I am confused in how to distinguish a syntactically oriented language exercise from semantically oriented language exercise.

For example, suppose a teacher gave the English exercise below to his student:

Question: Consider the sentence below.

"When I was at the party, I didn't drink much but I had a few sodas with ice. They were delicious. They didn't have many snacks available, but I found some chips and candy."

What does "they" refer to?

(from: https://www.helpteaching.com/questions/Parts_of_Speech)

Does this exercise test syntactic (because it's about finding meaning of the pronoun) or semantic (because the student would need to understand the sentence to solve this problem) knowledge of a student?

Thank you,

2 Answers 2


Does this exercise test syntactic (because it's about finding meaning of the pronoun) or semantic (because the student would need to understand the sentence to solve this problem) knowledge of a student

The question reveals a misunderstanding about the difference between syntax and semantics, and a more fundamental problem with the task of separating syntax and semantics. The basic idea is that there are certain aspects of sentence structure which are just about classes of words and how they combine (that's syntax); and then there are aspects of sentence structure which are about the meaning of sentences (that's semantics). In English, "John walked", "The boy walked", "Some wealthy women walked", "The rock walked" and "Individuality walked" are all syntactically well-formed: they have a general form, being composed of a noun-like expression and a predicate composed of just a verb. The last two examples have a problem that is related to semantics, that in the first case the situation described can't happen, and in the second you can't even imagine what that situation would even be like.

There are some cases where pronoun distribution and interpretation reference is somewhat syntactically influenced. For instance, in English you can't say "*Myself walked", "*Himself walked", also you can't say "*He likes myself", "*I like himself". You can't say "*I like me", "*You like you", "*We like us", and you also can't say "*She likes himself" or "*He likes herself". This inspires a rule to the effect that a reflexive pronoun has to be preceded by a subject that refers to that pronoun. "Reference" is a property of meaning. This is an example of a general problems in syntax and semantics, often referred to as "binding". The syntactic part of the puzzle is answering questions like "in what structural context can a reflexive pronoun appear?" which could sort out the above distinctions. But also there are pairs like "He saw himself" and "He saw him", where in the former the person being seen is the same individual as the person seeing, and in the latter the person being seen is a different person from the person seeing (i.e. the sentences mean different things: the pronouns in object position have different referents).

Syntactic theory has tended to resolve these issues by assigning indices to words, then stating syntactic rules for using the same index versus different indices. A reflexive pronoun appears iff it has the same index as the subject (very rough approximation). An ordinary pronoun can generally have any index, meaning that it can refer to some nominal term within the sentence, or to can refer to an arbitrary individual not mentioned in the sentence. There are some syntactic rules governing these options, so that in "Kim claim that she was sick", "she" could refer to Kim or to some other person not mentioned, but in "She claimed that Kim was sick", "she" cannot refer to Kim (because of the syntactic relationship between the words).

The thing(s) that a pronoun refers to is mostly a semantic problem, but there are syntactic restrictions on where pronoun-like expressions can refer to the same versus different individuals. The most solidly-syntactic distribution is use of reflexive pronouns as subject and direct object, and even then there are cases like "Screw you" vs. "Screw yourself". In the sentence "They were delicious", there is no nominal expression that "They" can be co-indexed with, so syntactic relations are irrelevant in regulating what "they" refers to. Given this short context, we can use practical reasoning to figure out that "they" most likely refers to sodas. I can rewrite the paragraph to make it refer to "shrimp chips". In that case, you could call it "semantics", but many people would say that this is "pragmatics" (practical reasoning about what speakers intended).


It's clearly expected that the answer is


although I don't think a fine line can be drawn. So I understand that the question can be confusing. It needs to be asked because some facts about syntax hold true invaruably of the semantics, and vice versa, so we want to disambiguate. Specifically:

The reader needs to apply sentence level syntax to find the verbs corresponding to each occasion of "they". Then it's a matter of semantics to understand what agent could fulfill that verb, and a matter of text-level syntax to find the nearest noun that can fulfill that role.

Sodas don't possess anything, so the second occasion of "they", that is "they didn't have many snacks available", must refer to something else. What else? Well, that's not really a matter of syntax for various reasons. One could conclude that backtracking is not a sentence-syntactic operation and chalk it up to pragmatics and context. One could also say that part of the semantics of they is introducing indefinite subjects. That is, we haven't even been told who threw the party.

PS: Although the word "party" derives from the description of a group of people that usually has a leader, semantics only can extract this throug contextual inference, in which syntactic production only plays a minor role, covering the meaning up under an adverb of place, "at the party", that is commanded by the verb of movement, arrive. The referent of the indefinite pronoun is hidden in the text-level semantics and needs to be extracted semantically.

PPS: Also, as a metter of semantics, we can tell from the negative polarity item, "didn't", that it doesn't matter who did not provide for many snacks: Nobody had many available for the speaker. We later infer that they had chips, but we don't really care who, as far as syntax is concerned. Thus we merely infer through syntax, that I ate chips at the party.

What's more, I'm not sure, but I'd guess in a regional jocular "I ate me some chips" there's a mediopassive. The chips are their own, not mine or anyones. Only through eating do they become a part of me. The category does not make much sense for English, though, because the verbs are not morphemically inflected.

Hope I didn't confuse.

  • 1
    BTW "didn't" isn't a negative polarity item.
    – user6726
    Commented Nov 16, 2019 at 16:19
  • 'A metter of semantics'? What is that, 'metter'? Metric? Matter? Meter?
    – Mitch
    Commented Nov 18, 2019 at 3:16

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.