Granger, Sylviane; Bestgen, Yves
[UCL]
The current study aims to replicate a study about the use of phraseological units by L2 learners that was conducted on the Michigan State University Corpus (MSU), on the basis of the International Corpus of Learner English (ICLE). The phraseological units investigated are collgrams, i.e. n-grams that have been assigned two association scores (Mutual Information (MI) and t-score) on the basis of a large reference corpus. The two studies led to one major convergent result, i.e. MI scores are significantly linked to text quality, while the correlations for the t-scores are weak. However, further analysis showed that the t-score could be used to filter the high MI collgrams so as to keep only those that are most closely linked to text quality. The most striking difference between the original and replication studies concerns the bigrams present in the learner texts but absent from the reference corpus, a category that is a strong proficiency predictor only in MSU. In our conclusion, we stress the importance of replication studies in learner corpus research and outline some of the challenges of the collgram measure.
Bibliographic reference |
Granger, Sylviane; Bestgen, Yves. Using collgrams to assess L2 phraseological development: A replication study . In: de Haan, P., De Vries, R., Van Vuuren, S., Language, Learners and Levels: Progression and Variation, Presses universitaires de Louvain : Louvain-la-Neuve 2017, p. 385-408 |
Permanent URL |
http://hdl.handle.net/2078.1/201651 |