Data-driven learning in informal contexts? Embracing broad data-driven learning (BDDL) research

Pérez-Paredes, P. (2024) Data-driven learning in informal contexts? Embracing Broad Data-driven learning (BDDL) research. In Crosthwaite, P. (Ed.). Corpora for Language Learning: Bridging the Research-Practice Divide. Routledge.

In this chapter, I argue that it is necessary to pursue an analysis of DDL practices in the broader language learning context (Pérez-Paredes & Mark, 2022), particularly in informal contexts outside the university classroom.

We need to push the boundaries of DDL praxis and research outside the classroom if we are to gain a more comprehensive view of the contributions of DDL to language learning in the first half of the 21st century. It is essential to expand the ecological research model that has dominated DDL research so far, and which has thoroughly examined higher education (HE) contexts.

While instructed, formal language learning continues to be central to language learners’ experiences, new sites of learning and technologies emerge sometimes unexpectedly (e.g. the impact of ChatGPT at the end of 2022 was surprising, and it is probably too soon to evaluate its impact on language education).

I use the term “prototypical DDL” (Boulton, 2015) to refer to DDL that is designed by an expert in corpus linguistics and which takes place in the context of instructed second language acquisition (SLA) as part of a module or an official programme, typically in a higher education institution (HEI).

The term “broad DDL” (BDDL) refers to pedagogical natural language processing resources (P-NLPRs) for language learning (see Pérez- Paredes et al., 2018). BDDL makes use of a wide range of existing resources such as online dictionaries, text analysis and text processing tools, vocabulary-oriented websites and apps, translation services, and artificial intelligence (AI) tools for language learning across a variety of contexts, including self-directed uses.

It also involves the use of informal language learning against the backdrop of digital learning, characterized by a new ecology of reading and writing, multitasking and the emergence of a new literate social formation (Pérez-Paredes & Zhang, 2022) where communication processes are transitioning towards “dialogic interactions [less] subject to the power of institutions to set standards of knowledge, procedure, and truth based on their control of written texts” (Gee & Hayes, 2011, p. 125).

In BDDL, corpora are one of the many resources available to language learners. While some research has examined the use of Google as a web corpus and a concordancer (Sun, 2007; Sha, 2010; Pérez-Paredes et al., 2012; Boulton, 2015), this has mostly happened in instructed SLA contexts. The impact of other P-NLPRs in informal learning remains largely unexplored (see Crosthwaite & Boulton, 2023 for a discussion of some of these resources).

User-generated activity using personal devices such as phones or tablets treasure the potential to inform designed activity and, most significantly, what we know about learners’ interactions with content online (Kukulska-Hulme et al., 2007). P-NLPRs have the potential to foster autonomy, personalization, induction and authenticity and may offer an alternative to prototypical DDL corpora when engaging with BDLL (Pérez-Paredes et al., 2018, 2019).

There are three areas, at least, that will benefit from an examination of BDDL practices in informal learning: The exploration of new sites of language learning engagement; New opportunities to increase our understanding of the cognitive processes involved in statistical language learning; and the study and analysis of the role of new corpora in informal settings.

Thanks to Carolina Tavares de Carvalho, Daniela Terenzi & Alejandro Curado Fuentes for providing their insights

New research on Data-driven language learning March 2023

Allan, R. (2023). Reserved for Research? Normalising Corpus Use for School TeachersNordic Journal of English Studies22(1).

Allan, R., Walker, T., & Langum, V. (2023). Data-driven learning: Tools, approaches, and next steps. Nordic Journal of English Studies22(1), 1-12.

Muftah, M. (2023). Data-driven learning (DDL) activities: do they truly promote EFL students’ writing skills development? Education and Information Technologies, 1-27.

O’Keeffe, A. (2023). A Theoretical Rationale for the Importance of Patterning in Language Acquisition and the Implications for Data-driven Learning. Nordic Journal of English Studies22(1), 16-41.

Şahin Kızıl, A. Data‐driven learning: English as a foreign language writing and complexity, accuracy and fluency measures. Journal of Computer Assisted Learning.

5 recent books for language teachers interested in corpus linguistics, DDL & language education

Crosthwaite, P. (Ed.). (2019). Data-driven learning for the next generation: Corpora and DDL for pre-tertiary learners. Routledge. (URL)

Jablonkai, R. R., & Csomay, E. (Eds.). (2022). The Routledge Handbook of Corpora and English Language Teaching and Learning. Routledge.. (URL)

Pérez-Paredes, P. (2020). Corpus Linguistics for Education. A Guide for Research. Routledge. (URL)

Timmis, I. (2015). Corpus linguistics for ELT: Research and practice. Routledge. (URL)

Viana, V. (Ed.). (2022). Teaching English with Corpora: A Resource Book. Routledge. (URL)

Recent DDL research & events: 5 tips

Really exciting times for DDL and corpus linguistics and education researchers. There’s some interesting new stuff that has just been published, including some interesting conference videos. Here’s my selection.

(1) Boulton, A., & Vyatkina, N. (2021). Thirty years of data-driven learning: Taking stock and charting new directions over timeLanguage Learning & Technology25(3), 66-89.

Abstract

The tools and techniques of corpus linguistics have many uses in language pedagogy, most directly with language teachers and learners searching and using corpora themselves. This is often associated with work by Tim Johns who used the term Data-Driven Learning (DDL) back in 1990. This paper examines the growing body of empirical research in DDL over three decades (1989-2019), with rigorous trawls
uncovering 489 separate publications, including 117 in internationally ranked journals, all divided into five time periods. Following a brief overview of previous syntheses, the study introduces our collection, outlining the coding procedures and conversion into a corpus of over 2.5 million words. The main part of the analysis focuses on the concluding sections of the papers to see what recommendations and future avenues of research are proposed in each time period. We use manual coding and semi-automated corpus keyword analysis to explore whether those points are in fact addressed in later publications as an indication of the evolution of the field

(2) Dr Peter Crosthwaite, The University of Queensland: Is Data Driven Learning dead? In this talk Dr Crosthwaite ****