This timeline looks at explicit uses of corpora in foreign or second language (L2) teaching and learning, i.e. what happens when end-users explore corpus data, whether directly via concordancers or integrated into CALL programs, or indirectly with prepared printed materials. The underlying rationale is that such contact provides the massive contextualised exposure needed for language learning, but in a more controlled way than purely haphazard exposure via regular reading or listening, thus promoting or enhancing noticing, language awareness, autonomy, and ultimately producing ‘better learners’. It interweaves with many other notions in language teaching, from authenticity and autonomy to induction and constructivism, taking in much of what we know about language from corpus linguistics itself – its fuzzy, probabilistic nature and the importance of lexical patterns, collocations, chunks, frequencies and distributions. Commonly known as data-driven learning or DDL (Johns 1990), the approach goes back several decades; yet despite its impressive pedigree, a frequent lament is that it remains a marginal practice. Teachers may not be aware of DDL if it is absent from their initial training, or may see it as a research activity confined to higher education. While corpus consultation may appear too demanding for school-age learners, there are connections with the web searches that are already a frequent practice for many. Though lack of uptake is sometimes attributed at least in part to a dearth of empirical research, this claim is increasingly difficult to defend, as witnessed by the number of entries in this timeline and the hundreds of others not included. For a balanced discussion of the limits of DDL as well as what it can reasonably achieve, see Kaltenböck & Mehlmauer-Larcher (2005).Footnote ¹

The goal here has been to provide an overview of the evolution of the field from the earliest publications up until the present day. The first objective was to focus on empirical evaluations of DDL based on an extensive trawl of empirical research which identified over 200 publications. In some cases, the choice is quite straightforward, especially where a name is associated with a single widely-cited and influential paper; in others, it is difficult to choose among an author's cumulative output of a dozen or more publications which could almost merit an individual timeline of their own. Few authors have more than one entry here as the aim is to provide as broad a sweep as possible. On occasion, an intermediate solution was adopted, with the same authors contributing a base study along with a follow-up or replication study in the same entry. As a failsafe, the reference sections of the various papers were analysed to identify authors of more than one empirical DDL study; the 25 most frequently cited are all included here.

This procedure on its own would not do justice to the full range of activity in the field, and some key players would be excluded since their most widely cited work is non-empirical. Since research is a fundamentally human activity, a second consideration was to focus on the individuals and their most inspirational work. Empirical research is thus complemented with a number of descriptive, argumentative or position papers, especially from the early days, along with a handful of more theoretical works that provide some foundation to DDL that would otherwise be missing, as well as the main surveys reported in the area. Only manuals and textbooks were ultimately excluded from the timeline itself, though some of these provide additional theoretical support for the approach; selected references for further reading are provided as supplementary materials on the Language Teaching website. Some difficult decisions still needed to be made, and though the final list inevitably has an element of subjective choice, some of my personal favourites have had to be omitted for reasons of space and coherence according to the various inclusion criteria.

DDL draws on many different strands of research, and some of the most frequently cited names from neighbouring fields could only be included in a heterogeneous timeline of several hundred entries. These range from corpus linguistics itself and the development of corpora, tools and methodologies that are widely used in DDL, including work on language for specific purposes, frequency lists and learner corpora, error-correction and contrastive analysis, to corpus uses in syllabus design, materials and resources, testing and translation, as well as fields from lexis to acquisition studies and more general language learning theory.

While much work has been reported in a few major journals (notably Computer Assisted Language Learning, ReCALL, Language Learning & Technology, System and CALICO Journal), the bulk appears as chapters in various collections, often arising from conferences or in less well-known journals or conference proceedings, often available on line. In favouring the most accessible and prestigious sources here, the result is inevitably biased towards publications in and on English and from certain regions of the world, and does not address reports in other languages, or using corpora or software specific to other languages.

Ultimately, the timeline papers fall into four main categories as outlined below. Since there is no overarching programme for DDL research, the balance is not perfect between the various categories, which often overlap.

1. A Theoretical underpinnings. As mentioned above, this category does not attempt to outline all the general language teaching and learning theory that the approach draws on (e.g. cognitive skills or autonomisation), and is limited to those landmark papers most directly connected to DDL itself. Much of this derives from research in corpus linguistics, which has influenced our understanding of language (e.g. in the nature of lexicogrammar, patterning or chunks).

2. B Descriptive papers. A potentially enormous category, so largely limited here to a few influential papers from the relatively early days.

3. C Empirical evaluations are divided into sub-sections.

   First, mainly emic papers, often featuring questionnaires, interviews, logs etc., which focus on:

4. C1 learners’ attitudes and representations about corpus use;

5. C2 learners’ behaviour and processes in corpus use.

   Second, largely etic papers, often with a pre/post-test or control/experimental design from tests or analyses of learner productions, which look at:

6. C3 effects of corpus use as a learning aid;

7. C4 effects of corpus use as a reference resource.

8. D Surveys and syntheses. This category includes papers that attempt to make sense of the increasing body of research in the field.

Several features emerge. In chronological terms, it is in the nature of the developmental curve in applied linguistics research that a new field is developed by a small number of visionaries first in descriptive and speculative papers (beginning around 1990), followed by increasing numbers of ever more focused and sophisticated empirical evaluations (especially from 2000 on, with a dozen or more such papers every year since 2005) until such time as enough are available for various types of research syntheses, with theoretical underpinnings appearing at various points along the way. Curiously, perhaps, while many empirical studies refer to theoretical and pedagogical foundations, few seek directly to test them, and theory has not been a major driving force leading to new practices. The main initial impetus came largely from different spheres in the UK (Birmingham, Lancaster and Nottingham, among others), but it has since widened into much of the rest of the world: other English-speaking countries, western Europe, the Middle East and Asia in particular. The relative paucity of DDL work in the United States of America, however, would bear further examination.

The methodologies have become increasingly rigorous and sophisticated in design – not to mention complex in the case of the many Ph.Ds. not covered here – with a range of data collection instruments. Interestingly, there remains a balance of emic and etic studies with questionnaires ranking alongside pre-/post-tests or control/experimental designs as the instruments of choice. This allows considerable insight not just into the outcomes but also into the participants’ representations and the processes involved. There has been a move over time from purely lexicogrammatical to more discourse-oriented studies, and from use of corpora as a learning aid towards their use as a reference resource, especially in language for specific or academic purposes, with a predictable dominance of English as the target language. Some issues prove controversial, such as the need for training, or the potential with lower proficiency levels or for oral skills. More longitudinal work is needed outside university classrooms: in autonomous contexts, in secondary education, in language schools, or in the workplace for ‘genuine’ life-long needs; as well as the possibilities of exploiting more everyday tools for DDL-compatible web searches.