国内外研究状况及水平: Translation Study on English Popular Science (EPS) in China 英文科普(EPS)在中国的翻译研究 The introduction and translation of English Popular Science (EPS) in China on a considerable scale has begun ever since 1980s and witnessed a boom in the next two decades. However, it was not until the turning of the 20th and 21st century that a systematic translation study on the genre was gradually put in place to clarify certain key concepts in this field as well as to discuss the translation strategies and methods regarding its writing style and the demands of its target readers. 中国自 20 世纪 80 年代起开始大规模引进和翻译英文科普作品,并在随后的二十年中迎来了蓬勃发展。然而, 直到 20 世纪初,才逐渐形成了系统的英文科普翻译研究,阐明了该领域的一些关键概念,并探讨了针对英文科普作品的文体和目标读者需求的翻译策略和方法。 1.1 Definition and Characteristics of EPS 1.1 EPS 的定义及特点 Among the related studies, the priority is to discriminate between EPS and English for science and technology (EST). As early as in 1989, Fang Mengzhi, the pioneer of Chinese translation study on EPS, prompted that EPS should be considered as a branch of EST in his genre-based analysis (1989:20). Fang pointed out that across the wide spectrum of English technical writing, EPS, featuring less formality and a mixture of scientific knowledge, engaging elements and vivid language (1989:20), distinguishes itself from the other sub-genres while sharing limited similarities like delving into a single or interdisciplinary topic within the large category. Sticking to his early classification, Fang further specified that EPS (or science article) is “scientist/journalist-to-layperson writing” a decade later, which differs from scientific paper (“expert-to-expert”) and technical prose/documents (“expert-to-expert/layperson”) within the realm of EST (1999:32). 在相关研究中,EPS 与科技英语 (EST) 的区别尤为突出。早在 1989 年, EPS 汉译研究的先驱方孟智就在其基于体裁的分析中提出,EPS 应被视为 EST 的一个分支 (1989:20)。方孟智指出,在广泛的英语科技写作中,EPS 的特点是形式感较弱,融合了科学知识、引人入胜的元素和生动的语言 (1989:20),这使其有别于其他子体裁, 同时也具有有限的相似之处,例如在大类别中深入探讨单一或跨学科主题。方孟智坚持其早期分类, 十年后进一步指出 EPS (或科学文章) 是 “ 科学家/记者对普通人的写作 ” ,这与 EST 范畴内的科学论文( “ 专家对专家 ” )和技术散文/文档( “ 专家对专家/普通人 ” ) 有所不同 (1999:32)。 Later, Guo Jianzhong deemed in Studies of Sci-Tech Translation, the seminal monograph on the subject matter, that EPS serves as a variation of EST incorporating both scientific and literary elements on the purpose of popularizing scientific and technological ideas and knowledge along with disseminating research methods and scientific spirits (2004: xxii). Seconded Fang and Guo’s definition, Wang Zhenping commented that underlying the long-term clarification on the concept, it reflected how Chinese translators had long confused EPS with EST (2006:35), and such a misperception prevented them from resorting to conducive guidance and obtaining effective feedback from the target readers. 后来,郭建中在其开创性的科技翻译研究著作《科技翻译研究》中认为,EPS 是 EST 的一种变体,它融合了科学和文学的元素,旨在普及科技思想和知识,传播研究方法和科学精神(2004: xxii)。 王振平赞同方建军和郭建军的定义,并指出,长期以来对 EPS 概念的界定,背后反映了中国译者长期以来将 EPS 与 EST 相混淆(2006:35),这种误解阻碍了译者对 EPS 进行有益的引导,也阻碍了他们从目标读者那里获得有效的反馈。 Then, in the latest overall review on the subject matter, Xu Bin and Guo Hongmei attributed the widespread confusion of EPS with EST to the “underdevelopment” of translation study on EPS at home with fewer periodical articles related (2012:42), which leaves room for translation practices and researches from various perspectives. Besides, the review supplemented that EPS, derived from EST, has presented a salient tendency of featuring more literary quality than before (Xu & Guo, 2012:43). In brief, while endeavoring to rectify the misunderstanding of non-professionals, a comprehensive definition of ESP through the relay of translators and researchers manages to offer the premise for a thorough analysis of the sub-genre in terms of its text type and writing style. 随后,在最新的关于该主题的综述中,徐斌和郭红梅将 EPS 与 EST 的普遍混淆归咎于国内 EPS 翻译研究的 “ 不发达 ” ,相关的期刊论文较少(2012:42),这为从不同角度开展翻译实践和研究留下了空间。此外,该综述还补充指出,源自 EST 的 EPS 呈现出比以往更具文学性的显著趋势(徐斌 、 郭红梅,2012:43)。总之,在努力纠正非专业人士对 ESP 的误解的同时,通过译者和研究者的接力,对 ESP 进行全面的定义 ,为深入分析该子体裁的文本类型和写作风格奠定了前提 。 Further, the existent studies focused on the characteristics of EPS on textual level. At the early stage, Fang reflected that scientific activities are of human beings and of the society, thus all those involved in the activities are also placed within the social environment, and can be regarded as “interpreters” with subjectivity like emotional inclination, bias and their own takes on the products of science and technology (1999:32). Naturally, those authors would prefer the means of literary rhetoric like analogue or allusion to reveal their attitudes other than facts and technical contents in their writing. Fang’s reflection originated from the wide-ranging EST works, thus is also applicable in the branch of EPS. 此外, 现有研究侧重于文本层面对 EPS 特征的探讨。早期,方志强认为科学活动是人的活动,也是社会的活动,因此所有参与科学活动的人都置身于社会环境之中,可以被视为具有主观性的 “ 阐释者 ” ,例如情感倾向、偏见以及对科技产品的个人看法 ( 1999:32)。