What it is
Every text type sits at a height: how formal it is, how far it holds the reader off, how much of its author shows in it. Communities differ on where a given type sits, and the difference survives translation only if someone adjusts for it.
The system leaves the height open. The devices for suppressing an author, passive, impersonal construction, nominal subject, plural of modesty, exist in all the languages this section works with. What varies is how much of that apparatus a readership expects in an abstract, a judgment or a leaflet, and the expectation is strong enough to be measured.
Corpus evidence is used here for a different purpose than in the phraseology section, where frequency proves how fixed a wording is. That question belongs to frequency and attestation. Here frequency establishes a direction of adjustment across a pair, which is a claim about the height of a text type and not about the fixity of any string.
Where the expectation comes from
No authority anywhere rules how formal a text type must be. The warrant is a parallel corpus, and it is the strongest quantified evidence in this section.
Pahor, Smodiš and Pisanski Peterlin (2021) built a corpus of 42 500 words: 150 Slovene research article abstracts published between 1999 and 2019 with their English translations, thirty each from geography, linguistics, medicine, mechanical engineering and sociology. Personal authorial reference falls from 1.49 occurrences per hundred words in the Slovene to 0.55 in the English, close to a threefold reduction. Of the personal references in the source, 62.3 per cent are omitted outright, and the omission rate ranges by discipline from 43.7 per cent in sociology to 85.5 per cent in mechanical engineering. Interviews with four experienced translators attribute the interventions to fluency and readability rather than to instruction.
Institutional guidance sits on top of the measurement without replacing it. The English Style Guide ties register to text type in its introduction: accuracy and clarity are paramount in legislation, while documents for the general public must be immediately understandable. Its 1.1 adds that "very colloquial Irish/British usage should also be avoided", the readership being largely non-native. The Stijlgids Nederlands takes the parallel decision at 12.1.1. Both bind one department and neither states what English or Dutch expects in general.
What the measurement does not establish has to be said in the same breath, because the number is quotable and the inference from it is not. One pair, one direction, one genre, a small corpus. It establishes that authorial presence is reduced when Slovene abstracts are translated into English, by more in some disciplines than others. It does not establish that English is more impersonal than languages in general, says nothing about the reverse direction, and fixes no threshold to measure a text against. The earlier contrastive studies the article names as its starting point were not opened here and are not cited from it.
Consequences for translation
Height is adjusted rather than converted, which changes what can be asked of a reviser and of a checker.
- Two compliant renderings can differ. If both hold the target's height, the more personal and the less personal are both right, and a reviser who prefers one is exercising judgement rather than finding a defect.
- The direction of adjustment is a fact about the pair, not about either language alone. Reversing a measured direction because it feels symmetrical is the commonest way of getting this wrong.
- Discipline moves the adjustment more than language does. A range from 43.7 to 85.5 per cent inside one pair is wider than most cross-linguistic claims a brief will contain.
- Nothing here can be flagged as an error. A text at the wrong height reads as foreign, or as arrogant, which is one interviewed translator's word for it.
Examples
SL>EN
- Source: Uvodni del smo oblikovali na podlagi dejstev, raziskovalni del pa s pomočjo strukturiranega vprašalnika...
- Target: The introductory part was designed on the basis of facts, and the research work was carried out using a structured questionnaire...
- Comment: Example 2 of the study, from medicine. The Slovene first person plural is replaced by two passives and the authors leave the sentence. The expectation belongs to the English-reading discipline and to no instrument. A translator who did not know it would have written "we designed" and produced a grammatical abstract at the wrong height.
EN>NL
- Source: you
- Target: u
- Comment: The department's own rule at 12.1.1, which prefers the polite form before a general audience and the familiar form for a young one. The English pronoun carries no height, so the distance is set on the Dutch side by the audience. The holder is a named body binding its own texts, which is why the same English word takes the other Dutch form in a leaflet for schools.
References
- Pahor, Tadej, Martina Smodiš and Agnes Pisanski Peterlin. "Reshaping Authorial Presence in Translations of Research Article Abstracts", ELOPE 18(1), 2021, pages 169 to 186, DOI 10.4312/elope.18.1.169-186. Open access, read in full. The figures are the article's own and were not recomputed; the interview wording passes through the article's rendering of what the translators said.
- English Style Guide. Directorate-General for Translation, European Commission, eighth edition January 2016, file consulted 21 November 2025. Introduction and section 1.1 (Language usage). The file's pagination does not survive extraction, so no page is cited.
- Stijlgids Nederlands. Directorate-General for Translation, European Commission, fifth edition May 2025, last amended 21 October 2025. Section 12.1.1 (Gebruik van het persoonlijk voornaamwoord "u").