In Search of Lexical Discriminators of Definition Style: Comparing Dictionaries Through n-Grams

This study investigates definition style in the range of English dictionaries through the analysis of distribution and frequency of most frequent words (unigrams) and word clusters (trigrams). The main goal of the study is to demonstrate how dictionaries can be compared quantitatively. The texts studied were the following six dictionaries: <it>Johnson</it> (1785), <it>Webster</it> (1865), <it>OED</it> (1888–1928), <it>Chambers</it> (1952), <it>LDOCE</it> (2005) and <it>COD</it> (2011). A selection of definitions were subjected to hierarchical cluster analysis and correspondence analysis. The former analysis confirmed the assumption that recent dictionaries, <it>LDOCE</it> and <it>COD</it>, are most distant to the other dictionaries with regard to the distribution of unigrams and trigrams. The latter analysis pointed to a few words and word clusters (lexical discriminators) which seemed to be distinguishing features of the dictionaries. It is hoped that the methods demonstrated in this paper will expand an inventory of instruments of lexicographical comparison, and will provide more insights into lexical composition of definitions.