Keynote speakers

Prof. Douglas Biber, Northern Arizona University

Grammatical complexity in L2 writing development:  Contrasting oral versus literate complexity across registers and developmental levels

The grammatical system of English comprises dozens of complexity features, including different types of grammatical structures (e.g., finite dependent clauses, non-finite dependent clauses, phrases) serving different syntactic functions (e.g., noun modifier versus clause modifier/complement).  Recent research has focused on the ways in which these features pattern in texts from general spoken and written registers (Biber, Larsson, Hancock 2024a,b).  Those studies provide strong evidence for two major groupings of complexity features: phrases functioning as noun modifiers, and finite dependent clauses functioning as clause-level constituents.  In contrast, though, a follow-up study (Biber et al. 2025) found considerably weaker support for these groupings in university student written texts. 

An unresolved question from these earlier studies was the large number of complexity features that did not covary in systematic ways with other features (e.g., non-finite dependent clauses functioning as noun modifiers, or phrases functioning as clause-level constituents). However, a recent study by Biber and Larsson (2025) shows that the full inventory of grammatical complexity features are organized at the register-level (as opposed to the text-level) as two major sets of covarying features:  ‘oral’ complexity features and ‘literate’ complexity features.  That is, all complexity features seem to covary in one of these two major groupings across registers, even if some of those same features do not regularly co-occur in the same texts.  These two groupings have a strong complementary relation to one another, so that a given register will tend to frequently employ the oral grouping or the literate grouping, but not both. 

The present talk begins by surveying the results of these previous studies, and then building on those findings to explore whether grammatical complexity is organized in the same way in the discourse domain of L2-English writing as in the broader domain of spoken and written registers. A large multi-level and multi-register corpus of L2 writing is analyzed, stratified by proficiency levels (CEFR levels A, B, and  C; Undergraduate disciplinary writing, Graduate disciplinary writing, and Professional L2-English disciplinary writing) as well as registers and academic disciplines (opinion-based registers versus information-based registers, in humanities, social science, and physical/life science).  Results are discussed in relation to three major issues:

1. Is the inventory of grammatical complexity features in academic L2-English written registers organized according to the same general distinction between oral complexity versus literate complexity found in general spoken and written registers?
2. Can those two general complexity dimensions be used to describe the developmental trajectories of L2-English writers?
3. Do these results provide convincing evidence that second language acquisition occurs at very advanced levels and through exposure to an expanding repertoire of academic registers of disciplinary writing?

Prof. Marije Michel, University of Groningen

Task effects in L2 performance

In any corpus research the composition and quality of the written or spoken texts that a corpus is made of play a major role in what can be investigated. Similarly, interpreting the findings requires insights into what language in terms of functional context, modality, a.s.o. forms the body of the big data being explored. In recent years, second language acquisition (SLA) researchers together with corpus linguists have started to draw on large-scale learner corpora that are collected from online language learning platforms, such as the EF-Cambridge Open Language Database (EFCAMDAT) or Duolingo. While these corpora provide treasure troves, as they allow for analyzing learner data at an unprecedented scale, interpreting the figures that come out of these inquiries necessitates a precise understanding of what tasks learners were working on when creating the texts that contribute to the corpus.

In L2 pedagogy, tasks are 'the hundred and one things people do in everyday life, at work, at play, and in between' (Long, 2015). For appropriate understanding of L2 corpus data gained from pedagogical environments, researchers need to take into account the instructional prompt, the embedding of a task within a lesson or whole course, as well as the learner characteristics.

In this talk, I will review existing investigations addressing these issues before I draw on my own work into task effects in learner corpora, with the aim to formulate desiderata for future L2 corpus linguists to consider pedagogical insights gained from the framework of task-based language teaching (TBLT). 

Prof. Shelley Staples, University of Arizona

Using the register-functional approach to examine development in learner writing: Working towards more situationally-aligned comparisons

While much learner corpus research has focused on examining differences in the use of linguistic features across L1 and L2 writers, as well as across different groups of L2 writers, this research has been criticized as setting up false dichotomies between these groups, with many questioning the comparability of corpora across L1 backgrounds and with some arguing that we shouldn’t even be making comparisons between these groups of writers (see Paquot, 2024 and Gilquin, 2022 for a discussion of this issue in the context of learner corpora). A register-functional perspective (Biber et al., 2022) including situational analysis of variation in communicative purposes and other factors across and within assignment types, provides a lens through which we can work towards comparability of samples, in order to make more informed decisions about whether the data is adequate for such comparisons. It also provides us with an approach for investigating how task variation interacts with development across L1 and L2 writers as well as across L2 writers from different linguistic backgrounds.

This plenary will explore data from the Corpus and Repository of Writing, showcasing a wide range of assignments (tasks) produced by learners (L1 and L2 English writers) in general education courses focused on developing academic and professional writing skills across disciplines. I start by examining my own previous research using a register-functional framework, showing how incomparability of samples led to conclusions that were incomplete. Then, we will explore variation within the communicative purpose of “to argue” (Goulart, 2024), including argumentative tasks which often are associated with learner writing, illustrating key situational features that may vary and thus lead to the use of different linguistic features. 

I will then use a situationally aligned sample to explore questions of whether L2 writers in the same course as L1 writers, producing the same assignments, are 1) able to exhibit register differences, given research that suggests L2 writers may not be as sensitive to these differences (cf. Gilquin & Paquot 2008, Larsson 2019); 2) produce similar linguistic features to those of L1 English writers (many of whom are multilingual themselves) in the same course. Finally, we will examine whether there is development across course levels compatible with previous research on academic registers, and how assignment types (tasks) impact linguistic trajectories. 

The talk will ultimately illustrate how a register perspective towards learner tasks can be used to bolster comparability of samples across different groups of writers, if such comparisons are to be made, and to understand variation within learner tasks, even within what are considered well-known categories (e.g., argument), in order to provide context for learner development.