Challenges And Reflections On Machine Translation In The Digital Age

Challenges And Reflections On Machine Translation In The Digital Age

by

Challenges And Reflections On Machine Translation In The Digital Age

Challenges And Reflections On Machine Translation In The Digital Age
Wang Zhenzhen, female, from Linyi, Shandong, lecturer at Capital Normal University, PhD, research direction is translation studies and systemic functional linguistics.
Challenges And Reflections On Machine Translation In The Digital Age
Challenges And Reflections On Machine Translation In The Digital Age

1. Introduction

With the innovative development of information technology, human society has entered the digital age. As early as 30 years ago, Nicholas Negroponte predicted various digital survival scenarios for humanity and described the impacts and thought-provoking issues brought by digital technology to our lives, work, education, and entertainment. Today, Negroponte’s prophecies have become a reality, with breakthroughs in new digital technologies such as big data, cloud computing, blockchain, the Internet of Things, and artificial intelligence, profoundly impacting human society. As an ancient human profession and social practice, translation is also undergoing a revolutionary change in this era. From automated online translation services to the widespread application of various translation software, and the birth of generative AI products like ChatGPT, digital technology is bringing changes to translation, which is considered “the most complex human activity since the evolution of the universe,” everywhere. These changes have a thorough and extensive impact on language, culture, and society.
The concept of Machine Translation originated from a memorandum titled “Translation” published by American mathematician Warren Weaver, known as the “father of machine translation,” in 1949. Weaver explored the possibility of translating from one language to another and utilizing modern computer devices with high speed, large capacity, and logical flexibility to achieve this process. Since the first successful experiment in machine translation in 1954, it has undergone nearly 70 years of development. Initially based on rule-based machine translation methods, it later evolved into example-based machine translation and statistical machine translation, centered around corpora and statistics. Since 2014, neural machine translation based on deep learning has gradually emerged, significantly improving translation capabilities compared to previous machine translation methods. At the end of 2022, the large model text-generating AI product ChatGPT exploded globally, with the machine translation capabilities of large-scale language models demonstrating stronger contextual understanding and information completion abilities than neural networks, along with good error correction and interactivity. In the future, it is possible that machine translation will achieve another leap in translation quality through large models.
Compared to traditional human translation, machine translation indeed shows many advantages in practical applications. For instance, the new translation model of “Machine Translation + Post-Editing” makes translation more efficient and has become the mainstream working mode for professional translators now and in the future; machine translation reduces the workload of translators, significantly lowering translation costs; various translation technologies are continuously innovating, with a wide range of applications, playing an increasingly important role in meeting social needs. Similar to the “double-edged sword” effect brought by new technologies at different stages of human history, the development of machine translation technology also causes excitement and anxiety among people, leading to reflections on whether “they are saviors or enemies?”
In recent years, in addition to research on the results of machine translation, the study of problems related to machine translation has also become a hot topic in academia, particularly research on the limitations of translation technologies and their applications. For example, scholars such as Han Jianguo, Yue Chunfang, Weng Huilin, Liu Yachang, Zou Li, Hu Yuhan, Zou Benjin, Zhou Chunfang, and Hu Hongjuan have pointed out from practical applications that machine translation lacks accuracy, readability, and coherence, is prone to low-level mistranslations and omissions, has significant cultural deficiencies, and cannot address issues arising from polysemy, ambiguity, and contextual influences. For instance, Jia Hongwei believes that the inability of machine translation to produce professional translations is due to the lack of experiential input for sense matching. Li Hanji and Chen Haiqing analyzed the external manifestations of the dilemmas faced by existing machine translation technologies and the internal causes of semantic and pragmatic issues, proposing countermeasures including recognizing limitations, highlighting advantages, and interdisciplinary collaborative development. Yao Fusheng raised issues regarding the construction of machine translation corpora and technical limitations, reflecting on three aspects: the instrumental, humanistic, and ecological distribution of translation.
Additionally, some scholars have explored the impacts and issues brought by machine translation. Cadwell et al. examined translators’ attitudes towards machine translation and concluded that the adaptive side generally recognizes that machine translation can provide inspiration, while the resisting side raises questions about it. Wang Huashu and Wang Xin pointed out practical issues that machine translation may bring, such as information security, intellectual property, translation subjectivity, translation ethics, and talent loss. Wang Yun and Zhang Zheng reviewed the implicit risks of machine translation, suggesting that the active agency and constraints of the subject, knowledge-driven and data-driven approaches, as well as symbolic and connectionist perspectives should be combined to promote the optimized development of machine translation. Wu Meixuan and Chen Hongjun examined ethical issues in machine translation technology and applications from aspects such as algorithm bias, data leakage, limitations in language processing, negative impacts on language development, the dissolution of translator subjectivity, ambiguity of responsible parties, and copyright disputes over translation results.
From the above review, it can be seen that previous research on issues has mostly focused on technology and applications, while the research on the impacts of machine translation is still limited. The dimensions of technological reflection should include not only the technology itself but also the relationship between technology and society, and between technology and humans. The relationship between technology and society can be reflected in various levels such as politics, culture, education, and life, while humans are an indispensable dimension in technological reflection and the ultimate focus and destination of technological thinking. This article will focus on three factors closely related to machine translation technology: foreign language education, public language life, and translators, discussing the related issues raised by machine translation.

