Facilitating the compilation and dissemination of ad-hoc web corpora

Since the World Wide Web gained prominence in the mid-1990s it has tantalized language investigators and ins ructors as a virtually unlimited source of machine-readable texts for compiling corpora and developing teaching materials. The broad range of languages and content domains found online also offers translators enormous promise both for translation by-example and as a comprehensive supplement to published reference works This paper surveys the impediments which s ill prevent the Web from realizing its full potential as a linguistic resource and discusses tools to overcome the remaining hurdles. Identifying online documents which are both relevant and reliable presents a major challenge. As a partial solution the author's Web concordancer KWiCFinder au omates the process of seeking and retrieving webpages Enhancements which permit more focused queries than existing search engines and provide search results in an interactive explora ory environmen are described in detail. Despite the efficiency of automated downloading and excerpting, selecting Web documents still entails significant time and effort. To multiply the benefits of a search, an online forum for sharing annotated search reports and linguistically interesting texts with other users is outlined. Furthermore, the orien ation of commercial sea ch engines toward the general public makes them less beneficial for linguistic research. The author sketches plans for a specialized Search Engine for Applied Linguis s and a selective Web Corpus Archive which build on his experience with KWiCFinder. He compares his available and proposed solutions to existing resou ces, and su veys ways to exploi them in language teaching. Together these proposed services will enable language learners and professionals to tap into the Web effectively and efficiently for instruction research and translation. t