Machine translation of Chinese internet literature: infringement, exploitation or empowerment

This article analyses the use of machine translation for the translation of Chinese internet literature within fan communities. It underscores the ethical dimensions of user-generated translation, considering fan perspectives. While recognising MT's role in enabling monolingual translators to partake in translation, ethical concerns such as potential infringement and exploitation are addressed. The proposed solution involves collaborative efforts with established platforms or even translation companies. The global market for Chinese internet literature presents prospects for MT integration and collaboration with fans. However, effective utilisation necessitates crucial educational support.


Introduction
In the past, machine translation (MT) has been applied more to pragmatic texts that focus more on informative or operational functions rather than to creative texts.Literature has "traditionally been viewed as fundamentally beyond the ken of Machine Translation systems as well as computer-assisted translation systems" (Hadley et al., 2022: 7).
However, the development of an additional commercial sphere with online shopping as well as the rise of online media have had a tremendous impact on the availability of various forms of literature; the democratising emergence of electronic literature and the participatory Web 2.0 have created the need for different forms of translation practice, such as user-generated translation (UGT), a meta concept which refers to translation "undertaken by unspecified self-selected individuals (…) in a voluntary manner without monetary reward" (O' Hagan, 2009: 97).This is very similar to online collaborative translation (OCT), another meta concept that embraces translation crowdsourcing, unsolicited online translation and various forms of online fan translation (Zwischenberger, 2022: 1).While these new forms of translation may have quite some overlap at first sight, they differ in their specific emphasis or scope, capturing different aspects of translational practices.Fan translation, a form of UGT, emphasises the unpaid translation of popular media texts, and is carried out by fans (Luong & Evans, 2021: 164).Meanwhile, when fan translation takes place in an online environment, it also represents a form of OCT, similar to crowdsourced translation (Jiménez-Crespo, 2017).However, the user generated nature of fan translation differs from crowdsourced translation: the latter is "generally solicited by the content owners" (McDonough Dolmaya, 2019: 125).The usergenerated approaches prompt us to re-examine the relationship between translation technology and literary texts, particularly as MT has become increasingly sophisticated and is considered an option for literary translation.Chinese internet literature can be considered a fertile ground for the reassessment of this relationship (Zhang, 2023).
Chinese internet literature is a digital entertainment genre that emerged in the late 1990s (Hockx 2015).It refers to "Chinese-language writing, either in established literary genres or in innovative literary forms, written especially for publication in an interactive online context and meant to be read on-screen" (Hockx, 2015: 4).While a broader definition would imply that Chinese internet literature also includes digitised print literature such as "classic works published in print before the new media age that have been digitised" (Chen, 2012: 542), this article will in fact adopt a narrower definition, focusing 162 on literary works "created by netizens 1 and published on the Internet, which is meant to entertain internet users or to engage them" (Ouyang, 2008: 4, in my translation).Although it can be assumed that the level of the texts involved is easier than Chinese literary classics, Chinese internet literature does share a readership with other literary works and cultural products such as light novels (novels of a lesser literary complexity), Japanese anime and manga, entertaining internet users with popular and down-to-earth content (Shao et al., 2018: 120).Following the unexpected success of the international dissemination of Chinese internet literature through the internet in the early 2010s, carried out by individual ethnic Chinese, overseas Chinese, and Chinese language learners who are passionate about this genre and frequent platforms and websites offering translations, particularly in English, there has been a proliferation of such platforms both domestically and internationally (Wu & Lin, 2022: 132-133).The overseas market for Chinese internet literature has expanded over the past decade (iResearch, 2021), with fan translation becoming one of the predominant methods of translating this genre (He, 2022: 104).As of 2022, the overseas market for Chinese internet literature has surpassed three billion yuan.Over 16,000 titles have been translated and distributed to international markets, attracting more than 150 million readers from over 200 countries and regions worldwide (Yu & Feng, 2023).The recent advancement of quality in MT output with neural machine translation (NMT) has further reduced the entry barrier to literary translation.This has given rise to the emergence of "monolingual translators".These monolingual translators have "no knowledge of the source language but aided by postediting and the display of translation options" (Koehn, 2010: 537) and by utilising machine output, their role as passive spectators has shifted to active "prosumers" or "produsers" (see Bruns, 2008;Tapscott & Williams, 2006;Toffler, 1980).This user-generated production represents "a liberating, empowering and democratising process" (Zwischenberger, 2022: 8).
