The sentence we’re looking at is
My response to this would be quite simply… what comes before the zero-が for いない? That’s the one zero-が you didn’t identify. If all we’re trying to answer is ‘what’s the subject?’, then yes, the answer is easy: it’s the sort of people we’ve just mentioned. However, if we stick strictly to the formalism of the model as it’s commonly presented, we’re obliged to insert the subject like so:
This raises a few problems. The first is of course that the sentence becomes much longer than it initially was for the purpose of conveying the same idea. It’s unwieldy. However, there’s more: rigidly applying this formalism interferes with natural Japanese parsing. In a normal, natural Japanese sentence, the appearance of another agent in the sentence would raise the possibility that that agent is the subject of the next verb in the sentence (i.e. AはBが[verb] typically suggests that B does [verb]). That’s not the case though, as you pointed out in your analysis of the sentence: 言っても is likely performed by the speaker or by some other person. Additionally, AはBが[verb] typically suggests that B is a subset or characteristic of A. That’s not always the case, of course, but since Bが[verb] provides information about A in a natural Japanese sentence of this form, it’s the norm. That sort of parsing is nigh impossible in this sentence. In essence, this sort of ambiguity, unwieldiness and confusion is the exact reason we frequently use pronouns in English and other European languages, and the reason Japanese doesn’t explicitly include the zero-が element systematically.
The matter of cases in which the subject is unclear has also been raised:
I think a good example of this is when the most probable subject is some sort of general situation, like
I think the closest English translation is
On that day, it ended up being that everyone’s bag [unfortunately] vanished.
I’d like to point out that this actually does already stretch the zero-が concept though, because this is an example of an ‘impersonal’ structure, which, as Oxford says, describes something
3 Grammar (of a verb) used only with a formal subject (in English usually it) and expressing an action not attributable to a definite subject (as in it is snowing).
The subject doesn’t clearly exist in these cases. We just have a general feel that something is happening, and we can assign it to the overall circumstances. I’d like to suggest that this poses a challenge to the idea that every sentence needs a subject to make sense, because never mind in my example: Oxford’s example of ‘snowing’ provides us with a case in which a subject is honestly unnecessary, since only the phenomenon really matters, not the presence of a subject.
Nonetheless, I guess I can agree that we generally need a source of action or something to which an adjective can be applied for a sentence to make sense. A ‘subject’ is not always necessary, but something that performs an analogous function is something we tend to seek, even if it’s just a vague sense of something occurring somehow. In the case of snow, for example, we might just refer to the weather or the sky as that source. Still, we could argue that this source or point of focus is sometimes closer to a simple topic, without a need for it to actually perform the function of a subject. (Yes, topics and subjects often overlap, but we’ve also seen that’s not always the case.)
Just as a final example though, even if we assume a sentence always needs a subject or something analogous in order to make sense, if we’re looking at the zero-が model as a parsing technique, I’d like to highlight a whole series of sentences that breaks it. The book 枕草子 opens with a series of sentences describing the most beautiful aspects of the four seasons. The sentences themselves, however, look like this:
For spring, dawn.
For summer, night.
For autumn, dusk.
For winter, early morning.
Just to be clear,「春は、あけぼの。」is literally how the book starts. There is no preceding sentence. Where do we insert the zero-が here? The typical analysis given by teachers of Classical Japanese to their students in Japan is that we’re meant to understand that each of the moments of the day identified by the author can be followed by a「がいい」. However, if we’re just considering the original structure of these sentences, none of this information is present, and the most we can do to render the sentences ‘complete’ is to add だ at the end of each one. Even so, we can’t just use the usual ‘take another element from the sentence and interpret it as the subject’ technique that often works with the zero-が model.
Again, this may not disprove the idea that every sentence needs a subject in order to make sense, but I think it should still give us pause because such sentences require quite a bit of additional interpretation in order to become ‘complete’ in the way we’re used to in English, with the subject for a hypothetical zero-が and the verb (and very honestly even all the rest of the predicate) needing to be invented based on context and yet they’re considered acceptable sentences in Japanese. By the way, these sentences still exist in modern Japanese, and aren’t just relics from classical works like 枕草子: a common structure for advertisement slogans is「AならB」, similar to a hypothetical ad in English that might go, ‘Computer? Windows.’ It’s still possible to seek out a subject and a verb/adjective here, sure, but there’s no simple way of applying the zero-が model in such cases.
This is also true for more common structures like といえば and 〜こと・ものだ (when used to indicate what’s important with regard to a certain thing, or general trends/habits):
- 山といえば富士山だ can’t be parsed as 山といえば 山が 富士山だ, and whether Mount Fuji is in the predicate or acts as the subject depends entirely on how one chooses to interpret or translate the sentence
- In 合格したかったら勉強することだ, the subject of だ is not clear, and cannot be simply resolved by treating what’s already present as a predicate into which a suitable zero-が can be inserted at the right place. In fact, the dictionary I’m using suggests that one interpret this 「〜ことだ」 as「〜ことが大事だ」, which is slightly different from the way one usually uses the zero-が model.
In essence, while I acknowledge that the zero-が model can be helpful, and I think it’s fine to use it or simply to look for a subject if it helps someone understand a sentence better, I think it’s sometimes more expedient to be aware of the relationships different particles or structures tend to imply. I’m not saying it’s wrong, but I do tend to agree with this
given that the model can’t always be applied neatly (as we’ve just seen), and all rules and models tend to fade into the background once one has got a feel for how a language works independently of how it can be explained or interpreted using other languages.