Unravelling the stylistic nuances: a comparative multidimensional analysis of amateur and professional translations of Legends of the Condor Heroes

Style research has always been an important ﬁ eld in corpus-based translation studies. However, the stylistic differences between amateur and professional translations of epic Chinese martial arts ﬁ ction have received limited attention in the existing literature. This study addresses this gap by examining the stylistic differences in amateur and professional translations of the novel Legends of the Condor Heroes , a representative work of this genre gaining popularity through fan-translation websites. Employing Biber ’ s multidimensional analysis, this study investigates ﬁ ctional speech and narration in both amateur and professional translations. The ﬁ ndings reveal that amateur translations exhibit less variation between speech and narration, characterized by lower levels of involvement in dialog and reduced informational density in narrative passages. Moreover, amateur translations tend to employ more abstract language across both speech and narration compared to their professional counterparts. These stylistic disparities may in ﬂ uence readers ’ comprehension and engagement with the text, potentially affecting their access to critical narrative elements. The study posits that the lack of professional training and experience, as well as the collaborative and open-source nature of amateur translation communities, may contribute to these observed differences in amateur translations. This research contributes to translation studies by examining stylistic differences between amateur and professional translations of Chinese martial arts ﬁ ction, potentially informing future studies on reader engagement and the reception of translated Chinese literature. The ﬁ ndings also highlight the importance of considering the impact of amateur translation practices on the dissemination and reception of popular genres in cross-cultural contexts.


