Text processing through LLM for gender-inclusive annotation and translation: A case study on informed consent documents

This study explores the ability of ChatGPT-4o to detect and classify gender-inclusive and exclusive language in Spanish informed consent documents for translation purposes. Combining a semi-automated annotation and a comparative translation study, this contribution assesses the impact of zero-shot and few-shot prompting strategies on the model’s outputs. Results reveal that while ChatGPT-4o can support large-scale annotation, it misses many inclusive and non-inclusive markers without explicit instruction. Regarding the use of ChatGPT-4o as a machine translation engine, prompt engineering significantly increases inclusive output, especially through dual forms and collective terms, yet masculine defaults persist, namely in zero-shot prompted translation. The findings highlight both the promise and limitations of large language models in gender-sensitive translation, emphasising the importance of human revision and tailored prompting.