Skip to main content

The が Particle: Subject Marker (and More)

The が particle is the Japanese subject marker. It attaches to a noun phrase to flag it as the grammatical subject (主語) of a clause.1 2 3 But "subject marker" only describes half the job. が also marks the noun phrase as new information, or as the exhaustively singled-out answer to an implicit "which one?".1 4 5

Overview

What が actually marks

が is a 格助詞 (kakujoshi, "case particle"). In traditional school grammar, 格助詞 are particles that mark grammatical roles on noun phrases (が, を, に, で, へ, と, から, より, まで). They contrast with 係助詞 (kakari-joshi, "binding particles") like は and も, which mark information-structure roles.2 4

The role が assigns is nominative case: it marks the grammatical subject (主語), the argument that the predicate is about.1 2 3

This is the syntactic baseline for the article. Every Japanese clause has a predicate. Its subject, when expressed, is marked by が unless the speaker has topicalized that subject with は. In that case, が is suppressed, not stacked, as explained below.4 3

あめる。6
"It is raining."

子供こどもいている。7
"A child is crying."

いぬにくべた。6
"The dog ate the meat."

Why "subject marker" is only half the story

が has two distinct functions that appear with the same kana. Kuno (1973) is the canonical reference for this split: が has a neutral-description reading (中立叙述) and an exhaustive-listing reading (排他, also called 総記).1 5

The neutral-description reading is the plain "report what is happening" use, as in 雨が降る. The exhaustive-listing reading singles the subject out as the one that satisfies the predicate, to the exclusion of other candidates, as in 私が行きます ("I, and no one else, am the one going").1 4

In both readings, が still marks the grammatical subject. The split is about an information-structure layer on top of the case marker, not about two different が particles. Makino and Tsutsui's Dictionary of Basic Japanese Grammar formalises the same two-way analysis with their "ga¹" / "ga²" split.4 5

Two readings, one particle

The Japanese-language reference grammar tradition uses the terms 中立叙述 ("neutral description") and 排他 ("exclusive") for these two readings.5 The English-language analytical tradition uses "neutral description" and "exhaustive listing".1 2 This article uses the descriptive English labels and uses the Japanese terms only where the analytical name is unavoidable.

Register, JLPT level, and pronunciation

が is register-neutral. It appears in every register from intimate speech to formal writing, and in every text type from spoken dialogue to legal prose.4 3

The official JLPT N5 practice workbook uses が in its first reading-comprehension passages without a separate gloss, confirming its N5 status.8 The analytical split between neutral-description and exhaustive-listing readings, and the が/の alternation in relative clauses, extend into N3 territory.1 4 5

Standard pronunciation is /ga/ with a voiced velar plosive [g]. In traditional Tokyo speech and conservative NHK broadcasting, mid-word が within a phonological phrase is pronounced as the velar nasal [ŋa], called 鼻濁音 (bidakuon). Because the case-particle が attaches at a word boundary inside the phonological phrase, it is one of the positions where the nasal variant historically applies.9 10 The "Good to know" section returns to this.

How が works: the case-marking job

The basic frame: [X が] [predicate]

The minimal Japanese clause is [subject-が] [predicate]; nothing else is structurally required.4 3

The subject slot is filled by a noun phrase or an equivalent form: a pronoun, a proper name, or a nominalized clause.3 The predicate can be a verb, an い-adjective, or a noun + copula phrase.4 When a direct object is present, it sits between the subject and the verb and is marked by : [X が] [Y を] [verb].4 6

あめる。6
"It is raining."

子供こどもく。7
"A child cries."

いぬにくべた。6
"The dog ate the meat."

ちゃつめたいです。7
"The tea is cold."

When は is absent, the subject takes が

In a clause with no topic phrase (no は), the grammatical subject is overtly marked by が.2 4 This is the default state of a Japanese clause: topicalization with は is an additional, optional step layered on top of the basic case-marked structure. If the speaker has not topicalized anything, the subject surfaces as [N が].2 4

This article assumes you have already met は as a topic marker. The full topic-marker mechanics are covered in "The は Particle: Topic Marker." Here only the negative half of the rule matters: no は means the subject takes が.4

とりんでいる。7
"A bird is flying."

電車でんしゃた。7
"The train came."

