News Translation as a Means of Fake News Dissemination on Social Media

Abstract

As fake news continues to plague the exchange of international news, a growing concern is evolving about the translation quality as well. Social media users exchange international news stories that are translated and manipulated by unknown sources. Inaccurate or untrue translations of international news may reproduce fake news and mislead readers. The present study examined the translation quality of news stories that have been further manipulated when translated from English to Arabic. The dataset comprises 10 verified international news stories on various events. The translation quality was evaluated by the bilingual evaluation understudy (BLEU) metric. The findings revealed that basic information and context were targets of manipulation in the Arabic versions of foreign news. Several manipulating strategies were applied, such as adding new or irrelevant information, eliminating important details, and recontextualizing. The results are thought to shed light on the role of translation in helping identify fake news.

Plain language summary

News translation as a means of fake news dissemination on social media

Misinformation or fake news has been one of the issues that challenge governments, organizations, and individuals in many parts of the world. A considerable part of fake news is translated into several languages by unknown social media users and spreads worldwide. Social media users and readers are one of the most vulnerable social groups who might be affected by false or inaccurate translations of international news stories. Some forms of the negative effects of fake news are making wrong decisions/impressions, violating rules, and violating social norms due to learning inauthentic content. The present study examined the translation quality of news stories that have been further manipulated when translated from English to Arabic. There were two objectives in this study: to explore the strategies of manipulating news translation in Arabic and to raise awareness about the role of translation in sharing worldwide news. This study showed that foreign news was translated and manipulated in Arabic by unknown users resulting in fake news. Basic information and context of foreign news were targets of manipulation using various strategies to mislead readers such as adding new or irrelevant information, eliminating important details, and re-contextualization.

Introduction

A large number of news stories about various events are circulated worldwide. Official news outlets, for example, news agencies, news media, and journals, compete with so-called citizen journalism, in which individuals disseminate news and information. Although official news outlets are assumed to publish information following professional and institutional standards, individuals often use social media platforms or websites to learn about or report news and information. Unfortunately, many users and readers still consume such news and information though official news outlets do not necessarily present them. The reliability of such news and information remains a continuing concern. Fake news from unknown sources infiltrates social media and reaches many users and readers. Several research studies examined the psychological practicalities of trust in rumor, fake news, and misinformation (Flynn et al., 2017; Pennycook & Rand, 2019). Multiple studies underscore such news spreading through the internet, web blogs and social media (Grinberg et al., 2019; Guess et al., 2019; Vosoughi et al., 2018) and influenced the information ecosystem.

The advancement in technology enabled people to use the internet, social media sites, and other related channels to gather and disseminate information which is also known as citizen journalism. This again created havoc and suspicion on most of the news circulated through various local channels multiplying concerns over whether citizen journalists are as reliable as trained professionals. Moreover, many people in different parts of the world post instant text and visual reporting of local events. Similarly, a few studies investigated the political and social impacts in various contexts (Fine & Ellis, 2010; Huang, 2017). However, people in countries affected by political upheaval and often in countries where print and broadcast media are controlled by the government have used a variety of technological tools to share information about hot spots. Besides, we find studies suggesting strategies for the effectiveness of correcting misinformation (Aird et al., 2018; Berinsky, 2017; Clayton et al., 2020). Despite its immense popularity, the debate on citizen journalism is still on to check its accuracy.

With the increasing demand for learning about world events instantly, social media users and readers may be unable to discern fake news, especially when written in foreign languages. The situation becomes more complex when news stories are translated and undergo further changes to reproduce fake news stories in the target language and culture. One of the classic cases is back to 1956 when Soviet Secretary Nikita Khrushchev said in Russian “Мы вас похороним” that was translated as “We will bury you,” while the intended meaning was “We will outlast you” (Macdonald, 2015). Another classic case is that of Willie Ramirez who has been misdiagnosed due to interpreting difficulty with the word “intoxicado” in Spanish which was misunderstood as “intoxicated” in English (Price-Wise, 2008). These classic cases of mistranslation caused misunderstanding at the time of traditional media, however, the phenomenon has become more critical with social networks nowadays.