Challenges And Reflections On Machine Translation In The Digital Age

It should be particularly noted that the impact of machine translation on foreign language education will also be reflected in international Chinese education. In July 2021, the Ministry of Education of China and the National Language and Writing Work Committee officially implemented the “Chinese Proficiency Standards for International Chinese Education” (hereinafter referred to as “Standards”). The “Standards” is the first national standard drawn up by the Language Commission for foreign Chinese learners, comprehensively depicting and evaluating learners’ Chinese language skills and levels. After more than 30 years, the “Standards” have reintroduced “translation” into the basic language skills. Therefore, international Chinese educators should also pay attention to the various issues discussed in this article during Chinese teaching and avoid the negative impacts brought by machine translation in the process of cultivating translation skills among Chinese learners. However, the theoretical research has not yet caught up with this issue, and it is hoped that the analysis in this article can contribute to the discussion on this topic.
Challenges And Reflections On Machine Translation In The Digital Age

2. Issues Raised by Machine Translation

(1) Impact on Foreign Language Education
1. Some foreign language learners may lose motivation and enthusiasm for learning
With the rapid advancement of machine translation capabilities, waves of change have also emerged in foreign language education. The voice of “I learned a foreign language, but with machine translation, I’m unemployed” is widespread. Gao Yihong and others conducted a survey on the learning motivation of undergraduate English learners and summarized seven types of motivations: intrinsic interest in the target language and culture; passing exams to obtain qualifications for further study or graduation; the influence of learning contexts such as courses, teachers, textbooks, and classes; seeking educational and job opportunities abroad, experiencing culture, or immigrating; serving parents and the country; achieving an ideal career, status, and accomplishments; and understanding information and learning other majors through English. These motivations also apply to learners of other foreign languages. Learning motivation largely stems from the necessity of learning. Learners of motivation type 2 are mostly elective students in general foreign language courses in colleges. Some of these learners choose to use machine translation to complete foreign language assignments, but they often cannot evaluate the quality of the assignments and may think, “Machines do it better and more conveniently.” Once they achieve the goal of accumulating credits for graduation, they will lose the motivation to learn foreign languages. Moreover, learners of motivation type 7 aim to acquire information, and current machine translation technology can roughly meet their needs for reading foreign literature; thus, the enthusiasm of these foreign language learners will inevitably decline.
2. Over-reliance on machine translation may inhibit foreign language learners’ language and innovative thinking abilities
Some scholars have conducted surveys on the usage of machine translation among foreign language learners, finding that over 80% of students use related software in their foreign language studies, with some classes even reaching a 98% usage rate. In light of the widespread use of machine translation among foreign language learners, Stapleton & Kin conducted a survey on teachers’ attitudes. The survey showed that all interviewed teachers opposed the practice of learners first writing articles in their native language and then translating them into a foreign language. However, teachers’ attitudes towards using machine translation as a learning tool showed divergence. Teachers opposing its use believe that machine translation reduces the enthusiasm for foreign language learning and suggest obtaining word meanings through dictionaries. On the other hand, teachers in favor argue that if using machine translation can improve learners’ foreign language abilities, they would not oppose students using it as a learning tool. The recent emergence and rapid implementation of ChatGPT have sparked discussions among experts and scholars about the impact of AI technology on education. Supporters view ChatGPT as a more advanced learning tool that can help students engage in personalized learning, benefiting learning efficiency and quality. Those against it primarily worry about academic integrity issues and the potential consequences of students becoming overly reliant on machines to solve problems, without improving their thinking abilities. Furthermore, excessive dependence on AI may lead to students’ passivity and mental inertia, as they tend to seek answers from AI tools rather than engage in independent thinking and judgment, which can undermine their initiative and creative thinking.
3. Foreign language learners using machine translation may raise application norms and ethical issues
The ethical issues of machine translation are a new form and manifestation of language ethics in the digital age. The application ethics of machine translation may also arise in the context of foreign language learning. In the technical application layer of translation ethics, Drugan & Babych classified the sources of machine translation data into two categories: translated texts published by governments and international organizations and translation resources and services shared online. The former generally does not involve serious ethical issues, while the latter presents ethical controversies. However, Dolmaya analyzed the professional codes of conduct from 17 professional translation associations in 15 countries and found that none of the codes addressed the use of software and technology. Aside from the normative issues related to the application of machine translation, foreign language learners using machine translation tools to complete assignments or classroom tasks also raise the question of whether this should be considered plagiarism or cheating. Previously, if learners directly downloaded and copied relevant content from the internet, teachers could easily verify this. However, now, even if using translation engines or ChatGPT to complete translation assignments, if the learners do not proactively disclose this and the translation is of high quality, teachers cannot verify whether the learners used machine translation, nor do they have any means to address this. The education sector is also exploring ways to apply relevant technologies. Currently, some universities both domestically and internationally have explicitly prohibited the use of AI-generated content (AIGC) for writing reports or papers, while others have issued warnings to students regarding copyright and information credibility issues. The Japanese Ministry of Education has publicly stated that it will formulate guidelines for the use of AI chat programs like ChatGPT in schools by the 2023 academic year.
