Spoken-Word vs. Written-Word Japanese: 話し言葉 vs. 書き言葉

Spoken-Word vs. Written-Word Japanese (話し言葉 vs. 書き言葉) is the register axis between conversation and formal writing. On one end are contractions, particles, and fragments. On the other are full sentences and Sino-Japanese vocabulary.¹ Native speakers move along this axis on purpose. Reading it wrong is a native-detectable error even when the grammar is correct.¹

Overview

話し言葉 ("spoken word") and 書き言葉 ("written word") are two ends of a single stylistic axis within one language, not two separate languages.¹ This axis sits inside the broader system of Japanese speech levels. This article maps that axis, gives you a concrete swap table of spoken-to-written word pairs, and shows where real texts mix the two.

Two registers, one language

The compound names can mislead learners. 話す ("speak") + 言葉 ("word") and 書く ("write") + 言葉 suggest that the first covers only spoken utterances and the second only written text. In practice, native speakers deliberately use 書き言葉 features when speaking and 話し言葉 features when writing.¹

A more accurate framing is cultural-convention-driven code-switching. The speaker or writer selects the register the situation calls for, rather than obeying a hard boundary between speech and writing.¹

The axis is a gradient, not a binary

森山 (2003) holds that the spoken/written difference "does not divide cleanly; it is a matter of degree" (ははっきり分かれるものではなく、段階的なものである). A single item can lean more spoken or more written depending on what it is compared with, so think in terms of a spectrum.²³

That spectrum runs from the most casual speech to the most formal literary prose. Everyday spoken and everyday written registers sit in between, with a wide mixed band in the middle.

One practical consequence: learners rarely flag the 話し言葉/書き言葉 distinction as "difficult." Yet native readers feel a "すわりの悪さ" (a sense of something not sitting right) when spoken-register items appear in an essay.¹ The distinction has historically been undertaught because the underlying research is thin.¹

Why the gap is wider than in English

Japanese has parallel native (和語 wago) and Sino-Japanese (漢語 kango) vocabulary strata, plus a mixed script. This gives writers extra lexical room to mark register.² The contrast is similar to the Latinate-versus-Anglo-Saxon layering in English, where inquire reads more formal than ask. It is a matter of degree, not a feature unique to Japanese.²

That extra room is why a single English idea often has two Japanese expressions at noticeably different points on the axis. The swap table later in this article collects those pairs.

How spoken Japanese (話し言葉) works

The spoken end of the axis is marked by sound reduction, speaker-oriented particles, and incomplete sentences. These are features of the register, not errors.

Contractions and sound reduction

Spoken Japanese routinely contracts forms that writing keeps full. The MIT conversion sheet lists ～ては→ちゃ, ～では→じゃ, ～なければ→なきゃ, and ～てしまう→ちゃう as standard spoken reductions.⁴ The progressive ～ている form also loses its い in speech, so ～ている becomes ～てる.²⁴

今いま、宿題しゅくだいをしてるんだ。⁴
"I'm doing my homework right now."

Here してる is the spoken reduction of している. The written-word equivalent restores the full form: 今、宿題をしている。

行いっちゃいけないよ。⁴
"You mustn't go."

行っちゃ contracts 行っては, and the written-word form is 行ってはいけない. This article does not re-list the full contraction catalogue. The dedicated casual-speech article covers that set, and the examples here stay illustrative.

Sentence-final particles and fragments

Sentence-final particles (終助詞) such as ね, よ, な, and わ are a core 話し言葉 feature. They encode speaker stance and addressee orientation, which formal writing suppresses.⁵ Spoken Japanese also tolerates ellipsis (省略) and incomplete clauses, since shared context and the present listener recover the missing material.¹⁵

明日あした、来くるよね？⁵
"You're coming tomorrow, right?"

The stacked よね marks the spoken end. A written-word version restores a full clause, such as 明日、来るのですか。

私わたしも。⁵
"Me too."

