Japanese Relative Clauses: Modifying a Noun With a Whole Sentence

Japanese relative clauses modify a noun by placing a complete plain-form clause directly in front of it, with no relative pronoun and no connecting word.¹² Once you can read that gap, "the book that I read" and "the town where I live" stop being puzzles and become a single, regular pattern.

Overview

A relative clause (also called a noun-modifying clause) is a kind of embedded clause. It works like an adjective: it sits before a noun and tells you which noun is meant.¹³ In Japanese, the clause comes first and the noun comes last, so the structure is the mirror image of English.

The Japanese grammatical term is 連体修飾節 (rentai-shūshoku-setsu), "prenominal modifying clause." It means a clause placed before a 体言 (a noun or substantive) to modify it.²¹

Why there is no "that," "which," or "who"

English puts the head noun first and then attaches a clause with a relative pronoun: "the book that I read." Japanese has no such pronoun and no connector at all. The clause simply attaches, and the head noun comes last.¹

寝ねている人ひと⁴
"a person who is sleeping."

流ながれる水みず⁴
"water that flows."

Where English needs "who" and "that," Japanese needs nothing between the clause and its noun. The slot the noun would have filled inside the clause is left empty. That empty slot is the "gap," covered in its own section below.¹

私わたしが作つくった椅子いす。³
"a chair I made."

Where this sits at JLPT N4

A relative clause is not polite or casual in itself. It is a structural building block you can use in any register, because the main predicate decides politeness, not the modifying clause.¹

The bare construction (a single plain-form verb modifying a noun, such as 読んだ本 "the book I read") appears early and is tagged N5 in at least one major reference.³ The fuller system taught here, including the gap, the が/の subject alternation, and clause-internal tense, is N4-level material. It builds on plain form and the た-form.

Why no single JLPT tag exists for "relative clauses"

References do not assign one uniform JLPT level to "relative clauses," because Japanese has no dedicated relative-pronoun marker to list as a grammar point. Instead, the construction is taught as "noun modification by a clause." Different references place the bare form at N5 and the full treatment at N4.³

How to build one: the head-final rule

The basic shape: [clause] + [noun]

Take a complete plain-form clause and place it directly before the head noun. The whole clause then behaves as a single modifier of that noun.¹²

Start from a full sentence: 私が読んだ watashi ga yonda "I read." Add the noun 本 hon "book." The result is 私が読んだ本, "the book that I read." The sentence becomes a modifier sitting in front of the noun. The reference uses 読んだ本 "the book I read" to introduce verb-based noun modification.³

Because the clause is head-final, the modifier can be a single word or a long string, and the noun always lands at the end.

洗あらった服ふく。³
"clothes that someone washed."

たくさん勉強べんきょうをした生徒せいと。³
"a student who studied a lot."

友達ともだちから借かりたペン。³
"a pen I borrowed from a friend."

The internal predicate is always plain form

Only verbs and adjectives in short, or plain form, may modify a noun. Do not use the polite ます／です forms inside the clause, even when the main sentence is polite.¹³²

The plain form spans the full paradigm. 読む (nonpast), 読んだ (past), 読まない (negative nonpast), and 読まなかった (negative past) all modify a noun directly.

日本にほんに住すんでいる外国人がいこくじん。³
"a foreigner living in Japan."

学生がくせいじゃない人ひとは、学校がっこうに行いかない。¹
"People who are not students do not go to school."

The topic particle は does not survive inside the clause

Inside a relative clause, the topic marker は is replaced by が. は does not appear as a clause-internal subject marker. The が-marked subjects in every object and adjunct example below confirm this.¹

Adjectives and noun predicates as modifiers

い-adjectives modify a noun directly in their plain form, exactly like a verb (寒い + noun).¹

A noun or な-adjective predicate modifies through だった for the past. The plain nonpast copula だ is the one form that does not directly modify a following noun; use the な or の linker instead. The past copula だった, however, modifies directly.¹

子供こどもだったアリスが立派りっぱな大人おとなになった。¹
"Alice, who was a child, became a fine adult."

先週せんしゅう医者いしゃだったボブは、仕事しごとを辞やめた。¹
"Bob, who was a doctor last week, quit his job."

The "gap": what the noun was doing inside the clause

The head noun fills an empty slot inside the clause. This is the slot it would have occupied as a full argument if the clause stood on its own. That empty slot is the "gap." It can be the subject, the direct object, or an adjunct / oblique.¹⁴

The diagram below shows the same head noun 本 standing outside the clause. Its grammatical role, the を-object, is empty inside the clause.

Subject gap

The head noun is the subject of the clause verb, so the が-slot is the gap. 走っている人 reads as "the person who is running," with the runner left empty inside the clause.

寝ねている人ひと⁴
"a person who is sleeping."

北海道ほっかいどうを旅行りょこうする田中たなかさん²
"Tanaka-san who will travel in Hokkaido."

