Published: May 23, 2025
Where Published: AKJournals: Across Languages and Cultures
Summary
This paper addresses the translation industry’s current crisis, where rapid technological advances in AI and machine translation have created widespread confusion about quality standards, while simultaneously driving down fees and pushing professionals out of the field. The paper argues that the solution lies not in debating specific technologies but in establishing clear, explicit specifications for the translation projects and systematic evaluation methods – essentially applying well-established quality management principles from other industries to translation. The paper introduces two new ISO standards that can help in this endeavor: ISO 11669:2024 for project specifications, and ISO 5060:2024 for translation quality evaluation, providing practical frameworks for clarifying requirements upfront, selecting appropriate workflows based on actual needs rather than assumptions, and measuring whether outputs meet the specified criteria. The paper also discusses the evolution of other standards, particularly the MQM (Multidimensional Quality Metrics) framework that has become widely adopted in the industry, with ASTM’s working draft WK46396 currently close to ballot as a requirements standard for analytic translation quality evaluation based entirely on MQM. In short, successful translation projects depend on making requirements explicit through structured specifications that address content, process, and risk parameters, then validating outputs against these specifications through consistent evaluation. In this way stakeholders can navigate the technological disruption by focusing on measurable quality outcomes rather than getting lost in debates about methods or falling for marketing hype about AI capabilities.
Key Takeaways
- The translation sector is undergoing rapid transformation due to technologies like neural machine translation (NMT), transformers, and large language models (e.g., ChatGPT). This has led to claims that technological advances have essentially “solved” translation, but this ignores important nuances, especially regarding use cases and language combinations. Despite automation, there is a need for “the expert in the loop” or “human at the core” rather than eliminating professionals altogether. Augmentation, not just automation, is the path forward.
- Precise specifications detailing requirements, use case, intended audience, acceptable error/risk levels, and production methods are critical for the success of translation projects. The lack of clear specifications leads to inefficiencies and disputes.
- The new ISO 11669:2024 standard emphasizes that both clients (requesters) and language service providers (LSPs)/translators share responsibility for defining needs, conducting risk analysis, and setting project parameters.
- The new ISO 11669:2024 standard emphasizes that both clients (requesters) and language service providers (LSPs)/translators share responsibility for defining needs, conducting risk analysis, and setting project parameters. Evaluation of translation output based on explicit specifications is essential for objectively determining if requirements have been met. Risk management—understanding risk and error tolerance levels—is a key part of modern translation projects.
- The path to sustainable translation quality in the face of technological disruption is through structured, explicit project specifications and regular, standards-based evaluation, maintaining a balance among all perspectives on quality.