Supporting Collocation Learning

Collocations are of great importance for second language learners. Knowledge of them plays a key role in producing language accurately and fluently. But such knowledge is difficult to acquire, simply because there is so much of it. Collocation resources for learners are limited. Printed dictionaries are restricted in size, and only provide rudimentary search and retrieval options. Free online resources are rare, and learners find the language data they offer hard to interpret. Online collocation exercises are inadequate and scattered, making it difficult to acquire collocations in a systematic way. This thesis makes two claims: (1) corpus data can be presented in different ways to facilitate effective collocation learning, and (2) a computer system can be constructed to help learners systematically strengthen and enhance their collocation knowledge. To investigate the first claim, an enormous Web-derived corpus was processed, filtered, and organized into three searchable digital library collections that support different aspects of collocation learning. Each of these constitutes a vast concordance whose entries are presented in ways that help students use collocations more effectively in their writing. To provide extended context, concordance data is linked to illustrative sample sentences, both on the live Web and in the British National Corpus. Two evaluations were conducted, both of which suggest that these collections can and do help improve student writing. For the second claim, a system was built that automatically identifies collocations in texts that teachers or students provide, using natural language processing techniques. Students study, collect and store collocations of interest while reading. Teachers construct collocation exercises to consolidate what students have learned and amplify their knowledge. The system was evaluated with teachers and students in classroom settings, and positive outcomes were demonstrated. We believe that the deployment of computer-based collocation learning systems is an exciting development that will transform language learning.