Information Density and Quality Estimation Features as Translationese Indicators for Human Translation Classification

Proceedings of NAACL-HLT 2016, pages 960–970, San Diego, California, June 12-17, 2016. c⃝2016 Association for Computational Linguistics Information Density and Quality Estimation Features as Translationese Indicators for Human Translation Classification Raphael Rubino Universit¨at des Saarlandes Saarbr¨ucken, Germany Ekaterina Lapshinova-Koltunski Universit¨at des Saarlandes Saarbr¨ucken, Germany e.lapshinova@mx.uni-saarland.de {raphael.rubino,josef.vangenabith}@uni-saarland.de Josef van Genabith Universit¨at des Saarlandes DFKI Saarbr¨ucken, Germany Abstract This paper introduces information density and machine translation quality estimation in- spired features to automatically detect and classify human translated texts. We inves- tigate two settings: discriminating between translations and comparable originally au- thored texts, and distinguishing two levels of translation professionalism. Our framework is based on delexicalised sentence-level dense feature vector representations combined with a supervised machine learning approach. The results show state-of-the-art performance for mixed-domain translationese detection with information density and quality estimation based features, while results on translation ex- pertise classification are mixed. 1 Introduction Translations, regardless of the method they were produced with, are different from their source texts and from originally authored comparable texts in the target language. This has been confirmed by many linguistic studies on translation properties commonly called translationese (Gellerstam, 1986). These studies show that translations tend to share a set of lexical, syntactic and/or textual features distin- guishing them from non-translated texts. As most of these features can be measured quantitatively, we are able to automatically distinguish translations from originals (Baroni and Bernardini, 2006; Ozdowska and Way, 2009; Kurokawa et al., 2009). This is useful for Statistical Machine Translation (SMT), as language and translation models can be improved if the translation direction and status of the data (trans- lation or original) is known (Lembersky, 2013). Research on translationese has recently focused on exploring features capturing aspects of transla- tionese such as simplification, explicitation, conver- gence, normalisation and shining-through (Volan- sky, 2012; Ilisei, 2012). Here we extend this work as follows: (i) we investigate the impact of information density and surprisal features, (ii) we explore the use of features used in machine translation quality estimation (Blatz et al., 2003; Specia et al., 2010), (iii) we explore classification between originally au- thored text and trainee and professional translation, as well as between professional and trainee transla- tion. In order to avoid biasing classification by topic content, throughout our experiments we use fully delexicalised features, resulting in dense vector rep- resentations (rather than sparse vectors, where the size of the vectors can be up to and in fact exceed the size of the vocabulary). We show that informa- tion theory as well as translation quality estimation inspired features achieve state-of-the-art results in mixed-domain original vs. human translation clas- sification. Languages provide speakers with a large number of possibilities of how they may encode messages. These include the choice of phonemes, words, syn- tactic structures, as well as arranging sentences in discourse. Speakers’ decisions regarding these choices are influenced by diverse factors: cognitive processing limitations can impact variation in lin- guistic encoding across all linguistic levels. Text production conditions, including monolingual vs. multilingual settings, can influence this variation: in …

Content truncated to 500 words