The effect of community knowledge elements on the effectiveness of a culturally-aware large language model in preserving intangible cultural heritage: Evidence from Turkana traditional medicine knowledge

Large Language Models (LLMs) increasingly encode and generate knowledge at scale, yet consistently underrepresent indigenous and community knowledge systems, a gap with acute consequences for critically endangered Intangible Cultural Heritage (ICH) such as Turkana Traditional Medicine Knowledge (TMK) in Kenya. Cultural grounding, Responsible AI, and Community Knowledge Elements have been studied in isolation, never combined and empirically justified within one architecture, leaving designers of culturally aware LLM systems without evidence for how Community Knowledge Elements specifically should be weighted. This paper reports an empirical evaluation of the effect of Community Knowledge Elements (CKEs), operationalised as Traditional Medicine Knowledge Elements (TMKEs), on the effectiveness of a culturally-aware LLM in preserving ICH. A sequential mixed-methods design combined a community case study with Design Science Research, drawing on 107 respondents from the Turkana community and domain experts, and 400 simulation outputs generated across two controlled LLM configurations. TMKE Integration Quality correlated strongly with LLM Effectiveness (r = .524, p < .001) and significantly predicted it in regression (R² = .290, p < .001), an effect driven specifically by Knowledge Integration Effectiveness (β = .389, p = .020). Under controlled simulation, TMKE integration produced significant gains in content overlap with the correct reference on BLEU-4 and ROUGE-L, at a significant cost to output fluency, and no significant improvement in literal Turkana-term retention. The findings establish that Community Knowledge Element integration produces a real, statistically supported, but partial and uneven effect on LLM effectiveness, with direct implications for how knowledge-grounded LLM architectures for indigenous heritage preservation should be designed and evaluated.