What it is
Equivalence is not a yes or a no. Between two concepts named in different languages there is a scale, and the scale has a small number of recognized positions. Two concepts may be identical, sharing all their characteristics. They may intersect, sharing some and not others. One may include the other, so that the target concept is a subordinate or a superordinate of the source concept rather than its counterpart. Or there may be nothing there at all, which is the void.
Every framework here is a way of cutting that scale, and the disagreements are about where to put the knives rather than about whether the scale exists. What follows is immediate: the question stops being whether a candidate is the equivalent and becomes how far short of identity it falls, and in which direction.
One distinction has to be held throughout, because the literature itself slips on it. The scale grades concepts; what a translator writes is a designation. A pair of designations can be perfectly stable while the concepts behind them sit at intersection, and the reader has no way of telling.
Where the authorities stand
The field agrees that equivalence is graded and disagrees on how many cuts to make. That is why the standing is contested rather than standardized.
Felber's scheme, as Vezzani reconstructs it, has four cases: exact equivalence with all characteristics matching, overlap, inclusion where one set of characteristics is contained in the other, and non-equivalence. Nielsen works with full and partial equivalence plus zero equivalence, and his refinement is the one with the most practical bite: partial equivalents divide into an acceptable group, which he calls near and approximate, and an unacceptable group, which he calls remote. The cut inside partial equivalence is where a reviser's decision actually lies. Sandrini reduces the apparatus to two, absolute and relative, and is candid about the cost: absolute equivalence requires two legal concepts identical in all their conceptual features and in their extension, and once you leave that case what takes its place is, in his own words, a blurred notion of more or less whose criteria are hard to define.
Vezzani's 2025 reformulation answers that complaint by making the scale computable. She represents each concept as a set of characteristics and applies a Jaccard index, the size of the intersection over the size of the union, so that a pair receives a number rather than a label.
A real and confusing mismatch runs underneath all of this. The scholarship grades concepts; ISO 1087:2019 defines equivalence as a relation between designations in different languages that represent one and the same concept, which makes equivalence a property of the designation layer and admits no degrees at all. ISO 704:2022 carries a subclause titled Equivalence among the relations between designations and concepts, and its content is not publicly readable.
Consequences for translation
The scale is worth using because it converts a disagreement about a word into a disagreement about a position, which can be argued.
- Naming the position tells the reviser what kind of objection is being made. Intersection and inclusion call for different repairs: one needs a note, the other needs a shift up or down the hierarchy.
- Inclusion is directional. A target concept that subsumes the source is safe in one direction and a loss in the other, and the term record should say which.
- A computed figure is not a verdict. It records how much of the intension is shared, and says nothing about whether the unshared part is the one that matters. A single unshared characteristic can be decisive.
- The void is a decision point, not a failure. Once a pair is at zero, the work moves from selecting an equivalent to selecting a strategy.
Examples
EN>FR
- Source: ecosystem
- Target: écosystème
- Comment: Identity, and worth an entry for that reason. Vezzani uses this pair, alongside Spanish ecosistema, as a case where the concepts count as equivalent in the environmental domain. Identity is rare enough in legal terminology that a genuine instance is a useful calibration point.
EN>DE
- Source: cricket
- Target: Schlagball
- Comment: Intersection, quantified. Vezzani computes a Jaccard index of 0.55 for this pair: both are team games played with a ball and a bat, and they diverge on the number of players and on how runs are counted. The number is useful precisely because it is unflattering. A pair sharing more than half its characteristics is still not an equivalent.
NO>EN
- Source: hentekjøp
- Target: no equivalent designation
- Comment: The void. Nielsen gives this as a Norwegian term for which English has no equivalent term, and the entry is instructive because there is nothing to put in the target column. The scale has run out, and what the translator chooses next is a strategy rather than a term.
References
- Felber, H. (1984). Terminology Manual. Paris: Unesco and Infoterm, section 3.544.
- Nielsen, S. (1994). The Bilingual LSP Dictionary: Principles and Practice for Legal Language. Tübingen: Gunter Narr.
- Nielsen, S. (2023). "Legal lexicography and legal information tools." In Ł. Biel and H. J. Kockaert (eds), Handbook of Terminology, Volume 3: Legal Terminology, 432 to 457. Amsterdam: John Benjamins.
- Sandrini, P. (1999). "Legal Terminology: Some Aspects for a New Methodology." Hermes 22, 101 to 111, at p. 102.
- Vezzani, F. (2025). "Pour une formalisation du concept d'équivalence en terminologie." mediAzioni 46, A138 to A156.
- ISO 1087:2019. Terminology work and terminology science: Vocabulary. Geneva: International Organization for Standardization.
- ISO 704:2022. Terminology work: Principles and methods, 4th edition. Geneva: International Organization for Standardization, clause 7.7.