Direct-object gap

The head noun is the direct object of the clause verb, so the を-slot is the gap. 私が読んだ本 reads as "the book that I read," with the thing read left empty inside the clause. The reference introduces this verb-modifies-noun pattern with 読んだ本, "the book I read."³

作家さっかが書かいた小説しょうせつ⁴
"the novel which the author wrote."

私わたしが作つくった椅子いす。³
"a chair I made."

忍者にんじゃが殺ころした人ひと⁴
"the person that the ninja killed."

The academic literature confirms the object-gap pattern with a minimal pair in which the subject may take either が or の. Ochi presents these in romanization with morpheme glosses (NOM = nominative が, GEN = genitive の) and marks the empty object position as [e]. That romanized form is the authoritative quoted form.⁵

Taro-ga [e] katta hon⁵
"the book that Taro bought" (NOM subject).

Taro-no [e] katta hon⁵
"the book that Taro bought" (GEN subject).

Adjunct / oblique gap (place, time, instrument)

The head noun can also correspond to an adjunct: a place (に / で), a time, or an instrument (で). The particle slot is emptied, just as with the subject and object. A place head such as 田中さんが旅行する北海道 reads as "Hokkaido where Tanaka-san will travel," with the place slot left empty inside the clause.²

田中たなかさんが旅行りょこうする北海道ほっかいどう²
"Hokkaido where Tanaka-san will travel."

金かねが降ふってきた日ひ⁴
"the day when money rained."

金かねが降ふってきた場所ばしょ⁴
"the place where money rained."

A place or adjunct head behaves differently from the subject and object gaps when the が/の alternation comes into play: if the clause also contains an overt を-object, the genitive の subject is blocked. The minimal pair for that restriction appears in the next section.⁵

The が/の subject alternation

When の can replace が

Modern Japanese permits a nominative-genitive alternation inside a prenominal clause. In basic cases, the subject may be marked with either が or の, "though the semantic contrast appears indiscernible."⁵ In the literature, this is called Ga-No Conversion, discussed since Harada (1971).⁵

The substitution is most natural with a short subject sitting directly before the predicate.²

The minimal pairs below are quoted from Ochi in romanization with glosses; the romaji is the authoritative form, and the kanji rendering is supplied as a faithful transcription.⁵

Taro-ga naita riyuu⁵
"the reason that Taro cried" (NOM).

Taro-no naita riyuu⁵
"the reason that Taro cried" (GEN).

An N4-register pair confirms the same meaning under either particle:

田中たなかさんが作つくるごはん / 田中たなかさんの作つくるごはん²
"meals which Tanaka-san makes" (both grammatical, same meaning).

When it cannot

This substitution is a nuance, not a free global swap. The central constraint is the Transitivity Restriction: the genitive の subject is blocked when the clause contains an overt accusative を object.⁵

Ochi's minimal pair makes the contrast sharp. The form marked with * is ungrammatical; the romaji and grammaticality mark are reproduced as printed.⁵

Taro-ga hon-o katta mise⁵
"the store where Taro bought a book" (NOM subject + ACC object).

*Taro-no hon-o katta mise⁵
same meaning with GEN subject, ungrammatical.

The restriction lifts when the object itself is the gap, that is, when the を-object is phonologically empty. Taro-no [e] katta hon "the book that Taro bought" is fine, because the object is the head noun and is not overt inside the clause.⁵ So の is available in the direct-object-gap case, but not when a second, overt を-object also appears in the clause.

Re-ordering the object does not rescue the genitive

This is not just a question of adjacency. Ochi notes, citing Watanabe (1996), that moving the object to the front does not help: *Hon-o Taro-no katta mise is still ungrammatical.⁵

The practical takeaway: の for が is a style option in short clauses that are intransitive or whose object is the gap. An overt を inside the clause forces が. Long or complex subjects also favor が for clarity.⁵²

Tense inside the clause

Clause tense is independent of the matrix verb

The tense of the clause predicate is relative, not absolute. Choose 買う本 (nonpast, "the book I will buy") or 買った本 (past, "the book I bought") based on the clause's own time reference. It does not have to agree with the tense of the main verb.⁴³

The cleanest demonstration is the same head noun with the predicate swapped between nonpast and past.

消きえるタトゥー⁴
"a tattoo that disappears."

消きえたタトゥー⁴
"a tattoo that disappeared."

The ている form marks an ongoing state inside the clause. Again, it does not depend on the main-clause tense.

日本にほんに住すんでいる外国人がいこくじん。³
"a foreigner living in Japan."

The た form, the ている form, and the plain nonpast are the forms most commonly seen modifying nouns. The choice encodes the time or aspect of the modifying event, not the time of the main clause.³

Nuance and usage contexts

Stacking length and readability

Because the modifier has no fixed length and is always head-final, long pre-noun modifiers are normal and natural in written Japanese. A whole multi-clause description can sit in front of a single noun.¹

晩ばんご飯はんを食たべなかった人ひとは、映画えいがで見みた銀行ぎんこうに行いった。¹
"The person who did not eat dinner went to the bank she saw in the movie."

A modifier can also contain its own object, even when the head noun is the subject of the clause.