The issue should not be reduced to social media users making translation errors while reporting international news. Rather, it seems more deliberate reproduction of international news as fake news. The consequences of exchanging manipulated translations of international news on social media are serious such as misleading readers or users and, eventually, affecting the flow and exchange of international news. This critical situation should receive urgent attention from researchers since it is also emerging in the Arab world. Thus, the study’s significance lies in the need to question the authenticity of the Arabic version of foreign news circulating among social media users and readers. To the best of the researchers’ knowledge, this study is the first of its kind to address news translation as a means of fake news dissemination in Arabic. By studying manipulated news translations in Arabic, the strategies of manipulating or falsifying translations can be unveiled as a preliminary step toward understanding the phenomenon. Researchers and decision-makers from media studies and news agencies can learn about some aspects of inauthentic translation by unknown sources. Accordingly, efforts may be directed toward developing solutions to detect false translations, preferably within the efforts to fight fake news and misinformation. Thus, it becomes of paramount importance to address fake news and translation as part of efforts to counter misinformation. This study has two objectives: to explore the strategies of manipulating news translation in Arabic and to raise awareness about the role of translation in sharing worldwide news. The research questions are as follows:

1. What is the overall translation quality of manipulated news in Arabic?

2. What aspects of the texts are the target of manipulation in the Arabic version of the news?

3. What strategies are employed in manipulating the translation of foreign news into Arabic?

Defining and Classifying Fake News

One challenge that arises when approaching the phenomenon of fake news is defining or describing the concept. Some scholars (Lazer et al., 2018; Southwell et al., 2017; Tandoc et al., 2018; Wardle, 2017) have attempted to capture the intricacy of the term and introduce a typology or operationalized definitions. Southwell et al. (2017) developed a typology of fake news that distinguishes between misinformation and disinformation based on truth. Tandoc et al. (2018) introduced a typology of six terms: Satire and parody are humor-oriented news stories based on real and fictitious information, respectively, with the audience often aware of such pieces’ intent. Fabrication draws on non-factual information presented similarly to typical news, with the readers unaware that it is untrue. Manipulation is usually associated with false narratives using images or videos. Finally, products or services advertise false news, whereas propaganda signifies news stories concerned with effectuating political ends. Similarly, Wardle (2017) created the seven-point typology to underscore the different types of information disorder that exist, and the typology includes: Satire/Parody, False Connection, Misleading Content, False Context, Imposter content, Fabricated, and Manipulated content. Lazer et al. (2018) stress the role of organizational process and intent in describing fake news. The overlap between definitions makes it difficult for individuals to discern the variants of fake news. In this paper, fake news refers to false news exchanged on social media, regardless of the user’s or reader’s intent.

Fake News and Social Media

Although fake news is not a new phenomenon, a significant surge of fake news on social media occurred during the 2016 U.S. presidential election. Since then, social media has played an important role in disseminating fake news among a large number of users. Free access to social media, almost without editorial control, encourages users to participate in publishing their own content regardless of authenticity, which is further enhanced by social media’s power to reach a wider public through the internet. Unfortunately, fake news stories go viral on social media and reach the public faster than original ones, as Vosoughi et al. (2018) found in a study of the spread of fake news over 10 years.

