Scrambling and Word-Order Flexibility in Japanese

Scrambling and word-order flexibility in Japanese means you can reorder most of a sentence's phrases without changing who did what, because case particles, not position, mark each phrase's grammatical role.¹² The one part that never joins the reshuffle is the predicate, which stays at the very end.¹²

Overview: What "Scrambling" Means

In linguistics, "scrambling" is the technical name for moving phrases away from their default order while keeping the same meaning.¹ In Japanese, the verb must come at the end of the sentence. The other phrases can move more freely, and the resulting orders mean virtually the same thing.¹

The Japanese label for this is かき混ぜ (kakimaze, "mixing up"). A processing study explains it directly: Japanese sentences can change their word order somewhat freely through the operation of scrambling.²

The starting point is the canonical, or unmarked, order. In English, the canonical word order is subject-verb-object (SVO). In Japanese, it is subject-object-verb (SOV).² Scrambling is any departure from that default SOV base.²

Scrambling is awareness, not a form to conjugate

Scrambling is not a verb ending or a grammar pattern you drill. It is a reordering operation: non-canonical orders come from the SOV base by moving the case-marked phrases and leaving each particle attached to its phrase.¹³

The clearest way to feel this freedom is to watch one sentence change order. The three pre-verbal phrases below (subject, a で-location, and an object) can appear in any order. All the orderings are grammatical and mean virtually the same thing.¹ The verb 買った stays last every time. The (こと) in parentheses is a nominalizing device. Nemoto uses it to treat the bare clause as a noun-like unit, "the fact that."¹

太郎たろうがあの店みせで本ほんを買かった（こと）¹
"Taro bought a book at that store." (canonical SOV)

Fronting the object creates a different surface order with the same roles.

本ほんを太郎たろうがあの店みせで買かった（こと）¹
"Taro bought a book at that store." (object fronted)

Fronting the location instead moves the で-phrase to the front of the clause.

あの店みせで太郎たろうが本ほんを買かった（こと）¹
"Taro bought a book at that store." (location fronted)

The object and the location can both move ahead of the subject at once.

本ほんをあの店みせで太郎たろうが買かった（こと）¹
"Taro bought a book at that store." (object, then location, then subject)

Two more orderings complete the set: 太郎が本をあの店で買った and あの店で本を太郎が買った. Together, they make all six permutations of the three pre-verbal phrases.¹ In every one of the six, the verb 買った stays last.¹

The reorderings are equivalent, not identical in effect

All six orders share the same propositional content, or basic truth meaning, but they are not interchangeable in every context. The canonical order is the genuine default. The marked orders even take measurably longer to process, an effect the literature calls the scrambling effect (スクランブル効果).² The nuance behind choosing a marked order appears further below.

Why Particles Free Up Word Order

Phrases can move because the role information lives on the phrase itself, not in its position. The case particles carry it. Tamaoka lists three possible sources of canonical-order information: case particles, semantic (thematic) roles, and grammatical functions. The experimental results showed that grammatical function, marked on the noun phrase, accounts for canonical order across active, passive, and potential sentences.²

Each phrase carries its own role marker: が for the nominative subject, を for the accusative object, に for the dative, で for location or instrument. Because the marker moves with the phrase, the role survives any reordering. The listener can recover who did what to whom from the particle, no matter where the phrase sits.¹²

The test that the meaning has not changed is truth-conditional equivalence: the reordering does not change the basic proposition. There is no difference in propositional meaning between an SOV order and its OSV scramble.⁴

Case Marking Travels With the Phrase

When you scramble, you move the whole が-phrase or を-phrase as a unit; the bare particle never moves on its own. In the reordering below, the accusative を stays welded to its noun.⁴

太郎たろうが次郎じろうを追おいかけた。⁴
"Taro chased Jiro." (canonical SOV)

次郎じろうを太郎たろうが追おいかけた。⁴
"Taro chased Jiro." (object fronted; same roles, same truth conditions)

The roles do not swap when the order changes. 次郎を is still the one chased, because を marks it, not its position. The published gloss is identical for both orders.⁴ Saito's movement analysis formalizes this: the object starts inside the verb phrase and moves to the sentence-initial position, leaving a trace behind. What moves is the case-marked phrase with its particle attached.¹³

Configurational vs Non-Configurational Languages

A configurational language signals grammatical relations by position. A non-configurational one does not need to, because something else (here, particles) carries the roles. English is the textbook configurational case: the noun phrase before the verb performs the action, and the noun phrase after the verb is the one acted on.⁵

The classic newspaper-headline pair makes the point. "MAN BITES DOG" and "DOG BITES MAN" differ in meaning entirely because of word order. English is an SVO language where order itself marks roles.⁵

