How do linguists define the idea of a grammatical subject?

The only place I could find on the Internet about linguists' definition of a grammatical subject was at the SIL glossary, here. (http://www-01.sil.org/linguistics/GlossaryOflinguisticTerms/WhatIsASubject.htm).

According to this definition, grammatical subjects often have...

  • "The grammatical characteristics of the agent of typically transitive verbs"
  • "The grammatical characteristics of the single argument of intransitive verbs"
  • "A particular case marking or clause position"
  • "The conditioning of an agreement affix on the verb"
  • "The capability of being obligatorily or optionally deleted in certain grammatical constructions, such as the following clauses: Adverbial, Complement, Coordinate"
  • "The conditioning of same subject markers and different subject markers in switch-reference systems"
  • "The capability of co-reference with reflexive pronouns"

The definition doesn't imply that all grammatical subjects in all languages have all of these characteristics.

But I was wondering whether this definition passes muster among the professional linguists on this list. If not, is there are better source of information about what grammatical subjects are?

  • see Levin & Rappaport's Argument Realization and references therein. also possible interesting is Hopper&Thompson (1980)
    – user483
    Jul 31, 2013 at 2:39
  • 2
    I don't think professional syntacticians spend much time worrying about definitions. If you have an urge to call something a subject, for whatever reason, just dub it "subject", and start looking at consequences.
    – Greg Lee
    Nov 1, 2017 at 3:02
  • 1
    @GregLee No way! Defining your terminology carefully is essential so that readers, who may have subtly different definitions in mind, can meaningfully understand your works. Anyone who doesn't bother to give or at least reference their definitions isn't worthy the label of "professional" syntactician.
    – curiousdannii
    Nov 1, 2017 at 7:41
  • I suppose that linguists will disagree among them. One that I can remember, Mário Alberto Perini, says somethink like "subject is that thing that makes the verb inflect". He also denies that subjects are (usually) agents; he says that "agent" is a semantic category, and "subject" is a sintactical one. In this he is right: in Portuguese at least, the subject is more often than not something else than the agent, many times even the pacient. But I suspect that this subject-verb agreement is a quirk of IE languages, not a linguistic universal. Nov 3, 2017 at 13:38
  • Unclear writing, close.
    – Lance
    Jan 15, 2020 at 12:02

2 Answers 2


Notice that the glossary say subjects often have these characteristics.
I.e, the characteristics are common, but not definitive.

In fact, there is no universal characteristic "Subject", except insofar as it may be demanded by various theories of grammar. In English, the concept of "subject" is very well-defined in most cases, but there are a few marginal situations where it's not so obvious, for instance with dummy-insertion rules like There-Insertion and Extraposition

  • There are two unicorns in the garden. (the verb agrees with unicorns)
  • There appear to be two unicorns in the garden. (there undergoes Subject-Raising)

  • It was unfortunate that he vomited during his aria. (That-clause subject or It subject?)

But in other languages the concept of Subject may be nonexistent, because it depends on a particular orientation of grammatical relations that doesn't occur everywhere.

What English (and Indo-European languages generally) mean by Subject is a relation between some noun phrase in a clause and the verb of that clause. In English, every clause has a Subject. We contrast the Subject with the Direct Object, which is specially marked in many languages (this is the "accusative case", of which English has only a few remnants like him and them), and which occurs along with the Subject, but only in some clauses; we call these clauses Transitive.

So every English clause has a Subject, but Direct Objects only occur in some clauses.
This is called an "Accusative" system. There are other systems.

One such system is the "Ergative" system, which is common in Mayan languages, Australian languages, Caucasian languages, and many others. In an Ergative system every clause has one kind of noun phrase (called an "Absolutive") and some clauses (the transitive ones) also have a specially-marked noun phrase (called an "Ergative") as well as an Absolutive. Not much different, except that the Ergative noun phrase in a transitive clause is what English would call the "subject", and the Absolutive is the "object".

This is very confusing for those who don't realize that there are other ways of marking NPS.
There's a diagram of the differences here.

  • 1
    I'm not sure how accurate it is to call "there" a dummy subject. It seems to make much more sense to analyze it as an adverb that mimics a subject syntactically in a syntax where the focal subject is forced to the right (toward the focal position at the end of the sentence). Aug 2, 2013 at 6:50
  • The third example seems like it could very plausibly be analyzed as a case of final correction that has become grammaticalized - "It was unfortunate - that he vomited during his aria". The complement clause is simply a rewrite of the dummy subject, not an independent nominal phrase. In this case the syntactic motivation is to move the cumbersome subject phrase out of the way to reduce ambiguity - you'll note that "That he vomited during his aria was unfortunate" has the same meaning but is more cumbersome to parse. Aug 2, 2013 at 6:56
  • Since it's too late to edit my second post: my (second) point was that it's nonsensical to ask whether "it" or "that ..." is the subject of the third sentence. They are one in the same. Aug 2, 2013 at 7:05
  • 1
    Only if they're referential. Dummies are not.
    – jlawler
    Aug 2, 2013 at 17:08
  • 2
    @JustinOlbrantz But "subject" is a syntactic term. If it is syntactically a subject it IS a subject. What part of speech it is doesn't have any bearing on anything at all (although there is certainly not an adverb in such situations) Dec 22, 2014 at 2:07

Yes, you are right, in traditional linguistics that is, linguistics before generative grammar, the idea "subject" is not well-defined, so every definition given by the traditional linguist can not cover all the cases.

So, generative grammar does not use it in the traditional sense, and gives a definition relating D-Structure. You may see some general works of generative grammar for the definition.

  • 1
    Some generative grammars use it -- Relational (Arc Pair) grammar makes "subject" a basic term. Only a few varieties of generative grammar use "D-Structures" to indicate subjecthood.
    – jlawler
    Oct 9, 2013 at 15:35

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.