Cambridge Handbook of English Corpus Linguistics

The growing interest in corpus linguistics methods in the 1970s and 1980s was largely enabled by the increased power of computers and the use of computational methods to store and process language samples. Before this, even simple methods for studying language such as extracting a list of all the different words in a text and their immediate contexts was incredibly time consuming and costly in terms of human effort. Only concordances of books of special importance such as the Qur’an, the Bible and the works of Shakespeare were made before the 20 century and required either a large number of scholars or monks or a significant investment in time by a single individual, in some cases more than ten years of their lives. In these days of web search engines and vast quantities of text that is available at our finger tips, the end user would be mildly annoyed if a concordance from a one billion word corpus took more than five seconds to be displayed. Other text rich disciplines can trace their origins back to the same computing revolution. Digital Humanities scholars cite the work of Roberta Busa working with IBM in 1949 who produced his Index Thomisticus, a computer-generated concordance to the writings of Thomas Aquinas. Similarly, lexicographers in the 19 century used millions of handwritten cards or quotation slips but the field was revolutionised in the 1980s with the creation of machine-readable corpora such as COBUILD and the use of computers for searching and finding patterns in the data. This chapter presents an introductory survey of computational tools and methods for corpus construction and analysis. The corpus research process involves three main stages: corpus compilation, annotation, and retrieval (see Rayson 2008). A corpus first needs to be compiled via transcription, scanning, or sampling from on-line sources. Then, the second stage is annotation, through some combination of manual and automatic methods to add tags, codes, and documentation that identify textual and linguistic characteristics. A snapshot of tools and methods that support the first and second stages of the corpus research process are described in sections 2.1 and 2.2. Retrieval tools and methods enable the actual linguistic investigations based on corpora: i.e. frequency analysis, concordances, collocations, keywords and n-grams. These tools are introduced in Section 2.3, together with a brief timeline tracing the historical development of retrieval tools and methods and the current focus on web-based interfaces for mega-corpora. Corpus tools and methods are now being applied very widely to historical