Subjects are often described as the person or thing that “does the action” in a sentence. Similarly, objects are commonly understood to be the thing undergoing the action of the sentence. These are simple and intuitive definitions, but unfortunately very inaccurate. The most obvious counter example to this is the passive voice:
Alice ate a poisoned donut.
A poisoned donut was eaten by Alice.
In the passive sentence, the second one, the subject is a poisoned donut, but it certainly isn’t doing the action. It’s in fact undergoing the action of being eaten, and undergoing the action is supposed to be what an object does.
Another case where the definitions don’t work is the set of verbs known as “ergative”. These are verbs that can be either transitive or intransitive, and where the object of the transitive form is the subject of the intransitive form. That sentence was mostly technical garbage, so here is an example to help. The verb shatter is ergative:
Transitive: The baseball shattered the window.
Intransitive: The window shattered.
When shatter is transitive, as in the first example, the window is the object. When it is intransitive, as in the second example, the window is the subject. In either case, it’s the thing undergoing the action of being shattered, which is supposed to be something only objects do.
Yet another problem for these traditional definitions are dummy subjects, as in the following:
(1) It is snowing.
(2) It seems the hobbit is upset.
(3) It isn’t nice to taunt the octopus.
it is the subject of (1) and the subjects of the main clauses of (2) and (3). It doesn’t fit the traditional definition of subject very well though, because it’s not exactly “doing” anything in these sentences. It doesn’t refer to any entity that would be capable of doing anything in the first place. it is only there for purely syntactic reasons. There needs to be a subject so English grammar inserts an it.
(Edit: Upon reflection, I don’t think that third example shows what I want it to show. I think that actually it is a proform for the verb phrase to taunt the octopus, and not a real ‘dummy it’. But in any case, a verb phrase is a theoretical entity, and it isn’t the kind of thing that can “do an action”, so these kinds of sentences are still problematic for the traditional definitions.)
Defining a subject as the do-er raises too many problems. It’s at best a generalization. Instead of defining a subject in terms of what it does, it’s better to define it in terms of where it is. Subject is best understood as a position in a sentence, and anything in that position is the subject. The same goes for object.
In more technical jargon, subject and object are “grammatical functions”, and they are a matter of syntax. Whether something is a do-er or an undergoer of an action is technically known as a “thematic relation“, and this is a matter of semantics. Thematic relations are determined in an entirely different way than subjects/objects are, and they depend a great deal on what particular verb is being used in the sentence. For example, the verbs “listen” and “hear” describe essentially the same event, but differ in the roles they assign to the subject. The subject of “listen” is an agent, because when you listen it’s intentional. The subject of “hear” is an experiencer, because when you hear it’s passive reception. (Check out that wikipedia link to see a list and description of some of these relations.)
So getting back to subjects – if subject is a position, what position is it? It’s a little messy to explain this because there are several word orders possible in English.
“I’m leaving” Mark said
“I’m leaving” said Mark
The noun Mark is the subject of both these sentences, but in one case it comes before the verb, and in the other case it comes after the verb. This makes it look ludicrous to try and define subject as a position, because there are obviously two possible positions for it. And it wouldn’t be very useful to define subject as something that can come before or after a verb.
Objects can be slippery too:
Laura sent a letter to George.
Laura sent George a letter.
*Laura sent to George a letter.
(Click here if you don’t know what that asterisk means.)
The objects of sent can apparently switch places, and also something funny happens with the preposition to. Again, how can one define “object” as a position, when apparently they can be in more than one place?
Linguists approach the problem by assuming that in these kinds of sentences, there is only one basic sentence structure, a “deep structure“, and others can be derived from this deep structure through something called syntactic movement, or transformations. So the subject or object is only ever in one place in the deep structure, even though it can appear in more than one place when you hear a sentence (in the “surface” structure).
The basic word order of English is usually given as Subject then Verb then Object if there is an object. That’s your simple affirmative sentence, as in The snake lied and Eve ate an apple. In slightly more technical terms, it means that the phrase which is the sister to the verb phrase is the subject. If we were to represent these sentences as syntax trees, they would look like this:
If you imagine these as family trees, you can see where the term “sister” comes from. Say that the sentence is the mother, and she has two girls: verb phrase and noun phrase. They are sisters. In one of those trees, the verb phrase also has a noun phrase daughter.
More generally, you can think of a basic English sentence like this:
Anything in that first noun phrase is the subject. Anything in that second noun phrase is the object. From here, movement might happen. For example, the verb might end up on the other side of the subject due to question formation: She is angry => Is she angry?. The word she is in that noun phrase marked [SUBJECT] before the verb moves to the front.
Fixing this bit of miseducation is not going to be easy. To get an understanding of what a subject is, you really need to know how to draw a syntax tree. But traditional sentence diagramming tends to treat sentences as “flat” structures, and focusses on labelling just the individual words. There’s also an annoying problem with the Reed-Kellogg system that the surface word order of the sentences is not always represented in the diagram (a subject for another post, I think). There’s little emphasis on how words group into phrases, which is crucial in understanding anything in syntax. So the traditional definition certainly is easier to understand, but easier isn’t better if it’s wrong.