(2) Impact on Public Language Life
1. New translation models may harm human language diversity and sensitivity of thought
The development and application of machine translation technology have had a profound impact on the translation and language service industry. The 2023 “Development Report of the Translation and Language Service Industry in China” indicates that as of the end of 2022, there were 588 companies in China engaged in machine translation and AI business, with 90.1% of related companies believing that the “Machine Translation + Post-Editing” model has improved efficiency. Over 90% of surveyed companies expressed willingness to enhance their machine translation-related technical capabilities in the future to reduce translation costs.
In the practical operation of the new translation model, to improve the accuracy of machine translation, translators often use controlled language for pre-editing the original text in addition to post-editing. The purpose of pre-editing based on controlled language is to effectively reduce the complexity and ambiguity of the original text, thereby enhancing the recognizability of machine translation. Complexity and ambiguity are the main characteristics that distinguish natural language from artificial language; thus, pre-editing can be seen as a process of eliminating the characteristics of natural language to some extent.
Cronin also mentioned the issue of “controlled language” in machine translation: he referred to new materials or novel expressions in language as “accidental content,” and reducing such “accidental content” can lower translation costs. Using controlled language makes texts easier to translate, which also means more translations can be reused. As Li Yuming said, “In the language cooperation between humans and machines, humans not only tame machines but also change themselves in the taming process, altering their writing habits, reading habits, language communication habits, and even language thinking habits.” In daily language life, new expressions that appear as “accidental content” are essential drivers for the continuous evolution of language. Fixing language in univocality and precision, although necessary in technical usage, is not suitable for protecting and promoting the development of language. This neglect of human language characteristics in favor of computer language can easily lead to the “homogenization and unification” of language, resulting in a loss of the original richness of natural language, which over time can harm language diversity and human sensitivity of thought. In fact, the impact of the new translation model under machine translation has already begun to manifest. At the first China Youth Literary Translators Forum on March 18, 2023, several writers and literary translators noted that technology and new media are subtly influencing writing. Texts are becoming increasingly simplified and flattened, especially in science fiction works, making it difficult to discern significant differences between the works of Chinese and foreign authors after translation. Whether Chinese literature or Western literature, they are becoming simpler and easier to translate, lacking multi-layered complexity.
2. Post-edited texts generated by machine translation lack personality, which may dampen readers’ enthusiasm for reading
Walter Benjamin stated in “The Task of the Translator”: “Translation carries the seeds of a certain language, positioned between creation and doctrine. The translator’s work is not as penetrating as doctrine, yet it is deeply etched in history.” This reflects the translator’s role; good translations can enrich the target language and influence the prosperity of national languages. Classic translations, especially literary translations, provide greater possibilities for language, sometimes even shaping a language more than original literature. However, the translated texts generated from pre-editing and machine translation, after post-editing by translators, may be fluent, but readers often feel they lack “warmth” and personality, and the phenomenon of repetition is severe. Even ChatGPT’s generated texts may exhibit limitations such as lacking specific details, using fancy and uncommon vocabulary, lacking emotion, and being unable to express appropriate feelings. As human writing increasingly relies on machines, the standardized text structures can lead to formulaic vocabulary, grammar, and discourse; the texts produced by machine translation may also exhibit formulaic issues. When large volumes of “formulaic” machine translations or post-edited works flood the market and the internet, they will gradually permeate the target language society and subtly erode the original natural language. The simplification of language lacking innovation and creativity may lead readers to experience “aesthetic fatigue,” ultimately resulting in a loss of interest in reading.
3. Cross-cultural communication empowered by machine translation may impact language life
German communication scholar Andreas Hepp defines transcultural communication as “cultural interactions and conflicts conducted through new and old media such as mass media and the internet.” Compared to cross-cultural communication, which emphasizes horizontal cultural comparisons and adherence to one’s own culture in interactions, transcultural communication refers to the vertical process of cultural exchange, emphasizing the interactivity of communication and culture. The communication between different languages is an unavoidable challenge in transcultural communication. With the popularization of machine translation, it has, to some extent, reduced or eliminated barriers to linguistic and cultural exchange, addressing the differences in communication between different languages. With machine translation as a convenient auxiliary tool, some scholars now consider how to adjust their writing style to be better translated into foreign languages, most of which are English, when writing papers. Authors adjust the generation of texts based on the acceptance of both native and foreign readers. Unlike pre-editing, the authors do not aim to eliminate the ambiguity and complexity of the original text but actively absorb the linguistic and cultural elements of the target language to integrate them into their own texts, making them closer to the expression characteristics of the target language and the reading habits of its readers. This phenomenon can be seen as a concrete manifestation of transcultural communication, with machine translation being a powerful facilitator of this process.
If individuals with different thoughts, languages, and cultures can communicate without barriers, mutually absorb the advantageous parts of each other’s cultures, promote cultural integration, and bring about cultural innovation that liberates thoughts, this is undoubtedly the ideal state of transcultural communication. However, in reality, we often have other concerns and worries. The “Sapir-Whorf Hypothesis” suggests that “the structure of language influences the way people understand the world, forming specific thinking patterns within that culture based on habitual and regulated language use; language shapes thought.” In the aforementioned transcultural communication, the native language, influenced by the target language over time, may deviate or change, or it may be assimilated by the target language to some extent, weakening the cultural consciousness and thinking patterns of the source language, while also being detrimental to the external dissemination of the native language’s cultural language.