自然,这些作者倾向于使用类比或典故等文学修辞手段来表达他们的态度, 而非仅仅关注事实和技术内容。方志强的思考源于广泛的科技作品,因此也适用于 EPS 这一分支。 Before long, Guo put forth the baseline that popular science, whether in natural or social subfield, is science, made simple for the public (Guo et al, 2004:5,6). From it, Guo distilled the four features of EPS, descending in importance, as being “scientific, literary, accessible and engaging” (2004:7-10), resonating to Fang’s “a mixture of scientific knowledge, engaging elements and vivid expression” (1989:20). It serves as a bridge between the previous and latter-day researches. Respectively speaking, Guo honored scientific scrutiny to be the top, which is embodied in the accurate contents and sound logics of EPS writing; second to it, literary delivery is verified as another indispensable feature based on the definition of EPS above, responsible for readers’ appreciation on aesthetic purpose; accessibility ranking the third since scientific knowledge cannot reach the general public who are unfamiliar with the subject without concise yet vivid expressions, especially by properly adopting rhetoric devices and a conversational tone; last but not least, EPS integrates engaging, even intellectually amusing elements in its writing for it is designed to attract readers with a general education level. So far, Guo’s “four features” have become the recognized guiding principles for Chinese EPS translators. 不久之后,郭沫若提出了科普的基本原则:科普 , 无论属于自然科学还是社会科学, 都是通俗易懂的科学(郭沫若等,2004:5、6)。 在此基础上,郭沫若提炼出科普的四个特征,按重要性依次递减,即 “ 科学性、 文学性、 通俗易懂、趣味性”(2004:7-10), 这与方励之的“ 科学性、趣味性、生动性”的提法相呼应 (1989:20)。 郭沫若将科学性审视放在首位,这体现在 EPS 写作的准确内容和合理逻辑上;其次,文学性表达被证实为基于上述 EPS 定义的另一个不可或缺的特征,它负责读者进行审美欣赏; 易懂性位居第三,因为如果不使用简洁而生动的表达,特别是不恰当地采用修辞手段和对话语气, 科学知识就无法传达给不熟悉该主题的普通公众 ;最后但并非最不重要的是,EPS 在其写作中融入了引人入胜、甚至具有智力趣味的元素,因为它旨在吸引具有普遍教育水平的读者。迄今为止,郭沫若的 “ 四个特点 ” 已成为中国 EPS 翻译公认的指导原则 。 Aside from the aforementioned, Fan Yu and Li Guoguo specifically focused on syntax structures. While witnessing how plain English movement had swept the scientific academics, they observed that authors have embraced the clarity and productivity by employing active voice and first person in their writing, hence the turgid style represented by passive voice and redundant components has been cast aside for its low efficiency in composition and acceptance (2004:88). Although Fang and Li’s observation rings true given the accessibility of EPS, Xu and Guo, taking more indexes like type of words and average sentence length (ASL) into account, gave a critical examination of readability based on quantitative studies (2012:43). By comparing EPS with other fictions and practical texts, it is found that the so-called “plain” style is aiming to reduce obscure jargon of a certain subject so as to streamline the delivery of logic or mechanisms to laypeople. In this case, they concluded that both researchers and translators should avoid misconceiving the plain and accessible traits of EPS as simplified English regardless of the complexity of the topic. Instead, it is encouraged to customize translation standards to various levels of EPS along with textual analysis. 除上述内容外,范宇和李果果特别关注语法结构。他们见证了简明英语运动席卷科学界,并观察到作者们为了追求清晰和高效的写作 ,纷纷采用主动语态和第一人称 ,而以被动语态和冗余成分为代表的冗长文体因其低效的写作效率和接受度而被摒弃(2004:88)。虽然方和李的观察在考虑到简明英语的易读性后也得到了印证,但徐和郭基于定量研究,更多地考虑了词汇类型和平均句子长度(ASL)等指标,对可读性进行了批判性检验 (2012:43)。通过将简明英语与其他小说和实用文本进行比较,发现所谓的 “ 简明 ” 风格旨在减少特定主题的晦涩术语 ,以便更高效地向普通读者传达逻辑或机制。在本案例中, 他们得出结论, 无论主题多么复杂,研究人员和译者都应避免将 EPS 的简洁易懂的特点误解为简化英语。相反,我们鼓励根据不同级别的 EPS 定制翻译标准,并进行文本分析。 1.2 The Standards and Challenges of EPS Translation 1.2 EPS 翻译的标准与挑战 Under a clear definition of EPS and a thorough discussion on its characteristics, the ultimate goal of previous researches is to set the standards for EPS translation while pinpointing possible challenges met by translators. Fang first voiced that translators of sci-tech works should be highly aware of the audience, including their roles, attitudes and demands, since those works highlight “persuasiveness” in its exposition (1999:32). He also mentioned how skillful decoding of rhetoric devices can ensue the arguments in the source texts to be more convincing (1999:34). Afterwards, aligned with the “four features” of EPS, Guo continued to propose the “three baselines” for EPS translation that comprise of faithfulness to disciplinary facts and information, stylishness compatible with logic coherence, and comprehensible expressions adapted to the target readers’ thinking pattern and cultural background (2004:59). 在对 EPS 进行清晰定义并深入探讨其特点的基础上,前人研究的最终目标是制定 EPS 翻译的标准,同时指出译者可能面临的挑战。方志强首先指出, 科技作品的翻译应该高度关注受众,包括他们的角色、态度和需求,因为科技作品在阐述时强调 “ 说服力 ” (1999:32)。他还提到,巧妙地解读修辞手段可以使源文本的论点更具说服力(1999:34)。随后,郭沫若针对 EPS 的 “ 四个特点 ” ,继续提出了 EPS 翻译的 “ 三条底线 ” ,即忠实于学科事实和信息 、 风格与逻辑连贯相适应、表达通俗易懂且适应目标读者的思维模式和文化背景(2004:59)。 