Although the international distribution of Chinese internet literature has increased owing to the activities of fan translation communities, the fact that most have not obtained permission to use the novels for translation purposes means that they are operating on the fringes of the law (Xiao et al., 2022: 10).This widespread phenomenon has been taking place in fan communities for other cultural products, such as audiovisual productions, see for instance, the largest fansubbing 2 website in China Renren Film and Television (Chen & Wei, 2021).However, with regard to Chinese internet literature, the underlying problems may be more complex due to the involvement of fan translators Small-scale studies have indicated that monolingual machine translators can enhance the quality of MT output (e.g., Koehn, 2010;Mitchell et al., 2013), especially when postediting is done by "a domain expert in the material being translated" (Schwartz, 2014: 41).In the case of Chinese internet literature, monolingual fans can be regarded as domain experts when compared with professional translators who might very well not be fans of this genre or who are not in the position to spend their valuable time on fan endeavours.Post-editing and having to rely on the machine translated output is not new and has been supported by scholars like Pym, who in 2014 already envisioned that "MT, along with its many hybrids, is destined to turn most translators into posteditors one day" (Pym, 2014: 488).This assertion holds even greater validity today, particularly with the enhancements seen in neural machine translation, surpassing the statistical machine translation that Pym originally addressed.Therefore, the influx of monolingual translators into the fan-based translation communities involved with Chinese internet literature, coupled with the democratisation of translation, raises concerns not only about copyright and translation quality but also about the professional survival of literary translators.
What if readers become accustomed to translating themselves using free MT systems rather than paying for licensed titles translated by professional literary translators?These concerns make the matter of taking a closer look at the online collaboration between humans and machines translating Chinese internet literature even more pressing.This article therefore aims to investigate the role of MT in the translation of Chinese internet literature by user-translators in an online collaborative context and its implications for future literary translation.

MT and literary translation
While translation technology like MT, and even CAT (computer-aided translation), has often been viewed "as either inappropriate or a threat to the skills and livelihoods of literary translators" (Youdale, 2019: 199), recent research has attempted to explore the potential behind the use of MT in literary prose translation (e.g., Moorkens et al., 2018;Toral & Way, 2015a, 2015b;Voigt & Jurafsky, 2012), as well as for the translation of poetry (e.g., Genzel et al., 2010;Greene et al., 2010;Lee, 2011).Some studies have demonstrated the possibilities of using MT for literary language.For example, Toral and Way (2015b) showcase the effectiveness of their tailored SMT system for a best-selling Coriolanus and The Merchant of Venice.Their findings reveal that "DeepL translations exhibit a certain degree of creativity in their use of translation methods" (p.1) and they argue that NMT "can thus be used in literary translation" (p.16).
As evidenced by the recent volume Using Technologies for Creative-Text Translation (Hadley et al., 2022) and the forthcoming volume Computer-Assisted Literary Translation (Rothwell et al., 2023), there are indeed optimistic discussions about the future of using MT for literary texts.However, it is crucial to approach this optimism with a degree of caution due to "the ethical and legal issues arising from the application of new technologies" (N.Wang, 2023: 4).Taivalkoski-Shilov (2019) discusses a number of ethical questions regarding the machine-driven translation of literary texts based on a discussion of translation quality, literary multivocality and textual ownership.Kenny and Winters (2020) examine ethical issues related to the human translator's voice in machinetranslated literary texts.However, as technological development lowers the threshold for accessing translation and demonstrates the potential for liberalising it, ethical concerns extend beyond traditional values like loyalty and -the most discussed within a literary context -the literary voice.Thought-provoking questions like "where would the present MT algorithms lead us when extensively applied to literary translation?"(H.Wang, 2023: 2) have been raised, soliciting reflection on the evolving relationship between humans and machines.The ethical concerns for MTs in non-literary texts, including the "ownership of translation resources, privacy and confidentiality of translation data, professional identity of translators, productivity and payment, translators' codes of ethics, and the potential contribution of tools to linguistic hegemony or linguistic diversity" (Bowker, 2020: 265), will also be considered in the context of literary translation.

