One object, one name

Terminology work is not a dictionary project. It is a decision-making process: within this document set, within this organisation, within this product line, this object has exactly one name, and every writer, translator and machine-translation engine that touches the material will use it. The decision is deliberate, it is recorded, and it is enforced. Everything else — the synonyms, the near-synonyms, the legacy names from the last product generation, the label someone coined in a different department — is suppressed.

That may sound administrative. It is, in the strictest sense. It is also the invisible structure that prevents a maintenance technician from looking for a part under the wrong name, a court translator from rendering two different terms as a single phrase, or a drug regulator from misidentifying a compound because two teams used the same word for different things. The stakes are not hypothetical. They are the reason the discipline exists.

The founding argument

The theoretical basis for modern terminology work was laid in Vienna, where Eugen Wüster completed his doctoral thesis in 1931. Wüster's argument, later formalised as the General Theory of Terminology, held that in technical language the relationship between concept and term should be one-to-one: one concept, one term; one term, one concept. Synonymy — two terms for the same concept — was waste. Polysemy — one term carrying two concepts — was ambiguity dressed as efficiency. Both, in technical communication, were failure modes.

Wüster went on to found Infoterm, the International Information Centre for Terminology, established in Vienna under UNESCO auspices, which became the institutional home for terminology methodology through the latter half of the twentieth century. The International Organization for Standardization built on this foundation: ISO Technical Committee 37, which Wüster influenced directly, has produced a suite of standards governing how terms are defined, recorded and exchanged. The intellectual architecture he built in the 1930s is still the skeleton inside every professionally managed term base.

What the work actually involves

A terminology project begins with the domain: not language in the abstract but a defined subject area within a defined organisation. The terminologist — or the team doing this work — identifies the concepts that need stable names. A concept is not a word; it is a thing in the world, or a relation, or a process, that the organisation needs to talk about without ambiguity. Once the concept is defined, usually in a formal definition statement plus a set of delimiting notes, one designation is chosen. That choice is documented in a term base: a structured database in which each entry records the preferred term, its definition, the domain, the source of the decision, and any deprecated alternatives.

The deprecated alternatives matter. They are not deleted; they are marked as forbidden or non-preferred, so that a translator looking up the legacy name finds it, is told it is not the chosen term, and is pointed to the correct one. A term base that lists only the approved terms is half-finished, because writers and translators arrive via the word they already know, not via the correct word they are trying to find.

Multilingual terminology compounds the decision. For a multilingual institution — the Directorate-General for Translation in Brussels works across twenty-four official languages of the European Union — each concept requires a designated term in every working language, and those designations must be made independently: not a word-for-word translation of the source-language term, but the established term in the target language for the same concept. Where no established term exists, one must be coined, defined and approved. The temptation to borrow directly from the source language — to produce a calque or a transparent loan — is real and often wrong, because the borrowed form may carry conceptual baggage from the source language that does not map correctly onto the target-language concept.

The enforcement question

Terminology decisions are only as good as the workflow that carries them to the point of writing. A term base that lives in a shared drive, consulted when someone remembers to open it, is not enforcing anything. Effective terminology management means integration: the approved terms are inside the translation environment and the authoring environment, surfaced automatically when a writer types a word in the forbidden list or when a translator encounters a source-language term with a recorded entry.

IATE, the Interactive Terminology for Europe database that serves the EU institutions, is a public-facing example of this: a terminological resource designed to be queried during the translation act, not afterward. The Court of Justice of the European Union maintains its own controlled vocabulary for legal concepts, for reasons that go beyond preference — a term used in a judgment has legal force, and the same concept must carry the same designation across languages and across time. A synonym is not a stylistic choice in that environment; it is a potential source of interpretive divergence.

Polysemy — one term carrying two concepts — was ambiguity dressed as efficiency

Outside the institutional context, enforcement depends on whatever lever the organisation can use. In technical documentation, style guides with binding terminology lists are the standard instrument. In regulated industries — pharmaceuticals, aerospace, medical devices — regulatory submissions impose terminology constraints from outside: the authority's published vocabulary defines the terms, and deviation from it signals non-compliance before the substance of the document is even read.

What makes a good definition

The definition is the part of the terminology record that does the most work and receives the least attention. A term can be memorised and used without understanding; a good definition cannot be faked. The ISO approach, following Wüster, requires that a definition be intensional: it must state the superordinate concept (the category the defined concept belongs to) and then the distinguishing characteristics that separate it from other members of that category. "A device that measures temperature" is not an intensional definition of a thermometer; it is a rough gloss. "An instrument that determines temperature by measuring a physical property that changes predictably with temperature, such as the volume of a liquid or the electrical resistance of a conductor" comes closer — it gives the concept a place in a logical structure.

In practice, definitions in term bases are often shorter than this, because the audience is specialists who share background knowledge. The criterion is still precision: the definition must include exactly the concepts that the designated term covers, and exclude the concepts it does not. Where a definition is uncertain, the terminology record is uncertain, and the designation built on it is unstable.