Following his line of reasoning, Guo discovered two major struggles shared by most Chinese practitioners. One concerns the misinterpretation of source contents due to lacking professional knowledge and language proficiency (2004:60) because the apt interpretation of meaning is the core across almost every dimension of language conversion (2007:1). The other touches on the rigid conformity to English syntax units rendering the opaque translation (2004:66). To address these problems, Guo hence held that a qualified Chinese translation of EPS should be content-centered as well as reader-centered (2004:75) with suitable rewriting to transfer meaning from English to Chinese (2007:2). In favor of the opinions above, Wang listed literary refinement as a must in EPS translation out of readers’ aesthetic need. Similarly, Guo himself supplemented later that translators ought to foster an intuitive grasp of the linguistic nuances in EPS works for difficulties often occur in the translation of seemly trouble-free phrases or sentences (2007:85). 顺着他的思路,郭发现大多数中国译者都面临两大难题 。一是由于缺乏专业知识和语言水平,容易对原文内容产生误解(2004:60),因为对原文意义的准确解读几乎是语言转换各个维度的核心(2007:1)。二是翻译中对英文句法单位的刻板遵循,导致译文晦涩难懂(2004:66)。针对这些问题,郭认为,合格的 EPS 中文翻译版不仅要以内容为中心,还要以读者为中心(2004:75 ),并进行适当的改写,将英文意思转化为中文 (2007:2)。王赞同上述观点,他认为出于读者审美的需要,EPS 翻译必须注重文学修养。同样,郭沫若后来也补充道,译者应该直观地掌握英语作品中的语言细微差别,因为翻译一些看似没有问题的短语或句子时,往往会遇到困难(2007:85)。 To finish, Xu and Guo emphasized the accuracy and consistency of terms and glossaries in EPS translation (2012:45). Meanwhile, they recapitulated that EPS writers tend to elaborate on scientific principles or technologies through preceding instantiations or well-known allusions to English natives (2012:46), which consequently obliges translators to bridge cognitive gap and prepare for the targeted readers. 最后,徐和郭强调了 EPS 翻译中术语和词汇的准确性和一致性(2012:45)。同时,他们还指出,EPS 作者倾向于通过先前的实例或众所周知的典故来阐述科学原理或技术,以方便英语母语人士理解(2012:46),这迫使译者弥合认知差距,并为目标读者做好准备。 In precis, the “four features” (being scientific, literary, accessible and engaging) of EPS works demonstrates how professional writers are apt to mobilize experience-based resources encompassing the accumulation of subject knowledge, thinking patterns and language cultures, i.e. multiple kinds of cognitive schemas, to accomplish the mission of enlightening the public. For that reason, it can be drawn from the “three baselines” for EPS translation that it’s essential for translators to activate the corresponding schemas of the target readers from a top-down perspective, which will be discussed in the following chapters. 概括而言, EPS 作品的 “ 四大特性 ” ( 科学性、文学性、通俗易懂、引人入胜 ) 体现了专业作家善于调动基于经验的资源,包括学科知识积累、思维模式和语言文化,即多种认知图式,来完成启迪大众的使命 。由此,从 EPS 翻译的 “ 三条基线 ” 可知,译者必须自上而下地激活目标读者的相应图式,这将在后续章节中讨论。 Schema Theory and EPS Translation 图式理论与 EPS 翻译 Unlike the prevalent bottom-up model seen in EPS translation, i.e. from words, sentences to passages, the author prefers a top-down framework that radiates from reader’s cognitive activation to contextual, linguistic and cultural compliments, which pilots the translation process coursing through the analysis of source texts, choices on translation tools and expected achievements in post-editing. On this purpose, the chapter will tease out the origin of schema theory, its implementation in translation and association with the characteristics of EPS. 与 EPS 翻译中普遍存在的自下而上 (即从词 、句到段)的模式不同 , 作者主张自上而下,从读者认知激活到语境 、语言和文化补充的翻译过程,以此指导翻译过程,贯穿原文分析、 翻译工具选择以及译后编辑的预期效果。为此,本章将梳理图式理论的起源、其在翻译中的应用以及与 EPS 特点的关联 。 2.1 Application of Schema Theory in Translation 2.1 图式理论在翻译中的应用 To date, numerous earlier researchers, including but not limited to Wei Wei and Liu Mingdong (2007), Kang Lixin (2011) and Fang Jing (2023), are credited for the inclusive appraisals on the progress of schema theory. Pertaining to their studies, the theory experienced three succeeding stages in terms of philosophy, psychology and application. 迄今为止, 包括魏巍和刘明东(2007)、康立新(2011)、方静(2023)在内的众多前辈研究者对图式理论的发展做出了较为全面的评价 。 在他们的研究历程中,图式理论经历了哲学、心理学和应用三个发展阶段。 As the ideological blueprint, Immanuel Kant (1781) in his Critique of Pure Reason identified schemas as intermediaries between perception and concepts, functioning as stimulants to help perceive the world. With the introduction of “schema” in philosophical field, British cognitive psychologist Frederick Bartlett (1932) steered it to the direction of psychological cognition. In this case, Bartlett identified "schema" as the mental structures that organize and interpret past experiences, influencing how new information is perceived, thus specified how schema operates in a theoretic sense for the first time (Fang, 2023:34). He posited that memory is not a straightforward reproduction of events but an interpretative process that constructs mental representations by integrating new information with pre-existing knowledge structures (Bartlett, 1932). Moreover, Jean Piaget (1952) expanded on the psychological aspects of schema theory, viewing schemas as structures or organizations of actions that result from the interaction between a subject and an object. 