The absolute majority of scholarly work in descriptive translation studies is product-oriented. In this article, the focus is moved from product-oriented to process-oriented translation studies by compiling an English – Lithuanian Phases of Translation Corpus (PT corpus). The PT corpus is analysed using quantitative and qualitative analyses. The quantitative analysis using frequency information highlights the difficult word types that either are missing or are inconsistently translated in successive Lithuanian translated versions. The qualitative analysis continues the quantitative research by help of parallel concordancing. The problematic cases of translation are extracted and cases of normalization, systematic replacement of terminology, and influence by the original language are reported.