When adjectives modify nouns, usually they restrict the class of objects that the noun refers to.

For example:

Red car

A red car is, in particular, an instance of a car.

However, in specialty fields, we often encounter objects named by an adjective+noun combination, which are not (necessarily) actually instances of the object described by the noun itself. An example from mathematics:

Weak topology

A "topology" is a type of mathematical object satisfying certain axioms, whereas a "weak topology" may satisfy only a subset of those axioms. A "weak topology" need not be an instance of a topology. In mathematics, there are even sometimes adjective+noun combinations which refer to objects that are necessarily not instances of the class the noun describes.

Outside of mathematics, I can find only a few examples that are only arguably examples. Usually if I encounter something that looks like this phenomenon, it can be explained away by an archaic meaning of the noun that would turn the adjective+noun combination of the first kind described above, for example:

Christian science

I have two questions:

  1. Does anyone know a name for this linguistic phenomenon?
  2. Are there examples of this phenomenon which are non-technical and can't be turned into the first type described above simply by appealing to an archaic or unusual meaning of the noun?

2 Answers 2


First, note that we are not talking about compound lexemes here. A blackbird isn't necessarily the same thing as a black bird. 'Easter egg' is a single lexeme; 'fresh egg' comprises two lexemes. A compound noun is a lexeme distinct from the obvious 'contributing words' and may have a meaning not obviously related to either. Thus, to give a famous misnomer as an example, the English horn originated in Poland (where they use a similar misnomer!) and is a woodwind instrument not a horn.

Addressing solely double lexemes:

A fake painting is likely to be a painting, but a fake gun is not a [true] gun.

Imitation diamonds aren't diamonds.

A former president is probably not a president.

An alleged criminal may or may not be a criminal.

When an A N is not both A and an N, the phenomenon is known as 'non-intersection'.

An interesting case is with frustrated, which has dual senses.

A frustrated teacher may be a person who regrets never having achieved their ambition to be a teacher (so who is not a teacher) or may be a teacher who is frustrated.

In particular, when the adjective doesn't really 'modify' the following noun [group], it is known as a 'non-semantically-predicative adjective'. Thus a fake gun is not a fake type of gun; a former president is not a 'president who is former'. A mere youth is not a youth who is mere, a heavy smoker is not a smoker who is heavy, and an invalid toilet is not a toilet that is invalid. Note that, in some cases here, the noun referent is still valid: a mere youth is a youth, whereas a future king isn't a king (yet). This article by Elizabeth Coppock provides a valuable analysis.

  • 2
    Nice answer! :)
    – F.E.
    May 31, 2014 at 22:55

Another name for these is privative or non-subsective adjectives. The Coppock chapter provided by the previous responder gives more details, but to sum up briefly: many adjectives are subsective because they create a subset of the noun, e.g., black dogs are a subset of all dogs, good skiiers are a subset of all skiiers, etc. So the kind of adjectives you mentioned are non-subsective because a weak topology is not a topology, Christian science is not science, etc. (Within subsective adjects one can also make a further distinction between adjectives that are intersective—a black dog is both a dog and black—and ones that are non-intersective—a good skiier is not both a skiier and good, rather she is "good for a skiier" but she might also be a bad painter, who knows—although, as is usually the case with language stuff, this distinction sometimes gets quite fuzzy and there is debate over whether these represent qualitatively different classes of adjectives.)

While this is the classical definition, note that there are alternative views. For instance, Partee (2010) argues that these sorts of privative adjectives actually are subsective (i.e., that a fake gun actually is a gun, in a way).

  • We could do with more contributions of this calibre (sorry) on ELU. / I asked a question some years ago about whether a dead metaphor was a metaphor. This is along the same line as 'is a fake gun a gun?' Dictionaries tend to add [once-?] metaphorical usages as new senses, broadening definitions. This plays havoc with analyses. Oct 11, 2016 at 9:40
  • Then there is electrical engineers, which does denote a subset of engineers, but does not do so by choosing only the engineers that are electric, the way black dogs chooses only the dogs that are black. Judith Levi's book Syntax and Semantics of Complex Nominals analyzes this class. Jun 16, 2018 at 16:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.