Consonant assimilation at word boundaries inFinnish


In the Finnish language, there is a relatively simple phenomenon that has caused much confusion due to inadequate explanations and terms. At boundaries between words, it often happens that the final con貞o要ant of the first word is completely assimilated so that it forms a geminate (double con貞o要ant) with the initial con貞o要ant of the second word. For example, the imperative form of a word that means “go” is (in standard speech) [menek] before a word that begins with k, [menep] before a word that begins with p, etc. Linguistically, this is a very simple example of a sandhi phenomenon.

In many situations, the assimilation is so regular that the first word does not appear in unassimilated form at all. Instead, when there is no next consonant to assimilate to, the final consonant is lost, e.g. mene. Moreover, the Finnish orthography does not write the con貞o要ant at all, i.e. the written form corresponds to the exceptional form that ends with a vowel, such as mene.

This phenomenon has many names in Finnish grammars, all of them more or less mis衍ead虹ng. Some names refer to aspiration, which is nonsense. Other names mean final or initial gemination, which is correct in a sense but does not actually describe the phe要om苟non.

I wrote this document mainly for people studying Finnish as foreign language and for linguists interested in Finnish. No particular prior knowledge about Finnish is assumed, but I mostly do not explain the sample words or the Finnish expressions in general. Thus, even without knowing Finnish you can hopefully get a general idea of this phenomenon.

The nature of the phenomenon

Let us first consider a somewhat different phenomenon. In colloquial Finnish, it is very common that in participles ending with -ut or -yt, the final -t is assimilated according to the first consonant of the next word. For example, the statement en mennyt sinne usually becomes [en mennys sinne] (or, with another assimilation, [em mennys sinne]) in spoken Finnish. Moreover, before a pause and before a vowel, the final -t in such forms is often lost, making e.g. en mennyt ollenkaan the spoken statement [en menny ollenkaan]. Such a phenomenon is normal spoken Finnish in many areas, including the Helsinki area. One might even say that it would be at least somewhat unnatural to use a final -t in such participles. Yet, the final -t belongs to standard written Finnish, and it is also used when reading written Finnish aloud as well as when speaking in public in more or less official situations.

The phenomenon we are discussing here is actually very similar, except for two features: the assimilation or omission of a consonant is part of standard spoken Finnish, and the standard orthography does not express the consonant at all. For example, the statement en mene sinne is read [en menes sinne] (or [em menes sinne]) in standard Finnish and in most dialectal pronunciations, too. It would be quite comparable to this if participles like mennyt lost their final consonant in the orthography and if the common pronunciation (assimilation or omission, depending on phonetic context) became officially accepted and required. To put it the other way around, if word forms like mene were written as menek (reflecting the history of the language) but pronounced as menej, menel, menem, etc., or perhaps as mene or menek, depending on the next sound, then there would probably be one mystery less in the Finnish language.

The phenomenon occurs at word boundaries in the following sense. First, as explained later, only a limited (though large) set of words has an assimilating final consonant. Second, the phenomenon only occurs when the word is immediately (without a pause) followed by a word that begins with a consonant sound. The phenomenon occurs

  1. between words, e.g. kone sammui [kones sammui]
  2. between a word and an inclitic particle (-kin, -kaan, -k鳵n, -ko, -k, -pa, -p, and possibly -han, -h鄚) attached to it, e.g. konekin [konekkin]
  3. between components of a compound word, e.g. konekauppa [konekkɑuppɑ]

However, some dialects completely lack the phenomenon (e.g., in the Kymenlaakso area and in the area around the city of Pori). In standard Finnish, it is more or less acceptable to omit the gemination in the third case (pronouncing e.g. [konekɑuppɑ]) and maybe even in the second case. Moreover, even in the first case, there is considerable variation. Pronouncing [kone sɑmmui] would not sound odd to most Finns, whereas in e.g. mene pois, lack of gemination (pronouuncing [mene pois] instead of [menep pois]) would be abnormal in standard Finnish, though normal in some dialects.

