arXiv:2605.30443v1 Announce Type: new Abstract: Multilingual large language models can generate figurative language, but whether the internal signals driving this behavior are language-specific or reusable across languages is unclear. Using activation steering as a probe, we estimate a direction for a figurative category from figurative--literal activation differences in one language and apply it during generation. Across five figurative categories, six languages, and four multilingual LLMs, these directions steer reliably within their own language, most robustly for metaphor and simile. More
This paper leverages recent advancements in multilingual large language models and activation steering to explore the internal mechanisms of figurative language generation.
Understanding how LLMs generate figurative language across languages could lead to more nuanced, culturally aware AI outputs and advanced cross-lingual communication tools.
The ability to 'steer' figurative language generation in LLMs suggests a new level of control and insight into their internal workings, improving their creative and communicative capabilities.
- · AI researchers
- · Multilingual NLP developers
- · Content creation platforms
- · Language learning applications
- · Monolingual AI development approaches
Improved control over stylistic elements in multi-language AI generated text will become possible.
More effective and culturally relevant AI tools for global communication and content localization will emerge.
The development of AI systems capable of synthesizing nuanced cultural expressions could lead to new forms of digital artistry and entertainment.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL