Structural features of collocations in translated and non-translated Lithuanian

Abstract

In translated language research, Lithuanian is still an underrepresented language, although there are several parallel and comparable corpora which could be used as data sources in corpus-based translation studies. This paper aims to compare the structural features of abstract noun collocations in translated and non-translated (original) Lithuanian across fiction and popular science genres. For this small-scale pilot study, eleven abstract nouns were selected, and their collocations were extracted semi-automatically from a 4-million-word monolingual comparable corpus ORVELIT. ORVELIT consists of four sub-corpora: original and translated fiction, as well as original and translated popular science texts. Collocations were classified by their parts of speech, and the analysis focused on both quantitative differences and the lexical diversity of the collocates. When considering the structural properties of collocations in our analysis, collocations that consist of two nouns emerged as being more diverse in original popular science texts. The results also show that there are more unique collocations with adjectives in original fiction, i.e. the lexical diversity of collocates is wider than in translation. In addition, some syntactic differences between translated and original popular science collocations were observed: nouns in translations are more often modified with an infinitive, while in original texts, the attribute function is more often expressed with a noun in the genitive case. Conclusions and recommendations suggest that more data needs to be included to examine the observed tendencies. Adding the data from the general corpus of Lithuanian for testing how typical and frequent observed collocations are could be a necessary methodological step to provide more solid research outcomes.