This study introduces the ecological Lotka-Volterra model into the analysis of language competition and develops a random dynamic framework to capture interactions between dominant and non-dominant languages. Through a system of coupled differential equations, it examines how regional language ecosystems evolve under the global diffusion of a shared communicative language. The results reveal some key dynamic mechanisms. In the natural evolution of a bilingual society, the system tends to form self-organized equilibrium states, marked by periodic coexistence. When social fluctuations intensify systemic phase transitions, the actual usage level of the dominant language may decline unexpectedly. This decline accelerates the process of language assimilation. In addition, chaotic oscillations may emerge if there are mismatches between language promotion policies and the system’s intrinsic dynamics. This leads to nonlinear deviations between policy targets and actual dominant language coverage. Based on these findings, the study suggests that effective coordination between language dissemination and diversity protection requires a dynamic adaptive governance. Such a strategy should involve three key actions. First, define critical disturbance thresholds to prevent abrupt declines in dominant language vitality; Second, optimize the timing and frequency of policy interventions to stabilize language coverage; Third, utilize the cultural resilience of non-dominant languages to strengthen intergenerational transmission within specific social domains. Overall, this framework provides a cross‑disciplinary reference for analyzing and governing multilingual ecosystems.