Translationese and Its Dialects

By Moshe Koppel, Noam Ordan. 2011

201 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

Citations in the corpus

Findings of the 2014 Workshop on Statistical Machine Translation (2014, background): “For discussions about trans-lationese please for instance refer to Koppel and Ordan (2011).”
Scaling Laws for Neural Machine Translation (2021, background): “Source-original data has translated text on the target side, which contains several artifacts of “translationese” that distinguish it from text originally written in that language, often lacking the diversity and complexity of “natural” text [24], while target-original data requires the prediction o”
Findings of the 2019 Conference on Machine Translation (WMT19) (2019, background): “This is of course not new and is related to translationese (Koppel and Ordan, 2011).”
Experts, Errors, and Context: A Large-Scale Study of Human Evaluation for Machine Translation (2021, background): “The impact of transla-tionese (Koppel and Ordan, 2011) on human evaluation of MT has recently received attention (Toral et al., 2018; Zhang and Toral, 2019; Freitag et al., 2019; Graham et al., 2020).”
XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning (2020, background): “…it is prone to compromise the latter goal, bending the target language to the structural and lexical properties of the source language: the resulting evaluation benchmarks thus fail to measure system performance adequately (Koppel and Ordan, 2011; Volansky et al., 2015; Artetxe et al., 2020a).”
Investigating Multilingual NMT Representations at Scale (2019, positive): “This is also in line with some findings of studies on translationese (Koppel and Ordan, 2011), where the authors show that that the translated text is predictive of the source language.”
BLEU Might Be Guilty but References Are Not Innocent (2020, background): “Translationese refers to artifacts from the source language present in the translations, i.e. human translations are often less fluent than natural target sentences due to word order and lexical choices influenced by the source language (Koppel and Ordan, 2011).”
The Low-Resource Double Bind: An Empirical Study of Pruning for Low-Resource Machine Translation (2021, background): “There may be interesting connections to recent ﬁndings about output hallucinations occurring on memorized examples (Raunak et al., 2021), or with respect to translation bias (Koppel and Ordan, 2011).”
Asymmetric Features Of Human Generated Translation (2014, methodology): “Standard machine learning algorithms like SVMs (Baroni and Bernardini, 2006) and Bayesian Logistic Regression (Koppel and Ordan, 2011) have been employed to train classiﬁers for one of the following tasks: i.”
On the features of translationese (2015, methodology): “Although Koppel and Ordan (2011) manage to train on all the T components vs.”
Unsupervised Identification of Translationese (2015, methodology): “d classiﬁcation of translationese (Volansky et al., 2015). Speciﬁ- cally, we use function words (FW), more precisely, the same list that was used in previous works on classiﬁcation of translationese (Koppel and Ordan, 2011; Volansky et al., 2015). Feature values are raw counts (further denoted by te”
Language Models for Machine Translation: Original vs. Translated Texts (2011, background): “A similar list of markers have been shown to be excellent discriminating features between original and translated texts (from several European languages, including French) in an independent study (Koppel and Ordan 2011).”
QAmeleon: Multilingual QA with Only 5 Examples (2022, background): “Both approaches rely on transfer from English, which leads to a dependence on translation artifacts (Koppel and Ordan, 2011; Artetxe et al., 2020a) and a bias towards the linguistic characteristics of English, which is not the best source for all target languages (Lin et al.”
APE at Scale and Its Implications on MT Evaluation Biases (2019, background): “This is in line with results from Koppel and Ordan (2011), who demonstrate that the mere fact of being translated plays a crucial role in the makeup of a translated text, making the actual (human) translation a less natural example of the target language.”
Translationese as a Language in “Multilingual” NMT (2019, methodology): “This follows several prior works attempting to identify translated text (Kurokawa et al., 2009; Koppel and Ordan, 2011; Lembersky et al., 2012).”
Identifying translationese at the word and sub-word level (2016, methodology): “…of translationese has been carried out on corpora containing data from restricted domains, e.g., geopolitics (Baroni & Bernardini, 2006), parliament proceedings (Kurokawa et al., 2009; Koppel & Ordan, 2011; Volansky et al., forthcoming ), or technical and medical data (Ilisei et al., 2010).”
OCNLI: Original Chinese Natural Language Inference (2020, background): “…even if the translations are correct, they suffer from “translationese”, resulting in unnatural language, since lexical and syntactic choices are copied from the source language even though they are untypical for the target language (Koppel and Ordan, 2011; Hu et al., 2018; Hu and K¨ubler, 2020).”
Source and Translation Classification using Most Frequent Words (2013, methodology): “In the past, researchers have used comparable corpora to validate these translation properties (Baroni and Bernardini, 2006; Pastor et al., 2008; Ilisei et al., 2009; Ilisei et al., 2010; Koppel and Ordan, 2011).”