ترجمۀ كتاب اخلاق در ترجمه:ملاحظات اخلاقي و حقوقي ترجمۀ ماشيني

Towards Responsible Machine Translation: Ethical an‎d Legal Considerations in Machine Translation

زبان و ادبيات انگليسي

Towards Responsible Machine Translation: Ethical an‎d Legal Considerations in Machine Translation is a multifaceted, interdisciplinary work that examines the diverse dimensions of responsibility in machine translation from technical, ethical, social, environmental, an‎d legal perspectives. The book collects research studies, viewpoints, an‎d case studies that each, in their own way, attempt to strike a balance between technological advancement an‎d the observance of professional ethical principles in machine translation. The editors’ main aim is to show that machine translation is no longer merely a technical tool but a social an‎d cultural phenomenon that directly affects people, data, an‎d communities. The opening chapters probe the concept of “responsibility in machine translation” from historical an‎d theoretical stan‎dpoints. Tracing the evolution of language models from statistical systems to neural networks, the authors draw attention to issues such as algorithmic transparency, data bias, an‎d users’ overreliance on automated outputs. This section demonstrates that technological progress—if pursued without ethical an‎d social oversight—can foster distrust an‎d even reproduce discrimination in translation workflows. Accordingly, the authors stress the necessity of retaining human judgment an‎d oversight at critical points in the translation process. Middle sections of the book address topics such as licensing an‎d data ownership, the role of human post-editors, an‎d the end-user perspective. The contributors argue that clarifying data provenance an‎d safeguarding content owners’ rights are preconditions for responsible system development. They also discuss the professional an‎d ethical challenges faced by human post-editors, including diminished autonomy, cognitive fatigue, an‎d time pressure associated with editing machine-generated translations. From the end-user perspective, the book shows that users can trust MT outputs only if they are sufficiently informed about how the outputs were produced an‎d about the technology’s limitations. Subsequent chapters explore more specific an‎d high-stakes concerns, including ethics in crisis situations, where the authors examine the use of MT in humanitarian emergencies, armed conflict, an‎d natural disasters. In such contexts, even minor system errors can have severe consequences for people’s safety; therefore, MT deployment should be accompanied by human mediation an‎d multi-layered validation. Another chapter focuses on gender bias, demonstrating how imbalanced training data an‎d algorithmic design choices can reproduce gender stereotypes. The authors recommend design an‎d training strategies that prioritize equity an‎d balanced representation of gender groups. A notable contribution of the book is the chapter on the ecological footprint of large MT models, which analyzes the environmental costs of training an‎d deploying large-scale neural systems. Drawing on empirical data, the authors show that large model training consumes substantial energy an‎d natural resources. They advise researchers an‎d industry to pursue more energy-efficient architectures, model compression techniques, an‎d the use of low-carbon infrastructures. This section frames environmental responsibility as an integral component of professional ethics in machine translation research an‎d practice. The final chapter, Speech as Personally Identifiable Information (PII), highlights privacy an‎d security concerns related to voice data. Given the increasing deployment of speech-enabled translation technologies, new ethical an‎d legal questions arise about storing an‎d analyzing users’ voices. The authors call for explicit policies to protect personal speech data an‎d to respect users’ right to anonymity.

11

150607