Theoretical considerations of applications and implications of concordance-based cloze tests

摘要

Concordance-based cloze (henceforth ConCloze) tests have rarely been researched over the past two decades. This paper explores some potentially practical applications of their manifold item formats to language testing. The formats can range from response-selected to response-constructed ones, all embracing Read's (Assessing Vocabulary. Cambridge: Cambridge University Press, 2000) context-dependent vocabulary assessment and likely tapping into multiple aspects and components of Nation's (Learning Vocabulary in another Language. Cambridge, UK: Cambridge University Press, 2001) word knowledge. These applications imply a fine-grained approach to vocabulary teaching and learning as well as word-knowledge profiling and evaluation in that the viability of this item type calls for attention to complex word-knowledge components rather than merely the usually recognized form-meaning dimension. This could in turn carry a broad implication for language-pedagogical paradigms in general and vocabulary assessment in particular. Moreover, as the ConCloze item formats may be constructed with the aid of corpus and concordance markings, the other implication is that corpora should be built purposively, and concordances annotated specifically yet multidimensionally. This will enable these corpus-linguistic tools to accommodate computer-based test construction and delivery of the ConCloze item type as well as other item types.