MT in an online collaborative context
The application of MT to literary texts within an online collaborative environment introduces further complications.Indeed, OCT, a product of the democratising, participatory Web 2.0, itself raises ethical concerns that require attention and is still a "lacuna" (Zwischenberger, 2022: 2) in the field of translation studies.Among the available studies, McDonough Dolmaya (2021) and Jiménez-Crespo (2021) discuss the ethical question of translation crowdsourcing in terms of its potential impact on professional translation.Drugan (2011) also explores how professional ethical codes of practice work for non-professionally produced translation, while Zwischenberger (2022: 8) acknowledges "the possibilities of text (co-)production offered by Web 2.0 and amateurism" but highlights the darker side of OCT, such as the growing employment of gamification by profit-oriented companies in order to make money from voluntary work, an approach initiated by Facebook where the user communities translate any possible string from the entire interface.In the case of fan translation, ethical dimensions regarding "how fans perceive their own contributions or how they reconcile this tension between empowerment and exploitation in their own lives" are still underexplored (Baym & Burnett, 2009: 435).
Despite attempts to explore the use of MT in OCT, ethical issues have often been overlooked.For example, Wongseree (2020) reports the case where digital technologies, including MT, are used by Thai fansubbing communities.Désilets and van der Meer (2011) discuss "post-editing by the crowd", projects that involved "a large crowd of mostly amateurs to correct the output of machine translations systems", intending to improve the accuracy of the system (Désilets & van der Meer, 2011: 30).These studies show that MT and non-professional translators form a symbiosis and enhance each other, but any discussion concerning ethical issues appears to be lacking.O'Hagan's ( 2022 The first type involves fan-based translation groups that employ MT with post-editing (MTPE).For analysis, LNMTL 5 , one of the earliest groups of this kind and once a leading MT website for Chinese internet literature (Deng, 2023: 123), has been selected and its "About" and "FAQ" webpages closely examined.The second type comprises fan-based translation groups that rely on MT without post-editing.ComradeMao 6 , known for its high visitor volume (Wu & Lin, 2022: 134), serves as a representative example and is analysed through content from its main webpage, the "Contact Us" webpage and the promotional post by its founder on NU.The third type of fan-based translation group uses both HT and MT, exemplified by smaller translation groups like KnoxT 7 , Travis Translations 8 and Galaxy Translations 9 .The webpages of "Join Us", "FAQ", and "Recruitment" from these groups were examined.There are also individual translators who have not yet joined or have no intention of joining any of these groups but are also active users of MT (MTLers).
Their relationship with MT is explored based on their posts to the forum, leading to the second phase of the study's research.
In the second phase, the study explores the opinions of fan communities regarding MT, and a keyword search is then conducted on the "Translator's Corner" and "Novel General" forums hosted on NU.These two forums were selected as one was specifically designed for translation-related discussions, while the other serves as a general forum that also contains translation-related discussions.The keyword searches targeted terms "MTL", "MTLer(s)", "machine translator(s)", and "machine translation(s)".To ensure a meaningful sample of opinions and discussions, posts with fewer than ten replies were excluded and those which only focused on language pairs other than Chinese -English were also excluded.
As one of the aims of the websites and forums is to inform the public about an organisation's agenda and goals, intending to attract new members, the posts within these forums are accessible to anyone (Holtz et al., 2012: 60).In other words, the data collected in this study were obtained from publicly available sources that can be accessed without any restrictions.This study therefore solely involves a non-invasive, non-interactive form of observation.Given the nature of these sources and the assumption that online spaces are anonymous and pseudonymous, informed consent is not needed for the content analysis of such boards (Vannini, 2008: 278).Since these are "intentionally public postings by the authors", it would be "unusual to seek permission to use direct quotations" (Langer & Beckman, 2005: 197).Furthermore, in the discussion of ethical considerations in other new media research methods, for example netnographic research, obtaining explicit informed consent, such as "participant signatures or selection of a button on a graphical user interface", can potentially facilitate "collect participants' personal information and bring chances of extra exposure to them" (Huang et al., 2023: 163).Similarly, in the case of this study, seeking informed consent was not applicable.
However, to protect their privacy, quotes from individual users were cited anonymously without disclosing their internet names when reporting the findings, unless the quoted content came from recruitment posts and promotion posts by translation groups or posts made by the administrators and staff of the groups or communities.

