Friday, August 23, 2019

Double Subject Constructions

In Japanese, a double-subject construction is when you have two subjects in a single sentence, called large subject and small subject, where the predicate of the large subject is a predicative clause containing the small subject plus the small subject's predicate.

In practice: zou ga hana ga nagai 象が鼻が長い, "elephants have long noses," or "about elephants: about [their] noses: long," has the large subject zou predicated by the clause hana ga nagai, which contains the small subject hana predicated by nagai.

The patterns ABがC, ABがC, ABがC, AにはBがC, BがC, C, ABC, ABC, among others often form double-subject sentences. Among these, some have two nominative subjects, and thus are also called double-nominative sentences.


Since double-subject sentences are pretty complicated to explain, and honestly most people don't really care, let's start by listing how they're used in Japanese for reference.

Note that if you have any questions about "why is this sentence a double-subject?" Don't worry because it's all explained in excruciating detail in the grammar section further below. Let's focus on the usage for now, and how it works later.


The most common use of double nominative sentences is to say that the large subject possesses the small subject.

This happens when you have a predicate including the verb aru ある, "to exist," its negative form nai ない, "nonexistent," or the similar verb iru いる, which is used with animals and people.
  • watashi wa {okane ga aru}
    {Exists is true about money} is true about me.
    {Money exists} is true about me.
    I have money.
  • watashi wa {okane ga nai}
    {Nonexistent is true about money} is true about me.
    {Money is nonexistent} is true about me.
    I don't have money.
  • watashi wa {kodomo ga iru}
    {Children exist} is true about me.
    I have children.
  • watashi wa {kodomo ga inai}
    I don't have children.

In general, you don't say watashi wa in Japanese, since it's implicit you're talking about yourself. So the sentence would become just okane ga nai, for example.

The double nominative construction can also be used to attribute a quality to a possession pertaining to someone.
  • kanojo wa {me ga kirei da}
    {Eyes are pretty} is true about her.
    Her {eyes are pretty}.
    • She - possessor.
    • Eyes - a possession.
    • Pretty - quality.

There are various idiomatic expressions in Japanese concerning body parts that take the form of double nominatives. Observe below how there's a literal translation and a more natural translation to every compliment and discompliment:
  • Tarou wa {atama ga ii}
    Tarou's {head is good}.
    Tarou is smart.
  • Tarou wa {atama ga warui}
    Tarou's {head is bad}.
    Tarou is stupid.
  • Tarou wa {ashi ga hayai}.
    Tarou's {feet are quick}.
    Tarou is quick on his feet. Tarou can walk or run fast.
  • Tarou wa {kuchi ga katai}
    Tarou's {mouth is stiff}. (in the sense it doesn't move.)
    Tarou is tight-slipped. Tarou can keep a secret.
  • Tarou wa {kuchi ga karui}
    Tarou's {mouth is loose}.
    Tarou can't keep a secret.
  • Tarou wa {se ga takai}
    Tarou's {back is high}.
    Tarou is tall.
  • Tarou wa {se ga hikui}
    Tarou's {back is low}.
    Tarou is short.

Not all possessives are double nominative. Some have dative subjects.
  • watashi niwa {sainou ga nai}
    {Talent is nonexistent} is true for me.
    I don't have talent.
  • ue niwa {ue ga aru}
    {Above exists} is true for above.
    To above, there's above.
    This sentence means there's always someone above you, in terms of skill, for example. In anime, it's often used in the big fish small pond way, where the bad guy usually thinks he's the strongest in the world, until he gets defeated by protagonist. There's always someone stronger.

Manga: JoJo no Kimyou na Bouken Part 5: Ougon no Kaze ジョジョの奇妙な冒険 黄金の風 (Chapter 458, ソフト・マシーンの謎 その①)
  • kono Joruno Jobaana niwa {yume ga aru}!
    {A dream exists} is true for this Giorno Giovanna!
    This Giorno Giovanna has a dream!


When talking about having a relationship with someone, the large subject is marked by ni に or niwa には. For example:
  • watashi niwa {kareshi ga iru}
    {A boyfriend exists} is true for me.
    To me, a boyfriend exists.
    I have a boyfriend.
  • watashi niwa {kanojo ga inai}
    I don't have a girlfriend.
    • kanojo 彼女
      She. (third person pronoun.)

Since having a relationship with people means the small subject is a person, and a person is an animate thing, the verb used is always iru いる, never aru ある.

This function applies to inalienable possessions such blood relationships, kinship.
  • watashi niwa {kyoudai ga iru}
    I have siblings. I have brothers.

The verb dekiru 出来る can mean "to make" a relationship.
  • watashi wa {tomodachi ga dekita}
    {Been made is true about friends} is true about me.
    {Made friends} is true about me.
    I made friends.

The word for "relationship" itself, kankei 関係, can also be used in this way.
  • watashi niwa {kankei φ nai}
    {Nonexistent is true about relationship} is true for me.
    {Relationship is nonexistent} is true for me.
    To me, there's no relationship.
    It's not related me.
    It has nothing to do with me.
    It's none of my business.

These relationships work just like the possessives we've seen before. You can qualify a family member just like a body part, for example.
  • Tarou wa {imouto ga kirei da}
    {Younger sister is pretty} is true about Tarou.
    Tarou's {younger sister is pretty}.


When possessives are combined with relative clauses qualifying koto こと, the sentence becomes about whether large subject possesses a given experience or not. In other words, whether they have ever done something (koto aru) or never done it (koto nai).
  • watashi wa {{mita} koto φ aru}
    {Exists is true about the act that is {to see}} is true about me.
    {{To see} has happened} is true about me.
    I have {the experience that is {to see}}.
    I have seen [it].
    I have seen [it] before.
  • watashi wa {{mita} koto φ nai}
    I have not seen [it].
    I have never seen [it] before.


Personally, I've never seen this actually happen, but it seems that theoretically you can have a triple nominative sentence in Japanese if you talk about the possessor's possession's possession.
  • Riko wa {kareshi ga {se ga hikui}}
    {{Low is true about back} is true about boyfriend} is true about Riko.
    {Boyfriend's back is low} is true about Riko.
    Riko's boyfriend's back is low.
    Riko's boyfriend is short.
    • Once again, normally you would use no の. Observe:
    • {Riko no} kareshi wa {se ga hikui}.
      {Back is low} is true about the boyfriend {of Riko}.
      {Riko's} boyfriend's {back is low}.
      {Riko's} boyfriend is short.

