하다 → 했어요: The Past of 하다-Verbs

하다 ("to do") is the most productive verb in Korean by a wide margin: attach it to a noun and you get a verb. 공부 (study) → 공부하다 (to study), 일 (work) → 일하다 (to work), 사랑 (love) → 사랑하다 (to love). There are thousands of these, and every one of them forms its past the same way — through 하다. So if you learn just one past stem, learn this one: 하다 → 했-. It is the single highest-frequency past stem in the language, and it hands you narration, storytelling, and reporting over a huge vocabulary at once.

Where 했 comes from: 하 + 였 → 했

하다 does not follow the ordinary vowel-harmony rule (it would predict ×하아). Instead 하- takes the special ending 여, so the present is 하 + 여 → 해, and the past is 하 + 였 → 했. You do not need to compute this each time — just bank the fused result. See 하다 → 해 (the 여 irregular) for the full logic; here we simply use it.

Memorize the trio and you can operate in every basic time frame: 해 (present, do/does), 했 (past, did), 할 (prospective, will do). Once 했 is automatic, the past of thousands of verbs is automatic with it.

Every 하다-compound inherits 했-

This is the leverage. You do not learn a new past for each 하다-verb — they all bolt their past onto the same 했.

The noun never changes — only 하다 carries the tense

A 하다-verb has two parts, and they have a strict division of labor: the noun part supplies the meaning and stays perfectly still, while 하다 alone carries the grammar. In 공부하다, 공부 never inflects; only 하다 moves to 했다. Do not try to put the tense on the noun.

Notice 이사 and 시작 sit unchanged; the whole past lives in 했. This split is a gift to the learner — it means you conjugate one verb (하다) rather than thousands.

Across speech levels: 했어요 / 했습니다 / 했어

The past stem 했 stays fixed; you just swap the ending to match the register. This too transfers across the entire 하다 class.

Descriptive 하다-words work the same way

Many adjectives are also 하다-words — 조용하다 (to be quiet), 유명하다 (to be famous), 편하다 (to be comfortable) — and they form their past through 했 identically. Korean adjectives inflect for tense on their own (see the past of adjectives and the copula), and for the 하다 class that means 했 again.

There is no separate "was" here: 조용했어요 by itself means "was quiet." The adjective carries its own past, and for 하다-adjectives the past is 했.

How this differs from English

English "do" is mostly a grammatical helper — did you eat?, I do agree — but it rarely fuses with a noun to mean the activity itself. Korean 하다 does exactly that, wholesale: 공부를 하다 is literally "to do studying," and it is the ordinary word for "to study." The upside for the past tense is enormous. Where English memorizes a separate past for each verb (studied, worked, called, moved, started), Korean routes almost all of them through one irregular past, 했. Learn 했 and you have simultaneously learned the past of study, work, call, move, start, clean, love, decide, exercise, reserve, cancel, and thousands more. Few single facts in Korean pay off this widely.

Common Mistakes

1. Regularizing the past to ×하았어요. 하- takes 여, not 아, so the past can never be ×하았-. The fused form is 했.

2. Using the un-contracted 하였 in everyday speech. 하였어요 is the historical un-fused form; it survives only in stiff literary or archaic writing. In normal speech and writing, always contract to 했어요.

3. Trying to inflect the noun instead of 하다. The tense belongs to 하다; the noun is frozen.

4. Mismatching the level midstream. Pick 했어요 / 했습니다 / 했어 and stay consistent within a conversation; don't slide from a formal 했습니다 into a casual 했어 with the same listener.

Key Takeaways

• The past of 하다 is 했: 해 (do) / 했 (did) / 할 (will do). Bank the trio like do / did / will do.
• Every 하다-compound inherits 했-: 공부했어요, 일했어요, 전화했어요, 이사했어요 — no per-verb memorization.
• The noun part never inflects; only 하다 carries tense (이사 stays, 하다 → 했다).
• Swap only the ending across levels: 했어요 / 했습니다 / 했어 — the stem 했 is fixed.
• Descriptive 하다-words behave identically: 조용했어요 = "was quiet."
• Never regularize to ×하았어요, and reserve the un-contracted 하였 for literary writing.