I
n recent years, amateur translation has gained considerable traction, driven by technological advancements and the proliferation of online resources.The rise of machine translation tools, such as Google Translate and DeepL, has made it easier for individuals without formal translation training to translate texts from one language to another (Jiménez-Crespo, 2017a).Amateur translation is also referred to as volunteer translation (Garcia, 2009;Wadensjö, 2007).Distinguished by its collective, self-initiated, and community-based nature, amateur translation, often referred to as "crowdsourced translation" or "wiki-translation," is characterized by a consumer-oriented model of internality where the potential audience becomes active producers or prosumers.This shift moves away from traditional distinctions between active translation agents and passive recipients (Cronin, 2010, p. 4).Research on amateur translation, primarily found in areas like fansubbing and gamification, has explored broader cultural and ideological dimensions.This includes the examination of sociopolitical agendas and influences (Wang and Zhang, 2017), intricate cultural negotiations (Liang, 2017), and the competitive dynamics with professional translators (Izwaini, 2014).Pérez-González (2010, p. 284) underscored the imperative for increased scholarly attention to amateur translation, recognizing its disruptive potential within dominant translation practices.
The rise of amateur translation has significantly impacted the global reach of Chinese martial arts fiction.Traditionally, this immensely popular genre in China had limited exposure among English-speaking readers.However, the advent of online platforms has catalyzed a groundswell of amateur translations, making these novels accessible to a wider international audience.This bottom-up approach to translation has bypassed conventional publishing norms, expanding access to Chinese martial arts fiction.As a result, millions of new readers worldwide have been introduced to the genre, underscoring the transformative power of amateur translation in reshaping the literary landscape and bridging cultural divides.This growing influence of amateur translations is exemplified by wuxiaworld.com, the largest platform for this genre, which claims a readership exceeding 5 million 1 .The impact of amateur translation on Chinese martial arts fiction has extended beyond online platforms to influence the traditional publishing industry.Some amateur translations have been professionally edited and published in print, thus blurring the lines between grassroots efforts and conventional publishing.A notable example is the works of 金庸 Jin Yong (Louis Cha, 1924-2018), a prominent figure in the genre.His novels have been translated multiple times by both amateur and professional translators, significantly increasing their accessibility to English-speaking readers.The coexistence of amateur and professional translations invites scholarly research into the impact of amateur translations on target cultures, as well as the linguistic and stylistic differences between amateur and professional renditions.The availability of these diverse translations provides a unique opportunity to explore these variations and shed light on the evolving dynamics of literary translation in the digital era.
A comparative analysis of professional and amateur translations offers an opportunity to reassess traditional understandings of translation and re-evaluate the role of the amateur translator (Cronin, 2010, p. 1).The primary aim of this study is to explore the linguistic differences between amateur and professional translations of Jin Yong's renowned work, Legends of the Condor Heroes (referred to as LCH hereafter).By examining these variations, the study seeks to uncover insights into the broader cultural and political implications of these different translation approaches.
Studies on the translations of Jin's martial arts fiction Martial arts fiction, or wuxia, is a prominent genre in Chinese popular literature, with a history spanning back to the Warring States period (403-221 BC) (Hamm, 2004).Wuxia stories typically revolve around the adventures of skilled martial artists, focusing on themes of martial prowess and chivalric ideals (Mok, 2001).The translation, introduction, and reception of wuxia works in Western contexts have become significant areas of research within translation studies.Scholars have taken a keen interest in investigating how these distinctively Chinese narratives are adapted and received in different cultural settings.This academic focus is primarily motivated by the lasting popularity of wuxia in Western literary and cultural spheres, which has endured despite substantial cultural and linguistic obstacles.
Jin Yong's martial arts novels have become a focal point in scholarly discussions on literary translation (Luo, 2011).As the world's 31st best-selling fiction author, Jin Yong's works have drawn the attention of both amateur and professional translators, leading to a variety of approaches and interpretations.In recent years, there has been a notable surge in amateur translations of Jin Yong's novels, with some works making their English debut through non-professional efforts.This development signifies a change in the translation practices surrounding martial arts literature.The table below presents a comprehensive list of Jin Yong's works that have been translated by both amateur and professional translators.This side-by-side comparison highlights the growing influence of amateur translations and the increasing diversity in the translation landscape of Jin Yong's martial arts fiction.
Table 1 shows that two of Jin Yong's works, The Legend of the Condor Heroes and The Deer and The Cauldron, have been translated by both amateur and professional translators.However, to our knowledge, no studies have directly compared these amateur and professional translations.Early research on Jin Yong's translations mainly concentrated on published versions and explored aspects such as translatability, cultural transformation, and translation strategies, as demonstrated in the works of Chen (2006), Wong (1997), Mok (2002), Shen (2007), and Zhao (2009).Wong (1997), analyzing John Minford's translation of Jin Yong's The Deer and the Cauldron, highlighted the challenges in translating martial arts fiction, particularly the rendering of imaginative Chinese martial arts techniques and intricate combat scenes.In the same volume, Lai (1997) examined the dichotomy of domestication and foreignization in translating The Book and the Sword, advocating for the preservation of Jin Yong's exotic elements through foreignization strategies.Zhao (2009) extended this line of inquiry to Russian translations, investigating the translation of martial arts terminology and character titles.Mok (2002), drawing on her experience translating Fox Volant of the Snowy Mountain, addressed the complexities of rendering appellations in Jin's works.She emphasized the importance of interpreting these within the context of traditional Chinese culture and the socio-cultural milieu of the genre.Shen (2007) questioned the very translatability of Jin Yong's oeuvre, proposing that the translation of his martial arts fiction be conceptualized as "the discourse of identity across linguistic boundaries and cultural differences."(p.207) Recent scholarship has shifted focus to the translators themselves.Xu and Zhang (2020) conducted an indepth interview with Gigi Chang, one of the English translators of Jin Yong's LCH, highlighting the critical role of translator subjectivity in both translation practice and cultural mediation.Similarly, Zhang (2020) examined the work of Anna Holmwood, another translator of the same novel, analyzing her translation strategies and interactions with the publisher.
Recent corpus-based studies have shed light on the translation practices of Jin Yong's martial arts fiction (Wu andLi, 2021, 2022).Wu and Li (2022) examined the normalization tendency in translations of Jin Yong's martial arts fiction by Olivia Mok, Graham Earnshaw, John Minford, Anna Holmwood, and Gigi Chang.They used indicators such as lexical richness, normalized POS distributions, high-frequency words, and the naturalizing percentages of special martial arts terminology.They found that Minford's translation exhibited the highest degree of lexical normalization and was better received in the overseas market.Chen and Dai (2021), in their study of the English translation of LCH, found substantial omissions of historical-cultural information, fight scene details, and character insights.They interpreted these omissions as deliberate interventions by the translator, aimed at reconciling differences in narrative conventions between the source and target cultures.
The growing prominence of amateur translation, particularly in certain contexts where it outpaces professional translation, makes it a critical subject for linguistic and stylistic research.While some studies have acknowledged the rise of online wuxia literature and amateur translator communities (e.g., Li, 2021), there remains a significant gap in scholarly research on amateur translations of Jin Yong's martial arts fiction.This gap is particularly noteworthy given the vast readership these online platforms have attracted, with millions of English readers worldwide engaging with these works.A comprehensive investigation into the amateur translation of Jin Yong's classics, compared systematically with professional translations, can offer valuable insights into the broader field of amateur translation.Such research has the potential to reveal important social, cultural, and political nuances embedded within these translations, contributing to our understanding of how amateur translators navigate complex linguistic and cultural challenges.