Since it can be double or triple, it's called just "multiple nominative sentences" sometimes. But you can call them Japanese Matryoshka predicates if you want.

I realize non-sense of this level mustn't go unsourced, so here's a sentence from an actual linguist:
  • bunmei-koku ga {dansei ga {heikin-jumyou ga mijikai}}
    {{Average-lifespan is short} is true about men} is true about civilized-countries.
    A civilized country's {men's {average lifespan is short}}.
    In civilized countries, the average lifespan of men is short.
  • —Kuno 1973, p.41-(16c)), as cited in Inoue, 2008, p.38.


The large subject can be the cognizer of a cognition expressed by the predicative clause. In simple terms, the predicate says what the large subject feels, thinks or knows about the small subject.
  • watashi wa {neko ga suki da}
    {Liked is true about cats} is true about me.
    {Cats are liked} is true about me.
    In my mind, cats are liked.
    I like cats.
  • watashi wa {inu ga kirai da}
    In my mind, dogs are disliked.
    I dislike dogs. I hate dogs.
  • watashi wa {kumo ga kowai}
    {Scary is true about spiders} is true to me.
    {Spiders are scary} is true to me.
    To me, spiders are scary.
    I'm scared of spiders. I'm afraid of spiders.
  • watashi niwa {shi ga kowakunai}
    To me, {death isn't scary}.
    I'm not afraid of death
  • watashi niwa {kono manga ga omoshiroi}
    To me, {this manga is entertaining}.
    I think this manga is fun.

One of the differences between the verbs shiru 知る, "to know (to have learned)," and wakaru 分かる, "to know (to comprehend)," is that shiru is a transitive verb while wakaru is part of a double subject construction.
  • watashi wa sore wo shitte-iru
    I know that. I've heard about it. I learned about it from somewhere. Someone told me about it once.
  • watashi wa {sore ga wakaru}
    I know that. I comprehend it. Not only have I heard about it, but I understand it perfectly.


When the verb is in the potential form, the direct object can become a small subject, and the subject becomes the large subject. For example:
  • Tarou wa manga wo yomu
    Tarou reads manga.
    • Tarou - topic, subject, agent.
    • Manga - direct object.
  • Tarou wa {manga ga yomeru}
    {Read-able is true about manga} is true about Tarou.
    {Manga is read-able} is true about Tarou.
    Tarou is able to read manga.
    Tarou can read manga.
    • Tarou - topic, large subject.
    • Manga - small subject.

In this case, the verb yomeru doesn't mean literally "can read," but something closer to "is read-able." Something is read-able. If something is read-able to Tarou, then Tarou can read that something.

The potential form suru する, "to do," is dekiru できる, by the way.
  • Tarou wa tabi wo suru
    Tarou "does" the travel. Tarou travels.
  • Tarou wa {tabi ga dekiru}
    {Travel is do-able} is true about Tarou.
    Tarou can "do" the travel. Tarou can travel.
  • watashi niwa {sore ga dekinai}
    I can't do that.
  • {watashi ni dekiru} koto
    Something [that] {I can do}.
    Things [that] {I can do}.


Similar to potentials, when a verb is in the -tai ~たい form, the direct object can become the small subject, and the large subject ends up "wanting" to do the verb.
  • Tarou wa {manga ga yomi-tai}
    {To read manga is desired} is true about Tarou.
    Tarou desires to read manga.
    Tarou wants to read manga.
  • watashi wa {piza ga tabetai}
    I want to eat pizza.
    • piza wo taberu
      To eat pizza.
  • watashi wa {ongaku ga kikitai}
    I want to listen to music.
    • ongaku wo kiku
      To hear music.


The predicative clause can be a statement about what the large subject requires or needs. In this case, the verb iru 要る, "to need," is sometimes used. Note that this is different from iru 居る, "to exist."
  • shujutsu wa {okane ga iru}
    {Needs money} is true about the surgery.
    The surgery requires money.
  • boku-ra niwa {kotoba ga iranai}
    {Doesn't need words} is true for us.
    We don't need words.
    Words aren't necessary to us. (we communicate without words.)
  • chiryou niwa {jikan ga hitsuyou da}
    {Time is needed}is true for the treatment.
    To the treatment, time is needed.
    The treatment needs time. It takes time for the treatment to work.
  • keeki niwa {ichigo ga fukaketsu da}
    {Strawberries are indispensable} is true for cakes.
    To cakes, strawberries are indispensable.
    Cakes can't be missing strawberries.

Note: these functions are more of what you call guidelines, not actual rules.
  • keeki niwa {ichigo ga kakasenai}
    To cakes, strawberries can't lack.
    Cakes can't be missing strawberries.
    • Is this a necessity? Well, yes. The verb is in the negative potential, so it's a potential too. Are strawberries part of the cakes' possessions? Also yes.

The verb kakaru かかる is used to talk about costs of time, money, and so on.
  • shujutsu wa {okane ga kakaru}
    {Money costs} is true about the surgery.
    The surgery costs money.
  • chiryou wa {jikan ga kakaru}
    {Time costs} is true about the treatment.
    The treatment takes time.

Sometimes this construction is used when ni に marks an objective.
  • tatakau niwa {yuuki ga iru}
    Fight {requires courage}.
    Courage is necessary in order to fight.
  • hito wo sukuu niwa {ai ga iru}
    Saving people {requires love}.
    Love is necessary in order to save people.

There might be other cases where double-subjects are used, but as you can see, the way it works is practically always the same, so you shouldn't have many problems with it.


Now that we've seen some examples of double-subject sentences, let's begin to understand the grammar behind why and how they work.


To begin with, the construction is sometimes called "double nominative" because, in case theory, subjects are marked with the nominative case. If there's two nominative cases, it's double nominative, double subject.

The ga が particle marks the nominative case, abbreviated NOM. The ni に particle marks the dative case, abbreviated DAT. The no の particle marks the genitive case, abbreviated GEN.

In Japanese, somehow all of them can mark the subject under certain circumstances. Consequently, there are different terms for the subjects depending on what's marking them:
  • ga
    Nominative subject.
  • ni
    Dative subject.
  • no
    Genitive subject.
  • wa
    Topicalized subject. (except when it's not the subject, but something else.)
  • niwa には
    Topicalized dative subject.