This is a bare fragment with the predicate elided. Written prose would restore it, for example: 私も同様である。

Spoken connectors: けど, でも, やっぱり

The colloquial connector and softener set sits firmly in the spoken column of the empirical survey. でも appears 31 times on the spoken side against 2 written, けど 23 times spoken, and やっぱり 13 times spoken.² けど is the spoken connective particle answering to written ～が. でも and だけど answer to written しかし and だが, and やっぱり is the spoken variant of やはり.²⁴

行いきたいけど、時間じかんがない。²⁴
"I want to go, but there's no time."

Spoken けど maps to written ～が. In writing, the sentence becomes 行きたいが、時間がない。

やっぱり、これがいい。²⁴
"This one's the best after all."

Spoken やっぱり maps to written やはり.

How written Japanese (書き言葉) works

The written end restores what speech drops. It also swaps in denser vocabulary and defaults to an assertive plain style.

Full sentences, fewer particles

Written Japanese restores full predicates and avoids the ellipsis and fragments that speech tolerates. The register survey treats the connective ～が and ～ため, used in place of spoken けど and から, as written-end markers.²⁴ One surface device of written news prose is 体言止め, ending a clause on a noun to compress information into limited space. The hearer-directed spoken register avoids this.¹

交通こうつうの便べんがよく、物価ぶっかが高たかい。⁴
"Transport is convenient and prices are high."

This is the written target of the い-adjective rule くて→く. Speech would say 交通の便がよくて物価が高い, and writing tightens it to よく.

昼ひるは大学だいがくに通かよい、夜よるは働はたらいた。⁴
"By day I attended university; by night I worked."

This is the written target of the 連用中止 rule (pausing a sentence with the verb stem), where spoken 通って becomes the stem 通い in writing. The same sheet gives other written reductions that speech avoids, such as ～ないで→～ず(に) and ～ていて→～ており.⁴

である / だ and the plain written style

Written news and essay prose defaults to the だ・である style. Spoken and broadcast news use です・ます.¹ The Aichi study attributes this to a present-addressee effect: radio and television address a listener directly, which prompts polite です・ます. Newspapers report past events and lean on the assertive だ and the past form た.¹

である is the formal written-style copula that the conversion sheet sets in its own column beside だ.⁴ The three-way distinction だ / です・ます / である is a recognized object of discourse study. The deeper mechanics of である belong to the dedicated copula article.⁶⁴

これは重要じゅうような問題もんだいである。⁴
"This is an important problem."

The である written style maps to the spoken or polite form これは重要な問題です。

Kango preference and denser vocabulary

Writing tends to select the 漢語 (Sino-Japanese) member of a lexical pair, which reads as denser and more formal. The empirical survey finds that 非常に, a kango adverb, is the single most-cited 書き言葉的 word at 28 citations. It is set against spoken すごく and とても.²

本ほん研究けんきゅうは非常ひじょうに重要じゅうようである。²
"This research is extremely important."

Here 非常に (kango, written) maps to spoken すごく重要だ. This kango-leaning tendency holds for nouns and many adverbs, but it does not hold everywhere. The Good-to-know section below records where it inverts.

The lexical swap table: spoken word → written word

This table is the article's main reference point. Every pair is drawn from the empirical literature survey ² and the MIT conversion sheet ⁴, with counts from the survey's 件数 (item-count) tables.² These are register-typical defaults, not absolute rules.

