Probably the most widely misunderstood and misrepresented feature of Finnish pronunciation is boundary gemination. It is a relatively simple phenomenon: there are word forms that end with a consonant, but this consonant varies by context so that it is the same as the initial consonant of the next word. This consonant is not expressed in standard writing in any way. For example, the phrase ota se (take it) is pronounced otas se.
Boundary gemination can be classified as a sandhi phenomenon. Generally, “sandhi” refers to phonetic changes that occur in speech at word boundaries.
Boundary gemination appears most regularly in the following word forms and endings, where it can be regarded as part of standard pronunciation:
There are many other situations where boundary gemination may appear in some language forms. For example, in negative imperative forms like antako, as in Älkää sanoko (Do not say), it is possible and probably goes unnoticed, whereas in passive past participle forms like annettu it is clearly a dialectal feature.
Boundary gemination does not appear when a word is pronounced in isolation or before a pause, as in the exclamation Anna! But in a normal flow of speech, the word is pronounced (though not written) so that ends with a consonant that is identical with the initial consonant of the next word. This means that anna se is pronounced annas se, anna tuo is pronounced annat tuo, etc. If the next word starts with a vowel, there is usually no boundary gemination, e.g. anna olla is pronounced as written. Boundary gemination is rare before h; e.g. anna heti is usually pronounced as written, less often as annah heti.
Boundary gemination also appears inside a word before a word-like suffix such as pa. For example, tulepa is pronounced tuleppa, and talonsakin is pronounced talonsakkin. It also appears at the junction of the parts of a compound word. For example, the word itse normally has boundary gemination, and this often applies to compound words like itsepäinen (stubborn; literally “self-headed”), which is pronounced itseppäinen. However, boundary gemination is much less systematic inside compound words. When the first part is a noun, lack of boundary gemination is probably more common, e.g. kirjekuori (envelope) is pronounced as written, rather than kirjekkuori. Similarly, hernekeitto (pea soup) is usually pronounced as written, but hernekkeitto is possible, too. This may extend to phrase-like open compounds like terve tuloa (welcome), which may be pronounced as tervet tuloa, but it is common to pronounce it as written.
The description of boundary gemination above is mostly sufficient for a language learner. There is much variation in the appearance of boundary gemination, by dialect, by speaker, and by word type, so stay tuned to hearing it in many other contexts than those mentioned. Some dialects do not have boundary gemination at all, though more often it appears to some extent.
Some dialects lack boundary gemination, so lack of it does not necessarily sound foreign—though somewhat odd outside the areas of those dialects.
In the context of consonant gradation, a word with boundary gemination is treated as ending with a consonant, irrespective of its actual pronunciation. Thus, the word liikex : liikkeen (movement) has weak grade in the nominative; the syllable ke is treated as closed. Similarly, the 2nd person singular imperative forms have weak grade, e.g. liikkuu (moves) : ei liikux (does not move).
Boundary gemination seldom affects the understandability of speech. However, many verbs have the infinitive and the 3rd person singular written identically (e.g. ostaa “to buy; buys”) but pronounced differently, because the infinitive has boundary gemination. In rare situations, this might affect the meaning of a sentence. More often, the gemination just helps the listener to recognize infinitives and therefore understand the sentence structure easier.
Many descriptions of Finnish characterize boundary gemination using words like “aspiration” or “glottal stop”, which is seriously misleading. No aspiration is involved, and a glottal stop only occurs in specific situations: in some dialects, boundary gemination appears before a word that starts with a vowel, causing e.g. anna omena to be pronounced anna? ?omena, where ? denotes the glottal stop.
Historically, boundary gemination is caused by the assimilation of a final consonant. The imperative form that is now written as anna was earlier probably pronounced as annak in all contexts. The final consonant was then assimilated (adapted) to be same as the initial consonant of the next word or, under certain conditions, completely omitted.
Such a process has relatively recently taken place in common spoken Finnish in participle forms ending with nUt, such as sanonut.
The assimilated consonant is not written at all in current standard orthography. In linguistic contexts, it is often marked with a superscript x, e.g. annax. This notation is used in this book to some extent.
When writing dialogue or other spoken words, whether common spoken language, a dialect, or a jargon, words are often written as spoken, e.g. Annam mullet toi kirja (standard Finnish: Anna minulle tuo kirja). However, it is more common to follow the standard language writing convention of not indicating boundary gemination, e.g. Anna mulle toi kirja.
Other sandhi phenomena in Finnish include the following.
The pronunciation misson mentioned above can be classified as legato, which means that successive words are pronounced as one word. Since there is normally no pause between words, legato means in this case that there is a double s, as opposite to pronunciation before a consonant, like mis Pekka on. Legato pronunciation can also be said to change the syllable structure; e.g. misson has the syllables mis and son. However, syllable division is largely a matter of convention rather than phonetic reality.
Legato pronunciation may also cause a short consonant to be doubled, e.g. nyt on may be pronounced as nytton and jos ei as jossei.
Some legato pronunciations have been established as a standard language words, as contractions, e.g. ettei (että ei), ehkei (ehkä ei), and jollei (jolla ei). In word division, they may be divided according to structure (ett-ei) or phonetically (et-tei), though both ways may look somewhat odd to some people.