A historical characterisation of American and Brazilian cultures based on lexical representations

Abstract

The goal of this study is to detect the historical distribution of representations of the United States and Brazil formed around the use of the nationality adjectives American and Brazilian. To achieve this goal, the study used a pre-existing multi-dimensional analysis of representations based on bigrams from the half-a-trillion-word Google Books bigram dataset of English writing ( Berber Sardinha, 2019 ), which provided the major representations of both cultures. The method was based on the text type approach developed by Biber (1989) , which uses cluster analysis to identify the groupings of texts (in this case, the years in which the bigrams occurred) that share similar linguistic characteristics (here, mutual bigrams). The results showed nine major clusters for the US and eight for Brazil. The clusters were interpreted historically and ordered by time to outline the predominant representations of each country over the 200 years’ worth of data analysed.