作为其思想蓝图,伊曼努尔·康德(1781)在其《纯粹理性批判》中将图式视为感知与概念之间的中介,作为刺激物帮助人们感知世界。随着“图式”概念被引入哲学领域,英国认知心理学家弗雷德里克·巴特利特(1932)将其引向心理认知领域。巴特利特将“图式”定义为组织和解释过去经验的心理结构,它们影响着新信息的感知方式,从而首次在理论意义上明确了图式的运作方式(方志强,2023:34)。他认为记忆并非对事件的直接再现,而是一个通过将新信息与先前存在的知识结构整合起来构建心理表征的解释过程(巴特利特,1932)。此外,让·皮亚杰(Jean Piaget,1952)扩展了图式理论的心理学方面,将图式视为主体与客体之间相互作用产生的行为的结构或组织。 Moving into the end of the 20th century, more experts invested attention in exploring the practical potential of the theory, especially in linguistics and discourse. Rumelhart (1980), a notable figure in artificial intelligence, described schemas as cognitive building blocks, encompassing knowledge units that combine and categorize all forms of events and objects. This idea accentuates the role of schemas in organizing people’s cognitive processes, which is crucial for understanding and interpreting language (Zhou, 2002:86). Later developments in schema theory were advanced by scholars such as Brown and Yule (1983), and Cook (1992). They defined schemas as the knowledge structures necessary for discourse processing, emphasizing that understanding a text involves activating relevant schemas that link new information to prior knowledge. 进入 20 世纪末,越来越多的专家开始关注图式理论的实践潜力,尤其是在语言学和话语分析领域。人工智能领域的杰出人物鲁梅哈特(Rumelhart,1980)将图式描述为认知的构建模块,涵盖了组合和分类各种事件和对象的知识单元。这一观点强调了图式在组织人类认知过程方面的作用,这对于理解和阐释语言至关重要(周志强,2002:86)。后来,布朗和尤尔(Brown and Yule,1983)以及库克(Cook,1992)等学者推动了图式理论的发展。他们将图式定义为话语处理所必需的知识结构,强调理解文本需要激活相关图式,将新信息与先前知识联系起来。 The varying perspectives above from philosophy, cognitive psychology, and other applicable disciplines highlight the broad applicability and fundamental importance of schema theory in understanding cognitive processes. Relying on these groundworks, contemporary research in translation studies has utilized schema theory to explore how translators process and reconstruct meaning. For instance, Zhu Ying and Cai Bin (2016:83) highlight that schemas in translation encompass linguistic, content, and rhetorical dimensions. They argue that the more extensive and refined these schemas are in a translator’s mind, the greater the likelihood and efficiency of comprehending the source text (Zhu & Cai, 2016:83). Naturally, the focus shifts to how schema theory is applied in translation practice. 上述来自哲学 、 认知心理学和其他应用学科的不同观点凸显了图式理论在理解认知过程方面的广泛适用性和根本重要性 。 基于这些基础研究 ,当代翻译研究已经利用图式理论来探索译者如何处理和重构意义。 例如, 朱颖和蔡斌 (2016 :83 )强调翻译中的图式涵盖语言 、内容和修辞维度。他们认为,译者头脑中的图式越广泛和精细,理解源文本的可能性和效率就越大( 朱颖和蔡斌,2016:83)。 自然而然,焦点转移到如何将图式理论应用于翻译实践。 Translation, as a cross-linguistic and cross-cultural communicative activity, requires the translator to understand the unique cultural background knowledge associated with the work. Schema theory, which is based on linguistics and psychology, emphasizes the role of this background knowledge in language comprehension (Peng, 2000:64). For translators, understanding is in twofold: one is the translator's understanding of the original text, and the other is making the translated text comprehensible to the target language readers (Peng, 2000:64). More specifically, the first fold is the premise and basis for the second; while the second fold is the purpose and necessary outcome of the first, and also serves as a test for the first understanding. 翻译作为一种跨语言、跨文化的交际活动,要求译者理解作品所蕴含的独特文化背景知识。 基于语言学和心理学的图式理论强调这些背景知识在语言理解中的作用(彭建军,2000:64)。 对译者而言 ,理解包含两个层面 :一是译者对原文的理解;二是使译文能够被目标语读者理解 (彭建军,2000:64)。 具体而言,理解是理解的前提和基础;理解是理解的目的和必然结果,同时也是对理解的检验。 Then, stemming from translators’ two-fold understanding, translation is essentially about constructing a "schema text" in the translator's mind, which involves correctly decoding the source language. The process includes activating and enriching the existing schemas in translators’ mind (Liu, 2002:56). When new information interacts with and is integrated into existing background knowledge, and is placed in the appropriate channel, the schema is activated, thus providing concrete samples to activate the target readers’ correspondingly schemas (Peng, 2000:66). Rumelhart's (1980) assertion that schemas involve any form of event or object and constitute a comprehensive memory organization illustrates their importance in translation. Translators must activate these knowledge structures to interpret and reconstruct the text accurately. This concept aligns with Liu Mingdong and Liu Kuanping's view (2004) that translating involves decoding schemas of source language and re-encoding them in the target language to maintain fidelity and elegance. 基于译者的双重理解, 翻译的本质是在译者头脑中构建“图式文本”,这涉及对源语的正确解读。 这个过程包括激活和丰富译者头脑中已有的图式 (刘,2002:56)。当新的信息与已有的背景知识相互作用、融合,并被置于适当的通道中时,图式就会被激活,从而提供具体的样本来激活目标读者相应的图式(彭,2000:66) 。 鲁梅尔哈特 ( 1980 ) 的论断表明了图式在翻译中的重要性,他认为图式涉及任何形式的事件或对象,并构成一个综合的记忆组织 。译者必须激活这些知识结构才能准确地解读和重构文本。这一概念与刘明东和刘宽平 (2004)的观点一致 ,即翻译涉及解码源语言的图式并将其重新编码到目标语言中以保持忠实度和优雅度。 