Sometimes with two ga is a double-nominative NOM-NOM, but something with ni plus ga is a double-subject construction with mixed case: DAT-NOM.

There are various ways to refer to each subject of a double-subject construction.
  1. First subject.
    dai-shugo 大主語
    "Large subject."
    Major subject.
    sou-shugo 総主語
    General subject.
  2. Second subject.
    shou-shugo 小主語
    Small subject.
    Minor subject.

Some papers call the small subject just "subject."
  • shugo 主語
  • shudai 主題
  • ni-juu shugo
    Two-layer subject.

Common abbreviations include MNC (multiple-nominative construction), DNC (double), and so on.


A predicate says something true about a subject.
  • yuki ga shiroi
    Snow is white.

Above, the i-adjective shiroi, "is white," predicates the subject marked by the ga が particle, the noun yuki, "snow."
  • "Is white" is true about "snow."
  • Snow is white.

In Japanese, if the subject is also the topic of the sentence, it's marked as the topic by the wa は particle instead. (see wa は vs. ga for details.)
  • yuki wa shiroi
    (same meaning.)
    • yuki wa - topic, subject.
    • shiroi - focus, predicate.

One way to get a sentence like yuki ga shiroi is by inverting topic-focus through asking a question with subject focus. This happens when you have an interrogative pronoun as subject, for example:
  • nani ga shiroi?
    What is white?
    • nani ga - focus.
    • shiroi - topic.
  • yuki ga shiroi
    Snow is white.
    • yuki ga - focus.
    • shiroi - topic.

As shown above, marking something with wa は is basically the same thing as marking it with ga が, except one is the topic while the other isn't.

Note that wa は can mark things that aren't the subject, like the object, but every time it marks the subject, it should be able to be replaced by ga が if certain criteria is met.

Next, we have something like this:
  • mimi ga nagai
    Ears are long.

Although the syntax of this sentence is the same as the previous one, we have one big problem: the meaning of the sentence.

If we say "ears are long," it sounds like we're saying that ALL ears are long. Just like "snow is white" says that ALL snow is white.

The large subject is then used to further limit the scope of the statement.
  • nagai
    • This sentence doesn't make any sense because we don't have a subject to which we attribute this "long" quality.
  • mimi ga nagai
    "Long" is true about "ears."
    Ears are long.
    • This sentence makes sense syntactically, but the meaning is too broad.
  • erufu ga {mimi ga nagai}
    {"Long" is true about "ears"} is true about "elves."
    {Ears are long} is true about "elves."
    Elves have long ears. The ears of elves are long.

When the predicative clause "ears are long" predicates the large subject "elves," we have to interpret the relationship between the predicate and large subject somehow.

One way to interpret this relationship is that the large subject (elves) possesses the small subject (ears). Thus, the sentence ends up meaning "the elves have long ears," even though it doesn't literally say that.

Double nominative sentences generally provide new information (focus) about the large subject (who's then the topic), so the large subject tends to get marked by wa は instead of ga が.
  • erufu wa mimi ga nagai
    Elves have long ears. (ears are inalienable.)

Just like before, we can force the large subject to be marked by ga が by asking a question with large subject focus.
  • donna doubutsu ga {hana ga nagai}?
    {"Long" is true about "noses"} is true about what kind of animal?
    {Noses are long} is true about what kind of animal?
    What kind of animal has long noses?
  • zou ga hana ga nagai
    {Noses are long} is true about elephants.
    Elephants have long noses.
    • zou - focus.
    • hana ga nagai - topic.
    • Note: the whole predicate clause hana ga nagai is the topic, not just hana, hence why hana isn't marked by wa.

When you see a double nominative with multiple ga が, like above, the first subject will be interpreted as the focus.(Inoue, 2008)

With this we've established that the normal way of constructing double nominative sentences like this is wa plus ga, not ga plus ga, because ga plus ga only happens in questions (and other more complicated situations), and such questions are very unlikely to happen.

This is the most infuriating part. Because it's much easier to just say we have a "topic and a subject," than to say "we have two subjects, but one of them happens to be the topic, plus there's a predicative clause or whatever mixed in."

Nevertheless, I think having two subjects makes more sense. After all, with questions like the above, you do end up with two ga が, two subjects, and that does put a hole in the "topic plus subject" theory that doesn't exist in the "large subject (that's also the topic) plus small subject" theory.

Vs. Other Possessives

In general, possessives sentences in Japanese aren't expressed with double-subjects, but through other means.

For example, there's a verb in Japanese that means "have," motsu 持つ, but it's used with alienable possessions—that can be aliened, taken away—not inalienable ones like body parts.
  • erufu ga mimi wo motte-iru
    The elf has the ears. He's holding the severed ears in his hand. Oh my kamisama, that's absolutely disgusting! It's horrifying!
  • Tarou ga sumaho wo motte-inai
    Tarou doesn't have a smartphone. (smartphones are alienable.)

The existential verbs aru and iru, and nai, are used when the existence or nonexistence of the small subject is being attributed the large subject;
  • hito wa {shippo ga nai}
    People's tail doesn't exist.
    People don't have tails.
  • neko wa {shippo ga aru}
    Cats' tail exists.
    Cats have tails.

Since the large subject can be understood as the possessor with verbs like above, and in phrases like Tarou wa {atama ga ii}, it makes sense that it can be understood as the possessor with basically any verb.

However, the genitive no の particle is normally used in such cases, and a phrase like the one above may even be considered ungrammatical by some native speakers who'll suggest the following one instead:
  • {Tarou no} saifu wa nusumareta
    The wallet {of Tarou} was stolen.
    {Tarou's} wallet was stolen.

Similarly, the no の particle is preferred instead of triple-subjects.
  • bunmei-koku ga {dansei ga {heikin-jumyou ga mijikai}}
    Civilized-countries' {men's {average-lifespan is short}}.
  • bunmei-koku ga {{dansei no} heikin-jumyou ga mijikai}
    Civilized-countries' {{men's} average-lifespan is short}.
    • {dansei no} heikin-jumyou ga - noun phrase marked as subject.
  • {{bunmei-koku no} dansei no} heikin-jumyou ga mijikai
    {{Civilized-countries'} men's} average-lifespan is short.
    • {{bunmei-koku no} dansei no} heikin-jumyou ga - noun phrase marked as subject.