Financial and ideological purposes can motivate the dissemination of fake news (see, for instance, Allcot & Gentzkow, 2017; Figueira & Oliveira, 2017; Lazer et al., 2018). In other words, user traffic on the platforms that publish misinformation has become a revenue generator through ads and marketing. Moreover, sharing false news is more likely to occur with group members who are ideologically defined. The ongoing flow of news on social media makes it difficult for users to ensure the credibility of each news story. This difficulty may be partly associated with the many important events or stories happening and developing across the world. Additionally, manipulated digital features, such as misleading links and fake websites, require advanced skills to verify their informational quality. This situation has prompted government departments, organizations, and public figures to respond to the credibility of news concerning them. Freeze et al. (2021) found that the quality and validity of misinformation warnings can vary widely. They studied social cognition and forensic psychology and discovered that retrospective warnings about misleading news can help people discard erroneous information. In addition, politicians and pundits worldwide use the term “fake news” to discount news reports and organizations they disagree with, to control political narratives and to shape public opinion (Kang, 2007; Qin & Zhang, 2020; Tandoc et al., 2018; Wardle & Derakhshan, 2017; Wong, 2019). Governments can now order platforms to remove content deemed “false” and “against the public interest,” and to post corrections (Wong, 2019). However, individuals are more vulnerable to fake news in that they may learn false information and make inappropriate decisions that negatively affect their lives.

News Translation

Social media users translate news into different languages and exchange it worldwide. The relevant literature on fake news and news translation focuses mainly on the manipulation of news translation by news agencies. Bielsa and Bassnett (2009) examined news translation by news agencies and found that “information that passes between cultures through news agencies is not only ‘translated’ in the interlingual sense, it is reshaped, edited, synthesized and transformed for the consumption of a new set of readers” (p. 2). Other researchers have explored how news stories and news reports are subjected to institutional control when translated into different languages and for particular audiences (see Ping, 2018; Valdeón, 2016). However, Stetting (1989) introduced the concept of trans-editing to explain the intervention between editing and translating and to justify the translator’s need to make a certain number of changes or modifications in the target translation. Some researchers view the changes in news translation as institutional practices due to political and ideological purposes (Darwish, 2006; Holland, 2006; Kang, 2007; Kuo & Nakamura, 2005; Orengo, 2005; Qin & Zhang, 2020), rather than unintentional mistakes. This line of research considers the broader framework of translation, ideology (Baker, 2018; Munday, 2007; Valdeón, 2009), and manipulation in journalistic translation (Bassnett, 2005; Schäffner, 2005).

Research scholarship on fake news and news translation is generally scarce. Amjad et al. (2020) attempted to employ machine translation for fake news detection using an English-Urdu language pair, but there was no significant improvement. Saghayan et al. (2021) followed a similar approach to detect fake news using Persian-English machine translation, but the study did not achieve an efficient classification model. By machine translating a dataset of fake news from English to Spanish using deep learning architectures, Ruíz et al. (2021) found some improvement, though they require high-quality translation. The aforementioned studies used machine translation to mediate the process of developing computational models to detect fake news, rather than evaluating the translation quality of manipulated news. A more relevant study by Phanthaphoommee (2023) investigated fake news as an outcome of translation within the Thai context and identified the linguistic features in the target texts, describing the phenomenon as incomplete translation due to the translators’ failure to translate source news properly. With the spread of global fake news, the quality of translation poses a challenge to detection-based solutions. Social media users are no longer receiving translations of international news stories only from news agencies; they also participate in translating and sharing unreliable news. This participation leaves the quality of the translation questionable and further complicates efforts to counter fake news.

Upon building up a strong foundation, this study examines how news translation serves as a conduit for fake news in Arabic media, focusing on uncovering manipulation strategies in translated content. The study breaks new ground by investigating translation quality, identifying targeted textual elements, and analyzing manipulation techniques in Arabic news translations. In addition, this research aims to equip media professionals with tools to detect false translations while contributing to the broader fight against misinformation in cross-cultural news dissemination. The arguments pave the way for research methodology, data collection, analysis, discussion and conclusion that are deliberated in the succeeding headings.

Methods

The translation quality was evaluated through two stages: using an automatic scoring system and qualitatively analyzing the strategies of manipulation or intervention. Automatic scoring systems can be employed to assess the quality of machine translations by comparing them to human-generated reference translations. They apply automated metrics to quantify the quality of machine-generated translations against human references. In the context of this study, the task involved two central factors: candidate translation which refers to the source translation intended to be evaluated, and reference translation which is used as a benchmark to compare against the candidate translation.