Methods
Corpus.This study employs a corpus-based approach to compare amateur and professional translations of LCH, a work often hailed as the Chinese equivalent of "The Lord of the Rings."The corpus, comprising both amateur translation (AT) and published translation (PT) of the novel, was meticulously compiled from online and electronic sources and subsequently converted to plain text format.To ensure a comprehensive comparison between AT and PT, we followed Egbert and Mahlberg (2020) in considering fiction as a hybrid genre encompassing fictional speech and narration.Fictional speech represents spoken interactions, while narration provides the situational context of storytelling (ibid, p. 74).Recent research has increasingly acknowledged that fictional dialog constitutes a distinct form of communicative activity (Bublitz, 2017).Despite not precisely mirroring everyday spoken interactions, fictional dialogs are curated versions of real speech, conveying a sense of authenticity akin to dialogs in daily life (Thomas, 2012).Previous research has shown that separating fictional speech and narration in translation analysis can reveal linguistic nuances and provide a more precise examination of translators' styles (Chou and Liu, 2024).By adopting this nuanced approach, the study aims to uncover subtle differences between amateur and professional translations that might be overlooked if fiction were treated as a homogeneous genre, thereby enhancing the depth and accuracy of the comparative analysis.
To achieve this, we developed a Python program to extract all dialogs from the two translations of LCH using speech punctuation (quotation marks).Consequently, the corpus consists of four sub-corpora: two for dialog and two for narration, each containing a total of 40 chapters.We manually reviewed the narration corpora to remove broken sentences, primarily in the form of reporting clauses such as "he said."This step ensured the overall integrity of the dialog corpus and maintained consistency across the four sub-corpora.By doing so, we guaranteed that the datasets used for analysis are consistent, reliable, and representative of the intended register variation patterns.This procedure resulted in a total of 160 files.The descriptive statistics of the four sub-corpora are presented in Table 2.
Multidimensional analysis.Multidimensional analysis (MDA), a framework pioneered by Biber (1988) for comparing written and spoken English registers, has become a versatile tool in linguistic research.This approach has been widely adopted for investigating both synchronic and diachronic aspects of language use across various genres, including business texts (Ren and Lu, 2021;Wang and Liu, 2024), news (Huang and Ren, 2020;Kohring and Matthes, 2007), and literary works (Chou and Liu, 2024;Gracheva, 2022).MDA employs advanced statistical techniques such as factor analysis and cluster analysis to identify patterns in the co-occurrence of linguistic features across different registers.
In the present study, we apply Biber's MDA model, which examines 67 specific linguistic features across six dimensions, to four sub-corpora: fictional speech of PT and AT for LCH, as well as fictional narration of PT and AT for LCH.Each linguistic feature contributes either positively or negatively to a dimension's overall score, which is calculated by summing the factor scores of features with positive loadings and subtracting those with negative loadings (McEnery et al., 2006, p. 288).This methodology allows us to quantify and compare different translation styles, potentially revealing subtle differences between professional and amateur approaches that might otherwise go unnoticed.By employing MDA in this context, we aim to provide a nuanced understanding of how different translators interpret and render the same source text, contributing to a deeper comprehension of translation practices in the realm of Chinese martial arts fiction.
MDA utilizes a set of dimensions, each defined by contrasting linguistic features, to examine language variations across different texts (see Table 3).While this framework has been widely applied in register studies, researchers often selectively use dimensions based on their specific research goals.For instance, Kruger and van Rooy (2016) used Dimensions 1, 3, 5, and 6 to analyze translated and non-native indigenized varieties of English, while Calzada-Pérez and Ramos (2021) focused on Dimensions 1, 2, and 6 to examine translated and non-translated parliamentary discourse.Given the objectives of the current study, we decided to apply all six dimensions of MDA to the four sub-corpora to facilitate a comprehensive comparison between the two translations.By utilizing the full spectrum of MDA dimensions, we aim to uncover subtle nuances in translation strategies that might be missed with a more limited analysis.
Research questions.This study aims to investigate the stylistic differences between amateur and professional translations in the context of fictional speech and narration.Using Jin Yong's acclaimed martial arts novel LCH as a case study, we aim to address the following research questions: 1.In which dimensions does the amateur translation diverge from the professional translation in both fictional speech and narration of LCH? 2. What are the specific linguistic features that characterize the differences in speech and narration between the amateur and professional translations?3. What potential factors contribute to the observed stylistic and linguistic differences between amateur and professional translations?
Data analysis.The analysis of linguistic features in the four subcorpora was conducted using the Multidimensional Analysis Tagger (MAT) developed by Nini (2019).MAT builds upon the Stanford Table 3 The MDA model (Biber, 1988).

