Non-fusional inflected languages are arguable, since it is possible to consider the verb affixes that indicate person and number to actually be pronouns, depending on how you look at it.

But Japanese doesn’t fit that model either, because there’s nothing about the form of the verb nor anything attached to the verb that indicates jack diddly squat about the subject or object (except, arguably, politeness-level marking, which with certain kinds of verbs, such as verbs of giving and receiving, can indicate something about e.g. the actor’s relationship to the speaker — but that’s clearly different from anything in English).

It seemed obvious to me once I understood even the most basic aspects of the grammar of the language.

Maybe it’s because I’ve studied a number of languages that *do* have the subject concept, and in every single one of them the subject appears either explicitly or implicitly (e.g., by choice of inflectional affixes on the verb) in every single sentence more or less without exception (barring interjection “sentences” that don’t have a verb either, like “Wow!”). The subject is specified for every (finite) verb, because the verb requires a subject: we have to know who is performing the action or the action doesn’t make any grammatical sense. (Yeah, yeah, passive voice, but even there we have to know who or what is performing the act of being acted upon.) If you look for something like that in Japanese, it’s patently absent.

You can also look in vain for an Indo-European-style active/passive voice distinction, wherein different forms of the *same verb* imply a different grammatical relationship between the subject and the verb. Japanese grammarians use the terms “passive verb” to mean something else entirely. These two things (subject and voice) are inextricably linked.

> Now add to that: “There is no such thing as a pronoun”

I’m not sure I’d go that far. 何 is arguably a pronoun, for example.

However, Japanese doesn’t use pronouns with anywhere near the frequency that English does, because most of the time when English uses them it’s to indicate the subject of the verb (i.e., who is performing the action), or the object of a transitive verb, in a relatively inspecific way. Japanese doesn’t have subjects, and few if any verbs actually *require* objects, so pronouns aren’t used for these purposes.

> and “there is no such thing as a copula.”

As best I can tell, the thing usually called a “copula” in Japanese grammar is in fact purely a medium-level politeness marker (in modern Japanese; how it was used historically, I don’t know).

However, it is arguable that the word “copula” is used somewhat loosely to describe a number of constructions in various languages where the “copula” in question does not behave terribly much like an Indo-European being verb. Add to that the fact that we don’t call our being verb in English (or most European languages) a “copula”, and I don’t think the use of this term causes nearly as much confusion as the use of the word “subject” for が.

> maybe they should just teach in all-japanese

The problem isn’t with using English (or any other language) to teach Japanese grammar. The problem is with trying to apply concepts from the study of English grammar to Japanese when they aren’t actually relevant. You can explain Japanese grammar in English just fine: you just have to describe what the various Japanese constructions actually do.

> You might be confusing subject with agent

He glossed over the passive voice (and the middle voice for that matter) because it’s not relevant, because Japanese doesn’t have grammatical voice either.

> In the sentence, “the game was won”

In the sentence “the game was won”, the verb “was won” is in the passive voice, which is markedly different from the active-voice form “won”, thereby implying a different grammatical relationship between the verb and its arguments. Both verb forms are of the same verb, “win”, but a different from is used when the subject is the actor versus when the subject is the acted-upon. Show me anything like that in Japanese.

It’s the same with 要る, which means 必要である, NOT (as our English-speaking brains want to think) 必要とする.

And potential forms of verbs are the same story. What they “clearly” mark with が *would* be the object of the *original* verb but *is* the subject of the *potential* verb. For example, the sentence 犬はチョコレートが食べられない doesn’t literally mean “Dogs can’t eat chocolate.” What it literally means is “As for dogs, chocolate can’t be eaten.” “Chocolate” is the subject, not the object, of that sentence. (And “dogs,” of course, is the topic, not the subject.) There are two reasons we think of chocolate as the object of that sentence: it’s what receives the action, and “As for dogs, chocolate can’t be eaten” is not a sentence we’d ever utter in English.)

In short, が *does* mark the subject in your examples, but that’s kind of hard for us to see because our brains, which automatically translate Japanese into the smoothest possible English, think of those subjects as objects.

Thanks for the opportunity to think that through! If you think of any other examples where が seems not to mark a subject, please put them on the table!

On the “subject” of が, I’m afraid I have to agree with Jay Rubin and disagree with you. が does mark the grammatical subject of a sentence, which as Charles pointed out doesn’t have to be the agent of the verb. And as Thomas explained, クレープが食べたい doesn’t prove that が isn’t a subject marker; what it proves is that 〜たい turns a verb into an adjective. It’s only because that adjective is derived from a transitive verb that が seems to mark an object rather than the subject it’s actually marking. That particular circumstance is what led you to comment that 〜たい might be a special case.