Furthermore, Zhou Dubao (2002:86) elaborates that the translation process involves filling in the gaps of the untranslated text by activating relevant knowledge structures from memory. In other words, the translation process is not merely passive decoding but an active and creative cognitive activity. Translators construct new schemas or modify existing ones to fit the target language and culture. If the information from the source text aligns with the translator's existing knowledge structures, the translation proceeds smoothly. Otherwise, translators may employ strategies such as modification, addition, or reordering to reconcile discrepancies and construct new schemas suitable for the target language (Zhu & Cai, 2016:85). 此外,周笃宝(2002:86)进一步阐述了翻译过程,即通过激活记忆中的相关知识结构来填补未译文本的空白。换言之,翻译过程并非仅仅是被动的解码,而是一个主动且富有创造性的认知活动。译者构建新的图式或修改现有的图式以适应目标语言和文化。如果源文本中的信息与译者现有的知识结构相符,翻译就会顺利进行。否则,译者可能会采用修改、添加或重新排序等策略来协调差异,并构建适合目标语言的新图式(朱文、蔡文静,2016:85)。 Hence, applying schema theory in translation hinges on the interaction between a translator's prior knowledge and new information presented in the source text. This interaction is facilitated more through top-down analysis (also known as conceptually driven processing) (Zhou, 2006:86) instead of bottom-up processing involves decoding individual linguistic units—words, phrases, and sentences. Top-down analysis allow translators to leverage their knowledge of the world, language, and subject matter to grasp the overall meaning of the text. 因此, 图式理论在翻译中的应用取决于译者先前知识与源文本中呈现的新信息之间的互动。这种互动更多地通过自上而下的分析 (也称为概念驱动处理 )(周 , 2006:86)而非自下而上的处理来实现。自上而下的分析涉及解码单个语言单位——单词、短语和句子。 自上而下的分析使译者能够利用他们对世界、语言和主题的知识来掌握文本的整体含义。 Analysis of Schemas in EPS EPS 中的模式分析 The ongoing development and refinement of schema theory continue to offer valuable insights into the cognitive processes underlying translation. According to the previous researchers, schemas fall into various categories, which covers almost all-around factors in translators’ decision-making. Zhou Dubao promoted structure schemas (or formal schemas) and content schemas (Zhou, 2002:86). In his opinion, structure schemas refer to knowledge about the structure and rhetoric of articles, including character symbols, phonetics, vocabulary, syntax, and semantic structures; content schemas refer to knowledge about the content and scope of articles, including cultural knowledge, pragmatic knowledge, and specialized knowledge. Such a division serves as an example of what is typically perceived to be the basics of top-down analysis in translation. 图式理论的不断发展和完善,为理解翻译的认知过程提供了宝贵的见解 。 前人认为 ,图式分为很多种,几乎涵盖了译者决策的各个方面 。周笃宝提倡结构图式(或形式图式)和内容图式(周笃宝,2002:86)。 他认为,结构图式是指关于文章结构和修辞的知识,包括文字符号、语音、词汇、句法和语义结构 ; 内容图式是指关于文章内容和范围的知识,包括文化知识 、语用知识和专业知识。 这种划分是翻译中通常被认为是自上而下分析基础的一个例子 。 Similar to Zhou’s structure schemas, Liu Mingdong and Liu Kuanping proposed language involving knowledge of vocabulary, grammar, and idiomatic usage, and deemed that it is essential to accurate conveyance during the translation process (2004:51). Additionally, they added context schema that pertains to all the occasion of language use, logical connections and various subjective and objective environmental factors that affect the meaning of discourse (2004:51). Moreover, it is important to convey the "stylistic schema," which refers to the stylistic features of the text. Different texts have distinct stylistic characteristics; some are colloquial, while others are formal. The stylistic schema is closely related to translation (Liu and Liu, 2004:52). 与周建军的结构图式类似,刘明东和刘宽平也提出了语言的概念,包括词汇、语法和惯用语的知识,并认为语言在翻译过程中对于准确传达意义至关重要(2004:51)。此外,他们还提出了语境图式,涵盖所有语言使用的场合、逻辑联系以及影响话语意义的各种主客观环境因素(2004:51)。此外,传达“风格图式”也至关重要,它指的是文本的风格特征。不同的文本具有不同的风格特征,有些是口语化的,有些是正式的。风格图式与翻译密切相关(刘建军和刘建军,2004:52)。 Likewise, Wang Lidi stated knowledge schemas that are sided with Zhou’s content schemas. In Wang’s explanation, knowledge schemas provide references for interpreting the content of texts, assist in the establishment and selection of word meanings, help predict subsequent text, guide information extraction, and aid in reconstructing information (2001:19). 同样,王立迪也提出了与周文内容图式相符的知识图式。在王的解释中,知识图式为文本内容的解读提供了依据,有助于词义的建立和选择,有助于预测后续文本,指导信息提取,并有助于信息重构(2001:19)。 Finally, Zhou added that cultural schemas should also be included as knowledge structures based on extra-textual cultural knowledge, including customs, habits, and traditions (2002:87). Cultural schemas are knowledge structure blocks about "culture," existing in human minds through previous experiences. During translation, the translator must correctly decode the cultural schemas in the source language and transplant them into the target language to facilitate the correct understanding by target language readers, thereby achieving the purpose of cross-cultural communication in translation (Liu and Liu, 2004:52). 