Spoken (話し言葉)	Written (書き言葉)	Class	Sourcing note
でも／だけど	しかし／だが	接続詞	でも 31× spoken vs 2× written; しかし 33× written; だが 22× written.24
だから／ですから	したがって／そのため／よって	接続詞	だから 27× spoken; conversion targets したがって・そのため・よって.24
それで	そこで	接続詞	それで 7× spoken vs 4× written; paired written form そこで.2
けど	が／だが	接続助詞	けど 23× spoken (single-classification spoken); written ～が.24
から (reason)	ため	接続助詞	conversion sheet pairs ～から → ～ため.4
すごく／超／とっても	非常に／きわめて（極めて）	副詞	非常に 28× written (top 書き言葉的 word); すごく 24× spoken; きわめて 7× written.24
ちゃんと	きちんと／正しく	副詞	ちゃんと 9× spoken; paired written きちんと／正しく.24
いっぱい／たくさん	多く(の)／(N が)多い	副詞	いっぱい 16× spoken; paired written 多くの N／N が多い.24
やっぱり	やはり	副詞	やっぱり 13–15× spoken; やはり 14–15× written.24
どんどん	次第に／徐々に／急速に	副詞	どんどん 11× spoken; paired written 次第に・急速に.24
だんだん	次第に／徐々に	副詞	だんだん 12× spoken; paired written 次第に・徐々に.4
もっと	さらに	副詞	さらに written-classified; paired against spoken もっと.24
～してる	～している	動詞活用	てる spoken-classified; written ～ている.24
～じゃない	～ではない	助動詞・否定	conversion sheet: 問題じゃありません → 問題ではない (じゃ→では).4

Connectors: でも→しかし, だから→したがって

でも→しかし and だから→したがって are the two best-attested swaps in the survey. でも and だから are the most-cited spoken items, while their counterparts しかし and したがって head the written lists.² These are sentence-initial 接続詞 connectors, distinct from clause-internal 接続助詞 (conjunctive particles) such as けど. The MIT sheet places でも／だけど → しかし／だが and ですから／だから → そのため／したがって in its 接続(助)詞 row.⁴

Adverbs and intensifiers: すごく→非常に, ちゃんと→きちんと

非常に is the most frequently prescribed written replacement for the spoken intensifier cluster すごく, とっても, and 超, with 28 citations.²

とても sits mid-spectrum, not cleanly on one side

The survey records とても on both sides: 8 times written and 11 times spoken. Written とても is offered as the replacement for spoken とっても and すごく. At the same time, spoken とても is itself flagged for replacement by written たいへん, 非常に, or 極めて.² Treat とても as a middle-register hinge, not a clean spoken pole.

A note on reading the table

These are register-typical defaults, not hard rules. The survey frames the spoken/written contrast as gradient (段階的). It also notes that different references can classify the same item in opposite directions, so a certain "判断にゆれ" (wobble in judgment) exists.²³

When texts mix both

Real texts rarely sit at one pole. The Aichi study's central claim is that written documents themselves contain a deliberate mix of 話し言葉 and 書き言葉. Cultural convention governs that mix more than the medium does.¹

Manga, blogs, LINE, and social media

The empirical survey motivates its work by noting that email and blogs have spread "話し言葉的な書き言葉" (spoken-style writing). As a result, spoken-register items now mix into contexts that would otherwise be 書き言葉的.² These are written media carrying spoken-style features by design.

Manga illustrates the split at the page level. Narration boxes lean to written style, while dialogue balloons carry the spoken-register markers catalogued above: contractions, sentence-final particles, and fragments. This is an illustration drawn from the feature catalogue, not a manga-specific corpus finding.²⁵

Getting the register wrong

Register mismatch is a real, native-detectable error, not just a stylistic nicety. Native readers feel a "すわりの悪さ" when a spoken-register item lands in an essay, even when the sentence is grammatical and its meaning is clear.¹

The trap also runs the other way through over-assertion. Even a "written" item can be wrong in a paper if it signals subjectivity or an over-strong claim. 必ず, meaning "100% of the time," is avoided in 論文 (academic papers) in favor of a hedged ほとんど.²⁷

Good to know

"Write how you'd say it" is a trap

The single most common learner error is transcribing speech directly into an essay. Dropping a spoken connector like だから into だ・である prose produces the native-detectable "すわりの悪さ," even when every word is correct.¹ The Aichi study's student-error analysis shows that the fix is to restructure the whole sentence into written-register grammar and vocabulary, not just to swap one word.¹