(3) Impact on Translators
1. Impact on translation practitioners, leading to changes in the structure of the translation workforce
The widespread application of machine translation technology has partially automated and scaled translation activities. Meanwhile, the rapid growth of translation demand spurred by globalization has also promoted the development of economies of scale in the translation industry, shifting the production mode of translation from manual to semi-automated industrial processes. Against this backdrop of development, translation practitioners must use machine translation as an auxiliary tool to pursue work efficiency. The traditional work model centered around translators has changed; the initial draft of language conversion is mostly completed by machines, and the identity of translators is gradually shifting towards reviewers and post-editors, making machine-assisted translation an ultra-human translation process.
For novice translators to become skilled professional senior translators, it often takes years to learn and hone their translation skills. In the traditional translation model, the initial draft is usually completed by novice translators and then reviewed and finalized by senior translators. However, due to the involvement of machine translation technology, many translation companies now prefer to use machine translation for initial drafts to save costs, which directly leads to some novice translators losing job opportunities. According to my understanding, at least 50% of written translation work in the market belongs to low-end translation, which is likely to be quickly replaced by machine translation. In the future, the structure of the translation workforce will change in the human-machine combined model: many novice translators’ translation levels are not even as good as machine translation, leading senior translators to believe that it is more efficient to directly perform post-editing than to correct drafts from novice translators. Therefore, those who could have acted as novice translators may only be able to perform basic data processing tasks, such as conducting pre-editing of texts. As the production capacity of machine translation can expand almost infinitely, it will inevitably reduce the overall demand for human labor in the translation service industry. Hence, the total demand for senior translators with post-editing capabilities will decrease compared to the current situation. It is foreseeable that the future development trend of translation practitioners will be: with the involvement of enhanced artificial intelligence, the threshold for entering the translation industry will become increasingly high, while the number of new entrants will decrease. Novice translators will be “deprived” of opportunities for practical training and growth, while high-level translators will become increasingly scarce, potentially leading to a talent gap in the future.
Moreover, the rise and development of the crowdsourcing translation model, driven by digitalization and internet information technology, have significantly increased the number of amateur translators. As a form of “language gig work,” crowdsourcing translation offers high-speed translation and flexible operation modes, creating significant competitive pressure for professional translators who prioritize quality. With the development of crowdsourcing translation, the boundaries between professional and amateur translators are becoming increasingly blurred, which may also squeeze the living space of professional translators, leading to an identity crisis.
In addition, the explosive development of machine translation technology aided by artificial intelligence also poses a crisis for senior translators, creating a sense of professional anxiety. The development of the language service industry has led to a gradual segmentation of the translation field, with increasing demand for translation in specialized fields such as medicine, law, and finance, raising the requirements for translators’ professional quality and background knowledge. Furthermore, from the perspective of language service providers (LSPs), the roles of translators have also changed due to technology-driven factors. For career development needs, more translation practitioners will consider transitioning from ordinary translators to more diverse industry roles, such as translation project managers, technical communication designers, terminology experts, localization specialists, translation engineers, language asset managers, and human-machine collaborative engineers, thereby escaping competition in the low-end market and the crisis of being replaced by machines.
2. Changes in the role of translators, with reduced subjectivity and influence
With the involvement of modern technology, the participants in translation activities have expanded from traditional translators to three categories: translation subjects, technical subjects, and management subjects. The role of individual subjects is diminishing, while the power of collective or institutional subjects is increasing. In traditional translation activities, translators had a certain degree of subjectivity in generating translated texts, allowing them to choose their translation styles based on factors such as text type, translation purpose, target audience, and publisher requirements. However, once the initial draft is completed by machine translation, translators must enhance and edit the post-machine-generated text according to the machine’s logic, transforming from “secondary creators” to “gap fillers” or “quality inspectors,” severely limiting their subjectivity. Long-term pursuit of efficiency and entanglement in “mechanical” corrections not only erases the translator’s subjective experience as a reader of the source text but also diminishes their creative spirit. Additionally, as Stupiello and Dos Santos noted, the use of machine translation may obscure, diminish, or underestimate the contributions of translators. Users may perceive translators merely as individuals who make minor modifications to the work already completed by machines, believing that the machine is the one that truly solves the problem.
As strong artificial intelligence products are gradually introduced in the future, when the training of language behavior to simulate humans reaches a certain level, AI products can achieve natural language generation capabilities comparable to humans, the space available for translators to “fill gaps” or “edit” will become increasingly limited. In a “cultural turn,” the subjectivity of translators, which was previously highlighted, may once again become “invisible” and could even face the risk of “disappearing” one day.

