A diphthong is a combination of two adjacent different vowels in the same syllable, as ie in the Finnish word tie. If two adjacent vowels do not form a diphthong, as au do not normally form in the Finnish word vapaus, there need not be any sound or any pause between them. The pronunciation still makes the vowels more separate, e.g. so that the tone of voice slightly falls at the end of a and then raises at the start of u. In contrast, the diphthong au in in the Finnish word kausi is pronounced so that a smoothly glides to u.
Since the details of pronunciation vary, the diphthong concept is relative and partly just a matter of conventional syllable division. However, the syllable division and hence the diphthong concept matters in metric poetry. In Finnish, it also matters as regards to secondary stress.
Diphthongs are very common in Finnish, but only a limited number of combinations of the eight vowels occur as diphthongs. In standard Finnish, according to conventional classification, the following diphthongs occur:
In dialects, many of these diphthongs have changed to other diphthongs. In particular, ie, uo, and yö largely appear as ia or iä, ua, and yä in Western dialects (e.g. hiano suara miäs instead of hieno suora mies), though such pronunciation is clearly substandard in public speech. The diphthong ai has changed to ae in Eastern dialects, au to ao in some dialects, etc.
In reality, there are many other combinations of vowels that are pronounced more or less as diphthongs in the first syllable of a word. For example, the words pian (soon) and luen (I am reading/I will read) are often pronounced as single syllables. This means that words like piankohan and luentoja have only three syllables, hence no secondary stress: pían.ko.han, lúen.to.ja; reading them as pí.an.kò.han and lú.en.tò.ja is less natural.
The above only applies to the first syllable, including the first syllable of a component of a compound word. In other syllables, vowel combinations other than those listed in the traditional view do not form diphtongs. For example, the word lukea (to read) consists of three syllables (lu.ke.a).
Two-vowel sequences with i as the latter vowel form a diphthong in any position within a word, e.g. in kirjoittaisimme (kir.joit.tai.sim.me), unless the i starts a component of a compound word, as in saunailta (sau.na.il.ta). A few loanwords have exceptions to this, e.g. ateisti (atheist), where isti is recognized as a unit and the syllable division is a.te.is.ti.
When the second vowel is u or y, pronunciation varies. The vowels do not form a diphthong if that would make the word have just two syllables. Thus, vapaus is pronounced va.pa.us and not va.paus. In longer words, diphthong pronunciation is common, e.g. vapaudessa is mostly pronounced va.pau.des.sa, though va.pa.u.des.sa, is possible, too. This variation was originally a dialect feature, but now varies by speaker, by word, and by context. The diphthong pronunciation is more common in open syllables than in closed syllables. In some words, the variation may affect secondary stress. For example, vapauden has no secondary stress in diphthong pronunciation, vá.pau.den, but may have it when pronounced as four syllables: vá.pa.ù.den.
Otherwise, two vowel sequences do not form a diphthong in other syllables than the first one. Rare exceptions to this include words ending with niekka (e.g. vitsiniekka), which could be classified as compounds rather than derivations, and the historical word lampuoti.
All diphthongs are written according to the general principles of Finnish orthography: each component is written using the same vowel sign as for the vowel when occurring alone. This is very different from English, where pronunciation has often changed after the written form was established. For example, in Finnish ai is always a combination of a and i, never like the “ai” in English “nail” in standard British or US pronunciation.
If you pronounce, say, suomi (Finnish language) as three syllables (su.o.mi), it sounds odd, but you will be understood. In native pronunciation, uo is clearly a diphthong, which means that the u sound smoothly glides to an o sound, with no break.
All Finnish diphthongs have the first vowel somewhat more stressed than the second one. This is different from e.g. Spanish and Italian diphthongs, where the second vowel is clearly more stressed.
There are no triphthongs (three consecutive vowels in the same syllable) in Finnish, except debatably in the interjection miau (miaow).
When there are three vowels in succession, there is a syllable boundary inside it. If the boundary is between two identical vowels, it is marked with an apostrophe, e.g. rei’issä. Otherwise, it is normally not marked.
However, it is permitted (though rare) to use an apostrophe to remove ambiguity in any three-vowel combination, e.g. hau’issa versus ha’uissa (which are inessive plural forms of hauki and haku, respectively). Usually the syllable boundary can be inferred from other forms of the word, since the three-vowel cluster results from the loss of a k in consonant gradation. For example, the syllables of kiuas are kiu and as, since other forms of the word have the stem kiukaa-. (Besides, the diphthong ua would be possible only in the first syllable of a word.)