でも, だから, or なので opening an academic sentence

でも, だから, and なので are spoken-end connectors. In 論文 and レポート (academic papers and reports), the written-end しかし, したがって, and そのため are expected at the head of a sentence. なので in particular is flagged on the spoken side of the survey at 16 citations, so it reads as conversational in formal prose.²

である in casual chat, です・ます in a report

である is the written plain style. In conversation, it reads as stiff or performative. The mismatch runs both ways: a polite です・ます chain dropped into a だ・である report clashes with the document's register just as audibly.⁴¹⁶

The kango member is not always the "written" one

Memorizing "kango equals formal" overgeneralizes, because the alignment inverts for conjunctions and adverbs of degree. 石黒 (2012) finds that the 漢語 form can be the more spoken one and the 和語 form the more written one, the reverse of the noun pattern. 全然, 多分, 絶対, 全部, and 一番 read as 話し言葉的, while まったく, おそらく, 必ず, すべて, and もっとも read as 書き言葉的.²⁷

とても and けど mark the middle of the spectrum

Not every item sits at a pole. とても is recorded on both the written and spoken sides of the survey, and けど differs from written が by only a single register notch.²³ Treat them as neutral middle anchors. They also remind you that the whole axis is gradient (段階的), not binary.²³

Email and chat blur the line

Digital text has pulled spoken features into writing as a standing fact of the medium. That is why email and blogs carry "話し言葉的な書き言葉."² A chat message that reads as natural can hold contractions and sentence-final particles that the same writer would strip out of a report.

See also

• When to Switch from です/ます to Plain Form
• The Japanese Verb Stem (連用形): The Masu-Stem and Its Uses
• The よね Particle: Asserting While Seeking Agreement
• Reading Japanese Social Media: Twitter/X, Instagram, and LINE
• Aizuchi (相槌): The Backchannel Sounds of Japanese Conversation

References

Footnotes

1. 山本雅子・大西五郎「話し言葉と書き言葉の相互関係 ―日本語教育のために―」『言語と文化』No. 8, 愛知大学語学教育研究室, pp. 73–. https://taweb.aichi-u.ac.jp/tgoken/bulletin/pdfs/NO8/07.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶

2. 柏野和佳子・田嶋明日香・平本智弥・木田真理「学術的文章作成時に留意すべき『書き言葉的』『話し言葉的』な語の文献調査」『言語処理学会 第22回年次大会 発表論文集』(2016年3月), 国立国語研究所 言語資源研究系・コーパス開発センター, 国際交流基金 日本語国際センター, pp. 1041–1044. https://www.anlp.jp/proceedings/annual_meeting/2016/pdf_dir/P18-1.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 ↩³⁴ ↩³⁵ ↩³⁶

3. 森山卓郎「話し言葉と書き言葉を考えるための文法研究用語・12」『國文學 解釈と教材の研究』48(12), 學燈社, 2003, pp. 15–22. (Cited in ² for the gradience claim.) ↩ ↩² ↩³ ↩⁴

4. MIT Japanese Program. "話し言葉と書き言葉" (Rules for Writing: です・ます体 / だ体 / である体 conversion sheet). Massachusetts Institute of Technology. https://web.mit.edu/kakikotoba/Rules%20for%20Writing%20%20copy.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

5. Makino, Seiichi, and Michio Tsutsui. A Dictionary of Basic Japanese Grammar (日本語基本文法辞典). The Japan Times, 1986. (Front-matter grammatical-terms section "Characteristics of Spoken Language": contracted forms, sentence-final particles, ellipsis.) ↩ ↩² ↩³ ↩⁴ ↩⁵

6. Maynard, Senko K. Principles of Japanese Discourse: A Handbook. Cambridge University Press, 1998. Chapter 4, "Style distinction: da, desu/masu, and dearu." ↩ ↩²

7. 石黒圭『この１冊できちんと書ける！論文・レポートの基本』日本実業出版社, 2012. (Cited in ² for the kango/wago conjunction-register observation.) ↩ ↩²