Dimension Label Description
Dimension 1 Tagger (Toutanova et al., 2003) to produce a grammatically annotated version of the selected corpus.This annotation facilitates the extraction of key statistics necessary for text-type or genre analysis.Due to the non-normal distribution of dimension scores and linguistic feature scores, the non-parametric Mann-Whitney test was utilized to identify significant differences between the two translations across each dimension.

Statistical results
Register differences in fictional speech translation.The scores for each dimension were calculated by treating each chapter in the two genres as an individual data point.Specifically, the mean score for each dimension is based on 40 data points, corresponding to the fictional speech sections across all chapters in both AT and PT.Table 4 presents the means and standard  deviations of the dimension scores for these sections in both AT and PT.A Mann--Whitney U-test showed that there was a significant difference between AT and PT in fictional speech in Dimension 1 (U = 593, p <0.05), Dimension 2 (U = 402, p < 0.01), Dimension 5 (U = 531, p < 0.01) and Dimension 6 (U = 67, p < 0.01).Figure 1 demonstrates the comparison of register dimensions between AT and PT in fictional speech.
Next, we analyzed the four dimensions to examine the typical linguistic features in fictional speech between AT and PT.Dimension 1, which represents the contrast between involved and informational discourse, has the most heavily loaded features among the six dimensions.Lower scores in Dimension 1 suggest a text with greater informational density, while higher scores indicate a more interactional and affective text.In this dimension, AT scored 3.12 and PT scored 3.98, suggesting that PT's fictional speech is relatively more involved and effective compared to AT's more informational focus.Dimension 2 focuses on differentiating between narrative and nonnarrative text.Both AT and PT showed scores below 0, suggesting a dominance of non-narrative text.However, AT displayed a greater tendency towards non-narrativeness compared to PT.Similarly, Dimension 5 captures the degree of abstractness in text production.Both AT and PT exhibit negative scores, indicating a non-abstract nature typical of speech data.However, PT tends to be more concrete than AT, as reflected by its lower score.Dimension 6 measures online information production, with higher scores indicating informational text produced under time constraints.According to Biber's model (1988, p. 105), the register of AT (0.56) aligns more closely with academic prose, whereas PT (−1.01) resembles romantic fiction.
Taking into account all the features across the four dimensions, the fictional speech of AT demonstrates lower levels of affectivity and interactivity, less narrative content, a lower degree of concreteness, and appears to be produced under greater time constraints compared to PT.Table 5 provides a detailed list of linguistic features that show statistically significant differences in these dimensions between the two translations of fictional speech.
All the features listed in Table 5 reveal statistically significant differences between AT and PT.Dimension 1 indicates an interactional and involved nature for both translations, as evidenced by the positive dimension scores in Table 4.However, PT shows a stronger tendency towards interactional and involved discourse compared to AT. Features with positive loadings in Dimension 1 are associated with an involved and noninformational focus, typically stemming from "a primarily interactive or affective purpose and/or to highly constrained production circumstances" (Biber, 1988, p. 105).For instance, PT uses first and second-person pronouns more frequently than AT.The frequent use of first-person pronouns indicates ego involvement and an interpersonal focus, while second-person pronouns suggest a high degree of engagement with the addressee.In fictional dialogs, this frequent use of personal pronouns indicates a more interactional register, as illustrated in PT (see Example (1)).
AT: "The stores here aren't half bad.Madame's clothing is old and worn, have to buy some new ones."PT: "The markets here are good and Madam's clothes are worn.We should buy you some new ones first." In Example (1), the AT does not include first or secondperson pronouns, resulting in a more detached tone.
Additionally, there is a grammatical error in AT 2 : the phrase "have to buy some new ones" lacks a subject, making the sentence incomplete.In contrast, the PT incorporates both first and second-person pronouns ("we" and "you"), which enhances the engagement between the characters and the readers.This inclusion of personal pronouns in PT creates a more interactional and involved dialog, reflecting the higher dimension score for involvement and affectivity.
The PT also maintains grammatical correctness and provides a clearer, more personal connection between the characters and the readers.
Dimension 5 measures the level of abstractness in texts: positive scores suggest abstract discourse, whereas negative scores indicate non-abstract discourse.In this dimension, both AT and PT show negative scores, reflecting their nonabstract nature.However, PT exhibits a more pronounced tendency towards non-abstract discourse.Out of the six linguistic features that show significant differences between AT and PT in this dimension, all are positive-loading features.The mean ranks of these features are higher in AT than in PT, contributing to a more abstract and formal register in AT compared to PT.In Example (2), the translation in AT utilizes the conjunction "while" to form an adverbial clause, contributing to a more abstract and formal tone.Conversely, PT employs shorter sentences and multiple adjectives, which results in a more conversational and immediate context.(2) Chapter 30 ST: "伯伯你瞧, 这两章柬贴中的字笔致柔弱秀媚, 图画中 的笔法却瘦硬之极。" Bobo ni qiao, zhe liangzhang jiantie zhong de zibi zhi rouruo xiumei, tuhua zhong de bifa que shouyingzhiji.AT: "Uncle, look, the writing on these two sheets of paper is delicate and elegant while the stroke on this drawing is very stiff 3 ."PT: "Look here, Uncle.The handwriting in these two notes.Soft, graceful, feminine.Now, look at the lines in the picture.Lean and hard, aren't they?" Dimension 6, which assesses on-line informational elaboration, indicates that a high score reflects an informational text produced under time constraints.Table 5 highlights four features, all with positive loadings in this dimension, that show significant differences between AT and PT.Notably, "that" clauses as verb complements carry the highest weight among all the features in Dimension 6 (Biber, 1988, p. 103).AT shows a higher mean rank for online informational elaboration than PT, indicating its more frequent use in dialog translation.In Example (3), AT uses "that" clauses as verb complements, while PT opts for a coordinate structure with two independent clauses joined by a dash.This suggests that PT has a lower degree of on-line informational elaboration compared to AT. (3) Chapter 16 ST: "那数十名敌手的武功招数, 他一招一式都记在心里, 于是苦苦思索如何才能破解, 他要想通破解的方法, 然后去杀 了他们报仇。" Na shu shiming dishou de wugong zhaoshu, ta yizhaoyishi dou jizai xinli, yushi kukusisuo ruhe caineng pojie, ta yao xiangtong pojie de fangfa, ranhou qu shale tamen baochou.
