Search

This paper explores a corpus-driven approach to studying the phraseological patterns of adjectives and nouns using data and methods from corpus linguistics. The rationale behind this approach is the strong resonance between how the meanings of adjectives have been defined and a major claim proposed in corpus linguistics; the meaning of a lexical item is typically defined both by its inherent lexical content and its relation to any accompanying words. For the analysis, the four most frequent English adjectives and their most recurrent noun collocates were chosen from the Corpus of Contemporary American English (coca.byu.edu/coca). The data are new/book, good/news, old/friend and high/levels. The occurrences of the six different phraseological patterns of each pair were counted and the internal structures of each pattern were examined. Then, the concordance lines containing the two words were analyzed to determine how the words are related semantically and structurally. The analysis reveals that the noun phrase consisting of an adjective and its noun collocate is the canonical form throughout the four collocations. The analysis also finds that the four adjectives differ in terms of their phraseological patterning with the noun collocates. The adjectives new and old are rarely used as predicates when they are selected after friend and book, respectively. On the other hand, good and high are canonically employed to serve the predicative function when they follow news and levels, respectively. The two groups also differ from each other in the way they contrast with their antonyms. The adjectives high and good are frequently employed in a contrastive context with their antonyms low and bad, whereas the adjectives new and old are seldom employed in that way.

Lee, Hye-Kyung. 2013. Antonymy and gradability: A corpus-based approach on English gradable antonyms. Linguistic Research 30(2), 335-354. This study examines the nature of antonymy relations of six gradable pairs (old/young, high/low, thick/thin, wide/narrow, big/small and tall/short) in terms of their frequencies and collocation in COCA (the Corpus of the Contemporary American English created by Mark Davies). The frequencies of the chosen twelve adjectives are counted and then the mean frequencies of the unmarked members and the marked members are calculated. The results reveal that the unmarked members of the pairs are used more frequently than their marked counterparts, which backs up the prediction that the unmarked adjectives are employed more frequently than their marked counterparts in general. The analysis of the collocation of the adjectives shows that the chosen pairs differ in terms of the extent to which they share common collocates. Four pairs (old/young, high/low, thick/thin and wide/narrow) have common collocates, which implicates that the antonyms can play in the same semantic fields. On the other hand, two pairs (big/small and tall/short) do not share any collocates. Hence in terms of the common collocates, some pairs are claimed to better qualify as antonyms than the others. The reason the latter pairs do not have common collocates is claimed to be the presence of other salient

Lee, Hye-Kyung, 2012, linguistic Variations between English News Headlines in the U.S. and Those in Korea, Linguistic Research 29(2), 329-350, This paper aims to address the similarities and differences between headlines of English political news in America and those in Korea. Acknowledging the contribution of the previous research towards the understanding of news headlines, the current paper investigates how the composing of news headlines is influenced by and interrelated with extra-linguistic factors such as the consideration of the target audience`s knowledge states or interests. For this purpose, two sets of data are collected from major online news sites within both countries and the collected data are classified according to the working criteria, especially in terms of employed structural patterns. The results regarding the comparison of the two sets of headlines reveal that headlines in Korean data employ shorter sentential headlines and a smaller variety of tense/aspect of sentences compared to American English counterparts. It is also discovered that cultural factors such as naming conventions practiced in the linguistic/cultural communities are taken into account. These differences are attributed to the consideration of the target audience`s multi-faceted states such as knowledge states, beliefs, or interests. At the same time, the two sets of data show similarities in terms of the variety of structural patterns they deploy.