Challenges And Reflections On Machine Translation In The Digital Age

3. Translators hold varying attitudes towards machine translation
From the era of tool technology to the era of machine technology, and now to the information technology era, humans have developed two extreme attitudes towards technology: technological optimism and technological pessimism. Technological optimists believe in vigorously developing technology and are fascinated by modern technology. In contrast, technological pessimists reject technology, emphasizing the negative effects such as repression of humanity and moral decline accompanying technology, while ignoring the significant material role and spiritual value of technology in driving social development. These two extreme views on technology also apply to machine translation. A certain translator was criticized by netizens for the “serious traces of machine translation” in their translation, and the translator responded by stating, “Machine translation is an issue of professional ethics; a manuscript translated word for word and revised repeatedly being labeled as having ‘serious traces of machine translation’ feels like a personal attack, not a matter of ‘poor translation.'” This dispute indicates that even today, some translators still resist machine translation, even considering its use a violation of professional ethics, while netizens’ comments about the translator’s work being “grammatically incorrect” also reflect the general public’s skepticism towards machine translation.
Marshall McLuhan once said, “The medium is the message,” later transforming it into “The medium is the massage,” believing that “it (the medium) gives us a heavy blow… in a barbaric way, massages us all.” Here, “medium” can be understood as technology; internal experience and knowledge do not always constitute favorable factors for learning new technologies. Often, the more experience one has, the greater the friction. The more solid our internal world, the more painful it becomes under the friction of new technologies. Issues such as the “digital divide” faced by the elderly in the technological age highlight this point; the uneven degree of technology proliferation may widen and deepen the digital divide. Therefore, experienced and slightly older translators may also develop anxiety and resistance towards machine translation technology, with some senior translators even feeling that using machine translation desecrates their years of hard-earned language accumulation, making it psychologically difficult to accept.
Conversely, there is also a tendency to “mythologize” machine translation technology within the translation industry. Some translators overly rely on and trust technology, blindly enthusiastically adopting or excessively using new technologies after their emergence. Of course, there are also machine translation manufacturers, language service providers, or media that exaggerate the quality effects of machine translation technology under the drive of profit to attract capital and market attention. This perspective not only leads translators to lose their sense of self in the use of technology but also inadvertently creates a panic that “machine translation will eventually replace human translation,” increasing psychological pressure on translation practitioners and aspiring new entrants into the field.
With the popularization of machine translation technology and the acceleration of globalization, user translation demands are showing a trend of diversification. For instance, some users of large commercial document translations only require an understanding of the general content of the text, and translators can provide “indicative translations.” In such cases, the translation effects generated by machine translation can largely meet user needs without requiring excessive intervention from translators. Even in the case of post-editing for users with higher quality requirements, as mentioned earlier, translators are often forced into a “formulaic” mode. The weakening of subjectivity may lead to some negative changes in translators’ attitudes towards their work, such as decreased enthusiasm, job satisfaction, and a sense of fulfillment and achievement. Moorkens & O’Brien conducted a survey on translators’ attitudes toward post-editing work, revealing that translators often feel frustrated due to the repetitive nature of post-editing tasks.
4. Translators may become dependent on machine translation, neglecting the improvement of their translation skills
Jean-Jacques Rousseau pointed out the issue of human dependence on technology during the Enlightenment, arguing that technology leads to the degradation of human physical capabilities, making it difficult for people to survive in nature without technology, thus rendering them vulnerable. In today’s technological society, people are increasingly showing strong dependence on technology. Machine translation technology provides many conveniences for translators, liberating them from a large amount of tedious repetitive work and significantly improving work efficiency. Some senior translators who proficiently master machine translation tools may feel that they save time and effort, no longer putting in the effort to delve into language and professional knowledge, but instead leaning towards exploring and improving translation technology. The new professional demands have diminished the humanistic qualities of translators while significantly raising the requirements for technological skills. Furthermore, with the increasing maturity of generative AI such as ChatGPT in the translation field, which can provide increasingly intelligent and personalized translation solutions, translators’ dependence on technology will only grow. Such excessive reliance on technological assistance may evolve into inertia or laziness, as they become immersed in the “comfort zone” created by machines, leading to a degradation of translation abilities and a devaluation of their own worth. Over time, they may risk becoming appendages to machines, distorting the value of humans due to the overemphasis on the utility of tools.
Challenges And Reflections On Machine Translation In The Digital Age