Japanese puts that signaling on particles, which frees word order from role-marking. This flexibility led Hale (1980) and Farmer (1980, 1984) to categorize Japanese as a typical non-configurational language, meaning a language whose constituents were claimed to have no hierarchical structure.¹

The "free word order language" label is a debated hypothesis, not settled fact

The non-configurational analysis is debated. Saito and Hoji (1983), Saito (1985), and Hoji (1985, 1987) argue instead that Japanese does have a verb phrase and is configurational. On this view, non-canonical orders come from scrambling movement rather than being generated flat.¹ Experimental work sides with the configurational view: the scrambling effect supports the hypothesis that even free-word-order Japanese sentences have configurational structure.² So "free word order" describes the surface freedom of the pre-verbal phrases, not a claim that the language lacks hierarchical structure.

The practical takeaway is simpler than the theory dispute. "Free" means surface freedom, not the absence of a default. The same sources that document the reordering also identify SOV as canonical and the reorderings as marked.¹²

The Verb-Final Hard Constraint

Against all that freedom, one point stays fixed: the predicate remains clause-final. Nemoto frames this as the boundary on the flexibility: the verb must come at the end of the sentence, while the order of the other phrases is free.¹

Tamaoka states the same limit on what may move: in Japanese, constituents may be reordered anywhere except after the verb.² That point is the right edge of scrambling. Pre-verbal arguments and adjuncts can change order freely. The predicate does not join the permutation in standard or written Japanese.¹²

Genuine verb-not-final cases are a separate topic, not scrambling

There are real situations in spoken Japanese where something trails after the predicate (right-dislocation, or 倒置 inversion). Those are a distinct phenomenon, not scrambling. Both core sources frame their data on standard, written order where the verb is final.¹² A dedicated treatment of when the verb is not final is its own topic, covered in the inversion article linked above.

Why the Predicate Anchors the Clause

The verb holds the final slot because Japanese is head-final: the head of a phrase closes it. The phrase-structure rule Hale and Farmer used to generate the free orders expresses the same point: Japanese is a head-final language, with the head (the verb, at clause level) closing its phrase.¹

Because the verb closes the clause, it is also where the clause shows tense, polarity, and mood. Moving it would not be a simple reordering. It would change the clause's structure, which is why the predicate sits outside what scrambling can touch.¹

What Stays Fixed vs What Can Move

The flexibility has a clean split. Anything pre-verbal that carries a particle can move. The predicate and the particle-to-noun bond cannot.

The diagram below shows the same three case-marked phrases changing order around a verb that stays pinned to the final slot. It is a picture of the verb-final anchor: the arrows reorder the left side while the predicate never leaves the right edge. The contrast between "moves" and "stays pinned" is spatial, so a layout makes it easier to see.

• Moves freely: case-marked and postposition-marked phrases (が-, を-, に-, で-phrases) and adjuncts. The six-order set above shows the subject, object, and a で-adjunct all changing order.¹
• Stays fixed: the predicate, whether verb, i-adjective, or copula, at clause-final position.¹²
• Never moves on its own: the particle. You reorder the whole に-phrase, never に by itself. The movement targets the case-marked phrase as a unit and leaves the particle attached.¹⁴

Information-Structure Effects of Reordering

If the orders mean the same thing, why scramble at all? Because the choice of order tracks information structure: how a sentence packages what is already known versus what is new. Kuno (1978) proposed that word-order choice in Japanese is determined by given-new ordering: given information comes at the beginning of a sentence, and new information comes toward the end.⁴

Consistent with this, scrambled objects in OSV orders tend to be given information.⁴ Fronting the object puts already-known material early and leaves the newest material near the verb.⁴

So the default SOV order is the neutral packaging. A scramble is a marked choice that signals something about what is given versus new, or what is being emphasized or contrasted.⁴

Fronting and Topicalization with は

Moving a constituent to the front and marking it with は is related to scrambling, but it is structurally different. Imamura and colleagues contrast plain scrambled OSV order with a topicalized variant.⁴

太郎たろうは次郎じろうを追おいかけた。⁴
"As for Taro, he chased Jiro."

The difference lies in what marks the fronted phrase. は replaces the case marking with topic marking and sets the phrase as the topic or contrast. A plain scramble keeps the case particle (が or を) and only changes the order.⁴ Both can put a constituent first, but they are distinct operations.⁴

The Pre-Verbal Focus Slot

Given-new ordering has one useful corollary. If given information comes first, new information tends to come toward the end. Since the verb is the end, the position just before the verb often carries the newest, most salient information.