AT: "He painstakingly pondered on how to defeat each and every one of them.He thought that as soon as he succeeded in developing the martial art, he would go back and seek his revenge." PT: "He had memorized every single martial move his enemies used on him, and spent every waking moment devising countermeasuresso he could kill them and avenge his family." As demonstrated by the examples provided, the linguistic features collectively contribute to the overall register and style of each translation.In comparison, AT exhibits lower levels of affectivity, interactivity, and concreteness than PT in the rendering of all dialogs in LCH, highlighting a systematic difference in translation approaches.This difference suggests that AT tends to produce more neutral, detached, and abstract language, potentially sacrificing some of the emotional immediacy and conversational authenticity present in the source text.PT, on the other hand, appears to prioritize maintaining the interpersonal dynamics and situational vividness of the original dialogs, resulting in translations that more closely mirror the affective and interactive qualities of natural spoken language.
Register differences in narration translation.In contrast to dialog in fiction, narration typically exhibits a distinct set of linguistic characteristics.Narration is expected to focus on conveying information and telling a story, rather than engaging the reader directly or mimicking conversational patterns.To investigate this, we applied the same analytical methods used for examining fictional dialog to the narrative portions of our corpus.We then calculated the average scores for each linguistic dimension across all 40 data points in our sample.These mean scores, along with their standard deviations, are presented for both AT and PT in Table 6, providing a comprehensive view of how the two translation approaches handle narrative text.This analysis allows us to compare not only how narration differs from dialog but also how AT and PT differ in their translation of narrative text.
The analysis of fictional narration revealed patterns similar to those found in dialog, with notable differences between AT and PT in Dimensions 1, 2, 5, and 6, as shown in Fig. 2. Focusing on Dimension 1, which measures the spectrum between informational and involved language, both translation types scored below zero, indicating that their narration is informationally dense.This aligns with the expected characteristics of fictional narration.Interestingly, while both AT and PT demonstrate this informational density, there is a statistically significant difference between their scores.AT scored −9.98, while PT scored −12.36.The lower score indicates a higher degree of informational density, meaning that PT produced more informationally dense narration compared to AT.
Examining Dimension 2, which measures the level of narrative content, both AT and PT exhibited positive scores, confirming their adherence to a narrative style typical of fictional prose.Dimension 5 revealed a notable contrast: AT scored relatively higher (1.60) compared to PT's negative score (−0.43), indicating AT's greater tendency towards abstract and formal information presentation in narration.For Dimension 6, which assesses online informational elaboration, both translation types received negative scores, suggesting that neither AT nor PT was significantly under time constraints.Summarizing these findings in light of Biber's (1988) dimension descriptions, the narration of AT is characterized by lower information density, less pronounced narrative features, higher levels of abstract information, and more online elaboration compared to PT.To provide a more detailed view of these differences, Table 7 outlines the specific linguistic features that showed statistically significant differences between AT and PT.These features contribute to the overall patterns observed across Dimensions 1, 2, 5, and 6, offering insight into  the subtle yet consistent stylistic differences between the amateur and professional translations in narration.The analysis of Dimension 1 register features reveals a significant difference between AT and PT, with PT exhibiting a more pronounced informational character.The deletion of "that" is a key factor, second only to private verbs in its significance.Example (4) illustrates that "that" is omitted in the AT version but retained in the PT version, contributing to AT's more involved style and PT's more informational tone.The retention or deletion of may seem minor, but it plays a crucial role in shaping the overall register of the translation.Its presence or absence can affect the flow and formality of the text.Deletion is often associated with a more conversational and involved style, while retention tends to create a more formal and informationally dense tone.AT: Guo Jing had just received a surprise attack; seeing Ouyang Feng in this weird position he believed [that] he was up to no good.
PT: He backed away, eyeing the Martial Great with caution, convinced that some infernal plot was unfolding.
Dimension 2 distinguishes narrative discourse from other types of discourse (Biber, 1988, p. 113).According to Table 7, while AT generally exhibits a higher degree of narrativity with features like past tense verbs and public verbs, the overall picture is more varied.PT shows higher ranks in several important narrative features, including perfect aspect verbs and synthetic negation.Notably, PT has a significantly higher use of present participial clauses, which add immediacy and dynamic action to the narrative.This highlights that although AT follows traditional narrative conventions more closely, PT employs modern narrative techniques that also contribute to a rich and engaging storytelling style.As suggested by Biber (1988, p. 109), the use of present participial clauses enhances the vividness of narration of past events.Example (6) demonstrates the use of a present participial clause "examining them in the moonlight" in PT, which is absent in AT.This clause adds immediacy and dynamism to the description of the woman's actions, creating a sense of simultaneous action that draws the reader into the scene.
( AT: The manager saw the expression on his face and actually thought that the scholar's words were true.Immediately his expression became less polite as he thrust his chest out and asked.
PT: The attendant straightened himself and puffed up his chest.So the scholar was right, and not just bitter.

