2Native Language Identification using Recurring n-grams – Investigating Abstraction and Domain Dependenceare of general importance or due to the nature of the Lang-8 corpus. Brooke and Hirst then explore the usefulness of artificial...
3Robust, Lexicalized Native Language Identificationin all three corpora, i.e. at least 1000 texts in the Lang-8 corpus, 200 texts in the ICLE, and 50 texts in the FCE. Table...