Skip to main content

The Plain Past た-Form in Japanese: Past, Perfective, and Beyond

The uses of the た-form in Japanese go far beyond the "past tense" label most courses attach to it. Native grammar files た under 過去 (past), 完了 (completion), and 想起 (recollection), and speakers use it for things that have not happened, will not happen, or are happening right now.1 This article maps each reading. The rules for building the form live in the dedicated construction article.

Overview: Why "Past Tense" Undersells た​

Standard modern Japanese splits its verbs cleanly enough at first glance: the ル-form (dictionary form) covers present and future, and the タ-form covers the past.1 The trouble is that た also carries uses that the concept of 過去 cannot explain.

NINJAL's descriptive material states the point directly. The same た can mark "the occurrence or discovery of a situation at the time of utterance" (発話時における事態の発生や発見), which is not a past event at all.1

The refinement that drives this article is this: た is better understood as marking not that the event happened in the past, but that the speaker acquired or experienced the information before now.1 The seal is on the speaker's knowing, not on the event's date.

The reading that reframes everything

NINJAL's analysis is that た marks "the past point at which the speaker came to know the situation," not the time the situation itself holds.1 That single shift explains why た can attach to a future appointment, a present discovery, or a wish that never came true.

Descriptive grammar sorts the form into five readings: 過去 (past), 完了 (completion), 発見 (discovery), 想起・確認 (recollection and confirmation), and 命令 (command). Of these, only 過去 is straightforwardly tense. 完了 is aspect. 発見, 想起・確認, and 命令 work as a kind of mood, expressing the speaker's stance at the moment of speaking rather than placing an event in past time.2

Native school grammar (国語) labels the auxiliary た（だ）a little differently, listing 過去・完了・存続・想起（確認）.2 This article foregrounds the learner-facing readings and folds 存続 (resultant state, as in 破れた本 "a torn book") into the discussion of the ている boundary. The two label sets overlap but are not identical.

Register and where た sits​

た is the plain, casual past. Its polite counterparts are でした for the copula and for noun and na-adjective predicates, and ました for verbs in the ます-form.34 The same event differs only in register: 見た (plain) versus 見ました (polite).4

The polite past lives with the ます-form material, so this section needs only the one pairing. The forms below are plain throughout unless marked otherwise.

Formation in one breath (recap, not the rules)​

The full conjugation rules belong to the sibling article on ta-form construction, which carries the complete tables. In one line: た uses the same sound changes as the て-form, swapping て→た and で→だ. 一段 (ichidan) verbs drop る and add た (食べる → 食べた). 五段 (godan) verbs fall into the clusters った, いた, いだ, んだ, and した. The irregulars are する → した and 来る → きた.4

映画えいがを見みた。4
"I watched a movie."

もう宿題しゅくだいをした。4
"I already did my homework."

The Core Split: Past (過去) vs. Perfective / Completion (完了)​

The single most useful distinction inside た is between locating an event in past time and marking an action as complete. English often fuses these, but Japanese does not have to.

Past time: an event located before now​

This is the uncontroversial reading. た places an event before the utterance time (発話時), the sense native school grammar calls 過去, "something already passed."12

私わたしは７時しちじに起おき、１２時じゅうにじに寝ねました。2
"I woke up at seven and went to bed at twelve."

昨日きのう、映画えいがを見みた。4
"Yesterday I watched a movie."

Of the five readings, the past reading is the only one descriptive grammar treats as straightforwardly tense; the rest are aspect or mood.2

Completion regardless of time: 終わった, できた, わかった​

The 完了 (completion) reading marks an action as finished. Because completion is aspect rather than tense, it can apply even when the time frame is present or future. Native grammar paraphrases the 完了 sense as 今〜した or ようやく〜した ("just now did," "finally did").2

Makino and Tsutsui frame the point this way: the past verb can express the completion of an action, but it expresses completion in the past, while the て-form plus しまう expresses completion regardless of the time of completion.5 Completion and past tense are different jobs that the same form can do.

The clearest learner case is the 〜たら time clause. In that pattern, the た-marked event has not yet happened at speech time, but it is presented as completed before the main clause.

会議かいぎが終おわったら帰かえります。5
"Once the meeting's over, I'll head home."

