, such as 中共病毒 (CCP virus), 病毒+ 习皇 (Virus + Emperor Xi), and 近平病毒 (Jinping virus).
“Such terms are used to criticize China’s zero-Covid policy or to attribute the outbreak of the coronavirus to a failure in Chinese governance. While at the time of this writing the exact origin of the coronavirus is still unknown, it is believed to have originated in China,” as per the study.
The research found that United States politics are also targeted.
“For example, 天佑川普 (God bless [Donald] Trump) is censored. We also found 川普+ 包子 (Trump + steamed bun), the second component being another derogatory reference to Xi Jinping and his steamed bun incident,” the study added. “We found references to American hard rock band 枪与玫瑰 (Guns N’ Roses) censored.”
The hard rock band’s 2008 album “Chinese Democracy” features lyrics which are critical of the Chinese government and which make sensitive references such as to Falun Gong.
Explaining its findings on censorship behaviour, Citizen Lab’s study said, “In our testing, we found that services had a variety of censorship behaviour upon the input of triggering content, including censoring the content’s sentence or line or all text within a character distance of the content.”
“We found that only Tencent varied its behaviour based on the triggering content found,” it said.
“While most triggering content censored only within line or sentence boundaries, we found 15 rules which would censor all input: (1) 习近平, (2) xijinping, (3) 习大大, (4) 习主席, (5) Xijinping, (6) 近平习, (7) 习书记, (8) 习总书记, (9) XiJinping, (10) XIJINPING,(11) JinpingXi, (12) jinpingxi, (13) 反习大大, (14) xidada, and (15) xIDaDa,” the research added. “These are all ways of referring to Chinese President Xi Jinping, suggesting that translations mentioning him were considered so sensitive by Tencent Translate’s operators that not only should the sentence mentioning him be censored but also the rest of the output.”
The Epoch Times reported, quoting Chen Shih-min, an associate professor of political science at National Taiwan University, that the Chinese Communist Party (CCP) censorship of online translation services, especially from other languages to Chinese, is because “it worries about Chinese people learning about some of the actual situations in China through foreign information.”
“For a totalitarian and authoritarian regime like the CCP, it’s trying its best to shape the whole of China by controlling media and speech to create a false illusion of prosperity,” Chen told The Epoch Times.
“Because the CCP is an atheistic regime, it’s very worried that Falun Gong, a belief following truthfulness, compassion, and forbearance, may break the brainwashing it’s done on the Chinese people and will pose a challenge to the legitimacy of the CCP’s rule,” Chen added.
Wu Se-chih, a researcher at the Cross-Strait Policy Association in Taiwan, told The Epoch Times that the heavy censorship of religion shows that the Chinese Communist regime is not confident in its own rule. “It’s worried that if people have the freedom of religious belief, the CCP will lose its absolute interests,” Wu added. “In other words, under the Communist party’s autocratic system, when leaders want to deify themselves, if there is religious freedom, it will not be conducive for them to do it or to exercise the ruling authority of an autocratic system or a totalitarian system.”