Excluding its possessive (as in 我が息子) and connective (as in 俺は知らないが、彼に聞いたら？) functions, が always indicates a subject. This confuses English speakers only when what we think of as a transitive verb is handled differently in Japanese (like an adjective, as with 食べたい above, or verbal adjective, as with 俺は彼女が好き). We don’t mind when が marks the object of a passive verb’s action (as in 俺が彼女に振られた), because as Charles pointed out, we’re used to using subjects that way in English.

As Rubin has explained, every Japanese sentence has a subject. The reason that many seem not to is that the subject is omitted when context makes it unnecessary to state. Thomas’s parenthetical curry-bread was a great example of this in action. In his example, カレーパン is the omitted subject of the adjective 食べたい.

Tae, I do appreciate your skepticism of explaining Japanese grammar in English-grammar terms. Your sentiment echoes that of a very smart Korean linguist who criticized the the ~하고있다 pattern for copying the Western notion of a present progressive tense and forgetting that Korean already handled this tense suitably with ~한다. His argument leaves a few bases uncovered–for example, what if I want to tell you with 요 politeness about something that’s going down right now?–but it’s still a legit critique of imposing Western grammar on an Eastern language.

However, I wouldn’t put this refutation of が-as-subject-marker in the same category. Unlike the present progressive tense, a subtle shade of chronology that a language may do well without, a subject is something that every language has to have. Without *someone* or *something* to *be* or *do* the predicate of a sentence, you have no sentence. What you’re calling an “identifier” is really just one of the functions of a grammatical subject.

I think that if all students of Japanese read the chapter on は and が in Rubin’s Making Sense of Japanese Grammar, there would be no ambiguity about this particle and more people would recognize “exceptions” like 〜たい and 好き as not exceptions at all, but mere differences between Japanese and English.

Though I disagree with you on this one point, I want to emphasize again that I love your work and find it an immensely beneficial resource for everyone grappling with this wonderful language. Thanks so much for the time and energy you’ve put in! I hope these words find their way to you and your audience even after this post’s 2.5-year sleep!

I was tought that が particle introduces the new topic, and is used in the folowing expressions: ~が好きです、~がわかります、~が上手です、~がほしいです、~が~たいです etc。

just a way how my teacher enplaned to me the use of がparticle in that form.

So the whole class just learned those few expressions.. it wasn’t that difficult at all.

I highly suggest the article I linked to. This excerpt is especially interesting:

「が」にしてさえ、英語などでいう主格を示すものではなく、動詞などで示された事態を引き起こしたものを示すものであり、いわば「由来格」とでもいうべきものだという人（山口明穂『日本語の論理』など）もいる。

It’s saying in essence that 「が」 is the trigger that caused the situation. That might be a better way to think of it and explains why you can’t answer “what are you eating” with 「が」 because the answer did not trigger the action.

Following your reasoning, the answer to “what are you eating?” would be パンが食べている because が “identifies something unknown”. However you can’t say that because what is eaten can only be an object of 食べる, not a subject.

One thing you might be confusing is that what is grammatically the subject or the object of a particular verb varies according to idiom for any given language. In Greek, Italian and Japanese, for example, “wait” takes a direct object, unlike English. In the same three languages, what you “like” is a subject, not an object as in English. Similarly 食べたい takes a subject for what one likes to eat. It’s a purely grammatical thing, distinct from semantics.

What makes Japanese tricky is that grammatically you can end up with two subjects. “Taro wa Hanako ga suki desu” can mean,
(a) tell you something about Taro, it’s Hanako that he likes, or
(b) tell you something about Taro, it’s Hanako that likes him.
It’s only context that can help you distinguish.

Similarly 食べたい grammatically has two subjects, the eater and the eaten. Obviously semantically the eaten is an object, but according to grammatical idiom (because the たい inflection is an adjective), there are two subjects. If someone uses が to identify one of those subjects (at the same time using が for its purpose of identifying something unknown), it is only by context that we can work out which subject it is.

What I’m trying to say is that が does plays the role of an identifier, but simultaneously also indicates the grammatical subject. 食べたい’s eaten is a subject by idiom, and が can is used for that reason. If が “can either represent the subject or the object” and “only identifies the unknown” as you claim, then why can’t you answer “what are you eating” with パンが食べている?

My name is Cyrus Farivar and I’m a reporter for The World, an American public radio show (www.theworld.org), a co-production of Public Radio International, BBC, and WGBH Boston.

I’m working on a story about how language learners and teachers are using Skype to conduct interactive langauge lessons via the Internet and I’d very much like to interview you.

When would we set up a time to chat — via Skype, even?

-C

When I explained the same stuff to my confused friends, suddenly they understood and gained a better understanding of the proper usage of the particles. Even learning Korean has given me a better understanding of particles in general.

It’s incredible that I’m one of the only people in my Japanese classes who understands the material, but then I realize how it’s being taught. It’s just not a practical teaching method. Fix it and people will understand.