Having outlined the integrated framework of WM–SLA and the phonological/executive (P/E) model, in this chapter I discuss how the basic tenets and general principles of this integrated perspective may help to inform and guide future WM–SLA explorations. More specifically, I discuss the possible effects of the two WM components and their associated mechanisms (PWM and EWM), independently or in combination with the various L2 task characteristics and conditions, on various dimensions of L2 speech production and performance. As such, in the first section I briefly review the key theoretical and methodological issues in current research on L2 task-based planning and performance. This gives rise to two (competing) theoretical hypotheses on the cognitive mechanisms underlying L2 task performance. In the second section I reconceptualize L2 task planning and speech performance from a WM perspective by incorporating and aligning the two WM components (PWM and EWM) and their putative mechanisms with the specific measures indexing the four dimensions of L2 speech performance (complexity, accuracy, fluency and lexis; i.e. the CAFL framework; Skehan, 2009). The third section outlines an integrated agenda for future research on the relationship between WM, tasks features/conditions and L2 speech performance. A summary section then concludes the chapter.

Key Issues and Debates in Current L2 Task Performance Research

An overview of the key issues in L2 task planning research

In terms of the L2 task-based speech planning research in SLA, Rod Ellis (1987) is generally believed to have first introduced the concept of ‘planning’ into applied linguistics. Ellis's paper subsequently led to a number of studies that manipulated the planning construct and explored its differential effects on L2 performance (Skehan & Foster, 2012). These studies culminated in a volume edited by Ellis (2005) and a special issue of the journal Applied Linguistics (2009). Several edited volumes of empirical studies following these early research traditions have also been published in recent years (e.g. Housen et al., 2012; Robinson, 2011; Skehan, 2014; cf. Bygate, 2015).