Some audiovisual translation (AVT) scholars redefine the source text (ST) to encompass all modes in the audiovisual text, and not just the verbal discourse, as in the still prevalent verbally driven ST notion. Focusing on subtitling, this conceptual paper explores the implications of a holistic redefinition of ST for AVT and TS at large. Two scholarly texts are analyzed, both proposing holistic ST concepts but approaching text and translation from different initial points of view—the one starting from whole texts in context and the other from parts of text. The paper examines the scholars’ argumentation regarding their holistic concepts and how it reflects their initial point of view, discussing its significance. The paper suggests, first, that both holistic concepts offer a theoretically more consistent approach to subtitling, compared to the verbally driven one, elucidating, for example, discussions on templates and machine translation in subtitling. Second, the paper argues that AVT provides a strong epistemological argument for a consistent initial whole-to-parts point of view for the study of translation as mediation. It calls for research within the multimodal framework to shift the primary focus from (ST) modes and meaning to the purpose and functions of whole texts in context.