Corpus-Concordance-Database-VARBRUL'

Abstract

Although concordances are in widespread use and have become the main output format of corpus analysis, they do little more than rearrange the selected data as a special kind of list. By yielding enough preceding and following context, however, concordances are usually sufficient for the purposes of classification and analysis. As items of high frequency have different senses, perform different functions, occur in differently constructed environments, and are thus used variably, classification is a major part of linguistic analysis. Up until now, however, it has. not been easy to store classificatory encodings alongside the concordance data, with a view to further exploitation of these encoded classifications. One solution is the importation of concordances into a database, where additional fields can then be created for mnemonic classificatory encodings, for further sorting on the basis of these encodings. This paper shows how this solution can be implemented in practice. It shows how concordances can come to be used in an ‘intelligent’ way as the basis of linguistic analysis; it also provides a practical tip for the use of the main software package for the analysis of interacting variables: VARBRUL For its quantitative and statistical correlation of co-occuring variants of different internal or external variables, VARBRUL depends utterly on a tokens file of encodings about the behaviour of each item. This paper shows how this tokens file can now be created by copying the classificatory encodings directly from the database.