3. Reflections Raised by Machine Translation

In the digital age, the rapid rise of artificial intelligence and the dizzying pace of technological iteration make it easy for people to become completely focused on technology, immersed in the myriad benefits it brings. At such times, we need to reflect on the various problems that technology may bring. Some issues may be unavoidable, but they will enable us to think more calmly and objectively about what we should “do” and “how to do it” in the context of significant technological changes. From the perspective of the relationship between technology, society, and humans, we see a series of issues and impacts that machine translation technology brings to foreign language education, public language life, and translators. Below, I will provide a brief reflection based on this introduction.
(1) Correctly view the role of machine translation in foreign language education, actively avoid risks, and transform “crises” into “opportunities”
As mentioned above, regardless of whether teachers agree or disagree, almost all students use machine translation in their foreign language learning. As a foreign language educator, one should not ignore or evade this phenomenon but should adapt to it and think about how to effectively utilize its advantages to compensate for the problems that arise, thus better transforming the “crisis” of foreign language teaching into an “opportunity.”
First, teachers should learn about the principles and latest developments of machine translation technology and familiarize themselves with its use, so they can guide learners to use it correctly and effectively. Additionally, they should establish rules for learners’ use of machine translation based on specific learning situations. For instance, they should remind learners not to use machine translation in the initial stages when their language foundations are not solid; require learners to assess whether the machine-generated foreign language text exceeds their current application level and modify it according to their foreign language abilities; in the absence of official guidelines, clarify to learners what situations constitute academic misconduct or plagiarism, emphasize the dangers of excessive reliance on machine translation, and establish corresponding penalties to prevent abuse; and guide learners to compare the effects of machine translation and human translation, allowing them to experience the “strengths” and “weaknesses” of machine translation and understand other useful functions, so that students can fully experience the “capabilities” and “limitations” of machine translation technology in practice.
Second, teachers need to comprehensively grasp learners’ learning motivations and transform their role from mere transmitters of language knowledge to guides in language learning. Teachers should center their teaching around learners, timely inspire them to find points of interest in the subject, enabling them to learn autonomously driven by interest and motivation, thereby enhancing their independent thinking abilities. At the same time, learners should experience the joy brought by the language learning process, which can reduce their dependence on machine translation. What learners obtain from machine translation is merely the conversion result of two languages, and they cannot learn the cultural connotations and humanistic backgrounds behind the language. The focus of foreign language education should be on acquiring cultural connotations, understanding the core of thinking, and developing the ability to express thoughts and communicate in a foreign language. Furthermore, education should not be narrowly understood as training and acquiring skills for a profession; humanistic education is a value orientation we should always adhere to. In the age of artificial intelligence, it is even more crucial to emphasize the cultivation of “human” qualities. Moreover, with the emergence of generative AI products, the demands on learners should also prioritize thinking over knowledge, questions over answers, and logic over enumeration. Foreign language educators should focus on cultivating learners’ problem-solving abilities, communication and collaboration skills, self-management abilities, adaptive learning capabilities, argumentative writing skills, knowledge innovation abilities, and critical thinking skills.
Machine translation and artificial intelligence technologies represented by ChatGPT bring more possibilities to foreign language education. Many foreign language educators are considering using them as auxiliary tools to innovate teaching methods and means. For example, Ritsumeikan University in Japan has recently introduced the English learning tool “Transable,” which combines ChatGPT and machine translation, attempting to use AI technology to provide new English education services and improve educational outcomes. Similarly, Khan Academy is using the online educational tool “Khanmigo” developed with GPT-4 as a virtual tutor for students, providing highly personalized learning options. Educators can utilize AI to assist in lesson preparation, automatically generate various exercises, evaluate and grade student assignments, diagnose issues learners encounter during the learning process, provide timely feedback on students’ questions, and offer personalized teaching guidance, among other tasks. Additionally, technologies such as the metaverse, robotics, and VR/AR provide more diverse solutions for language learning. In this technological development environment, machines are not just tools and aids but more like collaborators, leading educators to initiate a teaching model of “human-machine cooperation.” In this era of explosive technological development, we need to reconsider various aspects and factors related to education, such as technology, cultivation goals, teaching content, textbooks, teaching models, learning methods, teaching evaluations, resource construction, teacher roles, and teaching management, advancing systematic reforms and digital transformations in foreign language education to shape a new foreign language education system in the digital artificial intelligence era.