The present study utilized the Bilingual Evaluation Understudy (BLEU) metric to evaluate machine translation performance. BLEU is an automated metric that compares a machine-generated translation to one or more reference translations, providing an objective and standardized measure of translation quality. In addition, it calculates a score between 0 and 1, where 0 indicates a very poor translation and 1 indicates a translation that is identical to the reference. The metric assesses the n-gram overlap between the machine translation and the reference, placing greater importance on longer sequences of matched words. This helps BLEU capture both adequacy (how much of the meaning is preserved) and fluency (how natural the translation sounds). It offers more advantages compared to other evaluation methods that rely immensely on subjective human judgments. Moreover, it is more consistent, less prone to semantic gaps, and less dependent on the quality of reference translations. This makes BLEU a more reliable and efficient way to measure translation performance. While BLEU may not perfectly capture all nuances of translation quality, it provides a quantitative, data-driven approach that complements human evaluation.

Papineni et al. (2002) proposed the metric, which compares and counts n-grams in the candidate and the reference translations by the number of matches, regardless of word order. The larger number of matches suggests better translation quality, which is reflected in a higher cumulative score. The BLEU metric does not consider the meanings of words or synonymous relationships.

In this study, the Arabic anonymous translations of the selected news stories were the candidate translations, whereas the reference translations were obtained from a machine translation tool, namely Google Translate. A qualitative analysis followed the quantitative results, considering the interventions imposed on news translation as explicated by Bielsa and Bassnett (2009), such as changing titles and leads, eliminating unnecessary information, and adding background information. Although the modifications to the source texts described by Bielsa and Bassnett (2009) concern news agencies, some of them were pertinent to this study, where the translators are unknown. Other strategies of modification or alterations that translators applied to the original texts were also explored. Instances of low-quality translation were examined to reveal the changes or shifts in the real news stories, as well as the salient linguistic aspects.

Data Collection and Processing

The researchers reviewed several international news stories with falsified versions in Arabic social media and selected a small but verified sample to allow for a considerable quantitative and qualitative analysis. The dataset comprises 10 international news stories whose translations have been manipulated in Arabic social media. The selection criteria were as follows: the availability of both the original news and its corresponding falsified Arabic translation, verification of the falsified version by a recognized fact-checking organization, and each news story had to have been circulated in Arabic social media. The dataset was collected from the Anti-Rumors Commission (2024), which operates a fact-checking website to evaluate the credibility of news mainly in Arabic and partly in international news stories that are translated into Arabic. The dataset was collected and saved in a spreadsheet, where sentences for each news item were written from the verbatim texts. They were organized into three categories: the original fake news in English, the anonymous translations on social media (candidate translations), and the machine-translated version provided by Google Translate (reference translations), as shown in Appendix 1. The natural language processing tools (NLTK; Bird et al., 2009) were used to process the dataset, including tokenization and defining each news story to generate individual scores. The BLEU-sentence function from NLTK was applied rather than the corpus function due to the inaccurate punctuation marks of the Arabic texts, which makes identifying sentence boundaries difficult.

Results and Discussion

The evaluation of English-Arabic translations using the BLEU metric revealed overall low accuracy and indicated significant discrepancy. The analysis revealed concerning patterns in Arabic news translations, with BLEU scores ranging from 0 to .14 and a notably low average of .054. This quantitative assessment was complemented by significant word count discrepancies - original texts averaged 36 words, while candidate and reference translations averaged 25.5 and 32.6 words respectively, indicating substantial content modification. Further, the distribution of scores over the news stories ranges from 0 to .14, with an average overall score of .054. Table 1 below summarizes the quantitative results.

Table 1.

Translation Quality by BLEU Scores.

