My knowledge of Japanese grammar is in Japanese, so I'm not certain about the English term of 主語, to be honest. We use the same term to describe 'subject' in English grammar. I used 'topic' just because the original article used it.

Your explanation of 'ga'/'wa' is spot on as far as I can understand as a layman of native speaker with standard Japanese grammar education in Japan but no advanced linguistic degree.

I'd say that, because 'wa' emphasizes the introduced subject as the center of interest, it isn't used in the subordinate clause.

Tanaka ga nihon ni itta hi. (The day Tanaka went to Japan) ; ok - the interest is on 'hi'

Tanaka wa nihon ni itta hi. ; invalid

Thanks for your reply. I believe 主語 covers both subject and topic. Since an English sentence such as "John loves Mary." can be understood as e.g. "It is John (not James) who loves Mary." or "John loves Mary (not Lisa)", it might have several formal representations in Japanese via e.g. the use of wa/ga.

Also, see user gizmo686's excellent explanation for one approach below.

Not a native speaker, but have studied Japanese linguisticly (as well as as a second language).

Wa is a bit of a complicated topic. The prevailing thinking is (roughly) that it has two distinctive meanings: topic marking, and contrastive. As a topic marker, wa does not introduce the subject (although in many cases, there is a null anaphora referring to the topic).

In anycase, the common linguistic explanation for shiro's correction is that the subject of subordinate clauses resists topicalization.

(Is this where people start flaunting their phd:s, professor titles? j/k ;-) academic here as well - I do not hold a phd)

Well, I realise the topic + contrast bit but is it really treated as a null anaphora, rather than acting as both topic and subject marker in my example...? My examples referred to information structure more than anything.

Yes, I realise you can have sentences like "Ashita wa Tanaka ga..."/"Sou wa hana ga nagai." - I've even seen a discussion on double topics (some old, theoretical text by Yasuo Kitahara IIRC, probably more known for 'Mondai-na Nihongo'). I also realise that in some contexts where it seems to denote a subject its noun is only a topic ("watashi wa unagi desu").

Logically, it would indeed be quite difficult for a subordinate clause to contain the/a topic.

Anyway, I'm curious if you happen to have further explanations (or articles)!

(Unrelated note: why is it that Japanese of all things make us crawl out from under our rocks...? :-))

No PHD here either, just undergrad followed by some hobbiest reading (of scholarly sources) on Japanese linguistics.

To be clear, the comment about null anaphora was more of a throwaway comment anticipating the objection that sometimes the topicalizing wa does mark the subject. While I have seen this explanation presented, and it is my prefered explanation, I would not necessarily call it pervasive. Now, for the explanation itself (unfourtantly, I am on vacation, so cannot check any of my references).

Japanese is a clear example of a pro-drop language, so using pronoun dropping (aka, null anaphora) as an explanation requires less justification than it would in English, where we only see it in specific contexts. Additionally, we see the topicalizing "wa" in various contexts, not all of which can be understood as subjects, so a unified explanation that can account for all of them would be preferred.

For example, consider the sentence

1) Mary-ga ringo-o tabeta

We can topicalize Mary with the following derivation:

2) Mary-wa Mary-ga ringo-o tabeta

3) Mary-wa anohito-ga ringo-o tabeta (Pro-form substitution)

4) Mary-wa ringo-o tabeta (Deletion)

Simmilarly, we can topicalaize ringo with

2) Ringo-wa Mary-ga ringo-o tabeta

3) Ringo-wa Mary-ga are-o tabeta (Pro-form substitution)

4) Ringo-wa Mary-ga tabeta (Deletion)

We also have the following sentence (kudamono = fruit)

Kudamono-wa Mary-ga ringo-o tabeta

Admittedly, I struggle to think of a context where the speaker would not drop Mary due to context, but that should not be relevent here, and I am sure that there exists better examples.

Notice that, under the null anaphora explanation, all three of these examples could be explained in the same way. If we were to explain the first example as wa being a subject marker, then we would need to explain the second example as wa being an object marker, and the third example as wa being just a topic marker.

I have seen an alternative explanation that describes topicalization in Japanese as a transformation rule. I have mostly seen this by researchers who view Japanese non-configurationally, who argue that a rule such as [ga/o] -> [wa] in a non configurational language is directly analogous to a movement rule in a configurational analysis. Even under this approach, you still need to account for sentences where the topic has no co-referential place in the rest of the sentence.

Further, even under this alternative explanation, I would still not call wa a topic marker. Rather, I would say that when the listener reconstructs the deep-structure, he uses pragmatics to infer what syntactic role the topic plays. Indeed, If you consider a sentences such as ringo wa tabeta and Mary wa tabeta you can see that there is no syntactic way to identify where the topic falls in the deep structure.

Thank you so much for this writeup - it does ring a bell!

Don't worry about references. This is more than enough to get me re-started, dig through my old books/articles and find new ones.

> Further, even under this alternative explanation, I would still not call wa a topic marker.

-> "subject marker"? ;-)

Enjoy your vacation (and maybe pursue a phd)!