What it is
Part of what a text says is about the text. The sentence announcing what a chapter will do, the pointer back to an earlier paragraph, the hedge marking a claim as provisional: none of it advances the subject matter, and all of it tells the reader how to take what does.
The layer has two halves and they behave differently. One organizes: it announces, points forward and back, names where a statement came from and glosses the text's own terms. The other positions the writer: it hedges, insists, evaluates, and lets a person appear behind the prose. The first half is what most people mean by signposting. The second is what measurement keeps finding on the move, and the easiest thing in the world to carry across at source density.
Three boundaries. A lexical bundle is admitted by a count over a genre, an item here by what it does to a reader, so the classes cross without coinciding. Where a connective also tells the reader how to take what follows, the tie belongs next door and the instruction here. And a move is the stage itself, while the sentence announcing that a stage begins is this record's.
Where the fixity comes from
Nothing prescribes how much of this a text carries. The evidence is one body of texts counted against a comparable one, and its limits are stated rather than inferred.
- Chou, Li and Liu set 500 translated English texts against 500 native ones across news, general prose, academic prose and fiction. The translated half uses fewer stance features, with hedges, boosters and attitude markers all lower, and varies more across genres than the native half.
- What that does not establish is a rate for any other pair. The translations are from Chinese, and no comparable figure was found for German, Dutch, French or Turkish. Nor does every feature move together: self-mentions run the other way in general prose, so a claim that translated text uses less metadiscourse is already too coarse.
- Pahor, Smodiš and Pisanski Peterlin work the other side: 150 Slovene research article abstracts against their published English translations, where over 62 per cent of the personal authorial references are gone, from about half in geography and sociology to over 80 per cent in mechanical engineering. What they document is recasting rather than thinning, since the reference is replaced by an impersonal structure and leaves nothing to count in its old position.
Consequences for translation
Density is a property of the target community, the source cannot report what it is, and a measured underuse is no licence to add: the counts describe a corpus, not the sentence in front of you.
- Carrying source density across is the standard failure, and it is invisible sentence by sentence: nothing is wrong anywhere, the text is only more hedged or more insistent than its genre leads readers to expect.
- A hedge is not decoration where a text has effect. Hardening one into a plain assertion changes what the writer is on record as claiming, and that is a finding rather than a preference.
- A reviser can flag a signpost the target genre does not use, and can flag it only as unidiomatic, unless the change touches what the writer is committed to.
- Paragraph numbers and numbered headings do part of the same organizing work by other means, and they are notation: a conventions matter cited here, not this record's.
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: Pahor, Smodiš and Pisanski Peterlin, example 2, a published abstract and its published translation. The study prints a literal gloss alongside, We formed the introductory part on the basis of facts: good English, and what a translator following the words would have written. The published version takes the writers out of the subject position instead.
FR>NL
- Source: La présente directive s'applique au traitement de données à caractère personnel
- Target: De bepalingen van deze richtlijn zijn van toepassing op de ... verwerking van persoonsgegevens
- Comment: Directive 95/46/EC, Article 3(1). Both sentences point at the text they belong to, which is the organizing half of this record, and the pointer differs: French makes the instrument the subject, Dutch its provisions. A translator composing writes Deze richtlijn is van toepassing, which is Dutch and is not what the act prints. The displacement that moved the writers out of the abstract moves the instrument out of this sentence.
References
- Hyland, K. (2005). Metadiscourse: Exploring Interaction in Writing. London and New York: Continuum.
- Pahor, T., Smodiš, M. and Pisanski Peterlin, A. (2021). "Reshaping authorial presence in translations of research article abstracts." ELOPE 18(1), 169 to 186, example 2.
- Chou, I., Li, W. and Liu, K. (2023). "Representation of interactional metadiscourse in translated and native English: a corpus-assisted study." PLOS ONE 18(7), e0284849.
- Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995 on the protection of individuals with regard to the processing of personal data and on the free movement of such data, Article 3(1), French and Dutch language versions.