This type of transformation has been called many things, like "subjectivization," "nominativization," "possessor raising," and "genitive raising."(by Kuno, 1973, Shibatani, 1977, Ura, 1996, and Tateishi 1991 respectively, as cited in Seraku, 2017, p. 104)

I'm not sure, but the ga が particle also has a distinct genitive function seen in names of places and people, maybe it's related to this?
  • Senjou no Hara 戦場の原
    Senjou ga Hara 戦場が原
    Senjougahara 戦場ヶ原
    Plains of battlefield.

Contrastive Small Subject

The contrastive wa that happens when you have two wa は in a single clause, or a wa は in a subordinate clause, also happens in double-subjects.

Does the predicative clause counts as a subordinate clause? Or is the whole sentence just a single clause? Honestly, I don't know. Either way, the second wa は, the one marking the small subject, will be contrastive.
  • kanojo wa {atama ga ii}
    Her head is good. She is smart.
  • kanojo wa {atama wa ii}
    Her head is good. She is smart.
    • Implicature: {not-head is not-good} is true about her. She is smart, but: her personality is bad, her appearance is bad, her taste in anime is bad, and so on.


In double nominative sentences where both subjects are marked by ga が, the large subject always comes before the small subject.

This is a peculiar property because in Japanese the role of a word in a sentence generally only depends on the particle marking it, and not on its position.
  • neko ga nezumi wo kutta
    The cat ate the rat.
    • neko ga - subject.
    • nezumi wo - object.
  • nezumi wo neko ga kutta
    (same meaning, despite the arguments switching places.)
    • nezumi wo - object.
    • neko ga - subject.

For example, with double nominatives that express possession, the large subject A possesses the small subject B. A has B. If we invert the position we end up with B has A.
  • ?hana ga {zou ga nagai}
    The elephants of noses are long.

Above, we have a sentence with perfectly correct grammar syntax whose meaning makes absolutely no sense at all.

With cognitives, however, since the sentence is about what A thinks of B, inverting them gets you what B thinks of A. It will still make sense, however, it will also change the meaning of the sentence.
  • Tarou ga {neko ga suki da}
    {Cats are liked} is true about Tarou.
    Tarou likes cats.
  • neko ga {Tarou ga suki da}
    {Tarou is liked} is true about cat.
    The cat likes Tarou.


Some sentences that fit the double nominative pattern aren't really double nominative. This happens because the topic marker wa は can replace not only ga が, but wo を and even ni に, too.(Sugimoto 杉本, 1990, pp.167-168)

For example:
  • Tarou ga piza wo tabeta
    Tarou ate pizza.

Above, the subject is marked by ga が and the object marked by wo を.
  • Tarou wa piza wo tabeta
    {Ate pizza} is true about Tarou.
    Tarou ate pizza.
    • Tarou wa - topic, subject.
    • piza wo - object.

In the sentence above, the transitive verb tabeta already has an object argument (piza wo). The only argument left to be filled is the subject. Thus, Tarou wa is assumed to be the subject. Conversely:
  • piza wa Tarou ga tabeta
    {Tarou ate} is true about pizza.
    Tarou ate pizza.
    • pizo wa - topic, object.
    • Tarou ga - subject.

Now, the subject argument is already filled (Tarou ga), so it's assumed piza wa fills the remaining object argument of the transitive verb.

Note how it reassembles the double nominative pattern: AはBがC. However, it's impossible to ask a question that gets you piza ga Tarou ga tabeta, since piza would have object focus, not subject focus.
  • *nani ga Tarou ga tabeta?
  • Tarou wa nani wo tabeta?
    What did Tarou eat?
    • Tarou wa...tabeta - topic.
    • nani wo - focus.
  • Tarou wa piza wo tabeta
    Tarou ate pizza.
    • Tarou wa...tabeta - topic.
    • piza wo - focus.

The same thing happens with the dative case:
  • Tarou wa gakkou ni itta
    Tarou went to the school.
  • gakkou niwa Tarou ga itta
    (same meaning.)
  • gakkou wa Tarou ga itta
    (also same meaning.)

Notice that in this case niwa and wa are interchangeable, except not exactly.

You can mark gakkou with wa instead of niwa because the school doesn't go anywhere, so it's easy to interpret it as being the destination. By contrast, the following doesn't work:
  • Tarou wa Jirou ni butsukatta
    Tarou bumped against Jirou.
  • Jirou niwa Tarou ga butsukatta
    (same meaning.)
  • *Jirou wa Tarou ga butsukatta

Since Tarou and Jirou are both people, they're both capable of bumping against each other. It becomes ambiguous who is doing the bumping here, who is the agent, and who would be marked by ni に, so using just wa は becomes unacceptable.(Nagata, 1994)

In double nominatives, the topic also shows its special wildcard prowess. With a possessive sentence with two ga, the first ga is always the large subject (possessor). But in a sentence with wa plus ga, the wa, which comes first, can take the role of the small subject (possession) instead.(Seraku, 2017, p. 111)
  1. zou ga hana ga nagai
    Elephants' noses are long.
    • zou ga - large subject.
    • hana ga - small subject.
  2. zou wa hana ga nagai
    (same thing.)
    • zou wa - large subject.
  3. ?hana ga zou ga nagai
    Noses' elephants are long.
    • hana ga - large subject.
    • zou ga - small subject.
  4. hana wa zou ga nagai
    As for noses, the ones of elephants are long.
    Elephants' noses are long.
    • hana wa - small subject.
    • zou ga - large subject.

Sentences 1, 2 and 4 mean the same thing, but have different topics: topic-less, zou, and hana. The sentence 3 doesn't make sense. The sentence 4 has the words uttered in the same order as 3, but since 3 doesn't make sense, it's valid to interpret the topic as being the small (possessed) subject instead.

Usage of Nouns Phrases

Just like how it works in single-subject sentences, with double-subject sentences the subject-markers can only mark nouns and noun phrases. This applies to both large and small subjects.

To elaborate, if you were to say "apples are tasty" and "bananas are tasty," in each sentence you have one subject. A basic mistake would be thinking that "apples and bananas are tasty" has two subjects, then, because the predicate "tasty" applies to two things.

  • ringo ga oishii
    Apples are tasty.
  • banana ga oishii
    Bananas are tasty.
  • ?ringo ga {banana ga oishii}
    Apples' {bananas are tasty}. (what?)