News/document	No. of words			BLEU score
	Original news	Candidate translation	Reference translation
N1	36	20	40	.047
N2	30	52	24	.094
N3	9	14	11	.062
N4	22	44	23	.065
N5	17	25	12	.040
N6	86	45	84	.028
N7	21	17	21	.092
N8	21	12	17	0
N9	80	31	63	.045
N10	40	35	31	.142
Cumulative avg.				.054

The quantitative results were based on the number of matching words. There was a discrepancy in the number of words between the candidate and the reference translations. The average number of words in the original news was 36, whereas the average of the candidate and reference translations was 25.5 and 32.6, respectively. Although the meanings and word orders are typically not considered when computing BLEU scores, these factors were believed to show minor differences when validating the dataset by human translators.

The qualitative analysis uncovered three primary manipulation strategies: content addition, detail elimination, and contextual modification. It revealed the prominent aspects and strategies of manipulating foreign news translation in Arabic. The overall aspects of the text that were targets of manipulation can be described as manipulating basic information and contextual elements. Several strategies were used to execute these aspects, such as adding new or irrelevant information, eliminating important details, and recontextualizing by modifying the setting (time, place), participants, and events. Firstly, content addition involved inserting fabricated elements while maintaining minimal original content - exemplified by transforming a story about a 14-year-old criminal case into a sensationalized tale of bank hacking. Secondly, detail elimination proved particularly insidious, often stripping crucial contextual information that altered the fundamental nature of stories, such as removing simulation disclaimers from visualization reports. Lastly, contextual modification emerged as perhaps the most sophisticated strategy, involving the complete reframing of events within different geographical, temporal, or cultural contexts. As a result, the study observes that one or more of these strategies or techniques may be present in the translated version of the news in Arabic; however, the following analysis demonstrates selected instances of manipulated translations.

Addition of New or Irrelevant Information

In N3 below (See news items in Appendix 1 for the Arabic candidate translation of each news item), the basic part of the news about a child has been retained, and new information has been added: hacking a bank and transferring its balance. Thus, while the child remains the focus of the news, the rest has been significantly altered to suggest another story. Almost the same strategy was applied to N1, where news about a U.S. decommissioned ship participating in a sinking exercise has been changed and attributed to the U.S. Navy posting footage of an attack on a U.S. battleship by Houthis in the Red Sea. Here, the news about a U.S. battleship has been retained, and new information has been added that changed the news story significantly. N4 is an example of adding irrelevant details in the form of a statement by the Australian Prime Minister. Although the Prime Minister’s statement may be true, it was not part of the original news.

[N3]

Original news: 14-year-old charged with killing father sparks a protest movement.

Candidate translation in English: A 13-year-old American child hacked a bank and transferred all of its balance, 300 billion, to his account.

[N1]

Original news: #ICYMI Along with live fire from partner nation ships and aircraft, #YourADF participated in the sinking exercise of the decommissioned ex-USS Rodney M. Davis (FFG60) in waters 15,000 feet deep, 50 nautical miles North of Kauai during #RIMPAC20.

Candidate translation in English: U.S. Navy posts footage of the moment of hitting the U.S. battleship (USS Carney) after an attack by Houthis in the Red Sea.

[N4]

Original news: A little 3D visualization of the fires in Aus. This is made from data from NASA’s FIRMS (Satellite data regarding fires) between December 5, 2019 and January 05, 2020.

Candidate translation in English: NASA takes a photo of the fire outbreak in Australia. The effects of the large fires that broke out in Australia, where flames rise out of control. Australian Prime Minister Scott Morrison said that the death toll in the current fire season, which began in September, reached 23 people, including 12 in the fires this week.

Elimination of Important Details

Almost all the news stories in the dataset had important details eliminated. This technique may or may not be accompanied by replacing the details of the original news with other incorrect ones. N2 and N4 have an important detail in common that describes the nature of news as simulation or visualization. This key piece of information has been eliminated. On the other hand, N9, as shown below, is about a Chinese kid who could hack a school’s online homework system. Essential details in the news about the kid—that is, being young and ingenious—are missing. Whereas the original news states that the Chinese kid hacked a school’s online homework system, the Arabic version replaced the school’s online homework system with Facebook, WhatsApp, and Instagram.