The completion in that sentence is relative to going home, not to now. The meeting is still in the future.

レポートはもう出だしました。2
"I've already turned in the report."

あ、できた！1
"Oh, it's done!"

That last completion is realized at the moment of speaking, not reported from the past. This previews the discovery reading below.

Pair た with もう, not まだ

The completion sense is why た attaches in もう…した ("already did") and is rejected by まだ ("not yet"), which takes the negative ～ていない form instead. If you can say "already," た fits. If you mean "not yet," use ～ていない.25

The boundary with ている​

た with no auxiliary expresses the 完成相 (perfective or completive aspect) of an event. ている expresses the 継続相 (continuative aspect). NINJAL lays the contrast out as a four-cell paradigm: 読む is nonpast plus perfective, 読んでいる is nonpast plus continuative, 読んだ is past plus perfective, and 読んでいた is past plus continuative.1 In short, た and ている divide the labor between "done" and "in a resulting or ongoing state."

The following diagram shows that paradigm as two crossing axes, tense against aspect.

The canonical minimal pair is 死んだ versus 死んでいる: the event sealed as complete, contrasted with the resulting state.1

木きが倒たおれた。1
"The tree fell over."

木きが倒たおれている。1
"The tree is lying fallen over."

The resultant-state reading carried by ている is its own large topic. Here the point is only the aspectual split between the sealed event and its lingering result.

Beyond Past Tense: The Non-Past Readings​

This is the key point. Each reading below uses た for a situation that is not a past event. That is the strongest evidence that た marks something other than past time.

The four readings fan out from the single completion-or-knowing core.

Discovery and realization: あ、あった! ("I knew it / there it is")​

The 発見 (discovery) reading marks the speaker locating or confirming something at the moment of speaking, with no past event involved.12 NINJAL describes this as た expressing 事態の発生 ("the situation coming about") or 事態の発見 ("the discovery of the situation") at the utterance time.1

This is the cleanest proof of the thesis. The thing was there all along, so the た cannot mark the object's past; it marks the present instant in which the speaker came to know it.1

あ、ここにあった！2
"Oh, here it is!"

あ、ついた。1
"Oh, it's come on."

いた、いた。こんなところにいたのか。1
"There you are. So this is where you were."

The confirmation flavor, やっぱり…だった ("I knew it, just as I thought"), is the same reading: a present-moment verification reported with た.2 NINJAL's own example sets the scene in brackets, 「（太郎を探していて）あ、こんな所にいた」("[while looking for Tarō] oh, here he is"). The present situation, not the form alone, licenses the discovery reading.1

Recollection: そうだ、明日は休みだった ("oh right, tomorrow's off")​

The 想起・確認 (recollection and confirmation) reading marks a fact being retrieved from memory or checked. That fact can sit in the future.2 Native school grammar lists 想起 as a standard sense of the auxiliary た, glossed as confirming and recalling that something is so.2

Future-referring 明日…だった is grammatical because the た does not date the event. Tomorrow is still future. The た marks that the speaker is recalling a known fact in the present, which fits NINJAL's reading that た marks the past point at which the information was acquired.12

そうだ、明日あしたは休やすみだった。2
"Oh right, tomorrow's a day off."

明日あしたの会議かいぎは２時にじからだった？2
"Tomorrow's meeting was at two, right?"

お名前なまえは、えーと、田中たなかさんでしたっけ。2
"Your name was, um, Tanaka, right?"

The soft command: 寝た寝た, ちょっと待った, どいたどいた​

The 命令 (command) reading lets the た form issue a brusque or urging directive, often repeated.62 Descriptive grammar lists this as a marked, less common use of the form.2

Arita's academic analysis calls this a 差し迫った命令 (pressing command), citing Takahashi's term 直ぐの命令 ("immediate command"). It demands action at the moment of utterance and cannot specify a future time slot. That is why 「１時間後に、帰った帰った」("go home in an hour") is ungrammatical, unlike the ordinary imperative 「明日東京に行け」("go to Tokyo tomorrow").67 That present-moment demand is exactly why the た, not the dictionary imperative, appears here.