In reality, what happens is that you have a noun phrase "apples and bananas" marked as a single-subject, rather than two subjects marked individually.
  • {ringo to banana} ga oishii
    {Apples and bananas} are tasty.

The principle is that each clause has a number of roles (cases) to be filled. Each role can only be filled with one noun or Noun Phrase. If two NPs have the same role, it sounds weird. After all, it doesn't make sense for two NPs to be marked as the subject, right?

A parallel marked like to と, "and," is to join two NPs (ringo and banana), creating one larger NP that can be marked with a single case (ga).

With double-subject constructions, things get weirder. Double-subjects have two subjects, yes, but the large subject and small subject perform different functions (e.g. possessor and possession). It's a loop-hole in the "you can't have two subjects" concept, because the NPs don't have the same function.
  • kudamono ga {{ringo to banana} ga oishii}
    {Tasty is true about {apples and bananas}} is true about fruits.
    Fruits' {{apples and bananas} are delicious}.
    Among fruits, {{apples and bananas} are delicious}.
    • kudamono ga - noun large subject.
    • ringo to banana ga - NP small subject.
  • {ringo to banana} ga {iro ga chigau}
    {The color differs} is true about {apples and bananas}.
    Apples' and bananas' color differs.
    The color of {apples and bananas} is different.
    • ringo to banana ga - NP large subject.
    • iro ga - noun small subject.

As you can see above, neither the large subject nor the small subject needs to be a single word, they can be a multi-word Noun Phrase.

Dative Large Subject VPs

In some cases, with niwa には, it looks like you have a Verb Phrase marked as the large subject instead of a noun phrase. For example:
  • tabi suru niwa {okane ga hitsuyou da}
    {Money is required} is true for "to travel."
    In order to travel, money is necessary.
    [You] need money in order to travel.

Above, the VP tabi suru is marked by ni, or at least so it seems.

I'm not really sure how it works, but I guess it's the same principle behind ~ga ii ~がいい. In a classical Japanese, the attributive form of verbs, the rentaikei 連体形, could be marked by particles, just like a noun. In modern Japanese, the rentaikei is identical to the predicative form shuushikei 終止形. So it could be that what looks like a predicative clause, or a VP, is actually a NP that just happens to look exactly like a VP.

Transitivity Disparity

One very important and painfully obvious thing to notice about double nominatives, is that some of them translate to English as transitive verbs. For example:
  • Tarou wa neko ga suki da
    Tarou likes cats.
    • (Subject) likes (direct object).
  • Tarou wa eigo ga wakaranai
    Tarou doesn't understand/know English.
    • (Subject) understands/knows (direct object).
  • chiryou wa jikan ga kakaru
    The treatment takes/costs time.
    • (Subject) takes/costs (direct object).

If we were to translate these literally, we'd end up with "cats are liked," "English is understood," and something like "time costs," like just... costs. It costs. In the sense of it's being cost.

Naturally, this doesn't sound very normal.

Naturally, some linguists have thought it was better to analyze this as ga が marking the direct object in this case.(Kuno 1973, Shibatani 1977, 1978, Kageyama 1978 and others, as cited in Sadler, 2002, p. 31)

After all, in "Tarou likes cats" the word "cats" is the direct object, so why wouldn't ga be marking the direct object in Tarou wa neko ga suki da?
  • "...I shall show that ga is used not only for marking the subject but ALSO FOR MARKING THE OBJECT of all transitive adjectives and nominal adjectives...and of a certain class of transitive verbs."(Kuno, 1973:81, cited in Shibatani 1999, p.52)

I mean, sometimes the small subject for a suki predicate literally gets marked by wo を, becoming a literal direct object, so there's every reason in the world to believe ga が works the same way.(Shibatani, 1990, cited in Sadler, 2002, p. 30)
  • Tarou wa eiga wo suki
    Tarou likes movies.
    • To make matters even worse, suki 好き is actually the noun form of the verb suku 好く, which actually takes a direct object.
    • {e wo sukanu} kodomo wa mazu sukunai
      Children [that] {don't like drawings} are, to begin with, few.
    • Source: E no Kanashimi 画の悲み, by Kunikida Doppo 国木田独歩 (1871-1908)
    • And kirai is the noun form of the verb kirau 嫌う.

The same thing can be observe in the verb wakaru 分かる, which is sometimes used like rikai suru 理解する, "to comprehend," and even spelled wakaru 解る in such cases.
  • Evangerion wo wakaru
    To understand Evangelion.

Personally, I don't like the ga が being the direct-object theory.

I don't know a lot about linguistics—and obviously Kuno knows a lot more about it than I do—but if we go according to that theory, then ga が marks the subject, except when it marks the object sometimes, so now it has two functions instead of one.

With double nominatives, the ga が always marks the subject, except sometimes the subject is "large," sometimes the subject is "small," but, in principle, all subjects work the same way: the predicate always says something true about the subject.
  • Chikyuu ga hiratai
    The Earth is flat.
    • The predicate says the truth! Wake up sheeple!

So I personally prefer the subject-subject theory over topic-subject or subject-object or whatever. I think it's the one that makes most sense, and better explains things I've previously been extremely confused about, which is why I'm writing this article about it.


The double-subject approach makes it easier to explain some situations that would be harder to explain otherwise, like elliptical sentences. An ellipsis is those three dots (...) used when something is omitted. An elliptical sentence is a sentence that's missing part of it.

For example, if you've seen enough anime I'm sure you've seen this uttered before:
  • hayai!
    Is fast!

That's a predicate. It's a predicate, but there's no subject. That shouldn't make sense, after all, if you say "is fast," that means "SOMETHING is fast," and in the example above the "something" hasn't been uttered.

This is called an ellipsis, an omission. Below we have a possible full sentence:
  • kare wa hayai!
    He is fast!

So the subject was omitted, and now we uttered it. It's not like the sentence didn't have a subject, it's just that we didn't bother saying it.

This same elliptical situation can happen with double-subjects, except then we'll have two level of omission possible, because we have two subjects that we can omit. Observe:
  • suki da!
    Is liked!
    [I] like [it]!
  • kimi ga suki da!
    You are liked!
    [I] like you!
  • boku wa {kimi ga suki da}!
    I like you!

Before, "is fast" had to be true about "something," so we made it true about "him." He is fast.