The phenomenon creates even geminates that do not otherwise appear in Finnish (except perhaps in new loanwords and interjections). Considering just sounds that may appear at the start of an originally Finnish word or a fully adapted loanword, we can identify that the following geminates also appear in the middle of a word: [kk], [ll], [mm], [nn], [pp], [rr], [ss], and [tt]. The following appear as a result of an assimilating final consonant only: [ɦɦ], [jj], and [ʋʋ]. However, [ɦɦ] sometimes appears in interjections (like hihhei), and [jj] and [ʋʋ] appear in some phonetic variants inside words in cases where the orthography uses a single j or v or no consonant letter between vowels. For example, the word form aion (of the verb aikoa) is seldom pronounced as written; [ɑijon] or even [ɑijjon] is more common. Similary, the noun sauva is often pronounced as [sɑuʋʋɑ].

The phenomenon occurs before a foreign consonant sound, too. For example, ostɑ filmi is pronounced as [ostɑf filmi]. As usual in Finnish, the letters b, d and g are often pronounced as [p], [t] and [k] or just partly voiced. Thus, in educated and careful speech osta banaani might be pronounced as [ostɑb bɑnɑ:ni], though the actual pronunciation is mostly closer to [ostɑp pɑnɑ:ni].

For a description of the phonetic nature of the phenomenon, with spectrograms, see the document Suomen fonetiikkaa: Jäännöslopuke/ loppukahdennus/ alkukahdennus.

When a word that normally has an assimilating final consonant is pronounced in isolation or followed by a pause, the pronunciation ends with a vowel. However, the quality of this vowel, and perhaps the intonation of the word, may have some pecularities, which are has to describe. Shouting Anna! as an imperative of the word antaa (i.e., Give [it]!) might differ from shouting Anna! when addressing a person with this name. But I won’ try to describe the difference or claim that it would be recognizable. I’m just saying that some Finns might think there is a difference. However, it is not a glottal stop or any other consonant, though the vowel might end in a manner that slightly resembles a stop.

There are some verbal jokes that are based on the ambiguity of word forms that would be unambiguous if the assimilating final consonant appeared in some clearly recognizable form before a pause. The jokes are not good or new, but they might be interesting for linguistic reason. There is a worn-out joke about the question
Saako t鳵ll kiljua?
where the last word could be interpreted as a verb infinitive (meaning “yell, squeal”) or as the partitive singular of the noun kilju, which means a homebrew alcoholic drink of a kind. There is little or no difference in pronunciation, in any dialect. Similar jokes are also written (or told) in a form where the ambiguity is resolved thanks to the gemination, e.g.
Saatteko te muka kiljua ruokailussa?
In such cases, the joke implies that some people pronounce the verb infinitive without gemination.

Confusion of terms and explanations

There are so many wrong descriptions of this simple phenomenon that it would probably just add to the general confusion to list them in detail. The most common wrong approach is to describe the word forms under discussion as terminating with some mystical phoneme or sound, e.g. as “glottal stop”, which may “manifest itself” in different forms.

We could say that the spelling mene represents a word that ends with a consonant but so that the consonant is obligatorily assimilated or lost. The question then arises what this consonant is, since it never appears as such. Historically it is k, but let us not confuse the description of the modern language with historical forms or reconstructions, no matter how certain they are. It is better to describe the situation so that the spelling mene corresponds to a large set of pronounced words according to contextual rules. None of the forms is really primary, except by convention: when a word is mentioned in isolation (which is a quite special situation), then a form without a trailing consonant is used.

The term loppuhenkonen (final aspiration, final breathing) was used in old grammars, and it has mysteriously created an idea of a breathing-like sound. This term and its connotations are still known by many Finns, who thus have a quite wrong idea of a phenomenon of their language.

