I'm struggling to understand the restrictive/non-restrictive distinction for relative clauses modifying indefinite noun phrases.

The distinction seems very clear for definite noun phrases. It's a question of whether the relative clause is necessary to identify the noun phrase from the discourse/context or not.

A: The president…
B: Aah, the president.
A: The president, who had just toured the country, returned to Washington.

So non-restrictive.

A: The man…
B: What man?
A: The man who won the election…
B: Aah, that man.
A: The man who won the election was instated in the white house.

So restrictive.

For non-definite noun phrases, it seems to me that the same distinction doesn't apply, at least not in the same way.

I can think of two kinds of indefinite noun phrases

  1. Non-specific

    I'm looking for a TV that fits into my bathroom.

    This use of "a TV" is similar to "any TV" or "TVs". In this case, the relative clause seems restrictive in the literal sense of the word. I'm looking for a TV, any TV, but with the restriction that it has to fit into my bathroom.

    I cannot think of a non-restrictive relative clause that can modify this kind of noun phrase, at least in English. I think one would usually rephrase.

    *I'm looking for a TV, which is entertaining. (Ungrammatical with a non-restrictive reading)
    I'm looking for a TV, because TVs are entertaining.

  2. Specific

    I see that there are still cases where a restriction makes sense, as for non-specific noun phrases, e.g.

    I saw a TV that fit into my bathroom.

    I.e. I might have seen 100 TVs that day, but this was the only one that fit into my bathroom.

    But there are also cases where this kind of restriction doesn't apply, e.g. in descriptions of scenery:

    On the beach there were chairs that people could sit on.

    This is the case that really bothers me. It seems that all the relative clause can do is add new information about the specific noun phrase (it's new to the context/discourse, so there is no identification involved), still I see people discussing them as if there is a meaningful restrictive/non-restrictive distinction. They tend to explain it with "How integral is the relative clause" and such, but that does not seem like a useful discrete test to me.

So finally my question:

Does the restrictive/non-restrictive distinction make sense for relative clauses modifying all specific indefinite noun phrases? If so, please explain it and provide syntactic or semantic tests that can distinguish, not vague tests like "is the relative clause integral to the sentence?".

  • First, if you want to make a non-restrictive relative clause, don't use that; use a wh-pronoun like who or which (what is not allowed as a relative pronoun). That-relatives are always restrictive (put a different way, that only occurs in restrictive relative clauses). This post may help clarify the distinction.
    – jlawler
    May 2, 2014 at 17:11
  • Does anyone on here know a language other than English that distinguishes morphologically between restrictive and non-restrictive relative clauses?
    – fdb
    May 2, 2014 at 19:29
  • @fdb - Acc to Comrie in 'Linguistic Universals and Language Typology', in Persian the head of a restrictive rel clause takes the suffix -i, but the other type does not. Comrie also states that most languages do not make this distinction, or do so only through intonation.
    – neubau
    May 3, 2014 at 2:42
  • I know about Persian, but it is not all that clear-cut: the word “ke” is a relative pronoun (“who, which”), but also a conjunction (“so that, when”, etc), so when appended to a noun without the suffix /-i/ various interpretations are possible. This has not really been properly investigated.
    – fdb
    May 3, 2014 at 13:08
  • 1
    @jlawler, thanks for your input, but it's not really helping. You're explaining how to identify something from its symptoms, but I want to know the root cause. I want to know what considerations, semantic, anaphoric, whatever, determine which type is used. For the specific subcase I'm unsure about, as mentioned.
    – dainichi
    May 7, 2014 at 3:39

1 Answer 1


The answer to your question is 'yes', and, of course the restrictive/non-restrictive distinction makes sense for indefinite NPs, too! Why shouldn´t it?

You can easily generate as many examples as you want of non-restrictive relative clauses accompanying indefinite NPs (be they 'specific' or non-'specific') by choosing NPs that name unfamiliar objects and, therefore, if used, are likely to require a bit of explanation from the speaker to help the hearer get an idea of what is referred to (which, as you know, is the characteristic function of non-restrictive relative clauses; these cases are exactly the kind of cases we might expect to need such explanatory clauses, there is nothing special at all to say about them).

For example, if I had reasons to assume that you are not likely to know what a guqin (or any other, from our perspective, 'rare' object) is, I might perfectly well say to you something like a), where the indefinite NP is non-specific (i.e., 'intensional', non-referential: such an object existed, but might not be available anymore), or one like b), where it is specific (i.e., referential: there is, indeed, an old guqin in my music room):

a) I am currently looking for a guqin, which is a sort of Chinese sitar with seven strings [+ e.g., but I want an old one and they are very difficult to come by].

b) I have just bought a guqin, which is a sort of Chinese sitar with seven strings [+ e.g., and I am fascinated by its sound, but it is an old one and it has cost me a fortune. I'd better not tell my wife!]

You can even build examples that contain such non-restrictive relative clauses before restrictive ones, as parentheticals, as in c):

c) Tom told me about a former patient, who had never before done any harm to anybody, that one day, for no apparent reason, went into a school with his rifle and killed thirteen children.

Again, there is nothing in c) worth special comment: the non-restrictive relative clause must be adjacent to its nominal 'antecedent', but since the NP in this case also contains a restrictive relative clause that ends in an NP (= thirteen children) that could itself be interpreted as such, to avoid ambiguity the non-restrictive relative clause must be inserted next to 'a former patient'. Since, if it were inserted without commas, it would be interpreted as restrictive and that is not the intended interpretation, there is no choice but to insert it as a parenthetical.

Finally, if you are familiar with the syntactic and semantic tests that generally distinguish restrictive from non-restrictive modification, as I assume, given the way you have worded your question, there is nothing special to add about heuristics: apply the standard tests to a), b) or c) if you are in doubt (but you need not be).

  • Thanks, good point about "... a guqin, which is ...", I missed this case. For c), I do not see what the qualitative difference is which makes one relative clause non-restrictive and the other restrictive. And no, I do not know the syntactic tests that apply to the particular sub-case that I'm asking about. As I mentioned, almost all tests I can find focus on definite noun phrases.
    – dainichi
    Jan 28, 2015 at 3:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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