The intended conversational reference by speakers to entities or linguistic stretches of which knowledge is culturally shared among discourse participants to add an implied simile to entities within their conversation context is a challenge to on-screen film translation, as a multimodal content. Those units of linguistic input are termed as Extra-linguistic Culture-bound References (ECRs). This topic is interdisciplinary involving analysis tools from pragmatics mainly Relevance Theory, which presents a general framework for the description of the communication process, in addition to subjects such as translation strategies classification as well as terminologies from audio-visual cinematographic content annotation conventions. The main purpose for this research is to investigate the application of RT in describing the nature of multimodal (verbal and non-verbal) contextual effects by: a) understanding how the allusive items ECRs are interpreted by the native viewers of films; and b) how the same context can help the subtitlers in approximating the meaning of ECRs to non-native users of subtitles. The corpus for analysis applying RT method constitutes of eight extracts from four screenplays: “Crash" (2004); “Fences" (2016); “Collateral Beauty" (2016); “The Theory of everything" (2014). The terminology for describing the non-verbal context of the screenplay is derived from film making terminologies. Observations therefrom were used to understand the effect of multimodal context in understanding the ECR function and its implications on the translation choice.