[N9]

Original news: The kid is widely renowned in China as a computer genius. However, what stands out most about him is his ability to crack through codes and systems. Yes, Wang is one of the youngest hackers in the country, and his story is a remarkable one indeed. It all began for the kid, as an attempt to get away from spending hours on doing homework. And so Wang hacked into the school’s online homework system and took the easy way out.

Candidate translation in English: The Chinese hacker Sun Ji Soo, who is responsible for stopping Facebook, WhatsApp, and Instagram services, is 13 years old. A photo of the Chinese child hacker who worked to stop the programs (social media platforms)?! How about our children, What is the strongest/most powerful thing to them?

Re-Contextualization

Re-contextualization refers to modifying the non-linguistic factors of the news story and reproducing it in a different context. In N7, the original news is about a Jewish millionaire celebrating Hanukkah in Bahrain. The event has been changed to one celebrating the founding of a claimed Saudi-Israeli Friendship Association, and the place has been changed from Bahrain to Riyadh.

The event and where it took place have been changed in N8 from a visit to a farm in Iowa to a ceremony for opening a dairy company in Boston. In the original news, the participants were former President Trump and farmers, whereas in the Arabic version, they were former President Trump and a claimed Saudi businessperson.

[N7]

Original news: American Jewish millionaire Lazer scheiner and his friends throwing a Hanukkah party in Bahrain with the local Arabs isn’t it amazing.

Candidate translation in English: A celebration on the occasion of the founding of the Saudi-Israeli Friendship Association in Riyadh… Allah suffices me, and He is the best disposer of affairs.

[N8]

Original news: Earlier today, former President Trump visited a farm in Leighton, Iowa, and spoke to farmers and signed a John Deere combine.

Candidate translation in English: Donald Trump at the opening ceremony of the Saudi Dr. Awad Manahi Al-Balawi Dairy and Dairy Products Production Company in Boston.

The linguistic analysis of the dataset emphasized proper nouns, semantic aspects, and language use in relation to variety, genre, and style. Proper nouns referring to the names of specific people, places, and things were targets of manipulation, often associated with contextual information. Examples of this manipulation include inserting a false name of a claimed businessman in N8, changing the place of an event in N7, and changing the name of the U.S. battleship in N2. Using expressions with religious meanings was salient in the news, which includes commentaries often in the form of prayer or praise to Almighty (God); that is, Glory be to you, Lord; Allah suffices me; and He is the best disposer of affairs (see N2, N5, N7, N10). Additionally, words with negative connotations or meanings were present in the Arabic translations, such as attack, die, catastrophe, hoax, and hack. These words were not part of the original news and are suggested to express attitudinal meanings. Furthermore, the translated versions of the news in Arabic were characterized by using a mixed variety of modern standard Arabic and colloquial Arabic. Arabic translations of news combined formal and informal language and were informed by social media genres. There were also spelling mistakes and inaccurate punctuation.

It should be reiterated that this study assumed that social media users, rather than news agencies, professional translators, or organizations, translated the Arabic version of international news. The aspects and strategies of manipulated translation in the Arabic version of the news in the previous paragraphs agree partly with the modifications exerted by news agencies described by Bielsa and Bassnett (2009), particularly, the addition and summarizing of information. N7 adds an extra comment or explanation to the news, and N9 introduces an inaccurate summary of the original news. Such modifications not only result in new text or news, as Bielsa and Bassnett (2009) point out, but also reproduce false news.

The low Arabic translation quality scores, along with the associated strategies and linguistic aspects, reveal unprofessional and unethical work. This behavior can be explained in terms of whether it is deliberate. The manipulation and dissemination of the Arabic translation of international news primarily involves two parties: those individuals or users who intentionally manipulate the Arabic translation and are aware of its content due to their knowledge of the source language, often English, and those who lack knowledge of the source language and depend on the translations.