Now, "is liked" has to be true about "something," so we made it true about "you." You are liked.

However, "you are liked" is an opinion, a cognition, a love confession. Someone feels that way. It's not an universal statement. Therefore, "you are liked" must be true about "some cognizer," we make the cognizer "me," boku, so "I like you."

The important part is that the predicative clause, kimi ga suki da, requires a cognizer, it requires someone to feel that way.

This means every time you have suki da in a sentence, there must a small and large subject somewhere. Sometimes they're explicit, other times they're omitted and the sentence is elliptical.

This is the reason for a sentence like the one below to have a contrastive wa は instead of a non-contrastive topic:
  • neko wa suki da
    Cats are liked.
    • Implicature: not-cats are not-liked. Dogs are hated.

In general, when you have just one wa は, it's a non-contrastive topic. It's only contrastive when you have two wa は, so one topic contrasts with the other, or a small subject contrast with another small subject.

But a sentence like the above breaks this principle, because it only has one wa は despite having the contrastive function.

That's because it's an elliptical sentence.
  • watashi wa {neko wa suki da}
    I like cats.
    • Implicature: I not-like not-cats. I hate dogs.

Above, we have the full sentence with the large subject watashi. Now we end up with two wa は, and we can see that the small subject neko was marked with the second (contrastive) wa は all along.

This is particularly confusing since it's normal to omit watashi wa in sentences about oneself. Thus, to say you like something, you'd just say:
  • anime ga suki
    [I] like anime.

Without saying watashi wa. And then, if you're a beginner, you'd ask yourself: why do you always use ga が with suki すき? Why is it different from yuki wa shiroi, for example?

Even if someone told you: it's because it's the the omitted topic's opinion, you still wouldn't have the grammatical reason for this to happen. Now you do.

Similarly, a sentence like atama wa ii ends up with the contrastive wa は, because the atama must be possessed by a large subject somewhere: kanojo wa atama wa ii.

Null-Marked Small Subject

In Japanese, there's a sentiment among natives that using the ga が particle twice in a single sentence means having two subjects, and that having two subjects is ungrammatical.

After all, the normal situation is having wa plus ga. Questions with ga plus ga are regarded as a special situation. And, generally, people avoid saying two ga が in sequence, or two wa は in sequence.

Consequently, when you have a question with large subject focus, and a predicate clause like atama ga ii, the second ga marking the small subject may get omitted in order avoid having two ga one after the other.
  • dare ga atama ga ii?
    Smart is true about who?
    Who is smart?
  • dare ga atama ii?
    (same meaning.)

Linguistically, it's said atama is null-marked or zero-marked in this case. Since Japanese generally marks stuff with particles, having something not marked by a particle is analyzed as it being marked by an invisible null particle, from now on represented by φ (the Greek letter phi.)

For example, in Japanese the topic of the sentence must be part of the main clause. Therefore, something part of a subordinate clause, like a relative clause, can't be marked by wa は.
  • yuki wa shiroi
    Snow is white.
  • {yuki ga kuroi} hi
    Days [when] {snow is black}.
    • Here, yuki ga shiroi qualifies the noun hi.
  • {yuki ga shiroi} no de namae wo "shiro" ni shita
    Because snow is white, [I] made [its] name "white."
    • Here, yuki ga shiroi qualifies the nominalizer no の.

Observe that "because," node ので, is composed by the de で particle plus a nominalizer no の, which acts like a noun, and is qualified like a noun by a relative clause.

Therefore, we could end up with two ga が if we said "because someone likes something."
  • kodomo wa {neko ga suki da}
    {Likes cats} is true about children.
    The children like cats.
  • {neko ga suki na} kodomo
    Children [who] {like cats}.
    • Observe that suki da becomes suki na when it qualifies a noun.
  • {kodomo ga {neko ga suki na}} no de neko-cafe ni itta
    Because the children like cats, went to the cat cafe.
    Because [my] children like cats, [we] went to the cat cafe.

In this case, the following would also work:
  • {kodomo ga {neko φ suki na}} no de
    Because [my] children like cats.
    • Here, the small subject neko is marked by the null particle.
  • {kanojo ga {atama φ ii}} no de
    Because she is smart.

The phrases koto φ nai, koto φ aru, kankei φ nai, and kankei φ aru are generally null-marked regardless of whether there's a ga が or or not.
  • boku niwa {kankei φ nai}
    [It] has nothing to do with me.
  • sore wa {kankei φ nai daro}!
    That has nothing do to with [it], right?!
  • kankei φ oo ari da!
    [It] has a lot do to with [you]!
    [It] has a lot do to with [that]!

Predicate Suffixation

Sometimes, the predicate of the small subject can end up becoming a suffix and forming a new word. Observe:
  1. neko ga suki 猫が好き
  2. neko φ suki 猫好き
  3. neko-zuki 猫好き

The su す became zu ず due to rendaku 連濁, a change in pronunciation that occurs to the first syllable of suffixed morphemes.

Above, the ga-marked and null-marked sentences mean literally "cats are liked." The combined neko-zuki means "cat-liking" instead, and refers to someone who likes cats, or the act of liking cats. In practice, both mean the same thing:
  • {kodomo ga neko-zuki na} no de
    Because [my] children are cat-liking.
    Because [my] children like cats.

Except that now it's hard to say we have two subjects, because the would-be small subject neko is no longer marked by anything. It's fused with its would-be predicate suki. Similarly:
  1. kanojo wa {inu ga kirai da}
    She hates dogs.
  2. kanojo wa {inu φ kirai da}
    (same meaning.)
  3. kanojo wa inu-girai da
    (same meaning.)
    • ki き became gi ぎ.

This can also be seen in other noun-adjective compounds. Like na-dakai 名高い, "famous," "renowned," means one's "name," na 名, is "high," takai 高い, literally na ga takai 名が高い. But zuki and girai are probably the most common ones.

Genitive Subject

In Japanese, the no の particle can mark the subject in a relative clause. For example:
  • Tarou wa hon wo yonda
    Tarou read a book.
  • {Tarou ga yonda} hon
    The book [that] {Tarou read}.
  • {Tarou no yonda} hon
    (same meaning.)

The same thing works with double-subject clauses, with one caveat.