最后,周教授补充道, 文化图式也应作为基于文本外文化知识的知识结构,包括风俗、习惯和传统(2002:87)。 文化图式是关于“文化”的知识结构块, 通过先前的经验存在于人类的头脑中。在翻译过程中,译者必须正确地解读源语中的文化图式,并将其移植到目标语中,以便目标语读者正确理解,从而实现翻译中跨文化交际的目的(刘志军、刘志军,2004:52)。 EPS texts often contain specialized terminology and complex concepts requiring translators to have robust schemas encompassing linguistic, content, contextual, stylistic, and cultural knowledge. Therefore, all the schemas above can be classified into three major categories: EPS 文本通常包含专业术语和复杂概念,要求译员具备涵盖语言 、内容、语境、风格和文化知识的稳健图式。 因此,上述所有图式可分为三大类: Content Schemas: key scientific concepts, professional elements and terminology 内容图式: 关键科学概念 、 专业要素和术语; Linguistic Schemas: context, style, rhetorical devices 语言图式:语境、风格、修辞手法; Cultural Schemas: cultural references or examples that need adaptation for Chinese readers. 文化图式:需要适应中国读者的文化参考或例子。 Tailored to the three features of EPS (scientific, literary, accessible and engaging) , the three schemas will facilitate translators to have a holistic preview of the source text so as to be faithful to disciplinary facts and information, compatible with the style and logic coherence, and accessible to the target readers’ thinking pattern and cultural background. In this case, schema theory can effectively address the unique challenges posed on the translation of EPS. 三种图式理论针对 EPS 的三大特点( 科学性、 文学性、 通俗易懂和引人入胜 ),有助于译者对原文进行整体性解读,从而忠实于学科事实和信息 , 兼容文体风格和逻辑连贯性,并贴近目标读者的思维模式和文化背景 。 图式理论能够有效应对 EPS 翻译所面临的独特挑战 。 In conclusion, schema theory provides a comprehensive framework for understanding and improving the translation of EPS. By leveraging content, linguistic and cultural schemas, translators can effectively bridge the gap between source and target languages, ensuring that the interpretation and adaptation is accurate and accessible to the target readers. 总而言之,图式理论为理解和改进 EPS 翻译提供了一个全面的框架。译者能够利用内容、语言和文化图式, 有效地弥合源语言和目标语言之间的差距,确保译文的准确解读和改编 ,并使其易于目标读者理解。 AI Translation 人工智能翻译 As the society has marched from the Information Age into the AI Era, almost every walk of life requiring intellectual activities like information processing or creative production has been inevitably involved in the push-and-pull with the expansion and penetration of artificial general intelligence (AGI). Relying on a direct interaction with natural language, translation industry, the building block of language services, also need to proactively absorb in the frontier to grope for a path for its future survival against the ontological shock by AI, as well as the career wellbeing of nowadays translators challenged by highly capable algorithm. Since the report aspires to explore the technological integration in EPS translation under the framework of schema theory, the chapter will begin with the edges and shortages of AI translation between English and Chinese based on previous researches, then extend to a suitable translation mode feasible in the era of generative AI. 随着社会从信息时代迈入人工智能时代,几乎所有需要信息处理或创意生产等智力活动的行业都不可避免地卷入了通用人工智能(AGI)扩张和渗透的拉锯战。作为语言服务基石,翻译行业依赖于与自然语言的直接交互,也需要积极探索前沿技术,探索未来生存之道,以应对人工智能带来的本体论冲击 ,以及当今译员在高性能算法挑战下的职业发展。由于本报告旨在探索图式理论框架下 EPS 翻译的技术融合,本章将基于前人研究成果,探讨英汉人工智能翻译的优势与不足,并由此延伸至生成式人工智能时代可行的翻译模式。 3.1. Development of AI Translation 3 .1 . 人工智能翻译的发展 Although 2023, after the popularity of ChatGPT, was globally recognized as the “the dawn of AI”, the debates on the application of AI, or its predecessor Neural Machine Translation (NMT), were commenced long before the milestone year. In China, the previous researches on AI translation between English and Chinese can be roughly divided into three stages: prediction (before 2022), assessment (2023), and reflection (from 2024 onwards). Within the first two stage, there was always a focus on the scope of AI’s competence and the “dos and don’ts” of human translators. 尽管 2023 年在 ChatGPT 的流行之后被全球公认为 “ 人工智能的黎明 ” , 但关于人工智能或其前身神经机器翻译(NMT)的应用争论早在这一里程碑年份之前就已展开。在中国,此前对人工智能英汉翻译的研究大致可以分为三个阶段:预测(2022 年之前)、评估(2023 年)和反思(2024 年以后)。在前两个阶段,人们始终关注的是人工智能的能力范围以及人类翻译的 “ 该做与不该做的事情 ” 。 In the first stage, two noteworthy studies were served by Xiong Can (2020) and Xu Bin with the co-write Song Shuang (2021). Xiong contributed an early estimation on the possibility of AI performance in literary and non-literary translation. From the perspective of Connectionism, the scholar deemed the essence of AI as a simulation of the information processing in human’s brain (2020:107). In this case, NMT, like Google Translation, is able to mobilize its accumulated experience and couple with contextual understanding to improve accuracy and fluency (2020:108). Despite that what Xiong exemplified has become a relatively outdated derivative of AI, the argument did reflect that AI tools have the potential of accelerating translation process through analyzing from a big picture instead of driven by words or sentences. Such a working mechanism befits to the top-down pattern shown in schema theory. But Xiong also pointed out that cultural and pragmatic interpretation still worries AI translation (2020:112), which indicates the necessity of post-editing by human translators. Steering towards this direction, Xu and Song prompted a reasonable guess that post-editing will be widely accepted and even embraced by translators as a promising working pattern as AI development along with NMT is to receive much more recognition by professionals of language services and the general public (2021:115). 