Corpus-based comparison of the Corpus of Translational English (COTE) with the matched native-British-English corpus FLOB across 96 lexico-grammatical features isolated an 18-feature profile of translational English, organized by exploratory factor analysis around a dominant 'translation vs non-translation' dimension, and read as Simplification and Explicitation tendencies plus a tentative Formality tendency (over-use of relative structures and pied-piping, under-use of contractions, particles and stranded prepositions). Whether this profile is a crosslinguistic property of translated language or a language-, direction- and register-specific artifact of one English corpus pair remains open. Subsequent and companion work (2015-2026), concentrated on English-Chinese and on single registers, yields mixed, register-conditional results and increasingly favors source-contact and translator-behavior explanations; no typologically diverse confirmatory validation exists, and the formality tendency has never been tested with controls for source-language formality, editorial practice, and translator conventions.

Background

Corpus-based translation studies compares large machine-readable collections of translated texts with collections of originally written texts in the same language. Its central theoretical construct is the 'translation universal' (TU): a recurrent feature or tendency of translated text, as a text variety ('translational language' or 'translationese'), relative to non-translated text in the same target language. The most-debated tendencies are Simplification (translated texts are lexically and syntactically simpler than native texts) and Explicitation (translated texts spell out material that the source leaves implicit); a further candidate is elevated Formality of translated language. The standard method uses 'comparable corpora': a corpus of translations into a target language set against a corpus of native texts in that language, matched for sampling frame, genre, and register (register meaning the situational variety of language use, e.g. news, fiction, academic prose). In this framework, translated text is often modeled as a 'third code' shaped by cognitive load, risk-averse translator behavior, interference from the source language ('shining-through'), and normalization toward target-language norms.

Hu, Xiao and Hardie (2019) applied a multi-feature statistical model, in the tradition of register-variation studies that measure dozens of lexico-grammatical features per text and reduce them by factor analysis (a technique that groups co-varying features into dimensions), to English. They compared COTE (500 texts translated into English from many source languages) with FLOB (the Freiburg-LOB corpus of native British English, 500 matched texts) on 96 lexical, syntactic and textual features. Eighteen features met their joint criteria for a 'typical feature of translational English': a significant Mann-Whitney U difference, consistent direction across categories, registers and genres within stated tolerances, and a substantial loading on the factor-analysis dimension that most strongly separated translated from non-translated texts (Factor 2, interpreted as a 'translation vs non-translation' dimension; its mean factor-score difference was about four times that of any other factor). The 12 over-represented features included grammatical explicitness, Wh-determiners, the ten most frequent words, pied-piping relative clauses (relative clauses in which the preposition moves with the relativizer, as in 'the desk on which...'), average sentence and paragraph length, prepositions, articles, demonstratives, conjuncts, and sentence relatives; the 6 under-represented features were particles, stranded prepositions, subordinator-that deletion, lexical density, punctuation, and contractions. The authors grouped this profile into Simplification-like and Explicitation-like tendencies, and tentatively suggested a third tendency toward more formal syntactic realizations in translations.

The authors themselves delimited what the study established. The profile was derived from a single 1,000-text corpus pair with no hold-out, cross-validation, or replication on an independent corpus; the factor labels were exploratory interpretations needing confirmatory models; several features showed register- or genre-specific exceptions within the stated tolerances; the mapping from surface features to communicative functions was declared tentative; and the formality suggestion was left untested against obvious confounds such as source-language formality and publisher/editorial practice. Their companion monograph (Xiao and Hu, 2015) ran the mirror-image comparison for translational Chinese (the Zhejiang University Corpus of Translational Chinese, ZCTC, set against the native-Chinese Lancaster Corpus of Mandarin Chinese, LCMC) and re-evaluated the TU hypotheses 'from the Chinese perspective', but its universals discussion explicitly confines generalization to English-to-Chinese translation because of corpus composition. Whether the English-derived profile reflects properties of translated language as such, or of particular languages, translation directions, registers, and corpus designs, is therefore unresolved.

Problem Statement

Determine whether the distinctive multi-feature profile of translational language identified by Hu, Xiao and Hardie (2019) - the Simplification and Explicitation tendencies and the tentative Formality tendency - is a crosslinguistic property of translated text or a language-, translation-direction- and register-specific manifestation. The question concerns the profile as measured on balanced comparable corpora of translational and non-translational text in typologically diverse target languages beyond English (and beyond the already-studied English-Chinese pair), with matched register and genre composition and a controlled source-language mix, and with the feature inventory operationalized, or principled cross-language analogues of the English features defined, so the multi-feature profile is measurable in each target language. For each tendency, whether it holds as a crosslinguistic universal, a language-specific manifestation, or a condition on named factors (source language, typological distance, translation direction, register, mediation mode) is part of the question, as are the mechanisms (translator risk aversion and conservatism, source-language shining-through, normalization toward target-language norms, editorial intervention) that produce whatever profile is found. The three tendencies are posed jointly, as a single multi-feature profile.