赤あかいズボンを買かう友達ともだちはボブだ。¹
"The friend who buys red pants is Bob."

Long, heavily stacked modifiers read as written register. The "long modifiers are normal in writing" point is the directly sourced half of this observation.¹

Internally-headed relatives (preview)

In an internally-headed relative clause, the head noun sits inside the clause rather than after it. The whole clause is then closed with the nominalizer の and case-marked as an argument of the main verb.⁶ This is structurally distinct from the head-final clauses that are the focus of this article, and it is advanced material.

Kikuta notes that "the semantic head (target) of the IHRC is not the syntactic head of the clause." He also notes that the surface form is identical to a sentential complement.⁶ The natural English translation is usually not a relative clause at all, reflecting the construction's non-restrictive character.⁶

The examples below are quoted from Kikuta in his romanization and glosses (nom, acc, nmlzr; for example tabe-ta = 食べた). The romaji is the authoritative form, with kanji supplied as a faithful transcription.⁶

[[Oba-kara ringo-ga okutteki-ta] no] o tabe-ta⁶
"My aunt sent me apples and I ate one."

[[Koppu-ga ware-ta] no o katazuke-ta⁶
"A glass broke and I put it away."

Good to know

Don't insert a "that / which / who" word

The most common error is reaching for a linking word, a という, or a pronoun to translate English "that," "which," or "who." Japanese has none. The clause attaches bare, directly before the head noun, which simply comes last.¹²

読よんだ本ほん³
"the book I read."

Don't use the polite form inside the clause

Learners who are writing a polite sentence often carry the politeness into the modifier, producing forms like 田中さんが作りますごはん with a polite verb inside the clause.² Only the short, or plain, form modifies a noun. Politeness belongs to the main predicate.¹³ The correct form keeps the verb plain.

田中たなかさんが作つくるごはん²
"the meal that Tanaka-san makes."

Don't use a の subject when the clause has an overt を object

A の subject is tempting wherever a が subject sits before the verb. But it is blocked when an accusative を object is present in the same clause, giving the ungrammatical *太郎の本を買った店.⁵ The Transitivity Restriction on Ga-No Conversion forces が in that environment.⁵ The grammatical form uses が.

太郎たろうが本ほんを買かった店みせ⁵
"the store where Taro bought a book."

Read the modifier right-to-left into English

The head noun comes last in Japanese but first in the English translation. Reading 私が読んだ本 noun-first, as "the book ← that I read," maps the head-final Japanese order onto the English relative-pronoun order and makes the gap easy to locate. This is a reading aid for learners. The underlying head-final order it relies on is sourced.¹

The term 連体修飾 (rentai-shūshoku)

連体 means "attaching to a 体言 (a substantive or noun)," so 連体修飾(節) literally names a modifier, or modifying clause, that attaches to a noun.²¹ This is also why the dictionary / plain form is traditionally called the 連体形 (rentai-kei, "noun-attaching form"). It is the form that comes before a noun. The grammar term and the plain form share a root, which is a useful reminder that the plain form is the one used inside the clause.²¹

See also

• Japanese Complement Clauses with の: The Concrete Nominalizer for Perception and Feeling
• Japanese Quotation with と: How to Say What Someone Said or Thought
• Japanese Embedded Questions: How to Say "Whether or Not" with かどうか and か
• Attributive vs. Predicative Adjectives in Japanese: Why 静か Needs な But 大きい Does Not
• Japanese Particles (助詞): The Eight Categories Explained

References

Footnotes

1. Tae Kim (Kim, Tae). "Relative Clauses and Sentence Order." Tae Kim's Guide to Learning Japanese. https://guidetojapanese.org/learn/grammar/clause ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²² ↩²³ ↩²⁴

2. Wasabi. "Japanese Relative Clauses." Wasabi Japanese Grammar Magazine. https://wasabi-jpn.com/magazine/japanese-grammar/japanese-relative-clauses/ (limitation: language-learning publisher, not academic) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵

3. Bunpro. "Verb [た・ている] + Noun" (grammar point #357). Bunpro. https://bunpro.jp/grammar_points/357 (limitation: language-learning service; used for JLPT-level tagging and a corroborated example) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷

4. Leo (japanesewithanime). "Relative Clauses." Japanese with Anime. https://www.japanesewithanime.com/2018/12/relative-clauses.html (limitation: hobbyist reference site; used only for forms independently confirmed by [^1]–[^5]) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹

5. Ochi, Masao. "Ga-No Conversion and Overt Object Shift in Japanese." Nanzan Linguistics 2: Research Results and Activities, Nanzan University, 2004, pp. 61–66. https://www.ic.nanzan-u.ac.jp/LINGUISTICS/publication/pdf/NL2-5-ochi.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹

6. Kikuta, Chiharu Uda. "Qualia Structure and the Accessibility of Arguments: Japanese Internally-Headed Relative Clauses with Implicit Target." Proceedings of the 14th Pacific Asia Conference on Language, Information and Computation (PACLIC 14), Doshisha University, 2000, pp. 153–155. https://aclanthology.org/Y00-1016.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