This repository contains supplementary material for the publication:
Antonio Toral. 2019. Post-editese: an Exacerbated Translationese. Machine Translation Summit.
- bin/ code to run the experiments (bash, Python and R scripts). The script runme.sh runs all the experiments.
- datasets the three datasets used in the experiments. They are provided as they are and preprocessed (normalised punctuation and tokenised).
- results output of the experiments, in subfolders: length ratio, lexical density, lexical variety, perplexity difference on PoS sequences. Subfolder tagger-output contains PoS-tagged data, used for lexical density and perplexity difference.
- third third-party code. Moses' scripts are provided. Note that there are two other requirements: UDPipe (1.2.0 was used) and SRILM (1.7.2 was used).