References

  1. Xianyao Hu, Richard Xiao, Andrew Hardie (2019). How do English translations differ from non-translated English writings? A multi-feature statistical model for linguistic variation analysis. Corpus Linguistics and Linguistic Theory 15(2): 347-382. DOI: 10.1515/cllt-2014-0047. https://doi.org/10.1515/cllt-2014-0047
  2. Richard Xiao, Xianyao Hu (2015). Corpus-Based Studies of Translational Chinese in English-Chinese Translation. Springer, Berlin/Heidelberg. DOI: 10.1007/978-3-642-41363-6. https://doi.org/10.1007/978-3-642-41363-6
  3. Kanglong Liu, Muhammad Afzaal (2021). Syntactic complexity in translated and non-translated texts: A corpus-based study of simplification. PLOS ONE 16(6): e0253454. DOI: 10.1371/journal.pone.0253454. https://doi.org/10.1371/journal.pone.0253454
  4. Juqiang Chen, Hui Chang (2023). Testing simplification in translated and creative writing texts: The case of Robert van Gulik's Judge Dee detective stories. Across Languages and Cultures 24(2): 268-282. DOI: 10.1556/084.2023.00375. https://doi.org/10.1556/084.2023.00375
  5. Xu Cui, Dechao Li (2022). Exploring genre variation and simplification in interpreted language from comparable and intermodal perspectives. Babel 68(5): 742-770. DOI: 10.1075/babel.00289.cui. https://doi.org/10.1075/babel.00289.cui
  6. Jinru Luo, Dechao Li (2022). Universals in machine translation? International Journal of Corpus Linguistics 27(1): 31-58. DOI: 10.1075/ijcl.19127.luo. https://doi.org/10.1075/ijcl.19127.luo
  7. Hai Hu, Sandra Kübler (2020). Investigating translated Chinese and its variants using machine learning. Natural Language Engineering 27(3): 339-372. DOI: 10.1017/S1351324920000182. https://doi.org/10.1017/S1351324920000182
  8. Tove Larsson, Luke Plonsky, Gregory R. Hancock (2020). On the benefits of structural equation modeling for corpus linguists. Corpus Linguistics and Linguistic Theory 17(3): 683-714. DOI: 10.1515/cllt-2020-0051. https://doi.org/10.1515/cllt-2020-0051
  9. Dandan Sheng, Xin Li (2024). A multi-dimensional analysis of interpreted and non-interpreted English discourses at Chinese and American government press conferences. Humanities and Social Sciences Communications 11: 472. DOI: 10.1057/s41599-024-02968-9. https://doi.org/10.1057/s41599-024-02968-9
  10. Cui Xu, Dechao Li (2024). More spoken or more translated? Exploring the known unknowns of simultaneous interpreting from a multidimensional analysis perspective. Target 36(3): 445-480. DOI: 10.1075/target.22028.xu. https://doi.org/10.1075/target.22028.xu
  11. Jia Li, Xianyao Hu (2024). A better or worse communicator? Comparing human and machine translation in source language shining through across registers. Lingua 312: 103834. DOI: 10.1016/j.lingua.2024.103834. https://doi.org/10.1016/j.lingua.2024.103834
  12. Jia Li, Xianyao Hu (2025). Is human translation more conservative than machine translation? International Journal of Corpus Linguistics 31(2): 193-227. DOI: 10.1075/ijcl.24048.li. https://doi.org/10.1075/ijcl.24048.li
  13. Jia Li, Yuan Gao (2025). Variability of cohesion and coherence in Chinese-to-English translation: measuring the effect of translation variety and register divergence. Humanities and Social Sciences Communications 12: 1526. DOI: 10.1057/s41599-025-05814-8. https://doi.org/10.1057/s41599-025-05814-8
  14. Lin Shen, Haidee Kotze (2025). Linguistic indicators differentiating translated and untranslated diplomatic discourse: A diachronic analysis of the United Nations General Debate (1946-2022). Lingua 320: 103933. DOI: 10.1016/j.lingua.2025.103933. https://doi.org/10.1016/j.lingua.2025.103933
  15. Dandan Sheng, Xin Li (2026). Translated Versus Nontranslated Languages Again: A Multidimensional Analysis of State Addresses of Nine Countries. SAGE Open 16(3). DOI: 10.1177/21582440261463791. https://doi.org/10.1177/21582440261463791
  16. Jia Li, Yunhao Pan (2026). Does language prestige hypothesis hold true across typologically distant languages? A corpus-based study comparing shining-through effect in English-Chinese bidirectional translations. PLOS ONE 21(7): e0353687. DOI: 10.1371/journal.pone.0353687. https://doi.org/10.1371/journal.pone.0353687
  17. Yanmeng Liu (2026). Translation as language contact: a multidimensional perspective of syntactic variation. Frontiers in Language Sciences 4: 1712405. DOI: 10.3389/flang.2025.1712405. https://doi.org/10.3389/flang.2025.1712405
  18. Xianyao Hu, Man Zheng (2024). Corpora and Cross-Linguistic Research of Translation Universals. In: The Routledge Handbook of Corpus Translation Studies (1st edition). Routledge. DOI: 10.4324/9781003184454-16. https://doi.org/10.4324/9781003184454-16
  19. Muhammad Afzaal, Baorong Huang, Dina Abdel Salam El-Dakhs (2025). Decoding the digital: a corpus-based study of simplifications and other translation universals in translated texts. Frontiers in Psychology 16: 1517107. DOI: 10.3389/fpsyg.2025.1517107. https://doi.org/10.3389/fpsyg.2025.1517107
  20. Pengfei Bao (2025). Validation of Translation Universals Based on Comparable Corpus: Simplification and Explicitation. International Linguistics Research 8(1). DOI: 10.30560/ilr.v8n1p21. https://doi.org/10.30560/ilr.v8n1p21