Data and analysis
This section presents the data and the ensuing analysis regarding Chinese internet literature fan translation.Therefore, readers can access and enjoy novels (unlicensed translations) on the website without any financial obligations, while experiencing shorter waiting times for translations compared with commercial platforms.The website also benefits from the free post-editing provided by its readers, leading to an improvement in the quality of machine-translated content (Wang & Liu, 2021: 88).Clearly, this form of translation demonstrates how machines empower readers to become "translators".In this interaction between humans and machines, MT functions as a powerful assistant, though post-editing allows humans to retain the final authority and ensures their control over the translated output.).However, they do use "terms from the original translator" (Tyr, 2018) to improve the quality of the translation.In addition, unlike LNMTL, which has a considerable percentage of post-editing work finished by the crowd, ComradeMao has shut down the possibilities for them to participate directly in the translation process.Specifically, readers are not invited to make or propose any changes to the machine-translated novels directly on the website despite a few of its Discord channels being dedicated to the discussion of glossary and terms, so the only possibility of direct interaction with the website is for readers to leave comments below the translations.They are not given the same editorial rights as at LNMTL, nor do they have the chance of joining the management or translation team as members.Although readers of ComradeMao are welcome to join the community on Discord, their rights are clearly stated, including "suggest stuff", "get news about the translation" and "propose new things" (ComradeMao, n.d.).Nonetheless, the website still attracts readers who "can't wait until the novel gets properly translated" and "want to take a quick look into what the novel is about" (ComradeMao, n.d.).In this case, while the translation is still usergenerated, the interaction between humans and machines is to some extent dominated by machines since no post-editing work is conducted by humans.A possible explanation is that ComradeMao wants to avoid copyright infringement by limiting human involvement, as they claim "not [to be] a translation group but a free service similar to Google Translator, Yandex Translator or Bing Translator" (ComradeMao, n.d.).

Smaller fan translation groups using HT and MT
Other than these two pioneer websites, there are newcomers like KnoxT, Travis Translations and Galaxy Translations to name but a few.These websites, which provide both HT and MTPE, are also fan-based but operate differently from their precursors.Both 169 LNMTL and ComradeMao do not explicitly offer posts for their readers to join them, whereas the newly formed translation groups openly recruit translators, including MTLers (see Figure 1 for example), although some only express this subtly.For example, Travis Translations says it requires them to "be at least fluent in English" (Travis Translations, n.d.-b).These posts also mention the ad revenue and other forms of income that one can receive by working for them.The emergence of these translation groups has caused a trend in fan communities that is shifting from a non-profit fan-based model to a more profit-oriented one.Fan translators, editors and project managers work for these platforms as a team producing fan translations in a way that is similar to the commercial websites that produce official translations.However, problems arise when we think of the most urgent issue facing fan-based websites, that is, whether they are allowed to freely translate the original title and make a profit out of their translations.In these scenarios, MT serves a dual purpose.On the one hand, it lowers the cost of reading novels for readers and empowers them to take on a more active role, shifting from passive readers to translators and editors who have the autonomy to select what and when to translate.On the other hand, MT enables fan translation groups to leverage fan labour as "digital labour" (Fuchs & Sevignani, 2013), resulting in time and cost savings by eliminating the need to hire translators, ultimately benefiting these translation communities.
However, this raises pressing concerns for fan-based translation communities.The foremost concern is whether it is ethical for these fan-based translation groups to employ fan MTLers to translate titles without authorisation from the copyright holders, either with low pay or as volunteers.Is fan translation then a camouflage for legal infringement or exploitation?In both cases, the role that MT plays in this process may be more like an accomplice that supports and facilitates these "dubious" activities (O'Hagan, 2009: 94).Another concern is whether fan translators, including monolinguals, assisted by MT, will corner the Chinese internet literature translation market and finally usurp the position currently occupied by professional human translators.Fan translation groups like Travis Translations even claim they are providing "quality translations" in the hope that these translations (mostly edited MTs) will become official (Travis Translations, n.d.-a).Questions surrounding these translation activities merit attention, because they will directly affect, and are currently shaping, people's opinions not only of MTs, but also the status of human translators and fan communities, as well as the future of translation studies as a subject.

The underlying issues
Moving forward, the discussion and data analysis transitions to a critical examination of underlying issues.We explore the themes of empowerment within the fan translation community (4.2.1).Additionally, we scrutinise potential concerns related to infringement and exploitation, prompting a thorough analysis of the ethical considerations entwined with fan translation endeavours (4.2.2).Through these two comprehensive sections, we aim to unravel the multifaceted landscape of data and underlying issues inherent in the fan translation of Chinese internet literature.
To investigate these issues, the posts and threads involving discussions concerning MT, i.e., those with the identified four keywords in the title, have been closely studied.
The search yielded a total of 159 threads and posts across the two forums.These discussions can be loosely divided into six types (see The relationship between these stakeholders and ethical issues revolving round such translation collaboration can be interpreted by examining how user translators who are MTLers describe and reflect upon their translation activities, as well as by analysing the feedback provided by members of the NU community including readers who do not translate and translators who are not MTLers.