To begin with, the subject-marking no の can mark the small subject in a relative clause. So the qualified noun becomes the large subject.
  • Tarou wa {se ga takai}
    Tarou's back is high.
    Tarou is tall.
    • Tarou - large subject.
    • se - small subject.
  • {se no takai} Tarou
    Tarou [whose] {back is high}.
    Tarou [who] {is tall}.
    • se - small subject.
    • Tarou - large subject.

Furthermore, the subject-marking no の can ALSO mark the large subject in a relative clause. And then the qualified noun becomes the small subject.
  • Tarou wa {takoyaki ga suki da}
    Tarou likes takoyaki.
  • {Tarou no suki na} takoyaki
    The takoyaki [that] {Tarou likes}.

This is the same system seen in this phrase:
  • {watashi no suki na} hito
    The person [that] {I like}. My crush.

But wait, how can we be sure that this isn't the no の that creates possessives instead? After all, the following analysis is also perfectly valid:
  • {Tarou no} {takai} se
    The back [that] {is high} {of Tarou}.
    {Tarou's} {high} back.
    • {takai} se is a noun phrase qualified by Tarou no.

As we've seen before, the word suki requires a cognizer. If the word marked by no の isn't the cognizer, there would be another cognizer somewhere, and that would be very weird. For example:
  • {watashi no} {Tarou ga suki na} takoyaki
    ?{My} takoyaki [that] {Tarou likes}.
    • More likely:
    • {{watashi no} Tarou ga suki na} takoyaki
      The takoyaki [that] {{My} Tarou likes}.
  • {watashi no} {suki na} hito
    {My} person [that] {[someone] likes}.
    • Someone who? Not me???

So we can be pretty sure that the no の is marking the subject inside the clause, not the possessor outside of it.

Furthermore, a word marked by wa は can become relativized, that is, can be qualified by a relative clause.(Kuno 1973, cited in Nagata, 1994, p. 3)

As we've already seen, wa は can mark either large or small subject, so the noun after the relative clause should be able to fit either role, too.

Given this, you can end up with a situation where the meaning of a phrase is ambiguous because the subject inside and outside of the relative clause can be either the large subject or the small subject. For example:
  • Tarou wa {neko ga suki da}
    {Cats are liked} is true about Tarou.
    Tarou likes cats.
  • neko wa {Tarou ga suki da}
    {Tarou is liked} is true about cat.
    The cat likes Tarou.
  • {neko no suki na} Tarou
    (two meanings.)
  • If Tarou is the large subject:
    • {Cats are liked} is true about Tarou, thus:
    • The Tarou [that] {likes cats}.
  • If Tarou is the small subject:
    • {Tarou is liked} is true about cat, thus:
    • The Tarou [that] {the cat likes}.

The sentence above is ambiguous because no の marks the subject but there are two subjects so it can mark either subject.

But wait, if no の can mark the small subject, and it can also mark the large subject, then what is the caveat of no の in double-subject constructions?

The caveat is that it can't mark both of them at the same time.

As we've seen before, node ので comes after a relative clause. If that relative clause has a double nominative, you can't use the subject-marking no の in it. You can only use ga が.
  • {Tarou ga takoyaki ga suki na} no de
    Because Tarou likes takoyaki.

Why can't you use no の? Well, to begin with, if you mark both large and small subjects with no の, and the sentence ends with no の, you end up with a bunch of no の.
  • *{Tarou no {takoyaki no suki na}} no de

If you don't mark both, and only mark one subject with no の, there must be a reason for marking one with no and the other one without, and such reason simply doesn't exist.
  • *{Tarou ga {takoyaki no suki na}} no de
    (also wrong.)

Worse yet, if the large subject is marked with no の, it will probably be interpreted as possessive, since that's one of the usual functions of no の.
  • {{Tarou no} takoyaki ga suki na} no de
    Because [someone] likes the takoyaki {of Tarou}.
    Because [I] like {Tarou's} takoyaki.

A more smart-sounding reason for this is that relative clause is modifying the nominalizer no の, and the nominalizer isn't an argument for the verb inside the relative clause.

It isn't a subject or object or anything. It merely turns the relative clause into a noun. Since it's not linked to the verb in any way, the subject-marking no の can't link the subject inside the relative clause the noun outside of the relative clause.

In multiple-nominative sentences, the largest subject can be qualified by the Matryoshka predicate containing two smaller subjects. However, a small subject can't be qualified by a relative clause containing an even smaller subject.(Seraku, 2017, p. 108)
  • sono otoko ga {imouto ga {kami ga nagai}}
    That man's {younger sister's {hair is long}}.
    • This is valid.
  • {imouto ga {kami ga nagai}} sono otoko
    That man [whose] {younger sister's {hair is long}}.
    • This is also valid.
  • *sono otoko ga {{kami ga nagai} imouto}
    That man's {younger sister [whose] {hair is long}}.
    • This isn't valid, because the small subject imouto is qualified by a relative clause containing the even smaller subject kami.
    • Seriously, my head hurts.

Dative Subject

The large subject may be marked by the ni に particle, which marks the dative case.
  • ABがC
    A is a nominative nominal.
    About A, BがC is true.
  • ABがC
    A is a dative nominal.
    For A, BがC is true.
    To A, BがC is true.

In this case, turning the dative ni に into the topic makes it niwa には. Since we've already established that most the usual way is wa plus ga, not ga plus ga, it's safe to assume niwa is more common than ni, too.
  • watashi niwa {mahou ga tsukaeru}
    {Use-able is true about magic} is true for me.
    {Magic is use-able} is true for me.
    For me, magic is use-able.
    To me, magic is usable.
    I can use magic.
  • {watashi ni tsukaeru} mahou
    Magic [that] {is use-able for me}.
    Magic [that] {I can use}.

Above, we have a verb in the potential form, tsukau 使う, "to use," tsukaeru 使える, "use-able," "can use." The potentiality of the action, the capability of the action, is attributed to the large subject. Magic is use-able, to a select number of people, including, for example, me.

Besides potentials, the dative nominal is also found in cognitives.
  • watashi niwa {sore ga wakaranai}
    {Not understood is true about that} is true for me.
    {That isn't understood} is true for me.
    To me, that isn't understood.
    I don't understand that.

With wakaru the small subject is often omitted.
  • watashi niwa wakaranai
    I don't understand [it].
    I don't get [it].

