81006689

open access

Version of Record

XML Encoding for Spoken Learner (and Other) Corpora : A Modest Approach

English (英語)

Learner Corpus Studies in Asia and the World

2

49-62

神戸大学国際コミュニケーションセンター

2014-05-31

2014-06-25

Since the earliest days of corpus linguistics, markup has been used to represent features of corpus texts other than the actual words of the text. The first systems used were somewhat ad hoc, often based on using (sequences o!l punctuation marks to indicate para-linguistic properties of texts, over time the field has standardized on markup systems based on tags delimited by <angle brackets>: first SGML, and more recently the derived XML system. While the official standards for the encoding of corpora with XML - most notably the guidelines of the Text Encoding Initiative - are extremely heavyweight, and therefore most suitable for the development of large-scale reference datasets, I argue that a more modest level of XML can productively be applied within the context of corpora developed by individual researchers or small teams. To apply "Modest XML", it is necessary only to comply with certain fundamental rules of XML <Such as the layout of tags for regions and for points within the text, the use of attribute-values to add extra information, the use of a single document"level tag in each XML file, and the use of entities to represent certain special characters). However, it is also useful to be aware of certain de facto standard XML tags and attributes very commonly used in spoken corpora <Such as <U>, <vocl>. <pause/>, <unclear> and so on}. Finally, it is always possible to extend the XML vocabulary that one uses to support the specific needs of a particular corpus development project.

紀要論文