(2) The need for the role of “language guardians” in the technological tendency of language life
As mentioned earlier, the impact of the widespread application of machine translation technology on language life is already becoming apparent. Whether it is the “controlled language” required in machine translation, the “formulaic” post-edited works, or the texts produced for the convenience of cross-cultural communication that merge the characteristics of original and translated languages, they all highlight a common issue: texts are increasingly becoming simplified and flattened, with their complexity and richness diminishing. This indicates that language is being degraded and gradually losing its original polysemy and richness, increasingly becoming a technical construct. The technicalization of language equates to the degradation of language. These phenomena are undoubtedly detrimental to the protection and promotion of language development. To avoid the dilution of natural language by machines as much as possible, translators, who are closest to the original and translated languages, should take on the role of “language guardians.” Translators need to demonstrate their importance, constantly exercise their “thoughtfulness” and “creativity,” and critically engage with machine-generated translations to ensure that translated texts remain vibrant, warm, and layered. Additionally, these technological tendencies in language phenomena should draw the attention of linguists, prompting them to consider how to collaborate with technology developers to better protect language from being eroded by machines.
In the development history of machine translation, the rationalist phase based on rules played an important theoretical supporting role in linguistic knowledge. The subsequent phases of statistical machine translation, neural machine translation, and the current generative pre-trained model series of GPT have all adopted an empirical approach based on large language data. In the empirical phase, linguistic knowledge has been overlooked, and linguists have distanced themselves from the development of machine translation. Feng Zhiwei argues that focusing solely on large language data while neglecting linguistic knowledge makes it difficult to explain many issues from a rationalist perspective based on language rules, leading to potential developmental “bottlenecks” in the future. Li Yuming also mentioned that the current development of artificial intelligence neglects the intersection with linguistics, which is a manifestation of weak linguistic awareness. Research has shown that linguistic knowledge plays a crucial role in performance evaluation of machine translation, post-editing, and addressing linguistic ambiguity and polysemy. This knowledge could be used to enhance linguistic checks before data training results are output, serving as a basis for researchers to improve algorithms or models. The various issues faced by data-driven language intelligence are precisely because insufficient attention has been paid to the supporting role of linguistic theory. Furthermore, the development of machine translation also requires linguists to provide valuable suggestions on language structure, discourse analysis, language functions, and pragmatic rules.
(3) The need to highlight the “humanistic nature” of technology in the age of human-machine interaction, maintaining the competitive advantages of human translation capabilities
As previously mentioned, translators hold varying attitudes towards machine translation, with some even showing resistance. However, based on current trends, humans will eventually coexist with artificial intelligence. It is undeniable that machine translation has liberated some low-end labor, while translation memory, deep learning technologies, and others have enriched the connotations of translation professions and language service industries. Information technology has altered the traditional landscape of the translation industry, profoundly affecting translation work. The human-machine collaborative translation model of “Machine Translation + Post-Editing” has become mainstream, but the popularization of machine translation may weaken the subjectivity of translators, posing risks of technological alienation. In response to this issue, the translation industry has already taken action. The Common Sense Advisory (CSA), a U.S. language industry research and consulting organization, released the Augmented Translation model in 2017 based on extensive research, predicting that the future development trend of translation production models is to elevate the status of translators, make various technologies smarter and more coordinated, centering everything around translators, freeing them from time-consuming low-value tasks, and allowing them to focus on more challenging and innovative work. In the future, emphasizing human-machine interaction and protecting translator subjectivity should be an important direction for the research and development of machine translation systems.

