In linguistics, “phoneme” is a more abstract concept than “sound”. Different sounds are treated as being appearances of the same phoneme, if their differences do not make a difference in meaning of words. We can say that two sounds represent different phonemes if there is a minimal pair of words that differ only by the difference of those sounds. For example, the minimal pair of “sip” and “ship” shows that in English, “s” and “sh” represent different phonemes.
In standard Finnish words excluding relatively new loanwords, only the following phonemes appear (indicated here by their letters in normal Finnish orthography, except for the eng sound ŋ):
Almost all dialects lack the d sound, but it is well established in standard language as the “weak” counterpart of the t phoneme (between vowels), e.g. katu (street) : kadut (streets). The phoneme d also appears in different positions in loanwords, such as demokratia,
In new loanwords, and in some Western dialects, the consonants b and g may appear too. Moreover, š may appear in educated speech, and so may ž in the combination dž in a few words. Some minimal pairs like sakki (gang) – šakki (chess) for making š a distinct phoneme can be found, but it is questionable whether their difference is recognized from the difference of s and š sounds or from the context. (See also š versus s.) The sound ž can hardly be classified as a phoneme in Finnish, due to its limited contexts of use.
The phoneme status of b and g is more debatable. Minimal pairs like paarit (stretcher) – baarit (bars; cafés) can be recognized, but it is questionable whether the phonetic difference between p and b really makes a difference in meaning. This is discussed in more detail in section Unvoiced and voiced stops.
In foreign names and in newest loanwords, the consonant w (as in English) and the neutral vowel ə (as “a” in English “about”) as well as some nasalized vowels may be heard, but they can hardly be said to have phoneme status.
Vowels appear as short or long. A long vowel is normally written with two identical letters. The length of a long vowel is about two times the length of a short vowel in the same position.
The difference between short and long vowels is a phoneme distinction, i.e. it can make a difference in meaning. E.g. tuli (fire) and tuuli (wind) are distinct words, and so are of takka (open fireplace) and taakka (burden).
A long vowel could be described as a separate phoneme or as two identical phonemes in succession. However, the conventional approach in Finnish grammar is to treat the length of a vowel as a separate phoneme-like feature. Thus, a word like tuuli is analyzed as containing four phonemes, with the second phoneme as long.
In Finnish, long vowels may appear in unstressed syllables. This deviates from many European languages, where long (or half-long) vowels appear only in stressed open syllables. In the Finnish word paniikki (panic), the second syllable is unstressed and closed (ends with a consonant), but the vowel is still long.
Thus, it is important to make vowel length distinctions; they are not just a matter of sounding native. It is better to exaggerate the distinction, pronouncing a long vowel as really long, than make it too short.
When a foreign word contains a half-long (or long) vowel in a stressed position, Finns tend to interpret it as a long vowel. This is why there are long vowels, written with two letters, in loanwords such as gaala (gala) and miljöö (from Swedish “miljö”, which comes from French “milieu”). However, original spelling is retained in many words, e.g. ameba, commonly pronounced ameeba; see section Length of vowels in loanwords.
Consonants appear as single or double. A double consonant, normally written with two identical letters, is conventionally treated as two sounds, in different syllables. A word such as taakka (burden), with a long vowel and a double consonant, thus has the syllable structure taak.ka.
The syllable division is partly just conventional, but it may help to pronounce Finnish better: to produce a double consonant, try to pronounce first a syllable ending with a consonant, like taak, then immediately another syllable starting with the same consonant, like ka. Just remember that there should be no pause between the syllables. Instead, the consonant sound lasts longer; its length is typically two times the length of a simple consonant, but it can be even longer.
The difference between single and double consonants can be the only difference between two words, e.g. tuli (fire) and tulli (customs) are distinct words. Finnish grammars conventionally treat a double consonant as two phonemes.
According to some descriptions, consonants the d, h, j, and v do not appear as long. In reality, h, j, and v, can be pronounced (though not written) as long due to boundary gemination. For d, this is possible only when a new loanword is involved; it may appear as long in loanwords otherwise, too. Some interjections and descriptive words have a long h, e.g. hihhuli. Long j and long v are often pronounced, but not written, after diphthongs that end with i or u, respectively, e.g. leija [leijja]; see section Transitory sounds.
Vowels can form diphthongs, such as ai (pronounced like “eye” in English), but consecutive vowels may also belong to different syllables, like io in radio. This is described in section Diphthongs.
Except for new loanwords and some dialects, a word cannot begin with a consonant cluster. Old loanwords have adapted to this by losing all but the last one of the consonants, so that e.g. the word for beach or shore, from Swedish “strand”, is ranta.
Consonant clusters do not normally appear at the end of a word, except that some word-final clusters like ks are common in spoken language. Old koanwords have usually been adapted to this so that an i has been appended, e.g. pankki (bank).
Double consonants do not normally appear at the end of a word. A double consonant at the end of a foreign word is pronounced single, e.g. Mann [man]. However, a final vowel of a word may be omitted before a word that starts with a vowel, and in such a legato pronunciation, a double consonent before the lost vowel is retained. For example, missä on (where is) is often miss on in speech, pronounced as [mís.són], i.e. phonetically as two words in the sense that there are two main stresses, but as one word otherwise, with syllable break inside [ss]. Such a word combination may develop into a contraction word, such as ettei [ét.tei] from että and ei.
Some colloquial emphatic pronunciations such as toss (for tos, a shortened form of kiitos, ”thanks”) and jess (reflecting English ”yes”) have a word-final long (or even overlong) consonant, and it is shown in some forms of writing.
Even the appearance of a single consonant at the end of a word has restrictions. In standard language, not counting boundary gemination, word-final assimilation, and new loanwords, only l, n, r, s, and t may end a word, and l and r are rare in that position. Moreover, they rarely appear in the basic form of a noun, since a final is has usually been appended, e.g. Yet another explanation is that loanwords that have ended with a consonant have usually got a final i appended, e.g. paneeli (panel), paperi (paperi).
Finnish is rich in vowels, not by the amount of different vowel sounds but by the frequency of vowels. Finnish has eight simple vowels, which can appear as short or long and as diphthongs in many combinations. Especially if we count a long vowel as two vowels, corresponding to the orthography (a long vowel is written with two letters, like aa), the relative amount of vowels is larger than in most languages, typically about 50 %. This is one reason why Finnish somewhat resembles Italian phonetically; another reason is that the Finnish vowel signs a, e, i, o, and u have phonetic values similar to those of Italian (rather than English).
The relative amount of vowels is partly explained by the conservative nature of Finnish. A word like kala (fish) is, as far as we can know, very similar to the corresponding word in the Uralic proto-language thousands of years ago. In other Uralic languages, the final vowel has often been lost, in addition to other phonetic changes.
Another reason to the large proportion of vowels is that Finnish has not been conservative in one respect: it avoids consonant clusters, as described above.