Finding the sweet spot: Learners’ productive knowledge of mid-frequency lexical items

Abstract

Research into vocabulary knowledge often differentiates between breadth (how many words a person knows) and depth (how well the words are known). Both theoretical categories are essential for understanding language learners’ lexical development, but how the different aspects of vocabulary knowledge interconnect has not received the same attention as each individual dimension, especially in terms of productive knowledge. This study analyses lexis from mid-frequency lemmas in the K3–K9 frequency bands from the learner corpus PELIC (The University of Pittsburgh English Language Corpus). Critically for learners, mastery of lexis in this frequency range is essential for achieving the English proficiency required for university study. From these mid-frequency items, a dataset of 7,554 tokens were collected from word families with multiple derivations and manually annotated. The findings showed high rates of collocational and derivational accuracy for the forms learners opted to use. However, compared to expert speaker texts in the Corpus of Contemporary American English (COCA), learners overused the verb forms and underused the noun forms of these lexical items. These patterns provide evidence of the interplay between breadth and depth in learners’ productive vocabulary usage, suggesting that increased lexical depth will naturally lead to greater lexical breadth and vice versa. Pedagogical implications reaffirm the importance of developing learners’ explicit morphological awareness and collocational accuracy. Suggestions for mid-frequency lexical items to prioritize in language learning are also provided, with a view to helping learners achieve academic readiness.

'Her dreadful plight': A corpus-assisted analysis of the indexical and stance properties of <em>poor thing</em>

“…These corpora can be taken as representative of American English because of their size (Table 1) and because they encompass a wide variety of genres produced in the United States (described in Section 4.2.1). COCA, in particular, is possibly the most widely-used English corpus (Davies 2008-), appearing in many studies as representative of L1 English norms (e.g., Monteiro, Crossley, & Kyle, 2020;Naismith & Juffs, 2021). It was deemed necessary to use a combination of American English corpora to maximize the potential data because PT is a relatively low-frequency MWE and because each of the two corpora sample language from different time periods.…”

State of the art and practice in AI in education

“…More subtly, it might simply involve the language in which most classroom AIED tends to be trained—mainly American English (Cotterell et al, 2020). In any case, the impact of the English‐trained models used by AIED tools in non‐English contexts and on the children who use them remains unknown (Naismith & Juffs, 2021).…”