This follows from the given-before-new principle combined with verb-finality. It is the consequence of those two facts, not a separately worded source claim.⁴ Reordering uses that pattern: moving given material leftward leaves the focused, new constituent in the slot next to the predicate.

Comparison With Strict-Order Languages

Set beside a fixed-order language, the Japanese mechanism becomes clearer. The propositional content can be the same, but the machinery for expressing it differs. English must keep its SVO order to keep the roles. It reorders only through syntactic operations such as the passive or clefts, because in English order marks roles. That is what separates "MAN BITES DOG" from "DOG BITES MAN."⁵

Japanese can simply move the case-marked phrase. Because が, を, and に carry the roles, 太郎が次郎を追いかけた and 次郎を太郎が追いかけた keep identical roles and identical truth conditions, even though their order differs.⁴ What English does with a passive transformation, Japanese can do with a bare scramble plus an information-structure nuance.

The deeper contrast is configurationality itself, or how much position determines grammatical structure. English was the model of a configurational, position-based language. Japanese's particle-marking made its word order free enough to prompt the non-configurational hypothesis in the first place.¹⁵

Good to know

"Free word order" does not mean "anything goes"

The freedom is real for pre-verbal phrases, but it stops at the predicate, which must stay clause-final. A reordering that pushes the verb off the end is not a scramble. It is either ungrammatical or a separate inversion structure. The verb-final limit is stated directly in both core sources.¹² So a string that strands a leftover phrase after the verb, intending a neutral statement, is not a valid scramble. Keep the verb last and front the phrase instead.

本ほんを太郎たろうが買かった。¹
"Taro bought a book." (object fronted, verb still final)

When unsure, the canonical SOV order is the safe one

Scrambled orders still carry meaning effects. They track given-new ordering, so an unmotivated scramble can sound unnatural. When you have no reason to reorder, canonical SOV order is the neutral choice.⁴ Scrambled orders also carry a measurable processing cost, called the scrambling effect. This is evidence that they are the marked option rather than a free alternative.²

かき混ぜ and "scrambling" are linguists' labels, not a rule to drill

The term "scrambling" entered the field via Ross (1967) and was applied to Japanese by Saito (1985). The Japanese かき混ぜ mirrors it.¹² A learner does not need the terminology. What matters is the instinct: particles free the order while the verb stays last.¹²

A particle cannot be moved off its noun

You scramble the whole に-phrase, が-phrase, or を-phrase, not the bare particle. The case particle stays attached to its noun through the reordering, because the unit that moves is the case-marked phrase, not the marker by itself.¹⁴

Sentence-final particles sit after the predicate, outside scrambling

Sentence-final particles such as か, ね, and よ attach after the predicate. Because the predicate is final, they fall outside the pre-verbal domain that scrambling reorders.¹² They mark the right edge of the clause, beyond where scrambling can reach, so they are not part of the reordering at all.

See also

• Japanese Word Order: SOV and the Head-Final Principle
• Topic vs. Subject in Japanese: The Hidden Slot
• は vs が in Japanese: A Beginner's First Pass
• Japanese Particles (助詞): The Eight Categories Explained
• Japanese Focus Prosody: Pitch Widening, Contrastive は, and Information Structure
• Ellipsis and Implicit Reference in Japanese

References

Footnotes

1. Nemoto, Naoko. "Scrambling." Chapter 5 in The Handbook of Japanese Linguistics, edited by Natsuko Tsujimura. Blackwell Publishers, 1999, pp. 121–153. https://www.blackwellpublishing.co.uk/content/BPL_Images/Content_Store/WWW_Content/9780631234944/005.pdf ↩ ↩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

2. 玉岡賀津雄 (Tamaoka, Katsuo). 「文処理のメカニズム」. 第12回認知神経心理学研究会 (2009/8/22–23), 名古屋大学大学院国際言語文化研究科. https://www.lang.nagoya-u.ac.jp/bugai/kokugen/nichigen/0-kyouiku/research/gakkai/2009-08-22~23/01.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹

3. Saito, Mamoru. Some Asymmetries in Japanese and Their Theoretical Implications. PhD dissertation, Massachusetts Institute of Technology, 1985. (Cited via Nemoto 1999 ¹; original page numbers given inline where Nemoto quotes them.) ↩ ↩²

4. Imamura, Satoshi, Yohei Sato, and Masatoshi Koizumi. "The Processing Cost of Scrambling and Topicalization in Japanese." Frontiers in Psychology, vol. 7, article 531, 2016. https://doi.org/10.3389/fpsyg.2016.00531 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸

5. The Open University. "Describing language: Week 7, Section 2.1." OpenLearn. https://www.open.edu/openlearn/mod/oucontent/view.php?id=113155&section=2.1 ↩ ↩² ↩³ ↩⁴