Data-driven learning: a call for a broader research gaze

Abstract

Over the last three decades, data-driven learning (DDL) has been widely championed by those of us who see the exciting opportunities that it can bring to the language learner. From the initial days of DDL, there has been a sense of enthusiasm about turning language learners into researchers who will embrace language discovery (Johns, Reference Johns1986; Barlow, Reference Barlow1996; Tribble & Jones, Reference Tribble and Jones1997). We have believed that, as Pérez-Paredes (Reference Pérez-Paredes, Harris and Moreno Jaén2010) puts it, the methods of research in corpus linguistics can be transferred to the language classroom by turning linguists’ analytical procedures into a pedagogically relevant tool to increase both learners’ awareness of and sensitivity to patterns of language while also enhancing language learning strategies. Pedagogically core to DDL is the aim of fostering the independent acquisition of language knowledge (lexis, grammatical constructions, collocations, and so on). Within the ethos of DDL, learners are encouraged, in inductive processes, to discover patterns of language. It is widely claimed that such an endeavour aims to foster more complex cognitive processes such as making inferences and forming hypotheses (O'Sullivan, Reference O'Sullivan2007; Lee, Warschauer, & Lee, Reference Lee, Warschauer and Lee2019).

It is fair to say that the early enthusiasm was counter-balanced by some words of caution. Leech (Reference Leech, Wichmann, Fligelstone, McEnery and Knowles1997, p. 5) observed that while research is a natural extension of teaching and enables the learner to explore, investigate, generalize and test hypotheses, ‘it does not itself initiate or direct the path of learning’. Leech saw this as part of the teacher's role. Widdowson (Reference Widdowson and Alatis1991, p. 20ff.), referring to corpus insights, argued that ‘[s]uch analysis provides us with facts, hitherto unknown, or ignored, but they do not themselves carry any guarantee of pedagogic relevance’. Authors such as Römer (Reference Römer2006), Tribble (Reference Tribble2008) and Pérez-Paredes (Reference Pérez-Paredes, Harris and Moreno Jaén2010) have pointed to the need to find a plausible way of moving DDL from a research-oriented process suited to university settings (where learners analyse, hypothesize and discover language) to one with a broader pedagogical application and theoretical underpinning. As Römer (Reference Römer2006, p. 129) noted, a lot still remains to be done before arriving at the point where it can be said that ‘corpora have actually arrived in language pedagogy’.

Over a decade ago, while the late Stig Johansson lauded the potential of DDL for enhancing language learning because of the parallels between the natural processes of language acquisition and the processes involved in hypothesizing about language in DDL, he also called for a greater connection between DDL and second language acquisition (SLA) research (Johansson, Reference Johansson and Aijmer2009). Johansson foresaw connections that could be made with ongoing SLA work on attention and awareness as well as concepts such as input enhancement. Unfortunately, few of the many worthwhile DDL studies over the years have engaged with SLA theory and indeed few SLA studies have sought out DDL as a means of exploring their hypotheses. In this plenary paper, I wish to make a case for a broadening in our research gaze. Firstly, I want to look closely at the pedagogical and theoretical underpinnings of DDL. These are often inter-connected with SLA but under-explored by both DDL and SLA researchers. I want to focus on the question of how and where DDL fits within current SLA models and debates. And underlying all of this, I want to address why, as DDL advocates and enthusiasts, we should care about these issues. In summary, I will argue that while there has been a number of helpful meta-analyses, reflections and reviews of ongoing DDL work across many variables, there has been a dearth of focus on the learning theories that underpin DDL and on how this approach might inter-relate with SLA theories and vice versa. I will also argue that DDL is well-placed to be part of experimental research that could lead to cutting-edge insights into the cognitive processes of language learning and enhance ongoing SLA debates, especially in relation to implicit and explicit learning processes. Before we look at these issues, let us briefly summarize where the current meta-studies have brought us to in terms of our aggregated understanding of DDL.