The verb is typically repeated. The form can take よ in some cases, but it patterns differently from the true 命令形 with respect to particles. That marks it as a distinct, colloquial command type.6

さあ、子供こどもは帰かえった、帰かえった。6
"All right kids, off you go, off you go."

安やすいよ。買かった、買かった。6
"It's cheap. Buy it, buy it."

どいた、どいた！2
"Out of the way, out of the way."

Keep the た command out of polite speech

This use is rough and colloquial, at home in vendor calls, coaching, and clearing a path. It demands compliance now and cannot be deferred, which separates it from the standard imperative. It is out of place in polite or deferential speech, where requests use 〜てください and similar.62

Frozen idioms such as 寝た寝た ("go to sleep, go to sleep") and ちょっと待った ("hold on a second") belong to this same repeated, urging pattern. Treat them as the encouraging end of the register, by analogy to the directly documented 帰った帰った, 買った買った, and どいた class.6

Counterfactual and unrealized: 〜たらよかった, 帰ったほうがいい flavour​

In conditional and hypothetical frames, た can mark a situation as contrary to fact rather than as past. Ogihara argues that Japanese た is ambiguous between a temporal meaning (anteriority, or "before now") and a counterfactual or irrealis meaning. In the consequent of a subjunctive conditional referring to a future time, only the counterfactual reading survives.8

When た marks counterfactuality, its time reference is underspecified. It could be past, present, or future.8 The た in a regret or hypothetical frame is doing modal work, not locating a past event. Ogihara builds this on Iatridou's "true past versus fake past" distinction and argues that Japanese た is literally counterfactual, excluding the actual world.8

もっと早はやく帰かえったらよかった。8
"I wish I'd gone home earlier."

早はやく帰かえったほうがいい。8
"You'd better go home early."

In the first, 〜たらよかった voices regret over an event that did not happen. In the second, 〜たほうがいい uses a た that recommends an as-yet-unrealized action. Read these as "would have," "wish," or "had better," not as English past. The deeper machinery of the Japanese conditionals (たら, ば, と, なら) is its own topic and stays out of scope here.

Nuance and Usage Contexts​

The readings above are not chosen by the form alone. The same surface string can land in several readings. What selects one is everything around the form.

How context, not form, picks the reading​

The same string, such as あった, can be past, completion, discovery, or command depending on intonation, particles, and situation. The morphology alone does not pick the reading.12 NINJAL presents its discovery examples with bracketed situational context precisely because the context, not the form, licenses the non-past reading.1

A minimal contrast makes the point. 鍵があった can be "There was a key" (past report) or "Here's the key, found it" (discovery), depending on situation and intonation.1 帰った can be "He went home" (past) or, when repeated as 帰った、帰った, "Off you go" (command).6

（太郎たろうを探さがしていて）あ、こんな所ところにいた。1
"(While looking for Tarō) Oh, here he is."

太郎たろうは昨日きのうここにいた。1
"Tarō was here yesterday."

The same いた sits in both, yet 昨日 fixes the second firmly in the plain past.

た in adjacent constructions: たばかり and たところ​

Two common patterns use the completion sense of た. た plus ばかり means "just recently finished," and た plus ところ means "just did, a moment ago."9 Both are N4-level patterns covered fully in their own articles on the ばかり particle and the just-after constructions. The note here is only that the た inside them is the completion た.

The two differ in feel. 〜たところ marks an event that happened literally seconds or minutes ago (objective immediacy). 〜たばかり marks an event felt to be recent even if more time has passed (subjective recency, often carrying an "and that's why…" relevance).9

今いま、駅えきに着ついたところです。9
"I've just this moment arrived at the station."

さっき昼ひるご飯はんを食たべたばかりです。9
"I just had lunch a little while ago."

た on adjectives and the copula​

The past and completed reading extends beyond verbs. い-adjectives form the past by dropping い and adding かった (高い → 高かった). The plain copula だ becomes だった, and the polite です becomes でした (学生だ → 学生だった). な-adjectives behave like the copula (静かだ → 静かだった).3 The full paradigms live in the dedicated i-adjective, na-adjective, and copula articles.

去年きょねんの冬ふゆはとても寒さむかった。3
"Last winter was very cold."

父ちちは昔むかし、学生がくせいだった。3
"My father was a student long ago."