The dative nominal is also found in certain possessions.
  • ore niwa {imouto ga iru}
    {A younger sister exists} is true for me.
    To me, a younger sister exists.
    I have a younger sister.
  • otona niwa {{kodomo ni wakaranai} jijou ga aru}
    {Circumstances [that] {children don't understand} exist} is true for adults.
    Adults have circumstances [that] {children don't understand}.

One thing you may be wondering is: what is the difference between ni に and niwa には, and ga が and wa は?

Indeed. That's a tough one.

In many cases they're interchangeable. You can use either ni に or ga が, or the niwa には or wa は. In some cases they are not.

For example, you can't use suki and kirai with dative large subjects (ni, niwa). You can only use suki and kirai with nominative large subjects (ga, wa). This is despite the fact other cognitives, like wakaru and kowai, work with dative subjects.

According to Shibatani (1999, pp.68-69), the difference is that the suki and kirai have a higher dependency on the large subject than kowai and wakaru. A low-level of dependency between large and small subjects allows you to use dative subjects, while a high-level forbids it.
  • Tarou wa anime ga suki da
    Tarou likes anime.
  • *Tarou niwa anime ga suki da
    (this is too high-level for niwa.)

This difference in dependency also affects the interpretation of elliptical sentences and contrastive topics.

As we've seen before, X wa suki is elliptical, and ends up getting interpreted as a contrastive wa は with an omitted large subject cognizer, because suki depends highly on a cognizer so a cognizer ends up being required to exist somewhere.

On the other hand, kowai, "scary," omoshiroi 面白い, "entertaining," and so on, have low-dependency and do not require a cognizer. So saying X wa kowai isn't automatically interpreted as a contrastive wa は.
  • Tarou niwa {kumo ga kowai}
    Tarou is afraid of spiders.
  • Tarou wa kumo ga kowai
    (also acceptable.)
  • kumo wa kowai
    Spiders are scary.
    • All spiders are scary.
    • They're objectively scary.
    • The scary quality is inherent of arachnids.
    • Arachnophobia is just another word for common sense.
    • This is a stand-alone statement.
  • kumo ga kowai
    [To someone], spiders are scary.
    • This sounds elliptical.
    • It's X niwa {kumo ga kowai} without the X niwa.
    • So it's easily interpreted as a personal opinion.
    • They're scary to someone. Personally, not objectively.
    • Since generally the topic is yourself, it means "to me, spiders are scary." It's what you feel, personally.

With possessive double-subjects, as we've seen previously, no becomes ga, or vice-versa, through nominativization, or whatever it's called. It doesn't become ni or niwa.

  • Tarou (wa/ga/no) atama ga ii
    Tarou's head is good}
  • ?Tarou (ni/niwa) atama ga ii
    To Tarou, head is good.
    (this doesn't mean his head is good, so it doesn't mean he is smart.)

Since possessives can't be qualified by dative subjects, they end up being interpreted as cognitions, as personal opinions. This means you can end up in a situation where the using wa は marks the large subject as possessor, but using niwa には marks the large subject as cognizer. Observe:
  • watashi wa {doresu ga ao to kuro}
    {The dress is blue and black} is true about me.
    My dress is blue and black.
    • This is a possessive. I possess the small subject.
  • watashi niwa {doresu ga shiro to kin}
    {The dress is white and gold} is true for me.
    Personally, I think that the dress is white and gold.
    • This is a cognition. I cognize the predicative clause.

On the other hand, the existence or nonexistence of a possession can be either a general or particular statement. So nominative subjects and dative subjects are both valid.

  • hito (wa/ga/no) shippo ga nai
    Humans' tail is nonexistent.
  • hito (ni/niwa) shippo ga nai
    To humans, tails don't exist.
    • To humans, in particular, tails don't exist.

This explains why those idiomatic expressions like "smart," atama ga ii, always take the nominative wa or ga before the possession atama, while kinship relationships, like imouto ga iru, can take the the dative ni or niwa.

Even though both are inalienable possessions, one expression is qualifying the possession (atama ga ii), while the other is talking about its existence (imouto ga iru). Expressing existence can be done with niwa, but the quality can not.

Large Subject Control

In more complicated sentences, you often have to deal with relative clauses, so patterns that would be wrong in simple sentences become allowed because the small subject is in a different clause. For example:
  • Aには{Bがsuki da}
    This is wrong.
  • Aには{{Bがsuki na}CがD}
    This is alright.

  • watashi niwa {{anime ga suki na} riyuu ga aru}
    {A reason [why] {anime is liked} exists} is true for me.
    I have {a reason {to like anime}}.

Above, the skeleton of the sentence is:
  • watashi niwa {... riyuu ga aru}
    I have {a reason ...}.

This is a possessive double-subject construction. The small subject riyuu is qualified by a relative clause:
  • {anime ga suki na}
    Anime is liked.

But above we have a problem. As we already know, suki has a high level of dependency and requires a cognizer. One that's not marked by niwa. In the sentence, such thing doesn't exist. How is this supposed to work, then?

Linguistically, this would be called "control."

In English, the verb "to have" is transitive.
  • I have a cat.
    • I - subject.
    • Have - verb.
    • Cat - object.

Therefore, "I have a reason" is grammatically correct, because the object is the phrase "a reason."

Similarly, "to like" is transitive.
  • I like anime.
    • I - subject.
    • Like - verb.
    • Anime - object.

However, when I say "to like anime," there's an object, "anime," but no subject coming right before it. The word "to" is a preposition, it's not a subject.

So how does "I have a reason to like anime" works, grammatically-speaking? What's the subject of "to like anime"? Naturally, what we are saying is:
  • watashi wa {anime ga suki da}
    I like anime.
  • watashi niwa
    {watashi ga {anime ga suki na}}
    riyuu ga aru

    I have a reason {I like anime}.
    • The predicative copula da becomes the attributive copula na in attributive clauses.
    • The sentence topic must be in the main clause, so in subordinate clauses wa becomes ga.

The subject of "have" controls the subject of "like." At least that's how it's called, technically. In practice, it means the subject of "like" is the same thing as the subject of "have."

In Japanese, the same thing happens: the dative subject of riyuu ga aru controls the nominative subject of anime ga suki da. They're the same subject, despite being marked differently.


1 comment:

Leave your komento コメント in this posuto ポスト of this burogu ブログ with your questions about Japanese, doubts or whatever!

All comments are moderated and won't show up until approved. Spam, links to illegal websites, and inappropriate content won't be published.