Discussion
Stylistic variations between AT and PT.The comparison of AT and PT in fictional speech and narration reveals notable stylistic differences.In fictional speech, PT outperforms AT in capturing affective and interactional aspects, maintaining a more concrete and carefully crafted style.In narration, PT demonstrates a more narrative style with less informational elaboration, while AT tends to be more abstract and produced under time constraints.These findings indicate that professional translators better preserve the key characteristics of both fictional speech and narration, resulting in higher-quality translations that align more closely with the target language norms and expectations.To further contextualize the findings, we compare the speech and narration in AT and PT separately with other types of texts analyzed by Biber (1988).Figure 3 demonstrates stylistic differences in the translation of fictional speech compared to other registers in Dimensions 1, 2, 5, and 6. Figure 4 illustrates stylistic differences in the translation of fictional narration in these four dimensions compared to other registers.
Possible causes.The multidimensional analysis reveals that AT generally exhibits lower quality compared to PT, corroborating Izwaini's (2014) study.AT shows less distinctive variation between speech and narration registers, with lower involvement levels in speech and reduced informational density in narration.This inadequacy points to a common trend in amateur translations, where linguistic variation and narrative dynamics are not adequately captured.The blurred distinction between speech and narration in these translations contributes to a noticeable decrease in overall translation quality.In this regard, readers of amateur translations of Louis Cha's works may potentially overlook the inherent richness of information embedded in the original text.Furthermore, they might fail to perceive the depth of engagement and involvement of characters within the fiction, particularly as conveyed through dialogs.Several factors are believed to contribute to this phenomenon, including the differences in the translation processes, the distinct identities of amateur and professional translators, their divergent goals and priorities, and the varying levels of editing and collaboration.
Firstly, the stylistic divergences between AT and PT can be attributed to differences in their translation processes and the composition of their translation teams.According to Jiménez-Crespo (2017b), these disparities arise from how the translation process is initiated and the locus of control, whether within a selforganized community or a company or an organization.AT was completed by a group of seven anonymous translators, reflecting a common practice among fan translators who often work from their non-native languages into English.This approach deviates from the recommendation by Newmark (1988) that translators should ideally translate into their mother tongue.As a result, AT may be more prone to linguistic and cultural inaccuracies, leading to a less authentic representation of the source text's style and register.In contrast, PT involved a team of three translators working in their native language, allowing for a more nuanced understanding of the target language's norms and expectations.This enables professional translators to better capture the stylistic intricacies of the source text and maintain the distinct register variations between speech and narration.Furthermore, the collaborative dynamics of the translation teams contribute to the shaping of each version's unique translation style and register tendencies.The larger team size in AT may lead to inconsistencies in style and register, while the smaller team size in PT allows for greater coordination and consistency, resulting in a more cohesive and stylistically faithful translation.
Secondly, the stylistic divergences can be attributed to the distinct identities and backgrounds of the translators involved in each version.Investigations by Zhu (2022) suggest that the amateur translators are primarily individuals of Chinese descent or Chinese-speaking Americans, many of whom have resided in  English-speaking nations for extended periods.Despite their presumed high English proficiency and sound traditional Chinese cultural knowledge, the overall linguistic quality of AT is found to be inadequate, with instances of negative transfer from the translators' first language.This may be attributed to the fact that amateur translators are addressing a niche audience with a specific interest in martial arts works and Chinese traditional culture, rather than catering to "professional readers" (Chan, 2014) who evaluate the work in a scholarly manner.In contrast, PT likely involves professional translators with formal training and experience in literary translation, who prioritize the accurate rendering of fictional speech and narration, ensuring that the distinct registers are maintained.The divergent goals and priorities of amateur and professional translators contribute to the stylistic differences observed between AT and PT, with amateur translators focusing more on conveying central concepts to appeal to their niche audience, while professional translators aim to produce a more holistic and stylistically faithful translation that caters to a broader readership.
Moreover, the role of editing and quality control processes in professional translations further contributes to the stylistic divergences between AT and PT.Professional translations often undergo rigorous editing and quality control, ensuring consistency, coherence, and adherence to the target language's norms.In contrast, amateur translations may lack such systematic editing, leading to a higher prevalence of linguistic and stylistic inconsistencies.In addition, professional translation teams often have established channels for feedback and collaboration, allowing for regular discussions and improvements throughout the translation process.Amateur translation groups, while potentially more open and inclusive, may lack such structured feedback mechanisms, resulting in a less cohesive final product.

