10 ways to say yes in Finnish, ranked by use

In 412 million words of Finnish subtitles, the word kyllä begins a line 655,992 times. Joo begins one 201,618 times.

Three to one, for the word almost nobody says to you first.

That result is the article. Not the ranking underneath it, which is useful, but the gap between what the biggest available pile of written Finnish dialogue says and what happens when somebody answers you in a shop.

What was counted, so it can be argued with

The corpus is the Finnish monolingual file of OpenSubtitles 2024, distributed through OPUS at the University of Helsinki. It is the largest collection of transcribed and translated film and television dialogue that exists, which makes it the closest thing to a corpus of spoken Finnish that anybody can download.¹²

The pass counted 107,496,897 lines and 411,963,125 alphabetic tokens, lowercased, with digits and punctuation discarded. OPUS itself reports 570,353,193 tokens for the same Finnish file, and an earlier count here uses that figure: the gap is punctuation, numerals and symbols, which OPUS counts as tokens and this pass throws away. Same corpus, two token definitions, and the ratios below are unaffected either way. Two numbers were taken for every word: how often it occurs anywhere, and how often it is the first word of a line. A subtitle line is roughly one turn, so the second number is roughly the word used as an answer, which is the one this question is about.

The ranking

Line-initial counts, which is to say: how often each of these starts a line of dialogue.

Two of those rows change what you go looking for.

The first is that half of these are not translations of yes. Selvä accepts an instruction, aivan confirms a statement, no niin marks a transition. A phrasebook entry headed “yes” contains none of them, and between them they open more lines than kyllä does.

The second is that the standard grammar treats several of these as a category of their own. Dialogue particles are described as tools for managing the interaction rather than as answers, and they cannot be described at all without the surrounding turn.³ Joo is the most semantically settled of them: as the second half of a pair it answers a question or consents to a request, and juu is its rarer twin.³

Where the count is wrong

Now the part that makes the number interesting.

Nii is not a rare word. It is what niin sounds like in ordinary speech, in most of the country, most of the time. The corpus contains it 874 times against two and a half million for the full form, which is a ratio of about one in 2,900. Nobody speaks that way. Subtitlers write that way.

The same goes for the backchannel. Mm, the noise a Finn makes to show they are still listening, appears 5,423 times in 412 million words, which works out at once every 76,000 words. In a real conversation it turns up several times a minute. It is not rare in Finnish; it is simply not something a subtitler transcribes.

The divergence itself is old news to the people who study it. A lecture to the Finnish language board described written texts as having become steadily more speech-like since the 1950s, with authors handling spoken forms in dialogue however they personally chose.⁴ However they chose is the operative part. There is no convention, so each subtitler decides separately, and the safe decision is the standard form.

If you want the other measurement, it exists: ArkiSyn, a morphosyntactically coded database of recorded Finnish conversation held in the Language Bank of Finland, is built from actual talk rather than from writing.⁵ It is three orders of magnitude smaller than the subtitle file, which is the trade.

So what do you actually say

The ranking is still useful once you know which way it leans. Read it as the written league table and adjust.

The grammar puts kyllä in a group with tietysti, toki and totta kai, all of them confirming rather than merely agreeing.⁶

1. 1
   Default to joo

   For answering a question and for agreeing to a request, it is the ordinary word and it is never wrong socially. It is fourth in the written count and first in the room.
2. 2
   Keep kyllä for emphasis and for writing

   The grammar groups it with tietysti, toki and totta kai as a way of confirming rather than merely agreeing. Use it when you are contradicting a denial, and in every email.
3. 3
   Answer with the verb when there is one

   The most Finnish option on the list is not on the list. Tuutsä? gets tuun. Onks tää sun? gets on. No word for yes appears anywhere in the exchange.
4. 4
   Learn selvä as a separate thing

   It is second in the whole count and it is not a yes. When somebody tells you where to go, what time, or what they will do, selvä is the reply, and reaching for joo there is not wrong so much as slightly off.

One last number, because it reframes the whole exercise. The commonest line-opening word in the entire corpus is en at 3,265,188, and ei is third at 2,775,936, with hän between them. Finnish dialogue starts with a negation more often than with anything else, and both of those beat every word for yes combined.

Whatever you were expecting the hardest part of answering in Finnish to be, the corpus says to go and look at the negative verb first.