Concordance line sorting in The Prime Machine

Abstract

Abstract Corpus data provide evidence of the patterning of language, and one way word usage can be analysed is through the study of concordance lines. While popular concordancers provide different sorting methods, they are typically only able to display lines in the order in which they occur in the corpus, randomly, or alphabetically by words in slots to the left or right of the word of interest. Less sophisticated users may find recognising patterns from these orderings quite challenging. This paper considers possible needs of language learners in terms of concordance ranking and introduces two methods which have been adopted and developed for The Prime Machine. The first method uses repeated patterns, measuring the number of matches made with other lines in the set. The second method incorporates collocation scores, providing examples with strong collocations from the entire corpus at the top of sampled concordance lines.