Working memory (WM) generally refers to the human ability to temporarily maintain and manipulate a limited amount of information in immediate consciousness when carrying out complex cognitive tasks such as problem-solving and language comprehension. Though much controversy has surrounded the WM concept since its inception by Baddeley & Hitch (1974), an increasing number of cognitive psychologists have accepted WM as a multi-component system comprising both domain-specific storage mechanisms and domain-general executive functions (Miyake & Shah 1999; Baddeley Reference Baddeley2012; Williams 2012). Such a fractionated view of this cognitive construct manifests itself clearly in distinct strands of WM-language research, where two contrasting research paradigms have emerged (Wen Reference Wen2012).

WM researchers following the European tradition have sought to establish the critical role played by the phonological component of WM (henceforth PWM, which embraces a passive sound-based store and an active sub-vocal rehearsal process) in vocabulary acquisition and grammar development (e.g., Gathercole & Baddeley 1993; Gathercole, Baddeley & Papagno 1998; Baddeley 2003). To measure WM, these researchers usually administer a simple storage-only memory span task such as the digit span, word span or nonword repetition span task, and adopt participants’ total recall score as their WM capacity (Gathercole et al. 1994; Gathercole 2006). In contrast, many cognitive psychologists based in North America have tended to emphasize the executive functions associated with the WM concept (henceforth EWM) and focus on teasing out the implications of its attention-regulating mechanisms (e.g., information updating, task-switching, and inhibitory control; Miyake & Friedman 2012) for language learning and processing. The WM measure they choose is usually a complex memory span task that uses the storage and processing functions of WM, such as the reading span task designed by Daneman & Carpenter (1980). Extensive research following this paradigm has also pointed to significant effects of WM on selective facets of language activities such as comprehension processes, syntactic processing, and speech production (Miyake & Friedman 1998). Taken together, both research camps of WM have accumulated increased evidence for a close link between WM and L1 learning (Gathercole & Baddeley 1993; Baddeley 2003; Cowan 2011).

A growing number of second language acquisition (SLA) researchers are adopting these well-defined research paradigms of WM–language in cognitive psychology and increasingly subscribe to the view that WM should play an equal if not more important role in the learning and processing of a foreign or second language (L2). The rationale behind this assumption came mainly from the perceived fundamental difference between first language acquisition (L1A) and L2 learning. Viewed from the information-processing perspective, L1A unequivocally occurs early in life and proceeds automatically, as reflected in children picking up their mother tongue relatively easily. L2 learning, in contrast, usually comes at a later stage in life and relies on more controlled language processing and thus presumably places much greater demands on such cognitive resources as WM and/or attention (Harrington 1992; McLaughlin 1995).

An alternative approach to linking WM functions with the process of SLA can be deduced from the basic tenet of the connectionist account of language acquisition. Interpreted from this perspective, a central task of language acquisition (be it L1A or SLA) is to acquire, store, and use a functioning repertoire of all sorts of linguistic sequences ranging from lexis or multi-word formulae to morpho-syntactic constructions (Ellis 1996, 2012). In the process of chunking these linguistic sequences, WM, and particularly its PWM component, serves to sustain and rehearse (albeit sub-vocally) their phonological material in immediate consciousness while long-term representations are being constructed.

Such intriguing portrayals of the WM–SLA association have motivated an increasing number of empirical studies exploring the potential effects of WM in various aspects of SLA (Juffs & Harrington 2011; Wen forthcoming). Nevertheless, due to the elusive nature of the WM construct and the daunting array of assessment procedures in cognitive psychology, many of these WM–SLA studies are subject to limitations and caveats in terms of research design and methodology, thus posing great challenges to many SLA researchers (Juffs 2006; Gass & Lee 2011).

To resolve these theoretical and methodological issues surrounding the WM construct and its assessment, the SLA field requires a more principled way of conceptualizing and operationalizing the construct, one that is conducive to future research. In pursuit of that goal, I have proposed a conceptual framework for WM that integrates the general consensus around WM theories in cognitive psychology into existing SLA research paradigms (Wen Reference Wen2012, forthcoming). The present timeline aims to document and highlight selected seminal research from both fields that has contributed to the formulation of this integrated perspective for theorizing and measuring the WM construct in L1 and L2 research. These strands of theoretical and empirical research are organized and coded into the following four categories, which constitute the broad themes underlying this timeline:

1. A Conception of WM in cognitive psychology

   1. A1 Theorizing or conceptualizing PWM

   2. A2 Theorizing or conceptualizing EWM

2. B Assessment of WM in cognitive psychology

   1. B1 Designing and constructing assessment procedures for testing PWM

   2. B2 Designing and constructing assessment procedures for testing EWM

3. C Theorizing and measuring WM as it relates to L1

   1. C1 Theorizing and measuring PWM as it relates to L1

   2. C2 Theorizing and measuring EWM as it relates to L1

4. D Theorizing and measuring WM as it relates to L2

   1. D1 Theorizing and measuring PWM as it relates to L2

   2. D2 Theorizing and measuring EWM as it relates to L2

More importantly, when these lines of WM–language research from both cognitive psychology and SLA are put into perspective, they seem not just to converge on the fractionated nature of WM but to further corroborate the separate and distinctive roles of PWM and EWM in specific language-learning activities (Wen forthcoming). Emerging patterns of this PWM–EWM dichotomy seem to suggest that the PWM component, with its associated cognitive mechanisms, is most closely related to the acquisition and developmental aspects of language-learning domains such as vocabulary and formula acquisition and grammar development. In contrast, the EWM component, with its embedded executive functions, is particularly relevant to the processing and real-time performance aspects of language-learning activities, such as certain demanding cognitive processes involved in language comprehension and production (Kail & Hall Reference Kail and Hall2001; Engel de Abreu & Gathercole Reference Engel de Abreu and Gathercole2012; Szmalec, Brysbaert & Duyck 2013). This proposition and other associated hypotheses regarding the different effects of PWM and EWM, as they relate to specific L1A and L2 learning domains and activities, are encapsulated in the Phonological/Executive Hypothesis (the P/E model; Wen forthcoming). To incorporate this demarcation between PWM and EWM into this research timeline, each of its four broad themes is further categorized into two subordinate themes.

To sum up, once PWM and EWM, the two key components of WM for language, are pinned down and their associated functions/mechanisms aligned with specific L1A and L2 learning domains, novel, specific, and testable hypotheses regarding their intricate relationships can be formulated in a more effective manner. In light of these proposals for the WM–language nexus, future research can set out to either support or refute hypotheses outlined in the P/E model and to further tease apart all their possible theoretical and pedagogical ramifications.