Are pronouns in English resolved syntactically or semantically? Do they always refer to the closest matching noun?

A wikipedia article has these examples:

  1. We gave the bananas to the monkeys because they were hungry.
  2. We gave the bananas to the monkeys because they were ripe.
  3. We gave the bananas to the monkeys because they were here.

My understanding was that pronouns are resolved syntactically, and so sentence 2 is semantically incorrect (unless the author really meant that the monkeys were ripe) and sentence 3 is not ambiguous. Is this wrong?

Citations and references are most welcome.

  • 2
    Syntactic agreement does not mean the pronoun must agree with the closest noun.
    – Kosmonaut
    Dec 20, 2010 at 13:22
  • 2
    @Roman: No, they aren't. The index for the coreference would be different.
    – Kosmonaut
    Dec 20, 2010 at 15:24
  • 1
    @JSBangs: Well, Roman said that his understanding was that pronouns resolve syntactically. In any framework where this is thought to be true, you have to have some form of indexing, right? With that in mind, it is completely relevant to point out that his conclusion that (2) is semantically incorrect is untrue even in a world where pronouns resolve syntactically. It is important because it means that this set of sentences can neither prove nor disprove how pronouns are resolved.
    – Kosmonaut
    Dec 20, 2010 at 15:37
  • 2
    @Claudiu: One can only theorize how this actually works, but it is a fact that frameworks have existed for decades that deal with this exact process as a syntactic one. (Of course others look at it in varying degrees as semantic too.) See my comment to Roman lower down on the page. There are syntactic restrictions on what pronouns can link to what other nouns — semantics only tells you which possible configurations make logical sense.
    – Kosmonaut
    Dec 20, 2010 at 19:21
  • 2
    Example from article: "He asked Mary to wash John." Even if the context tells you that only Mary and John are there, and that the only person he could possibly be is John, you still can't link he and John because of a syntactic violation. It is just not grammatical.
    – Kosmonaut
    Dec 20, 2010 at 19:31

4 Answers 4


Kosmonaut seems to be spot on: semantics are absolutely necessary for tracing back personal pronouns to their antecedents. Sentences 1 and 2 are perfectly acceptable in formal and informal English, both old and new. The same applies to most other Indo-European languages. But it is important to distinguish between the different kinds of pronouns; relative pronouns, for example, have an unmistakable syntactic link to their respective antecedents.

A. I gave some bananas to the women, who were rather hungry.

This is standard.

B. ?I gave some bananas to the women, which were larger than expected.

Somewhat bumpy, but I'd say still acceptable, because the distance between pronoun and antecedent is short, and because the pronoun "which" is restricted to inanimate objects.

C. I gave the bananas to the monkeys that were there.

Somewhat ambiguous, but, lacking strong semantic clues, the reader will not have much doubt that "that" refers to the monkeys. In speech, the use of accents could make the pronoun a bit more flexible.

D. *I gave the bananas quickly, before the storm reached the forest, to the friendly but fastidious monkeys that I had kept stored in the freezer.

Impossible, because the distance between bananas and "that" is far too great.

  • @PeterShor: Thanks! I think the limits of antecedents become most clear when you are forced to a single antecedent that would make sense, but the sentence still doesn't work: then it is clear that distance or other intervening factors are too strong. Apr 26, 2011 at 21:59
  • For what it’s worth, I find sentence B unacceptable, too. My first thought was that it’s rather rude to say that the women were larger than expected, quickly followed by the awkwardness of using ‘which’ to refer to women. Only because of the context of the post am I even able to arrive at anteceding ‘which’ with ‘monkeys’. “I gave the bananas to the woman, which were larger than expected” causes even more brain meltdown in my head. Sep 8, 2013 at 11:37
  • @JanusBahsJacquet: Noted. Do you think it deserves a question mark? Sep 8, 2013 at 16:13
  • I think it does, yes. Sep 8, 2013 at 16:16
  • @JanusBahsJacquet: There. Sep 8, 2013 at 16:49

The first sentence makes sense to me so I am resolving that pronoun semantically. I don't think I'm unique in that regard.

Furthermore the second sentence has (at least) two potential interpretations:

We gave the bananas to the monkeys because [the bananas] were [ripe].


We gave the bananas to the monkeys because [the monkeys] [smelled strongly].

Most people are not going to pick the second interpretation without some other context. So semantics is extremely important in understanding the sentence and assigning meaning.

