What it is
A false friend in terminology is a pair of designations that look like a translation pair and are not. The resemblance is not incidental to the problem; it is the problem, because it supplies a candidate before any analysis is done and makes the analysis feel unnecessary.
This is the negative-evidence counterpart to the acceptability test. That test asks whether a candidate may be used; this record concerns candidates that are never tested at all, because the form has already answered the question. It is the failure mode a working translator meets most often, and one of the few in this field whose cost has been counted.
The detection method is the one that settles equivalence generally, applied where nobody thinks to apply it: compare the characteristics of the two concepts, not the shapes of the two designations. Cao supplies the reason it is necessary. Legal concepts are often system-bound, so a shared form across two systems is evidence of shared etymology and of nothing else.
Where the authorities stand
Nobody disputes the phenomenon, which is why the standing is descriptive. There is now measurement.
Prieto Ramos and Guzmán scored renderings of national court names across United Nations, European Union and World Trade Organization corpora on a three-point adequacy scale, where 0 marks a translation departing significantly from the essential meaning of the source term, 1 an acceptable but imprecise rendering, and 2 an accurate one. Almost every term and institution averaged above 1. The exception is the Spanish handling of magistrates' court, which averages 0.73 against 1.51 for French, with level 0 renderings accounting for 62.63 per cent of the Spanish sub-corpus. They attribute the collapse to the false friend magistrados inside literal renderings, noting that in Spanish the term names senior judges.
One detail complicates the moral, and reporting only the headline would mislead. The French figure is better because tribunal de première instance is a conceptual rather than a literal rendering, but the same phrase is itself scored 0 when used for a different court, the high court, in a United Nations text. The lesson is not that one language avoids the trap: it is that a rendering is adequate for a given source concept, and a solution that works for one court name is a level 0 error for another.
The constructive half comes from drafting rather than translation. Fuglinszky and Somssich record that in preparing the English version of the Hungarian Civil Code the drafters rejected unjust enrichment, the obvious common law counterpart, for jogalap nélküli gazdagodás. The Hungarian concept descends from the German ungerechtfertigte Bereicherung and turns on the absence of a valid legal title, whereas the common law concept requires the claimant to point to a particular unjust factor. Their conclusion is worth stating as they put it: unjust enrichment appeared as a false friend, and the correct translation was a formulation expressing the difference from the common law concept.
Consequences for translation
- Treat formal resemblance as a reason for suspicion rather than support. The cognate arrives without evidence, so it needs the evidence most.
- The measured damage is concentrated in literal renderings. Where a house style or a deadline pushes toward the literal, court and institution names are where that pressure costs most.
- A solution is adequate for a concept, not for a language. One phrase can be the right answer and a serious error inside a single corpus.
- Deliberate divergence is a legitimate move. Choosing a designation that does not resemble the obvious counterpart, to signal that the concepts differ, is a defensible drafting decision and should be recorded as one.
Examples
EN>ES
- Source: magistrates' court
- Target: magistrado, in literal renderings
- Comment: The measured case. Spanish magistrado names senior judges, so a literal rendering inverts the level of the court. Prieto Ramos and Guzmán score the Spanish handling at 0.73 average adequacy against 1.51 for French, with 62.63 per cent of the Spanish sub-corpus at level 0.
HU>EN
- Source: jogalap nélküli gazdagodás
- Target: unjustified enrichment
- Comment: The counter-move, made by drafters rather than translators. Unjust enrichment was available and was refused, because the Hungarian concept descends from the German one and turns on the lack of a valid legal title, while the common law concept requires an unjust factor to be pleaded. The designation is deliberately not the cognate, and its unfamiliarity is the point.
EN>FR
- Source: high court
- Target: tribunal de première instance
- Comment: A trap with no formal resemblance at all, which is why it belongs here. The same French phrase is the recommended rendering for one court name and a level 0 error in five occurrences of a United Nations text referring to New Zealand's High Court. Nothing in the forms warns anybody: only the two court systems do.
References
- Cao, D. (2007). "Legal Terminological Issues in Translation." Ch. 4 of Translating Law, 53 to 73. Clevedon: Multilingual Matters, at pp. 53 to 54 and p. 56.
- Fuglinszky, Á. and Somssich, R. (2020). "Language-bound terms, term-bound languages: the difficulties of translating a national civil code into a lingua franca." International Journal for the Semiotics of Law 33(3), 749 to 770, at pp. 754 to 755.
- Prieto Ramos, F. and Guzmán, D. (2023). "Measuring the quality of legal terminological decisions in institutional translation: A comparative analysis of adequacy patterns in three settings." In Ł. Biel and H. J. Kockaert (eds), Handbook of Terminology, Volume 3: Legal Terminology, 375 to 396. Amsterdam: John Benjamins, at pp. 378, 381 and 384 to 385.