The paper compares systematically the utility of specially-made text corpora and the textual resources of the World Wide Web for linguists and language learners. Different modes of access are discussed, including via dedicated concordancing software, web-concordancing and the universal search engine.