Loading...
Abstract
Extensive reading is an effective language learning technique that involves fast reading of large quantities of easy and interesting second language (L2) text. However, graded readers used by beginner learners are expensive and often dull. The alternative is text written for native speakers (authentic text), which is generally too difficult for beginners. The aim of this research is to overcome this problem by developing a computer-assisted approach that enables learners of all abilities to perform effective extensive reading using freely-available text on the web.
This thesis describes the research, development and evaluation of a complex software system called FERN that combines learner modelling and iCALL with narrow reading of electronic text. The system incorporates four key components: (1) automatic glossing of difficult words in texts, (2) individualised search engine for locating interesting texts of appropriate difficulty, (3) supplementary exercises for introducing key vocabulary and reviewing difficult words and (4) reliably monitoring reading and reporting progress. FERN was optimised for English speakers learning Spanish, but is easily adapted for learners of others languages.
The suitability of the FERN system was evaluated through corpus analysis, machine translation analysis and a year-long study with second year university Spanish class. The machine translation analysis combined with the classroom study demonstrated that the word and phrase error rate generated in FERN is low enough to validate the use of machine translation to automatically generate glosses, but is high enough that a translation dictionary is required as a backup. The classroom study demonstrated that when aided by glosses students can read at over 100 words per minute if they know 95% of the words, whereas compared to the 98% word knowledge required for effective unaided extensive reading. A corpus analysis demonstrated that beginner learners of Spanish can do effective narrow reading of news articles using FERN after learning only 200–300 high-frequency word families, in addition to familiarity with English-Spanish cognates and proper nouns.
FERN also reliably monitors reading speeds and word counts, and provides motivating progress reports, which enable teachers to set concrete reading goals that dramatically increase the quantity that students read, as demonstrated in the user study.
Type
Thesis
Type of thesis
Series
Citation
Walmsley, M. (2015). Learner Modelling for Individualised Reading in a Second Language (Thesis, Doctor of Philosophy (PhD)). University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/10559
Date
2015
Publisher
University of Waikato
Supervisors
Rights
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.