第一阶段,熊灿(2020)和徐斌与宋爽(2021)合作完成了两篇值得关注的研究。熊灿对人工智能在文学和非文学翻译领域应用的可能性做出了早期评估。他从联结主义的视角出发,认为人工智能的本质是对人脑信息处理过程的模拟 (2020:107)。在这种情况下,神经机器翻译(NMT)系统,例如谷歌翻译,能够运用其积累的经验,结合语境理解来提高翻译的准确性和流畅性(2020:108)。尽管熊灿所举的例子已成为相对过时的人工智能衍生物,但该论点确实反映了人工智能工具具有通过从宏观角度进行分析而不是由单词或句子驱动来加速翻译过程的潜力 。这种工作机制符合图式理论所展现的自上而下的模式。但熊先生也指出, 文化和语用解释仍然令人工智能翻译感到担忧(2020:112),这表明人工翻译进行后期编辑的必要性 。 朝着这个方向,徐和宋提出了一个合理的猜测:随着人工智能的发展以及神经机器翻译( NMT)将得到语言服务专业人士和普通公众的更多认可, 译后编辑将被翻译人员广泛接受,甚至作为一种有前途的工作模式而受到欢迎(2021:115)。 After the advent of ChatGPT, AI has evolved into the generative version, which consequently leads to the interactive exchanges in translatio ChatGPT 出现后,AI 进化到了生成式,从而带动了翻译领域的互动交流。n practice 实践oth 其他er than the one-way output by traditional NMT. In the assessment stage, practitioners and researchers were able to get firsthand experience from applying generative AI tools to translation projects. 比传统 NMT 的单向输出更具优势。在评估阶段,从业者和研究人员能够获得将生成式 AI 工具应用于翻译项目的第一手经验。Impressed 印象深刻 by ChatGPT 作者:ChatGPT’s revolutionary featur 的革命性特点es, Wang Huashu with Xie Ya sketched the edges of it as prominently improved accuracy and fluency, more automatic and creative generation, 例如,王华书与谢娅勾勒的边缘线条明显提高了准确性和流畅性,更加自动化和富有创意地生成,interact 相互影响ive synergy with translators by real-time prompters as well as continuously enhanced multi-modality (2023:81- 通过实时提示器与翻译人员进行有效协同,并不断增强多模态性(2023:81-82). In view of that, they argued from both technical and educational perspective that 82)鉴于此,他们从技术和教育角度认为“how to integrate with technology to raise the standards of language services 如何与技术相结合,提高语言服务的标准” has outweighed 已经超过“whether “无论 technology will replace human translators 技术将取代人类翻译” to be the hot issue at 成为热点问题present (2023:83). Furthermore, it is 现在(2023:83)。此外,concluded 结论 by Wang and Xie that AI with its upgrade and expansion is joining knowledge of natural science and liberal arts together, causing 王和谢认为,人工智能及其升级和扩展正在将自然科学和人文科学的知识结合在一起,从而导致interdisciplinary 跨学科 effect on translation 对翻译的影响industry 行业 (2023:84). Hence, it is inevi (2023:84)。因此,这是不可避免的table that post-editing, the by-product of technological involvement (2023:85), will become the next subject of translation studies. Similarly, Yang Yanxia and Wei Xiangqing approved 译后编辑,作为技术介入的副产品(2023:85),将成为翻译研究的下一个主题。杨艳霞和魏祥庆也同样赞同that 那 generative AI is transforming the workflow from 生成式人工智能正在改变工作流程machine-assisted hum 机器辅助嗡嗡声an translation to AI translation managed and edited by human translators (2023:94). But then, the two 翻译由人工翻译管理和编辑的人工智能翻译(2023:94)。但是,两者outlined 概述 that 那 currently speaking, the comfort zone of AI 目前来说,人工智能的舒适区translation 翻译 is mostly confined to scientifically informative texts loaded with facts (2023:95). L 主要局限于充满事实的科学信息文本(2023:95)。Likewise, the 同样,compatibility 兼容性 of AI translation and 人工智能翻译和“hard text 硬文本” is also 也是acknowledged 承认 by Ye Zinan (2023:177). In other words, Ye 作者:叶紫楠(2023:177)。换句话说,叶subscribed 订阅 that it still calls 它仍然调用for human’s 对于人类 understanding and judgement when translating expressive 翻译表达性文章时的理解和判断“soft text 软文” flexible 灵活的 in cultural 在文化方面 and 和emotional 情绪化 nuances (2023:177), which explains to 细微差别(2023:177),这解释了wh 什么y human decision-making is central to EPS, the combination of hard and soft textual features. 人类决策是 EPS 的核心,是硬文本特征和软文本特征的结合。 As more intricate ChatGPT 4 appeared by the end of 2023, in-depth studies on AI translation has since then touched on more untried areas including but not limited to style convention (Zhang & Wang, 2024), syntactic variation and structure refinement (Yu, 2024) and terminology extraction and management (Long & Zhou, 2024). It is via these efforts that a more reliable evaluation of AI’s translation competence was founded on critical thinking out of quantitative researches, which enters the stage of “reflection”. Speaking of AI’s undeniable privileges, Yu complimented that attributed to rich sources of texts input in its training, ChatGPT is able to diversify the use of vocabularies and perform better in constructing long and difficult sentences (2024:305). In addition, thanks to its Internet-based real-time search function, it’s underscored by Long and Zhou that ChatGPT can quickly retrieve professional terms and list background suggestions related to the source texts (2024:7), which is conducive to translators for their reference work. However, when comes to the style of source texts, Zhang and Wang claimed that even the advanced ChatGPT 4 is yet in want of ideal command of transcoding shifting context influenced by social conventions and adapting for reader’s expectation (2024:12-13). 