が cannot stack with は on the same noun

When the same noun phrase is both topic and subject, は overrides が. The case marker is suppressed, not stacked. The doubled sequences はが and がは do not exist in modern standard Japanese.2 4

The mechanism, following Heycock (2008) and Makino and Tsutsui (1986), is this: が and を mark the two least-oblique, or most core, grammatical roles (subject and direct object). The information-structure layer realised by は overrides those slots rather than co-existing with them.2 4

Oblique case particles behave differently. , で, へ, と, から, and まで all stack with は: には, では, へは, とは, からは, までは are all grammatical. The suppression rule is specific to が and を.4

わたし学生がくせいです。4
"I am a student."

The doubled marker はが does not exist

A learner who knows both は and が may try to combine them on the same noun ("私はが学生です") on the model of "I, the subject, am a student." This is ungrammatical. When the subject is also the topic, write は alone; the が is suppressed.4

東京とうきょうにはったことがある。4
"I have been to Tokyo."

The locative に stacks with は to form には ("as for Tokyo"). The subject of ある keeps its own が because that が is on a different noun phrase, not on the topic.

が as a focus marker: new information and exhaustive listing

The new-information rule

When a referent is introduced into discourse for the first time, it is new information (新情報) to the listener and is marked by が. Once the same referent has been established, later mentions usually shift to は.1 4

Kuno's formulation: the contrast between が and は at the opening of a discourse is the contrast between "this referent is being put on stage for the first time" (が) and "this referent is already known and is what I am going to comment on" (は).1

The textbook demonstration is the opening of a Japanese folktale, where the protagonists are first introduced with が and then carried forward with は.1 4

昔々むかしむかし、おじいさんとおばあさんがいました。おじいさんはやま柴刈しばかりにきました。4
"Long ago, there were an old man and an old woman. The old man went to the mountain to cut brushwood."

The shift from が to は across the two sentences is the clearest single demonstration of the rule in the language.

Exhaustive-listing focus: "X, and X alone, is the one"

The exhaustive-listing reading of が singles the subject out from a candidate set: "X, and not the other candidates, is the one that satisfies the predicate."1 4 5

Kuno (1973) treats this as a focus function and gives it the analytical name 総記 (sōki, "exhaustive listing"). Noda (1996) uses the alternative term 排他 (haita, "exclusive"). Both names point to the same effect.1 5

This reading is the natural answer to a who- or which-question. Such questions implicitly establish a candidate set and ask the addressee to single one out. Tae Kim's "identifier" framing of が is a pedagogical restatement of the same fact.11

わたしきます。4
"I am the one going."

ジェニーが犯人はんにんだ。12
"Jenny is the culprit."

アリスがべた。11
"Alice ate it."

This exhaustive force supports English glosses such as "I (not anyone else) am going," "Jenny (and no other suspect) is the culprit," and "Alice (the answer to your question) ate it." が is doing focus work on top of the basic case marking.

Neutral-description vs. exhaustive-listing reading

Predicate type strongly conditions which reading a sentence-initial が-phrase receives.1 2 5

Eventive predicates (descriptions of events or temporary states such as rain, cry, arrive, or open) tend to license the neutral-description reading. 雨が降る, 子供が泣いている, and パン屋が開いた are reports of what is happening. The が-phrase is not being singled out from a candidate set.1 2

Stative or identifying predicates (descriptions of inherent properties or identity, such as "be the new one," "be Tanaka," or "be the library") tend to force the exhaustive-listing reading on a sentence-initial が-phrase. 私が田中です, パン屋が新しい, and あの白い建物が図書館です all single the subject out as the one that matches the property.1 2

Heycock (2008) reformulates this in modern terms: with an individual-level predicate, a sentence-initial bare が-phrase cannot naturally be interpreted as a one-off event report. The only available reading is exhaustive focus.2

パンいた。2
"A bakery opened."

The eventive predicate 開いた licenses a neutral-description reading: this is a report of an event, not a contrast against other bakeries.

パンあたらしい。2
"The bakery is the new one."

The stative predicate 新しい forces the exhaustive-listing reading: the bakery, as opposed to other candidates in the conversation, is the new one.

あのしろ建物たてもの図書館としょかんです。1
"That white building is the library."

The identifying predicate です forces exhaustive listing: that one, not the others, is the library.

Where が is non-negotiable: four environments

There are four environments where が is the only correct choice and a learner cannot fall back on は. These are not four scattered idioms. They are four predictions of a single underlying rule: each one is a position where the topic marker は cannot meet its semantic requirements. The referent must already be established, or at least known and commentable on. In these positions, that requirement cannot be met, so the case marker has nowhere to be overwritten from.1 2 4 3 J-Compass groups them together for diagnostic convenience.

