Working Memory (WM) is a central structure maintaining information at short term in face of temporal decay and interference for its processing in ongoing tasks. As a consequence, WM is strongly involved in learning, especially in learning first or second languages. The Time-Based Resource Sharing (TBRS) model describes the functioning and development of WM, in particular by integrating the role of executive attention and the time constraints that affect cognitive processes. After a brief overview of the model, this chapter focuses on the distinction between a domain-specific system of maintenance in charge of verbal information and a domain-general system relying on attention. Hence, we show how verbal information is maintained in WM from childhood to adulthood, and how linguistic features of verbal information impact its short-term memory. Conversely, we explain how WM mechanisms in the TBRS model affect the creation of true and false verbal long-term memory traces.

The phonological component of the working memory system is specialized in maintaining a sequence of verbal items (digits, letters, words, pseudowords) over a very short period of time. Therefore, a central issue has been why we are provided with such ability, and what is its functional role. A series of studies on healthy people, on children learning their mother tongue, on children and young adults learning a second language and, crucially, on neuropsychological patients with a selective deficit of auditory-verbal short-term memory has clearly shown that a fundamental function is to maintain a new phonological representation for a period of time long enough to build permanent phonological representations. This is exactly what happens when we learn a new language. In this chapter I will report converging evidence involving different languages showing how this important result has been obtained.

The multicomponent model of working memory developed during the period when psycholinguistics was dominated by Chomsky’s transformational grammar and its potential implications. The original model had assumed a limited capacity attentional control system, the central executive, aided by temporary verbal storage from the phonological loop and visuospatial storage from the visuospatial sketchpad. Over the decades, each component of the model has been systematically explored by language studies, which have repeatedly resulted in challenges to earlier versions of the model and led to the addition of the fourth component of the episodic buffer and the recent incorporation of the concept of binding. Overall, the multicomponent model was developed using a different approach than Popper’s emphasis on falsification, and the model continues to evolve and has proven successful both in accounting for a broad range of data and in its application to the understanding of a wide array of language phenomena and populations. L11

The capacity for temporary storage and manipulation of information, i.e., working memory (WM), was first reported to be related to vocabulary acquisition over 30 years ago (Daneman & Green, 1986, for general WM capacity; Gathercole & Baddeley, 1989 and Service, 1989, for phonological WM). Although a relationship with L2 vocabulary knowledge has been highlighted repeatedly among different populations, its strength seems to be a function of the aspect of WM (central executive, phonological short-term storage) and vocabulary knowledge (breadth, depth) investigated, and relatedly, the methodological choices made (e.g., measurement instruments, populations – children, adults). In our chapter, we aim to untangle these intersecting effects through a methodological lens. In order to do so, we discuss some published and unpublished results, comparing them according to their methodologies. We finish by proposing new perspectives on the interpretation of some of the commonly used tasks in WM and L2 vocabulary studies.