Empowerment
Interestingly, during the search, no posts were found from monolingual translators asking for job opportunities like bilingual translators.User translators tend to be cautious about disclosing their identity as MTLers, especially as 'pure' MTLers (i.e., people who use MT systems without pre/post-editing, usually monolingual translators).To further investigate this preliminary finding, an additional search was conducted.This additional search on posts in "Translator's corner" tagged as "Looking for Work" (a suggested tag by NU).
Among the 186 identified profile, only one refers to being "MT/MTL(ers)", here between Chinese and English.In contrast, many people tend to express their willingness to undertake MT/MTPE by replying to recruitment posts, sometimes even to posts that are not about hiring MTLers.Therefore, when examining the role of machine translation, it is important to consider not only the posts and threads from Category I that directly express opinions towards MT but also those from Category IV.This includes fan translation groups' opinions as expressed in their recruitment posts, as well as the replies from individual fans.
The hostile replies a pure MTLer can expect to receive probably explain their reluctance to openly assume the label of pure MTLer and post as solo MTLers.For example, a member who admitted to not knowing Chinese or Japanese and having average English skills faced negative feedback when expressing an interest in becoming an MTL translator, with one reply bluntly stating, "MTL is bad.If you want to improve[,] learn the language(s) properly" (NU Member, 2021b).However, if someone claims to do post-editing work in addition to MT, the replies are usually not as disheartening.For example, a member (monolingual translator) with no knowledge of Chinese but contemplating translating Chinese novels using an MT system but with serious editing before posting received encouraging replies.Suggestions were also given.For example, one reply suggests that beginners with no knowledge of Chinese should stay away from translating cultivation novels (a genre of Chinese internet literature) or stories set in ancient Chinese settings where the unusual keywords and the fact that the plot takes place in a palace might confuse MT systems.The post also receives a positive reply from a current MTLer, saying "I'm also a translator who uses MTL as a reference and translates it in an understandable way!" (NU Member, 2022a) and attracts other readers who want to be MTLers as well.
In general, people are more tolerant of solo MTLers who also offer post-editing.However, an MTLer posting a machine-translated version online and asking for donations is a more sensitive issue, and that MTLer can expect to receive more criticism, as the discussion thread "Opinions on MTLers?" shows.For example, one member expressed concerns that the fan translators did not own the copyright to the novels, so it was not surprising that some people were uncomfortable seeing MTLers potentially profiting from those works (NU Member, 2020).This discomfort intensified when the MTLs were minimally edited before being used for profit.
Compared with the varying opinions expressed by individual members, translation groups are more straightforward in terms of showing their inclusive acceptance of MTLers.
This can be seen from their recruitment posts.In the forum Translator's Corner on NU, there are 15 threads and posts tagged "recruitment" which contain the identified keywords in the title.These are from fan translation groups looking for translators, editors and proofreaders.They adhere to the position that for translating Chinese internet literature no prior experience is needed, and proficiency in English is a must.Some posts also require applicants to be familiar with tools such as Google Drive and Discord (see Figure 2), as digital communication tools "facilitates collaboration" (Vandepitte et al., 2016: 5).
A majority of MTLer recruitment posts have emerged after 2020.This increase can very likely be attributed to the COVID-19 pandemic, which has resulted in people spending more time reading online and potentially attracting more fan translators, including monolingual fan translators.The growing number of recruitment posts suggests the emergence of MT as a highly prevalent phenomenon in the field of Chinese internet literature translation.More than half (8 out of 15 threads and posts with the tag "recruitment") openly welcome MTLers provided that their MTs are edited (see Figure 3), while the remaining posts do not explicitly mention any editing requirements for the MTs (see Figure 4).These posts also emphasise that the work is paid and mention the opportunities for individuals to earn through platforms like Ko-Fi or Patreon.The replies and comments to the recruitment posts mainly revolve around salary and working conditions.Fans who master English well are motivated to contribute, as they can leverage MTL to maximise their capabilities.It is apparent that there is little objection to individuals earning an income by collaborating with translation groups in this manner.In fact, it is not surprising to see people react negatively when individual MTLers solicit money from readers, as fan translations are generally only tolerated if they are "not undertaken for profit" and "do not compete with an official translation" (Evans, 2019: 178, 180).The introduction of a paid aspect challenges the recognition of fan translation, which traditionally focus on the absence of monetary reward.This money soliciting fan translation deviates from the framework of UGT but represents a new form of translation within the realm of OCT.However, when MTLers, including monolingual ones, start soliciting money together with translation groups, it becomes more "legitimate" and receives no objections within fan communities.In fact, double standards seem to be at play here.