The diagram shows the decision path for picking が. The four required-が cases are on the right branch. The topicalization escape hatch is on the left branch.

Question-word subjects always take が

When the grammatical subject of a clause is itself a question word, it must be marked by が, never は. This includes 誰, 何, どれ, どこ as the subject of an existential, and いつ as the subject of a time predicate.1 4

The mechanism is simple: question words refer to unknowns, and an unknown cannot be the established discourse topic that は requires. The information-structural slot marked by は is incompatible with the open-variable semantics of a question word.1 2

The mirror rule applies to answers. A noun phrase that answers a who- or which-question inherits the exhaustive-listing reading from the question, so it is marked by が.1 11

だれましたか。4
"Who came?"

なにがいいですか。7
"What would you like?"

どのかたがトーフグの社長しゃちょうさんですか。12
"Which person is the president of Tofugu?"

田中たなかさんがました。4
"Tanaka-san came."

The last example shows the mirror rule. An answer to 誰が来ましたか takes が on 田中さん because the answer slot is exhaustive ("Tanaka, of the candidates").

Stative predicates take が

A class of stative predicates marks the object of the state with が rather than with the を that English-trained learners often expect. Shibatani (1990) analyses this as nominative-object marking: the second argument of these stative predicates is structurally an object but is realised with the nominative particle.3

The inventory:4 3

Predicate classMembers
Preference / evaluative なadj好き, 嫌い, 上手, 下手, 得意, 苦手
Desiderative いadj欲しい
Cognition / abilityわかる, できる
Existentialある (inanimate), いる (animate)
Desiderative on verb stem〜たい

The first argument of these predicates is the experiencer, the person who likes, understands, or wants something. It is typically realised as the topic with は, or as a dative experiencer with に or にとって. が attaches to the object-of-state.4 3

日本語にほんごきです。7
"I like Japanese."

漢字かんじがわかる。12
"I understand kanji."

かねがある。4
"I have money."

小犬こいぬしい。12
"I want a puppy."

寿司すしべたい。4
"I want to eat sushi."

For English-trained learners, this is the single most common N5-level mistake: they reach for を on the model of English transitive verbs. The "Good to know" section returns to it as a labelled pitfall.

Subjects of subordinate and relative clauses

Inside a subordinate clause, the embedded subject is marked by が, not は. This includes temporal clauses with 時, 前, and 後; adverbial clauses with ながら; and adnominal or relative clauses.2 4 5

This is the clause-level versus sentence-level asymmetry: thematic は operates at the level of the whole utterance and is restricted to matrix clauses. が is the case marker of the embedded subject regardless of clause depth.2 5

The diagnostic value for a learner is direct. When you see a multi-clause sentence, the subject of any non-matrix clause will almost always surface with が.2 5

わたしかえったときあめっていた。4
"When I came home, it was raining."

あめまえに、洗濯物せんたくものんだ。4
"I brought in the laundry before it rained."

わたしったほん面白おもしろい。1
"The book I bought is interesting."

The third example shows the relative-clause case. 私が買った is the embedded clause modifying 本, and its embedded subject takes が. The matrix-level topic は then attaches to 本, "the book."

The が/の alternation in relative clauses

Inside an adnominal (relative) clause, the embedded subject can be marked by either が or の. The two forms are typically interchangeable when the right conditions are met. This is the が/の alternation (ガノ交替). It was documented by Harada (1971) and reanalysed by Hiraiwa (2001) as licensed specifically by the adnominal form of the predicate, the environment where a clause is modifying a noun.13 14

The diagram shows the contrast: が is always available. の is licensed only when several narrow conditions line up.

The conditions, drawn from Harada and Hiraiwa:13 14

  • The embedded clause must be in adnominal form, modifying a head noun.
  • The embedded clause is typically short.
  • There is typically no direct object inside the clause. A を-marked object makes the の variant less acceptable for many speakers (the so-called transitivity restriction).
  • The embedded subject must not carry contrastive or emphatic focus. の disprefers focus, so when the embedded subject is being emphasised, が is the only natural choice.

The の form is more frequent in writing and in fixed phrases, such as 私の知る限り "as far as I know" and 母の作った料理 "the food my mother made." It is recessive in spoken Japanese, where が is the unmarked choice.13

わたしったほん1
"the book I bought"

わたしったほん13
"the book I bought"

The two phrases are equivalent in reference; the の variant reads as slightly more written-register.

ははつくった料理りょうりはおいしい。13
"The food my mother made is delicious."

子供こどもほん14
"books that children read"

Substituting の into the last example reads as marginal: children-as-a-generic-subject takes が by default.

