HOW DOES INPUT AFFECT OUTPUT? LINKING PEDAGOGIC AND LEARNER CORPORA

Abstract

Several studies report differences between learner and first language (L1) use in written corpora in terms of overuse (e.g., discourse marking features), underuse (e.g., hedges), and misuse of certain forms (Altenberg & Granger, 2001; Chen & Baker, 2010; Granger, 1998). However, the output that Foreign Language students produce is obviously connected to the input and teaching they receive in the classroom (Ellis, 2002; Zapounidis & Mattheoudakis, 2024). Despite this, only two studies to date have attempted to draw direct comparisons between a pedagogic corpus (Hunston, 2002) and learner output (Vyatkina, 2013; Zapounidis & Mattheoudakis, 2024). This study fills this gap by examining differences between learner writing and the input received in the learning materials with respect to learner formulaic language (FL) use.  Using a learner corpus and a pedagogic corpus, it compares learners’ use of FL (also known as multi-word sequences (Wray, 2002) – e.g., This is because, as a result) with how the language is presented in the pedagogic corpus. The frequency, range, and accurate use of discourse organizing FL were calculated from a longitudinal learner corpus. FL was then identified in the pedagogic corpus and labelled in terms of the frequency of use and intensity of teaching for each.  Findings showed a significant uptick in use of FL directly after teaching, but not a concurrent improvement in accuracy of use, consistent with research showing form uptake precedes mastery (Ellis, 2003; Schmitt, 2010). The accuracy of some FL, however, improved between draft submissions.  Results are discussed in terms of the task-induced involvement load framework (Laufer & Hulstijn, 2001), and the applications to pedagogy are elucidated. Future research in the area will test the theoretical claims suggested by this research in controlled conditions.