Now I'm wondering whether the Sapir-Whorf hypothesis applies to NLCs: Does the natural language that the LLM is trained on affect its cognition?