Good to know

"が means but" is a different particle

Japanese has a separate particle, also spelled が, that links two clauses with a "but, however" reading: コーヒーは好きですが、お茶は嫌いです ("I like coffee, but I dislike tea"). In school-grammar terminology, this is a 接続助詞 (setsuzokujoshi, "conjunctive particle") rather than a 格助詞 (kakujoshi, "case particle").4 3

The two are spelled with the same kana but have unrelated distributions. Case-marking が attaches to a noun phrase and assigns nominative case. Conjunctive が attaches to the end of a clause and links it to the following clause. The conjunctive use is out of scope for this article; the conjunctions-and-connectives subcategory of the curriculum covers it.4

The [ŋa] nasal variant of が

In conservative Tokyo speech and traditional NHK broadcast pronunciation, the kana が is realised as the velar nasal [ŋa] (鼻濁音, bidakuon) in word-internal positions. At the start of a word and in loanwords, it remains the voiced plosive [ga].9 15 10

The case-particle が attaches at a word boundary but inside a phonological phrase, so it is one of the positions where the nasal variant historically applies. A conservative speaker pronounces 私が as /watashi-ŋa/.15 10

The distinction is recessive among younger speakers, and the recession is more advanced at commercial broadcasters than at NHK. The Japanese Wikipedia article on 鼻濁音, summarising broadcast-industry reports, describes the contrast as lost on air for most younger commercial-broadcaster announcers. Among younger NHK announcers, it describes the loss as affecting only a sizeable minority.15 NHK's pronunciation dictionary still records the [ŋ] realisation as the recommended form for announcers, but institutional enforcement has softened.9

The practical takeaway for the learner is this: producing [ga] in all positions sounds natural and modern. Producing [ŋa] mid-word is what older Tokyo speakers and traditional broadcasters do, and it is worth being able to hear without being expected to imitate.

The は/が exchange in story openings

The clearest, most memorable demonstration of the new-information / established-topic split is the opening of a Japanese folktale. The protagonists are introduced with が because they are entering the discourse for the first time. From the next sentence onward, they are referred to with は because they are now the established topic.1 4

昔々むかしむかし、おじいさんとおばあさんがいました。おじいさんはやま柴刈しばかりにきました。おばあさんはかわ洗濯せんたくきました。4
"Long ago, there were an old man and an old woman. The old man went to the mountain to gather brushwood. The old woman went to the river to do laundry."

Memorising this template gives a learner the new-information versus established-topic split as a single pattern.

Using を after a stative predicate

The first instinct for an English-trained learner faced with 好き, 欲しい, わかる, or できる is often to mark the object of the state with を, on the model of English "I like X." The correct particle is が. The wrong form 日本語を好きです is the canonical N5 mistake.4 3 7 12

日本語にほんごきです。7
"I like Japanese."

The full inventory of stative predicates that take が on the object-of-state is in the "Stative predicates take が" section above: 好き, 嫌い, 上手, 下手, 得意, 苦手, 欲しい, わかる, できる, ある, いる, plus 〜たい on verb stems. Learning the list as a closed class is how to defeat the English-transfer error.

Using は instead of が on a question-word subject

A second common N5 mistake is to mark a question-word subject with は: 誰は来ましたか. The correct particle is が.1 4

だれましたか。4
"Who came?"

Question words refer to unknowns, and an unknown cannot be the established topic that は requires. The slot marked by は is incompatible with the open-variable semantics of 誰, 何, and どれ. Question-word subjects must take が.1 4

Using は inside a relative clause

A third N5 mistake is to use は on the embedded subject of a relative clause: 私は買った本.1 13 The correct particle inside an adnominal clause is が. の is a stylistic alternative inside short clauses without an internal object.

わたしったほん1
"the book I bought"

Thematic は is restricted to matrix clauses. A は inside a relative clause cannot be interpreted as marking the embedded subject. The parser reanalyses it as the matrix topic, and the relative-clause reading breaks.2 13

The は vs が distinction belongs to its own article

This article explains what が does: it marks a grammatical subject, with neutral-description and exhaustive-listing readings, and it is required on question-word subjects, stative-predicate objects-of-state, and subordinate-clause subjects. The harder question of which particle to pick when both are grammatical belongs in the beginner-level overview article on は versus が. The full topic-marker mechanics are covered in the article on the は particle.1 5

