International Journal of Corpus Linguistics - Volume 1, Issue 1, 1996

In this paper intuition-based studies of reflexive forms such as myself are contrasted with a corpus-based investigation of actual usage of reflexives. The examination of reflexives in English in several corpora reveals a variety of patterns, which are analysed within a schema-based approach to grammar (Barlow and Kemmer 1994). This approach follows the cognitive/functional tradition of grammatical analysis in viewing all grammatical units as composed of form-meaning pairings.The paper demonstrates that a schema-based approach is well-suited to the task of describing the major and minor patterns of use revealed by corpus analysis. The importance of text analysis in language teaching is highlighted and connections between the schema-based grammatical formalism and data-driven approaches to second language learning (Johns 1991b) are briefly explored.

The interaction between corpora and lexicons in compiling a monolingual learner dictionary, based on the experience of writing the Cambridge International Dictionary of English (1995). Exploitation of search facilities, statistical collocation analysis and analysis of syntactic patterns in a grammatically tagged corpus. Some limitations of corpus-based lexicography. Ways in which dictionary data was exploited in the analysis and enhancement of the corpus. In this project the corpora, software and dictionary were developed simultaneously rather than sequentially. An advantage of this way of working was that the evolution of the software was constantly guided by the needs of lexicographers, and a number of desirable avenues of development for the software tools and for the corpus itself were identified.

This paper describes an experiment in statistical analysis of corpora with respect to the temporal changes in language use. The technique approximates the notion of temporal relevance of usage evolution by analysing and evaluating the frequency distribution of a set of indicators over time and by isolating string configurations with unlikely temporal distribution.

Preliminary findings from corpus analysis suggest that the semantics of each verb in the language are determined by the totality of its complementation patterns. Accurate description of those patterns requires a level of analytic delicacy which was not possible until the advent of large bodies of data, along with techniques for distinguishing significant patterns from mere noise. Such analysis is in its infancy, but it is already clear that, in order to analyse the semantics of verbs empirically, it is necessary to identify typical subjects, objects, and adverbials and to group individual lexical items into sets within those clause roles. The nature of lexical sets is discussed and an attempt is made to indicate the range of semantic and syntactic phenomena likely to be encountered in lexical analysis of this kind.

This paper1 contrasts two views on the analysis of language. In one view, language is primarily seen as a carrier of messages in sentences whose propo-sitional content can be retrieved, and symbolised in a knowledge base. In the other, language is seen as a means of communication that deals in much more complex matters than just carrying messages.In relation to vocabulary and the design of lexicons, the model of terminology suits the first position, while in the other the lexicon is considered empty at the start and is gradually filled with the evidence of usage.Similar contrasts are made in other areas relevant to natural language processing. In one approach, the expectation is of tidiness and conformity to rules; the other stresses the inherently provisional nature of the organisation of language and, therefore, the meanings.As these two approaches encounter the vast amount of evidence stored in today's corpora, their methods and responses contrast in interesting ways.

An overview is given of work on the creation, collection, preparation, and publication of electronic corpora of written and spoken language undertaken at the Human Communication Research Centre at the Universities of Edinburgh and Glasgow. Four major efforts are described: the HCRC Map Task Corpus, the ECI/MC1, the MLCC project and work on document architectures and processing regimes for SGML-encoded corpora.