The above analysis shows that MT is more acceptable if some human effort is involved; this is also attested by the rules for those who want to release translations with NU, "Novel Updates does not list pure machine translation.All translations listed must pass through human hands -whether manually translated, or visibly edited to correct as many errors caused by machine translation as possible" (NU Staff Member, 2020).In addition, MT is more acceptable when a collective activity is involved.This is especially so when it becomes a means of generating an income.

Infringement and exploitation?
While the tone of the website is generally supportive towards MT and MTLers, there are instances where the forum members can be critical, generating heated debate, which reveals a number of ethical issues seen from different angles.
One such issue revolves round an accusation concerning a translation group's potential exploitation of MTLers.The original post, now deleted, is a recruitment post hiring MTLers who also need to carry out post-editing at a rate of $1 per chapter, with three chapters having to be finished daily.The original poster is questioned regarding his or her motivation for initiating such a translation project, as is shown in this member's reply: [W]ith no website to display what you have, with no real guarantees of payment and in [sic] such abysmal low rates… you expect people in good faith to join your potential scam sweatshop for translators and translate, edit and proofread three chapters per day, which even through MTL is still an excessively tall order.(NU Member, 2021a) It is worth noting that the comments in the discussion pay limited attention to the original poster's (who has since deleted his or her account) explanation that the wage is negotiable and that they have bought the titles from b.faloo.com 11(a platform for Chinese internet novels), which means they have the licenses necessary for the books they are translating unlike many other fan translation groups.Those who reply seem not to care whether the translation group is in violation of the copyright law or not but focus on what they perceive as "absurdly low wages".The strange thing is that, whereas there are also posts hiring MTLers as volunteers, or with limited pay only, such posts are welcomed more readily, and there are no hostile replies accusing them of taking advantage of free labour.A potential explanation for this could be the prevalent popularity of MTLers within the fan-based translation community, as a result of the low compensation they receive.This popularity might cause discomfort among bilingual reader-translators who expect higher remuneration.For monolingual readers, MT is their helper, empowering them with the ability to read and translate texts, whereas for bilingual translators, both professional and non-professional, MT systems are now also enabling MTLers to play an important role in enriching the translation market, particularly the market for Chinese internet literature.Without machine translation, a substantial number of titles would remain untranslated.
Moreover, the power of MT may also be endangering licensed translations.This issue has sparked considerable debate on copyright, as seen in a post published in July 2018 by the founder of ComradeMao.Its primary purpose is to promote the website, but it soon receives accusations, with someone claiming that ComradeMao "is infringing on the publication rights of four different publishers/rights holders on a large scale.It's likely only a matter of time before one of them takes action against the site" (NU Member, 2018).Some warn the founder not to get money out of this, and some blame him for ruining many decent translations.When both a fan-based MT translation website and a commercial website are working on the same title, the fan-based website, often with faster translation speed, tends to attract readers who might have otherwise chosen to read the licensed translation produced by the commercial website.This may ultimately impact the work and status of translators involved in the licensed version.However, there are also people who support him.To quote an NU staff member, who writes: "After all, this is just machine translation.Anyone can do this.... Unedited machine translation is just garbage.The presence of it [sic] does not detract from the value of a real translation at all" (NU Staff Member, 2018).The founder explains the website's legitimacy by saying that it is not an aggregator website that uses original translators' works without permission."We as MTL websites don't do that" (Tyr, 2018).Nonetheless, the core question remains unresolved, as stated by a member in the following quote: In the case of ComardeMao, simply disclaiming themselves as not being an aggregator website does not necessarily make them more ethical or respectable.It raises questions about the distinction, if any, between fan translators who "steal" translation copyright usually held by commercial websites to make a profit and aggregator websites that exploit fan translation communities by appropriating their work without permission.One might also question whether there should be copyright protection for the translations produced through these fan-based translation websites.As a matter of fact, both activities might involve unauthorised appropriation and it is arduous to ascertain definitively who is more erroneous in this situation.The complex ethical and legal issues surrounding these practices necessitate further examination and discussion.