あの店みせは静しずかだった。3
"That shop was quiet."

Good to know​

Translating every た as English simple past​

The habit of rendering た as English simple past breaks on the discovery sense. わかった is not "I understood, back then". It is "I get it now" or "Got it," a present-moment realization.12 た can mark the moment the speaker acquires information rather than a past event, and the English past loses that. The same error hits できた ("there, done") and あった ("here it is").1

わかった。1
"Got it. / I get it now."

た comes from the classical 完了 auxiliary たり​

NINJAL traces modern た to the 連体形 (attributive) たる of classical たり, with る dropped.10 Classical Japanese had six past and perfective auxiliaries: past き and けり, and completion つ, ぬ, たり, and り. As their distinctions blurred, the high-frequency た consolidated their functions, and the old process-and-state expression migrated to ている and てある.10 That た is historically a completion auxiliary explains why "past" never fully fit it.

た = sealed, not dated​

A compact way to hold all five readings together is this: た seals an action as complete or known, and context decides where that seal sits. It can land in the past (過去), at the moment of speaking (発見・想起), or over a wish that never came true (反実仮想). The form marks the point at which the speaker acquired or experienced the information, not the date of the event itself.1

See also​

References​

Footnotes​

  1. 木部暢子・沖裕子・井上文子.「テンス・アスペクト」. 国立国語研究所（NINJAL）, 方言文法全国地図解説／日本語記述文法（DGG 教材）. https://www2.ninjal.ac.jp/takoni/DGG/05_tensu_asupekuto.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

  2. 庭三郎（niwa saburoo）.「23. テンス」『現代日本語文法概説』. https://niwasaburoo.amebaownd.com/posts/5738196/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13 ↩14 ↩15 ↩16 ↩17 ↩18 ↩19 ↩20 ↩21 ↩22 ↩23

  3. Makino, Seiichi, and Michio Tsutsui. A Dictionary of Basic Japanese Grammar. The Japan Times, 1986, pp. 581–582 (Appendixes: adjective and copula conjugation tables). ↩ ↩2 ↩3 ↩4 ↩5

  4. "GrammarBook N5: Plain Past Form In Japanese." JPLT Online Japanese Language Training School. https://jplt-dialogplus.com/n5-g-050-plain-past/ (limitation: learning-school source; used only to confirm the JLPT N5 placement, not for the semantic claims.) ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  5. Makino, Seiichi, and Michio Tsutsui. A Dictionary of Basic Japanese Grammar. The Japan Times, 1986, pp. 405–406 (entry "shimau", Related Expression I). ↩ ↩2 ↩3

  6. 有田節子（大阪樟蔭女子大学）.「『差し迫った命令』に関する覚え書き ―日本語のテンスとモダリティの接点―」. 九州大学言語学研究室 KUPL 35. https://www2.lit.kyushu-u.ac.jp/~linguist/kupl/doc/kupl35_Arita.pdf ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  7. 高橋太郎.『現代日本語動詞のアスペクトとテンス』. 国立国語研究所報告 82, 秀英出版, 1985（「直ぐの命令」の用語; 有田6が引用）. ↩

  8. Ogihara, Toshiyuki. "The Semantics of '-ta' in Japanese Future Conditionals." In Luka Crnič and Uli Sauerland (eds.), The Art and Craft of Semantics: A Festschrift for Irene Heim, vol. 2, MITWPL 71, 2014, pp. 1–21. University of Washington. https://semanticsarchive.net/Archive/jZiNmM4N/Ogihara.pdf ↩ ↩2 ↩3 ↩4 ↩5

  9. "Just Finished: 〜たばかり vs 〜たところ." japanistry.com. https://www.japanistry.com/just-finished/ (limitation: language-learning site; used only for the た-adjacent N4 patterns the blueprint marks as out-of-scope cross-references, not for core claims.) ↩ ↩2 ↩3 ↩4

  10. 福澤将樹（愛知県立大学）.「古典文法では過去や完了の助動詞がたくさんあるのに、現代語ではなぜひとつしかないのですか」ことば研究館（国立国語研究所 / NINJAL）. https://kotoba.ninjal.ac.jp/qa/yokuaru/qa-68/ ↩ ↩2