Conclusion
This study has highlighted the stylistic differences between amateur and professional translations of Jin Yong's martial arts novel, Legends of the Condor Heroes.By employing Biber's multidimensional analysis to separately examine fictional speech and narration, we uncovered nuanced variations in the linguistic features of these translations.These stylistic disparities can impact readers' comprehension and engagement with the translated text.The lack of variation between speech and narration in amateur translations may affect readers' ability to distinguish between character voices and narrative perspectives, while the use of more abstract language could hinder the vividness and immediacy of the storytelling.This research contributes to the growing body of work on amateur translation by systematically comparing amateur and professional translations in the context of Chinese martial arts fiction.The findings underscore the importance of considering the stylistic impact of amateur translation practices on the dissemination and reception of popular genres across cultural boundaries.
Despite the valuable insights gained from this study, it is important to acknowledge its limitations and propose avenues for future research.The current analysis focused on a single source text, Legends of the Condor Heroes, and its translations, which may limit the generalizability of the findings to other texts and translation contexts.The study primarily examined register features using Multidimensional Analysis, leaving room to explore other linguistic or stylistic elements that could impact the overall quality of the translations.To further enhance our understanding of the divergences between amateur and professional translations, future research should consider expanding the scope of investigation by incorporating a wider range of texts and employing diverse quantitative linguistic methods, such as activity and descriptivity measures (Zhou et al., 2023), dependency distance measures (Xu and Liu, 2023;Ma et al., 2024), andentropy-based approaches (Liu et al., 2022;Wang et al., 2024).Moreover, integrating qualitative research methods, such as interviews or surveys with translators, could provide valuable insights into the decision-making processes and intentions behind their linguistic choices.By combining these methods, researchers can gain a more comprehensive understanding of the factors influencing the observed translation styles and the translators' perspectives and experiences, ultimately contributing to a more holistic view of the divergences between amateur and professional translations.

