摘要：AbstractAs considerable attention has been paid in recent years to variability within fixed sequences (e.g. Biber, 2009; and Gray and Biber, 2013), this paper describes the use of a Corpus Linguistics technique, the Part-of-Speech-gram (usually abbreviated to PoS-gram), that allows potential variability across all slots, and is extremely effective for the discovery of phraseologies that might otherwise remain hidden. A PoS-gram is a string of Part-of-Speech categories (Stubbs, 2007) the tokens of which are strings of words that have been annotated with these PoS tags. Hence, in each slot of the PoS-gram, any word can occur as long as it belongs to the PoS category of that particular position. Despite the vast potential of this technique, it has up to now been largely underused.This paper will illustrate the utility of PoS-grams by way of analysis of a 450,000 token corpus composed of travel journalism texts from the BBC website. The PoS-grams extracted are compared with a database of PoS-grams obtained from the 100M token BNC. While a large number were found to be statistically significant, in-depth analysis was conducted on PoS-grams containing the inflected superlative adjective form AJS, a feature previously recognized as being central to tourism/travel-writing though without reference to corpus-based techniques (e.g. Dann, 1996).