These two cases indicate that fan communities are engaging in self-reflection regarding their translation activities.However, it also shows a complex contradiction within these communities.Many fans, especially those who are also user translators, tend to selectively overlook the fact that most fan translations, whether machine-translated or humantranslated, may infringe upon the original titles.Indeed, the growing overseas market for Chinese internet literature presents opportunities for the widespread integration of MT and increased collaboration with fans for commercial companies.It should be noted, however, that this consideration does not account for the progress in large language models (LLMs), which could expand the potential for fan translation of Chinese internet literature.LLMs have the potential to "mimic the human translation process that takes many preparatory steps" in comparison to "traditional machine translation that focuses solely on source-target mapping" (He et al., 2023: 1).LLMs can interpret prompts not only as translation requests but also as In conclusion, this paper has attempted to present diverse opinions from various stakeholders in the system.However, as an initial study that is purely based on observation, it is possible that certain points of view have not been fully explored.
Furthermore, this paper has only examined posts on NU.To gain a comprehensive understanding of the topic, in-depth netnographic research with surveys and interviews will follow up on this study.
who are machine translators (MTLers, translators who use MT systems), and in some cases, even monolingual MTLers 3 .The term MTLer(s) is a specific term used within the discourse of fan translation communities of Chinese internet literature: it refers to both 1 Netizen refers to a person who often uses the Internet, and here also as an active participant in a dedicated online community 2 Fansubbing concerns the phenomenon whereby fans are actively involved in the translation and subtitling of audiovisual content into languages other than the original.3In forum conversations fans use MTLers and "machine translators" interchangeably.But there are also fans who use "machine translators" to refer to Google Translate or Bing.Shuyin Zhang Machine translation of Chinese internet literature: infringement, exploitation or empowerment Revista Tradumàtica 2023, Núm.21 163 monolingual and bilingual individuals, who rely on MT systems for their fan translation activities.
author when translating from Spanish to Catalan.A more recent study conducted by Hu and Li (2023) compares the human translation (HT) undertaken by a renowned Chinese translator with the MT output by DeepL from English to Chinese for Shakespeare's plays ) is one of the few studies that address ethical implications stemming from MT and amateur or fan translation.It mentions issues such as the "unacknowledged secondary use of human translations as the training data in corpus-based machine translation (MT)", "exploitation of translation as free labour", "undercutting professional translation" and "the unsolicited re-distribution of copyrighted materials by fans with their own translations"(O'Hagan,   2022: 425-440).Nonetheless, the ethical issues associated with MT and amateurism have been discussed separately, but a comprehensive analysis of the ethical implications of using MT in user-generated translations in an online collaborative context is still needed, especially in a context of such numerous followings in the case of Chinese internet literature.Furthermore, the application of MT to literary texts within an online collaborative context equally is an underexplored topic, although it has become a widespread phenomenon in the fan-based translation communities involved with Chinese internet literature.The use of MT in this field has the potential to redefine the nexus of translators, authors, editors and readers, thereby giving rise to ethical issues worth discussing.This article is an attempt to rethink the two sets of dichotomous relationships between humans and machines and professional translators and reader(user)-translators, placing them within the context of the translation of Chinese internet literature.Therefore, in viewing fan translation of Chinese internet literature as a social collaborative practice, this article not only describes the interaction and intervention of different participants in newly emerging online collaborative translation activities in fan communities, but also seeks to answer a wider range of ethical questions addressing the influence exerted by the MT-powered, user-generated, online collaborative literary translation on professional translators, the profession as a whole, the literary translation industry and intercultural communication.3.Research method and procedureThis article presents a descriptive case study that examines fan-based translation activities involving Chinese internet literature and analyses fans' comments and posts related to the use of MT in these activities on internet forums.The study aims to explore the role of MT in the user-generated translation of Chinese internet literature and highlight the underlying ethical issues with regard to MT in the field of literary translation.It adopts a content analysis approach, with the data being collected from five fan-based translation websites (15 webpages) and the Novel Updates (NU) 4 website (159 threads and posts) on March 6 th , 2023.NU is a comprehensive directory of Asian Translated Shuyin Zhang Machine translation of Chinese internet literature: infringement, exploitation or empowerment Revista Tradumàtica 2023, Núm.21 166 Novels, which also serves as the largest English forum dedicated to enthusiastic fan translators of Asian web novels (Shao et al., 2018: 121), making it an ideal choice for this study.It begins by exploring the role of MT in fan-based translation activities associated with Chinese internet literature.Four major types of translation activity are identified within the fan-based translation communities of Chinese internet literature.These activities involve both monolingual and bilingual fan translators, including MTLers and non-MTLers.

