Assimilations and Dropped Sounds

Relaxed Finnish does not just drop sounds off the ends of words (that erosion — the lost -i, -n and -t — is the subject of dropped endings). It also reshapes sounds where words meet and where endings attach: one consonant takes on the qualities of its neighbour, a d dissolves between vowels, two vowels fuse into one long one. These are assimilations, and the single most liberating fact about them is that they are regular, predictable processes, not random slurring. Fast speech sounds like a blur until you learn the handful of rules on this page — after which the reduced forms become fully decodable, and you stop mishearing Otaksä kahvii? as an unknown word.

A useful split runs through this page. Some of these changes are purely phonetic — they happen in the mouth but are never written, not even in casual texting (onpa is always spelled onpa). Others have become written spoken forms that do appear in chat and dialogue (tiiän, taloo). We flag each as we go.

Final -n takes the place of the next consonant

Word-final -n is a chameleon: in connected speech it assimilates to the place of articulation of the consonant that follows it across the word boundary. The rule is mechanical:

• Before p, m, v (lip consonants), -n → m: onpa → ompa, menen pois → menem pois, sen mukaan → sem mukaan.
• Before k, g (back consonants), -n is pronounced as a velar nasal (the ng sound), as in sen kanssa → seng kanssa.
• In fast speech before j, l, r, s, the -n assimilates completely, doubling that consonant: sen jälkeen → sej jälkeen, menen lujaa → menel lujaa, sen sijaan → ses sijaan.

The vanishing d: tiedän → tiiän

Standard Finnish has a weak-grade d in words like tiedän "I know", meidän "our", yhden "of one". This d is itself something of a written-language artefact — it was chosen as a neutral compromise for a sound the dialects handle in wildly different ways. In ordinary spoken Finnish across much of the country, the intervocalic d simply drops, leaving the vowels to meet directly.

Unlike the -n assimilation above, this reduction does get written in casual text and dialogue — tiiän, meiän and tänään are common in messages and in fiction rendering speech. But it is still firmly (informal / spoken); an essay keeps tiedän and meidän. Note also that this is a dialect flashpoint: where standard has d and this spoken register has zero, other regions use j, r, l or a tapped sound (western meirän, some dialects meijän), a variation covered under regional pronunciation. The dropped d itself is the weak grade of t (see consonant gradation t → d); relaxed speech simply erases the compromise consonant that standardisation inserted.

Vowel coalescence: taloa → taloo

When a partitive or similar ending brings an -a / -ä up against a preceding vowel, fast speech coalesces them: the ending vowel is swallowed and the preceding vowel lengthens. The commonest case is the partitive of vowel-final stems.

The pattern is regular: taloa → taloo, kahvia → kahvii, euroa → euroo, tietoa → tietoo. It also hits the -eA adjectives, where -ea/-eä becomes a long -ee: korkea → korkee "tall", vaikea → vaikee "difficult", pimeä → pimee "dark".

Now the trap the brief-writers of Finnish courses always flag: coalescence preserves vowel harmony. The long vowel that results is the word's own vowel, so a back-vowel word produces a back long vowel and a front-vowel word a front one. There is no leaking of a into a front word or ä into a back one.

Compare back taloa → taloo with front hyllyä → hyllyy: both simply double the stem vowel, and because the stem already carries the harmony, the result can never break it. You will hear taloo (back) and pöllöö "owl" (partitive, front) but never a harmony-scrambled mix.

Clitic fusion: otatko sinä → otaksä

The question clitic -ko / -kö reduces to -ks in speech, and the pronoun that follows fuses onto it. This is how onko becomes onks and otatko sinä collapses into otaksä.

The mandatory comparison, feature by feature:

Notice in Otaksä kahvii? how several processes stack in one short question: the clitic reduces (-ko → -ks), the pronoun fuses (sinä → sä), and the object coalesces (kahvia → kahvii) — all at once. That stacking is why fast Finnish is hard to parse at first and easy once the rules are in place: you are not hearing one blur, you are hearing three regular reductions. The clitic itself is treated fully on the question clitic -ko/-kö page.

Comparison with English

English does exactly this. Did you eat? becomes "d'jeat?", going to becomes gonna, want to becomes wanna — regular fast-speech assimilations that native ears decode without effort and that no one writes in a report. The parallel is precise: onks sulla is to onko sinulla what whatcha is to what are you. The classic English-speaker error is the opposite of slurring — it is over-articulation: pronouncing every written segment of Otatko sinä kahvia? with equal weight produces a stiff, robotic, unmistakably foreign delivery, because no Finn actually keeps all those segments in casual speech. Learning the assimilations is not learning to be lazy; it is learning to sound like a person rather than a text-to-speech engine. (For the reductions that reshape sounds at word boundaries in careful speech too — like boundary gemination — see sandhi.)

Common Mistakes

Key Takeaways

• Assimilations reshape sounds; dropped endings delete them — this page and dropped endings are two halves of fast speech.
• Final -n copies the place of the next consonant: onpa → ompa (before p), sen kanssa → velar nasal (before k), sen jälkeen → sej jälkeen (before j). Pronunciation only — never written.
• Intervocalic d drops: tiedän → tiiän, meidän → meiän. This one is written in casual text but stays d in formal writing; the d is a standardised compromise the dialects render as j, r, l or zero.
• Partitive vowels coalesce: taloa → taloo, kahvia → kahvii, korkea → korkee — and the long vowel keeps the word's harmony (hyllyä → hyllyy).
• Clitics fuse: -ko/-kö → -ks, pronouns attach (otatko sinä → otaksä). Several reductions stack in one word — regular, so decodable.
• The English-speaker error is over-articulation; these forms are rule-governed reductions, not sloppiness.