随着 2023 年底更加复杂的 ChatGPT 4 的出现,对人工智能翻译的深入研究自此触及更多未曾涉足的领域,包括但不限于风格规范(Zhang & Wang, 2024)、句法变异与结构细化(Yu, 2024)以及术语提取与管理(Long & Zhou, 2024)。正是通过这些努力, 对人工智能翻译能力的更可靠评估建立在对定量研究的批判性思考之上,进入 “ 反思 ” 阶段 。谈到人工智能不可否认的优势 , Yu 称赞道,由于训练中输入了丰富的文本来源 ,ChatGPT 能够多样化地使用词汇,并在构建长句和难句方面表现更出色(2024:305)。此外 , Long 和 Zhou 强调,得益于基于互联网的实时搜索功能,Ch atGPT 可以快速检索专业术语并列出与源文本相关的背景建议 (2024:7),有利于翻译人员进行参考工作。 然而,当谈到源文本的风格时,张和王声称,即使是先进的 ChatGPT 4 仍然缺乏理想的控制能力,无法根据社会习俗的影响转换不断变化的语境并适应读者的期望(2024:12-13)。 3.2 Post-editing (PE) in the Era of Generative AI 3.2 生成人工智能时代的译后编辑(PE ) The prevalence of post-editing (PE) and its effectiveness is largely correlated with the quality of translation output by MT. From 1980s to the early 21st century, the post-editing had restricted application due to the poor first draft generated by underdeveloped MT entailing manual overhauling rather than polishing (Xu & Song, 2021:112). It was not until the worldwide launch of Google Translator’s Toolkit that PE finally held its position as a widely-adopted workflow in translation industry (Xu & Song, 2021:112). Despite it, the PE back then is now considered as traditional mode, mainly referring to “bottom-up” proofreading of MT translation, thus the polished version cannot completely outdo some “from-the-scratch translation”, which runs counter to its purpose on high efficiency with good quality. However, since the dawn of generative AI is shedding light on all translators, PE has also evolved from “correction” to “integration”, i.e. selecting useful elements from the machine pre-translation and integrating (modification, interpretation or rewrite) them to fit with the ideal version (Xu & Song, 2021:115). In this case, as Xu and Song précised, successful PE demands translators to exhibit long-term capacities regarding professional, linguistic and cultural knowledge as well as decision-making and problem-solving (2021:125), which is associated with activating and implementing content, linguistic and cultural schemas from a holistic perspective in this report. 译后编辑 (PE) 的普及及其有效性与机器翻译的译文质量密切相关。从 20 世纪 80 年代到 21 世纪初 , 由于机器翻译技术落后,生成的初稿质量较差,需要人工大修而非润色 , 译后编辑的应用受到限制 (Xu & Song,2021:112)。直到谷歌翻译工具包在全球范围内推出,PE 才最终成为翻译行业广泛采用的工作流程 (Xu & Song,2021:112)。尽管如此, 当时的 PE 现在被认为是一种传统模式,主要指对机器翻译进行 “自下而上 ” 的校对,因此润色后的版本无法完全超越一些 “ 从头翻译 ” ,这与其高效优质的宗旨背道而驰 。然而,自从生成人工智能的曙光照亮所有翻译人员以来,PE 也从 “ 校正 ” 发展到 “ 集成 ” ,即从机器预译中选择有用的元素并进行集成(修改、解释或重写)以适应理想版本(Xu & Song,2021:115)。 在这种情况下,正如徐和宋所指出的 , 成功的体育教育要求译者展现出专业、 语言和文化知识以及决策和解决问题的长期能力 (2021:125),这与本报告从整体角度激活和实施内容、语言和文化图式有关。 In essence, the biggest flair of generative AI represented by ChatGPT is producing content customized to user’s instantaneous needs through chatbot-like interface and retrieval system link to the Internet (Hu & Geng, 2023:41). As a result, it’s crucial to know how human translators mange to effectively interact with AI tools and drive it to spawn desirable feedback on aspects of situation context, register, syntactic structure, etc., and the key to this problem is prompt engineering, through which translators can enhance their decision-making ability and form a knowledge system in PE (Wang & Wang, 2023: 18). In addition, the adjustment of generative AI to the alternative prompt once again proves that nowadays PE has extended from sheer correction after reviewing the draft by MT to pre-translation editing (Yang & Wei, 2023:93), as schema preparation for the source text in this report. Nevertheless, it’s advised by other experts that AI like ChatGPT 4 empowered by emergent ability and chain of thought (Zhang & Wang, 2024:15) can expose to the risks of hallucination and batch processing (Li & Niu, 2023:30) that do harm to both the credulity of content schema and style and tone of linguistic schema. 本质上,以 ChatGPT 为代表的生成型人工智能的最大亮点在于通过类似聊天机器人的界面和与互联网连接的检索系统,生成满足用户即时需求的定制内容 (Hu & Geng,2023:41)。 因此 , 了解人类译者如何与人工智能工具有效交互,并驱动其在语境、语域、 句法结构等方面产生理想的反馈至关重要,而解决这个问题的关键在于提示工程,通过提示工程,译者可以提升自身的决策能力,形成译前知识体系( Wang & Wang,2023:18)。此外 , 生成型人工智能对替代提示的调整再次证明,如今译前工作已经从单纯的机器翻译审稿后修改扩展到译前编辑(Yang & Wei,2023:93),如本报告中对源文本的图式准备。然而, 其他专家建议,像 ChatGPT 4 这样的人工智能,如果被赋予了涌现能力和思维链(Zhang & Wang,2024:15)的能力,可能会面临幻觉和批处理的风险(Li & Niu,2023:30),这不仅会损害内容图式的可信度,还会损害语言图式的风格和语调。 In general, AI with its massive memories of multidisciplinary knowledge and great learning skills can avoid repeated mistakes (Wang, 2024:6), but to accomplish a qualified EPS translation, such a technological facility must abide by human translator’s interpretation and revision, which renders the new post-editing pattern in the era of generative AI. 总体而言 ,人工智能凭借其海量多学科知识的记忆和强大的学习能力,可以避免重复性的错误(Wang,2024:6),但要完成一篇合格的 EPS 翻译,这样的技术设备必须遵循人类译者的解读和修改,这就为生成性人工智能时代的译后编辑带来了新的模式。 |