Translationese in Swedish novels translated from English

By Martin Gellerstam. 1986

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

Tagged Back-Translation (2019, methodology): “Under this lens, quality gains from TaggedBT/NoisedBT could be re-framed as transfer learning from a multi-task set-up, where one task is to translate simpler “translationese” (Gellerstam, 1986; Freitag et al., 2019) source text, and the other is to translate true bitext.”
BLEU Might Be Guilty but References Are Not Innocent (2020, background): “The product of human translation is assumed to be ontologically different from natural texts (Koppel and Ordan, 2011) and is therefore often called translationese (Gellerstam, 1986).”
MURI: High-Quality Instruction Tuning Datasets for Low-Resource Languages via Reverse Instructions (2024, unknown): “Thus, around 60% of their training data relies on translation which may introduce systematic translation artifacts known as transla-tionese (Gellerstam, 1986; Yu et al., 2022) in the model outputs.”
Identifying translationese at the word and sub-word level (2016, background): “The term translationese was coined by Gellerstam (1986), who compared texts originally written in Swedish with texts translated from English into Swedish, and concluded that the striking differences between them do not indicate poor translation but rather a statistical phenomenon, a systematic…”
Translationese and Its Dialects (2011, background): “Some (Baker 1993) have emphasized general effects of the process of translation that are independent of source language and regard the collective product of this process in a given target language as an „interlanguage‟ (Selinker, 1972), „third code‟ (Frawley, 1984) or „translationese‟ (Gellerstam, 1”
Transforming machine translation: a deep learning system reaches news translation quality comparable to human professionals (2020, background): “It has been observed that text translated from language X into Y has different properties (such as lexical choice or syntactic structure) compared to text originally written in language Y 36 .”
On the features of translationese (2015, background): “This may bring about, as noted by Gellerstam (1986), a different distribution of elements in translated texts, which he calls 'translationese', keeping it as a pure descriptive term (cf. Santos (1995)).”
Personalized Machine Translation: Preserving Original Author Traits (2016, background): “Authorial traits in translationese A large body of previous research has established that translations constitute an autonomic language variety : a special dialect of the target language, often referred to as translationese (Gellerstam, 1986).”
Translationese as a Language in “Multilingual” NMT (2019, background): ““Translationese” is a term that refers to artifacts present in text that was translated into a given language that distinguish it from text originally written in that language (Gellerstam, 1986).”
APE at Scale and Its Implications on MT Evaluation Biases (2019, methodology): “We formalize our APE model as a translation model from synthetic “translationese” (Gellerstam, 1986) text in one language to natural text in the same language.”
Unsupervised Identification of Translationese (2015, background): “r a dialect) of the target language, presumably reﬂecting both the artifacts of the translation process and traces of the original language from which the texts were translated (the source language). Gellerstam (1986) called this sub-languagetranslationese, and suggested that the differences between”
Is Textual Similarity Invariant under Machine Translation? Evidence Based on the Political Manifesto Corpus (2026, unknown): “However, research from 2018 to 2025 reveals that machine-generated translations still exhibit distinct linguistic footprint known as translationese [24, 31, 56].”
High-Dimension Human Value Representation in Large Language Models (2024, unknown): “The two question sources do not originally provide natural questions for LLMs, hence we employ Mixtral 8x7B to generate questions based on the situations provided in the two sources.”