The last few years have seen a dramatic shift in translation technologies, with neural machine translation (NMT) and large language models (LLMs) increasingly promoted as removing the need for human translation, despite evidence of public concerns around artificial intelligence (AI).
Such tools serve a purpose in certain contexts, but used without expert human input, they carry risks of harm and can produce poor-quality results, particularly in terms of consistency of terminology, cultural nuance and real-world knowledge. Separately, the environmental costs of AI are a cause for concern.
While no tool is a substitute for the experience, subject-matter expertise, cultural understanding and linguistic skills of a human translator, computer-assisted translation (CAT) tools combine translation memory, terminology management and quality assurance functions to help produce high-quality, consistent translations and improve productivity, without compromising confidentiality. They are particularly valuable for:
- producing updated versions of existing documents
- ensuring consistent use of terminology over time and across related documents
- reducing turnaround time for very repetitive texts
- providing an extra level of quality assurance.
I use a range of industry-leading tools, including SDL Trados Studio and memoq, which handle a wide variety of file types, including industry-standard translation memory (TM) and terminology database (TB) formats, I also offer machine translation post-editing (MTPE) services for clients who request them.
