This thesis uses a contextual case study approach covering the period 1985 to 2011 to examine the construction of Trinidad and Tobago's ICT4D policy as discourse. The guiding theory of method is contextualism as described in Pettigrew (1990), according to whom a contextual analysis can be characterised as: processual, by emphasising the evolution of actions embedded in specific contexts (structural and otherwise) over time; multi-stakeholder, by recognising the competing viewpoints of reality perceived by actors at different levels; longitudinal, by considering both historical and contemporary views of actions and events. Consistent with this approach a framework has been adopted here that views policy as an iterative process involving the generation of texts from events, the translation of texts into narratives, and competition between alternative narratives resulting in institutions, which, in turn, enable and constrain events. This framework facilitates understanding interactions between actors at multiple levels across time. There are three original contributions to knowledge made in this thesis. Firstly, I have proposed an analytical framework that integrates three separate bodies of literature. The discursive model of institutionalization of Phillips, Lawrence and Hardy (2004) and the ‘policy as discourse' approach of Shaw & Greenhalgh (2008) and Shaw (2010) are integrated by way of a bridge, the ‘trading zones' concept of Galison (1997) as extended by Collins, Evans, & Gorman (2007). Thereby, I developed a series of analytical constructs that can be used for contextual policy research, especially in developing countries where dominant policy narratives constrain and moderate discursive exchange when those policy narratives - which were originally articulated in advanced economies - are subsequently transferred into developing countries. Secondly, I have empirically applied the framework to the study of ICT4D policy construction in Trinidad and Tobago, generating new insights in the process. In so doing I critically examined the process of constructing policy as discourse with the aim of identifying ways in which policy could be done differently. A key finding is that the process of discursively transferring previously existing policy narratives into new contexts can result in one of three outcomes: no change - if the introduction of policy narratives had no impact whatsoever on institutions (either by creating new ones, or disrupting existing ones); the construction of policy pidgins (semi-specific yet incomplete proto-languages that mediate discursive transfer) - when discursive transfer, imitation and assemblage of narratives partially occurs; or the construction of policy creoles (full-fledged languages that facilitate not only discursive transfer, but social action) - if the discursive transfer is complemented by translation, editing and social embedding. Thirdly, I developed a model of policy creolization through which the two main factors that influence the emergence of policy pidgins and, eventually, policy creoles (both viewed as particular forms of institutions) in a setting of discursive construction were identified, namely:  The length of the temporal window over which policy actors have an opportunity to develop interactional expertise to transfer, imitate and assemble narratives, and eventually to translate, edit and embed those narratives into social actions;  The degree of intentionality of the discursive action, and subsequently the social action, that policy actors engage in, noting that there are three categories of social action: o Intentional action – which deliberately conveys particular ideas through texts. o Consequential action – which is generated as a by-product of ongoing dialogue among actors during which they may draw on broader narratives. o Emergent action – which arises through discursive contestation and struggle in ways that were not necessarily intended or predicted. This highlights that both intentionality and time are required to bridge the knowledge gaps present between the different contexts, and even so, that the policy construction process in the new context requires practitioners to develop non-trivial levels of interactional expertise. This thesis has implications for policy practice on two fronts. Firstly, the framework can be employed to assist policymakers in creating policy creoles through coordination and interaction between external mainstream narratives and alternative narratives, including those that are locally derived. In doing so, policymakers and policy analysts can unpack the conceptual constructions of their subject domain, learn how to engage with new domains (and thereby gain interactional expertise) and uncover the latent power dynamics that are reinforced by lack of critical analysis. Secondly, application of the framework provides a means of assessing institutional dynamics. This is important because of the powerful normative, cognitive and regulative functions institutions play on the development of new institutions, and ultimately on social action.