A
high score represents an effective and interactional text.A low score represents an informationdense text.Dimension 2 Narrative versus non-narrative concerns A high score indicates the text is narrative, a low score non-narrative.Dimension 3 Explicit versus situationdependent reference A high score indicates the text is context-independent, a low score context-dependent.Dimension 4 Overt expression of persuasion A high score indicates that the text explicitly marks the author's point of view and their assessment of likelihood and/or certainty.Dimension 5 Abstract versus non-abstract information A high score indicates the text is highly technical, abstract, or formal.Dimension 6

Fig. 1
Fig. 1 Comparison of register dimensions between AT and PT in fictional speech.This figure illustrates the comparison of register dimensions between amateur translation (AT) and professional translation (PT) in fictional speech across four selected dimensions.Each panel represents a different dimension.

Fig. 2
Fig. 2 Comparison of register dimensions between AT and PT in narration.This figure illustrates the comparison of register dimensions between amateur translation (AT) and professional translation (PT) in fictional narration across four selected dimensions.Each panel represents a different dimension.

Fig. 4
Fig. 4 Stylistic differences in the translations of fictional narration in the four dimensions compared to other registers.This figure compares stylistic differences in the translations of fictional narration of amateur translation (AT) and professional translation (PT) with other registers across four selected dimensions.Each dotted line represents a different register.

Fig. 3
Fig. 3 Stylistic differences in the translations of fictional speech in the four dimensions compared to other registers.This figure compares stylistic differences in the translations of fictional speech of amateur translation (AT) and professional translation (PT) with other registers across four selected dimensions.Each dotted line represents a different register.

Table 1
Amateur and professional translations of Jin Yong's works.

Table 2
Descriptive statistics of translations of LCH.

Table 4
Dimension scores in the fictional speech of AT and PT.

Table 6
Dimension scores in the narration of AT and PT.
Na nvzi shenchu yizhi ranman xianxue naojiang de shouzhang, zai yueguang xia yimian xiao yimian qiao.AT:She stretched her hand, spotted with blood and brain matter, and examined it in the moonlight still laughing.PT: Her fingers were smeared with blood and spots of the brain.She stretched them out before her, examining them in the moonlight.She was still laughing.Dimension 5 distinguishes between abstract and nonabstract discourse.Features with higher mean ranks and sums of ranks indicate a more abstract, technical, and formal register.As shown in Table6, AT exhibits a positive score, suggesting an abstract nature, while PT leans towards a non-abstract register with a negative score.Dianxiaoer jian ta lianse ganga, zhidao qiongsuan de hua bucuo, shense dengshi buru shicai gongjin, ting yao tu du de dao.