Voice is not a new concept in writing; however, it is relatively new in the field of academic writing. The main aim of this research is to determine how voice as a social construct is understood and perceived by doctoral students and supervisors from the faculties of Humanities and Social Sciences at a South African university. The focus is on the challenges of exhibiting an authorial voice in doctoral writing in particular, with the aim of informing a pedagogical framework of voice that might serve as a foundation for further development of an instructional framework. The term ‘voice’ started to appear in North American composition writing in the mid-1960s as a mark of self-discovery, individualism, and expressivism. However, the emergence of social constructivism led to a marked decrease in the emphasis on individual voice in favour of regarding voice as socialised and constructed. The post-2000 voice era became more nuanced and established a definite niche for voice in academic discourse. The three approaches that influenced written voice most significantly are individualised voice, powered by the expressivist approach; socialised voice, which embraces voice as multi-dimensional and dialogic and embedded in Bakhtin’s heteroglossia; and voice as empowerment, represented by the Academic Literacies Approach. Except for its historic evolution the notion of voice was impacted by Systemic Functional Linguistics (SFL) as a theory of language. Two partially operationalised models, grounded in social constructivism and SFL, provided the substance for designing a heuristic framework for voice: the Engagement Framework, situated in the Appraisal Framework of Martin and White (2005) and Hyland’s (2008a) model of stance and engagement. With the decline of the expressivist approach a number of theoretical and empirical studies propagating a pedagogical approach started to appear. Although these studies validate the need for a visible voice pedagogy, voice has yet to be operationalised as student friendly pedagogical tool. The following research questions guided the research: 1. How is authorial voice theorised in linguistics and applied linguistics? 2. Has the notion of ‘voice’ been adequately operationalised in academic writing contexts? 3. What guidance on developing a voice pedagogy is found in the scholarly literature on writing instruction in higher education? 4. How is the notion of voice understood by supervisors and doctoral students? A qualitative case study was conducted to determine the understanding and perceptions of voice by supervisors and doctoral students by means of semi-structured interviews. The data were systematically analysed and coded using qualitative content analysis. The qualitative data analysis software program ATLAS.ti.2 was used for this purpose. The data yielded four main categories: 1. Assumptions about voice as non-negotiable in doctoral writing; 2. Enablers of voice; 3. Impediments of voice, confirming voice as complex and unstable; 4. Opinions on voice as construct that substantiated gaps in the literature. As the findings point to a need for a pedagogy of voice these categories were translated into parameters for a pedagogy of negotiated voice. The pedagogical model integrates the theory-based heuristic as well as pedagogical attempts at measuring voice and the findings of the empirical study.