Automated L1 identification in English learner essays and its implications for language transfer

Abstract

This talk gives an overview to our study: it is based on a corpus of TOEFL English test essays written by learners of 11 different first language backgrounds. In our research we use automated machine learning techniques to automatically classify the learner texts according to the L1 of their authors. Furthermore, we take a closer look at some of the most informative features for the classifier regarding L1 German and L1 Italian speakers. Some of these features show a possible origin in processes of L1 transfer.

Date
2014-04
Event
Work in Progrss Series
Location
Kompetenzzentrum Sprachen, Freie Universität Bozen, Bolzano, IT
Next
Previous