4. 1
Contextualising MT in the fan translation of Chinese internet literature The initial section (4.1) of the data analysis focuses on contextualising machine translation (MT) within fan translation dynamics.It is subdivided into distinct categories and explores the role of fan translation groups utilising MTPE (Machine Translation Post-Editing), those relying solely on MT, smaller groups incorporating Human Translation (HT) alongside MT, and the distinctive practices of solo Machine Translation enthusiasts (Solo MTLers).4.1.1Fan translation groups using MTPE LNMTL, which stands for Light Novel Machine Translation, is a fan-based website that provides the MT of Chinese internet literature.LNMTL was founded in 2015 by a fan of Chinese internet literature with no knowledge of Chinese.The website allows readers to edit the MT outputs indirectly by proposing terms to be included in its glossary 10 (LNMTL, n.d.).These proposed terms can then be used to replace the original terms in the MT outputs, and this is where the amateurs' expert knowledge can help.The glossary includes terms such as proper nouns, genre-specific terms, and culture-specific terms, and phrases and sentences can also be included.By proposing more refined or contextually translated terms or phrases to the glossary, fan translators can improve the quality of the machinetranslated texts.In this sense, the website is generous as it offers editorial rights to its readers, including a post-editing option that is open to all and an acceptance of retranslation based on the number of edits.This devolution of power on LNMTL is proving to be a mutually beneficial arrangement for both the website and its readers.The website does not acquire translation licenses from the copyright holders of the novels it 10 https://lnmtl.com/termtranslates, which means it operates without any financial obligations in this regard.

4. 1 . 2
Fan translation groups using MT ComradeMao, a fan-based translation website founded in 2018 by a group of avid readers, offers pure MTs of novels in Chinese, Korean and Japanese.Novels posted on this website, which are translated with limited human intervention, involve the team making a script for the translation programme, adding novels to the script and creating terms and propositions.The translation process does not include any post-editing such as "proofreading, quality check, or anything else that would change the quality of translation directly" (ComradeMao, n.d.

Figure 2 .
Figure2.Recruitment post-a(Travistann, 2020) They have been used to consuming these free materials and have even started to expect to profit from them, despite being originally motivated by pure passion and a love for what they do.Moreover, while some fans accuse translation groups of exploiting translators and MTLers by offering low wages, they simultaneously attempt to avoid paying the title holders by relying on reading fan translations or even becoming MTLers.5.The outlook for the futureThe above analysis shows that it is important to acknowledge the contributions that MT has made in allowing even monolinguals to act as translators and editors in usergenerated MT for Chinese internet literature.However, it is only natural for those within fan communities who have benefited from the popularisation of MT and have been empowered by MT to face questions concerning potential infringement and alleged exploitation stemming from their activities.Nonetheless, the ability of monolingual readers to become translators and the increased dissemination and availability of Chinese internet literature online have the potential to make this type of human-machine collaboration one of the most promising modes of translation once these ethical issues have been adequately addressed.This could as well be applied to other Asian internet literature.A feasible option for user translators including MTLers might either be to collaborate with well-established platforms that are original rights holders, like the collaboration of Wuxiaworld (a fan translation website providing HT) with 17k.com, or to join larger translation companies like funstory.ai,a Chinese platform for overseas digital publishing featuring MT.
instructions regarding what and how to translate, a feature which possibly allows for more customised translations.However, to fully harness the potential of digital and fan labour, educational support is needed.Initiatives for MT literacy courses should be rolled out for students enrolled in language-oriented training programmes who aspire to a nonprofessional career in fan translation.Such courses could benefit students interested in language-related professions(Ehrensberger-Dow et al., 2023: 407), and individuals "who ha[ve] not received any formal training in translation and who do not practise translation professionally" (Bowker, 2023: 3).Reader translators, particularly monolingual translators, would be ideal candidates for such courses.Moreover, in the Chinese internet literature fan translation market, with instant messaging platforms like Discord becoming a new means of communication, and crowdfunding platforms such as Ko-Fi and Patreon generating income from fans for fans, translator training programmes would do well to keep abreast of these new trends and adjust their curricula accordingly so as to give graduates the best possible chance of success.

Table 1
Collected Data (See Appendix for a full list) If you are MTLing a licensed translation[,]then you are still illegally publishing a translation as those who have the license have exclusive publication rights for an English translation of that novel.Meaning, no one else is legally allowed to publish any English translation of that novel.Doesn't matter if it's 100% original.