The first group may be ideologically motivated and seek to promote or criticize certain ideologies for sociopolitical or other purposes. This explanation agrees with the findings from the related literature (see Allcott & Gentzkow, 2017; Au et al., 2021; Figueira & Oliveira, 2017; Lazer et al., 2018; Tandoc et al., 2018). Inserting personal comments, false descriptions, and expressing attitudinal meanings are instances of implicit tactics to influence readers’ perceptions of the news. Likewise, adding irrelevant information or eliminating important details from the original news can be deliberate acts to guide the explanation or interpretation of the news to predefined points.

The second group of individuals or users rely on the available translations and have little, if any, knowledge of the source language of the international news. Citizen journalism motivates this group, which seeks to contribute to the dissemination of news and information to the public. A participatory culture may influence such a group (Gretter & Yadav, 2016), which is eager to create and publish media online. Although this behavior may be well-intended, it contributes to the spread of fake news and misleading readers. Unfortunately, the strategies and aspects of manipulated translations of international news in Arabic often go unnoticed by readers or social media users. Worse, they spread widely, exposing readers and users to the risks of imperfect learning, making inappropriate decisions, and forming negative impressions or perceptions. Thus, it is time to pay attention to the role of foreign news exchanges, especially on social media, and whether they are properly translated. Additionally, efforts to counter fake news and raise public awareness should include foreign news translation.

Against the backdrop of this discussion, these manipulation patterns suggest complex underlying motivations operating at multiple levels. At the individual level, two distinct groups emerged: deliberate manipulators with source language knowledge who appear motivated by ideological or political agendas (Fine & Ellis, 2010; Huang, 2017) and well-intentioned but potentially naive citizen journalists who lack source language expertise. The former group’s activities often reflect sophisticated manipulation strategies aimed at shaping public opinion, while the latter’s contributions, though well-meaning, inadvertently perpetuate misinformation. Furthermore, at a broader societal level, several factors appear to influence these practices. The rise of participatory media culture has created an environment where rapid information sharing is valued over accuracy (Qin & Zhang, 2020; Tandoc et al., 2018; Wardle & Derakhshan, 2017; Wong, 2019). Political polarization may drive the deliberate manipulation of translations to serve specific narratives, while cultural factors—evidenced by the insertion of religious expressions and local colloquialisms—suggest an attempt to localize content for greater resonance with target audiences.

The study’s findings and their discussion at length underscore a critical vulnerability in cross-linguistic information flow, particularly relevant in today’s interconnected media landscape. The amalgamation of low translation quality and informed manipulation strategies creates a perfect storm for misinformation propagation, posing significant challenges for media literacy and information verification efforts. Thus, efforts should start to respond to the dissemination of false translations. The present study suggests two solutions: first, social media platforms should assume a more active role in curbing false translations by applying advanced technologies to verify news translations. Second, professional translators can contribute to countering the spread of false translations by debunking or exposing false translations to social media users and, simultaneously, raising their awareness about the problem.

Conclusion

Social media users have become more responsive to world events; thus, official news agencies and media are no longer the only news publishers. The study found that the overall translation quality of manipulated news stories in Arabic was very low, with BLEU scores ranging from 0 to .14 and an average of .054. The main strategies used to manipulate the translations included adding new or irrelevant information, eliminating important details, and recontextualizing the news by changing key elements like locations and participants. Moreover, these findings are significant because they reveal how translation can be exploited as a means of disseminating fake news on social media. The low-quality, manipulated translations can mislead readers who are unable to verify the original sources. This is especially problematic for non-English speakers who rely on these translations. The results can benefit translators, news media researchers, and fact-checking websites by increasing awareness of translation-based fake news tactics. Therefore, identifying these manipulation strategies can inform the development of tools and policies to detect and mitigate the spread of translation-fueled misinformation. This study was limited to a small dataset of news written in English and translated to Arabic by unknown users, yet widely circulated in the Arab world. Nevertheless, further research may commence to explore this phenomenon across other language pairs and in larger datasets and investigating the motivations and processes behind the intentional manipulation of news translations would also provide valuable insights. Largely, this study underscores translation as an overlooked trajectory for fake news that warrants greater attention.