The third sentence is ambiguous. There isn't enough information to know if the bananas were conveniently located (i.e. "here") or the monkeys, or both.

  • "The first sentence makes sense to me so I am resolving that pronoun semantically." It also makes sense to me, although I'm resolving it syntactically ("monkeys" is the closest matching noun). My question is whether the grammar permits to assign your first interpretation of the second sentence to it.
    – user2978
    Dec 20, 2010 at 15:29
  • @Roman: when you say "closest matching noun", how are you matching the noun, and does this process change for sentence 2 and why? Dec 20, 2010 at 16:22
  • : by "matching" I understand pronoun-antecedent agreement of person, number and gender as described e.g. here: leo.stcloudstate.edu/grammar/pronante.html. Since this matching is purely syntactical and I fail to see any syntactic difference between sentences 1 and 2, the process should not change.
    – user2978
    Dec 20, 2010 at 16:31
  • @Roman: I don't see why "they" has to match "The monkeys" and can't match "The bananas". That page even says about matching that you need to know "It is the best source available. What source is that?" Here, "best" is resolved by inferring the meaning. If the meaning doesn't make sense (or is unlikely, e.g. ripe monkeys) then you infer a better meaning (hungry monkeys, ripe bananas). (Aside: that page is wrong about "they" being always third-person-plural. It can also be third-person-singular.) Dec 20, 2010 at 16:44
  • The word that I'd use here is 'pragmatically'. How would most people understand / use it (when neither syntax nor semantics clearly disambiguate). Aug 18, 2016 at 18:45

My understanding was that pronouns are resolved syntactically, and so sentence 2 is semantically incorrect (unless the author really meant that the monkeys were ripe) and sentence 3 is not ambiguous. Is this wrong?

Yes, this is wrong. Intuitively, it is clear that none of these sentences are incorrect, and sentence 3 is ambiguous.

Linguists who believe that pronouns are linked to their antecedent in syntax do not claim that the antecedent must be the immediately preceding noun. So, concluding that sentence 2 must link "they were ripe" with "monkey" is operating under a framework that does not resemble any framework that claims that pronouns are resolved syntactically. These frameworks would also not make the claim that sentence 3 is unambiguous without semantic intervention. They would say that there are different indexes applied to the pronouns and nouns, which they would also consider to be part of the syntactic structure.

I do not even necessarily agree with the idea that pronoun resolution is syntactic, but these sentences do not disprove those theories.

  • But those indexes must be somehow inferred from the sentence, right? (Since they are not a part of the language.) And if they belong to the syntactic structure, then they must not depend on the semantic meanings of the words in the sentence, just their syntactic roles. So I can't see how pronouns and nouns in sentences 1 and 2 can be unambiguously assigned different indexes. Could you please recommend some (online) reading regarding the frameworks you mentioned?
    – user2978
    Dec 20, 2010 at 16:12
  • @Roman: Those indexes, under that kind of a framework, are part of the language. For more info, you would want to read about [binding](en.wikipedia.org/wiki/Binding_(linguistics)) and go from there. Semantics does have to be involved somewhere in order for the speaker to decide which links are the correct ones, but semantics must also be involved in deciding whether to say "dogs like bones" or "bones like dogs", even though the process is still syntactic.
    – Kosmonaut
    Dec 20, 2010 at 17:03
  • I'm new. May I respond to Kosmonaut's answer? "None" takes a singular verb: "none is." Same with "each" and "every." I am not a linguist. I'm not sure about Syntax and Semantics, but I think I have a handle on English syntax. What about "between...and vs "between to?" I think any one who uses the latter should be sentenced to 10 years with a dozen ripe monkeys.
    – user7812
    Apr 26, 2011 at 15:06
  • @Chuck: No, none does not necessarily take a singular verb. Also, please do not leave comments as answers. Thanks.
    – RegDwigнt
    Apr 26, 2011 at 15:18

Pronouns can be syntactically ambiguous, but relative clauses can't (at least when the subject is a noun). The difference can be subtle:

  1. "The secretaries will destroy the documents after being digitized."
  2. "The secretaries will destroy the documents after they've been digitized."

Sentence 1 clearly says it's the secretaries that will be digitized. Sentence 2 doesn't. And when there are two relative clauses in the same sentence, lots of people make mistakes. I'm pretty sure these sentences don't mean the same thing:

  1. "She'll tell us why the program keeps crashing after she gets back from lunch."
  2. "She'll tell us after she gets back from lunch why the program keeps crashing."

Your Answer

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