Tracing marked collocation in translated and non-translated literary language

This study explores marked collocation in translated text in the COVALT corpus and relates the facts and patterns observed to properties of translation. Data have been drawn from a comparable and parallel corpus: English source texts, their corresponding Catalan translations, and texts originally written in Catalan. Its method unfolds in three steps: a quantitative and a qualitative cluster analysis and a manual analysis of concordances. Results for the second and third steps suggest that source text interference is an important factor at play, even though it is not the only one. An important conclusion that emerges from the study at different levels is that markedness is a matter of degree, not kind.