See also

References

Footnotes

  1. Kuno, Susumu. The Structure of the Japanese Language. MIT Press, 1973. (Current Studies in Linguistics 3.) ISBN 978-0-262-11049-5. Chapter 2 on は vs が, including the exhaustive-listing / neutral-description distinction. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

  2. Heycock, Caroline. "Japanese -wa, -ga, and Information Structure." In The Oxford Handbook of Japanese Linguistics, edited by Shigeru Miyagawa and Mamoru Saito, Oxford University Press, 2008, pp. 54–83. Author manuscript: https://www.lel.ed.ac.uk/~heycock/papers/topic-draft.pdf 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

  3. Shibatani, Masayoshi. The Languages of Japan. Cambridge University Press, 1990. ISBN 978-0-521-36918-3. Chapters on case-marking, nominative-object constructions, and stative predicates. 2 3 4 5 6 7 8 9 10 11 12

  4. Makino, Seiichi, and Michio Tsutsui. A Dictionary of Basic Japanese Grammar. The Japan Times, 1986. ISBN 978-4-7890-0454-1. Entry: "ga¹" (subject marker) and "ga²" (exhaustive-listing / contrastive marker). 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

  5. 野田尚史 (Noda, Hisashi). 『「は」と「が」』("Wa" to "Ga"). くろしお出版 (Kurosio Publishers), 1996. ISBN 978-4-87424-122-7. Standard Japanese-language reference on the は / が distinction; covers neutral-description が (中立叙述) vs exhaustive-listing が (排他). 2 3 4 5 6 7 8 9 10 11 12

  6. JLPTsensei. "JLPT N5 Grammar: が (ga) Subject Marker Particle." https://jlptsensei.com/learn-japanese-grammar/%E3%81%8C-ga-subject-marker-particle/ (limitation: language-learning publisher; used for sourced N5-level example sentences.) 2 3 4 5

  7. Bunpro. "が (JLPT N5 Grammar Point)." https://bunpro.jp/grammar_points/%E3%81%8C (limitation: language-learning publisher; used for JLPT-level confirmation, structure box, and beginner example sentences with verified readings.) 2 3 4 5 6 7 8 9

  8. 国際交流基金・日本国際教育支援協会 (Japan Foundation & Japan Educational Exchanges and Services). Japanese-Language Proficiency Test Official Practice Workbook N5 / 『日本語能力試験 公式問題集 N5』. 凡人社, 2012. ISBN 978-4-89358-840-1. Official JLPT N5 practice materials assume が as subject marker from the first reading.

  9. 日本放送協会 (NHK). 『NHK日本語発音アクセント新辞典』. NHK出版, 2016. ISBN 978-4-14-011345-5. Reference treatment of ガ行鼻濁音 in standard broadcast pronunciation. 2 3

  10. Vance, Timothy J. The Sounds of Japanese. Cambridge University Press, 2008. ISBN 978-0-521-61754-3. Chapter 6 on the allophonic distribution of /g/ and [ŋ] in standard Japanese. 2 3

  11. Tae Kim. "The Difference Between 「は」 and 「が」." Tae Kim's Guide to Learning Japanese. https://guidetojapanese.org/learn/difference-between-wa-and-ga/ (limitation: language-learning blog; used only for the "identifier" framing and the question-word answer pattern.) 2 3

  12. Tofugu. "Particle が: Subject Marker." https://www.tofugu.com/japanese-grammar/particle-ga/ (limitation: language-learning publisher; used for the "as opposed to" gloss convention, the new-information framing, and several pedagogical example sentences.) 2 3 4 5

  13. Harada, S.-I. "Ga-No Conversion and Idiolectal Variations in Japanese." Gengo Kenkyū, no. 60, 1971, pp. 25–38. Reissued in Harada, Syntax and Meaning: S.-I. Harada Collected Works in Linguistics, edited by Naoki Fukui, Taishukan, 2000. 2 3 4 5 6 7

  14. Hiraiwa, Ken. "On Nominative-Genitive Conversion." In MIT Working Papers in Linguistics 39: A Few from Building E39, edited by Elena Guerzoni and Ora Matushansky, MIT Department of Linguistics, 2001, pp. 66–125. https://ojs.ub.uni-konstanz.de/jsls/index.php/jsls 2 3

  15. Wikipedia contributors. "鼻濁音 (Bidakuon)." 日本語版ウィキペディア. https://ja.wikipedia.org/wiki/%E9%BC%BB%E6%BF%81%E9%9F%B3 (limitation: encyclopedic reference; used for the cited percentages on younger announcers and the word-initial / word-medial rule statement.) 2 3