Challenges And Reflections On Machine Translation In The Digital Age

In the translation model involving machine translation, translators can easily “lose themselves” in technology, making it difficult to stimulate their creative inspiration. Over time, their language literacy and aesthetic sense may degenerate. Therefore, translators must never forget their primary role as “translators” and should value their worth as “humans,” striving to ensure a certain amount of translation output without relying on machine translation, thus maintaining the advantages of human translation: the ability to think independently and sensitivity and creativity towards language.
Challenges And Reflections On Machine Translation In The Digital Age

4. Conclusion

Andrew Feenberg once said: “Every major technological change will have impacts on multiple levels, including economic, political, religious, and cultural.” The advancement of the digital age is unstoppable, and the rapid development of machine translation technology in this era inevitably affects various aspects of society and humanity. Foreign language education, public language life, and translators are just one facet of this, reflecting many common issues. Critically reflecting on technology does not mean denying it but rather aims to understand and recognize the essence of technology more objectively. In the future, as artificial intelligence technology continues to develop, the relationship between humans and technology will no longer be a binary opposition of subject and object; technology will not merely be a tool for humans, but rather both should exist in a symbiotic relationship of interactive cooperation. Building a good relationship requires human-machine collaboration, leveraging each other’s strengths. The problems brought by technology can be seen as contradictions and frictions between humans and technology, and we need to recognize these contradictions in advance, reduce friction, and leverage our respective advantages to promote healthy interactions between the two, thereby better coexisting in harmony.

(Note: The references and annotations in the article published in the public account are omitted; see the paper edition for details.)

Challenges And Reflections On Machine Translation In The Digital Age

The article is published in the Journal of Yunnan Normal University

(Philosophy and Social Sciences Edition)

2024th Issue

Challenges And Reflections On Machine Translation In The Digital Age

Initial Review｜He Zhili

Re-review｜Qiao Xiaoming

Final Review｜Xiong Liran

Previous

Issues

Return

Review

Zhao Libo Wang Chunhui｜Language Gig Workers in the Digital AgeHuang Lihe Che Yiran｜Reflections on the Development of Linguistics for the Elderly Empowered by Artificial IntelligenceTable of Contents and Abstract of the Journal of Yunnan Normal University (Philosophy and Social Sciences Edition) 2024, Issue 5

Challenges And Reflections On Machine Translation In The Digital Age

Leave a Comment