Another, less serious confusion has been caused by the term loppukahdennus (final gemination, end gemination), which is probably the most common name for the phe要om苟要on or for the unwritten consonant that appears at ends of words. In a sense, the term corresponds to the nature of the phenomenon, since it is total assimilation, resulting in a geminate, i.e. a double consonant. Describing the pronunciation as en menes sinne (as above) might be misleading, since there are normally no pauses between words. Since Finnish always has stress on the first syllable of a word, word boundaries are recognized from stress, so there is especially little reason to leave pauses between words. A spelling like enmenessinne (or maybe emmenessinne, due to a possible assimilation not discussed in this document) would better express what happens at the word boundary. (However, it might be argued that the geminate resulting from “final gemination” is not necessarily quite the same as normal geminates. It might be shorter, perhaps due to uncertainty, perhaps due to the orthography.)

Yet, “final gemination” is somewhat misleading, since the phenomenon appears where words join. Moreover, it also appears inside a word in some situation, such as before an enclitic particle (e.g., menekin is pronounced menekkin).

This document uses the expression consonant assimilation at word boundaries to refer to the phenomenon, whereas a word that participates in the phenomenon as the first component can be said to have an assimilating final consonant.

The large descriptive grammar of Finnish published in 2004, Iso suomen kielioppi, uses the term rajageminaatio, which could be translated as boundary gemination. (Such a term is actually used to some extent, though in the combination morpheme-boundary gemination, which is questionable, since this is not about morphemes in general but about words and enclitic particles.) The glossary of the grammar is available online: Sananselityksi: Ison suomen kieliopin termej.


Sometimes notations like menex (i.e., with superscriptx) or mene’ (i.e., with an apostrophe) are used to indicate that a word has assimilating final consonant. Such notations can be useful if understood as auxiliary notations indicating the presence of an assimilating (or conditionally absent) consonant, but quite often, they have created the impression of a real sound, not just an abstraction.

Strangely enough, even most Finnish dictionaries do not usually indicate an assimilating final consonant in any way. This is understandable for word forms and words that belong to some patterns that always have an assimilating final consonant, such as adverbs ending with -sti. It is less understandable when the existence of the phenomenon cannot be deduced from any general principles but can be classified as lexical information. However, Nykysuomen sanakirja (an old dictionary, published in the 1950s) sometimes indicates the phenomenon, using pronunciation information as in the following:

alemma [-a’]

The notation is explained in the dictionary as follows:

’ merkitsee ns. loppuaspiraatiota (joka eri asemissa 鳵ntyy eri tavoin tai j鳵 鳵ntym酹t鄢in). Sit ei merkit tapauksissa, joissa se kuuluu muototyyppiin kauttaaltaan: e-loppuisten, taivutustyyppiin 78 [hame], 79 [terve] ja 82 [askele] kuuluvien sanojen nominatiivissa, allatiivissa, nsa-, ns-suffikseissa, Iinfinitiiviss, sti-, lti-, nne-, ti- ja tse-loppuisissa adverbeissa.

In the quotation, the words in brackets are my additions, showing the model words of the declination types identified in the text by numbers. Note the highly misleading term loppuaspiraatio, final aspiration.

Beware that in older Finnish orthography, a final apostrophe is used to indicate omission of letters from a standard form, with no specific implication on pronunciation. Thus, in an old poem verse Se aik’ ol’ ajoist’ ankarin the apostrophes just indicate omission of final vowels (with the full forms, it would be Se aika oli ajoista ankarin), and in poetry the word form mi’ indicates a contraction of mik, with no final consonant in pronunciation. In modern orthography, apostrophes are not used that way; modern poetry writes e.g. just mi. (Modern orthography uses the apostrophe for other purposes, inside words.)

In informal writing, or in texts by unexperiences authors, an assimilating final consonant is sometimes written according to the pronunciation, e.g. menep pois, especially before an enclitic particle, e.g. menepp (standard orthography: menep) and e.g. olekkaan (standard orthography: olekaan). The usual reason is that the writer does not remember the rules (although they are taught at school), rather than any conscious choice.

On the other hand, in old texts, in texts written in dialects, and in proper names, words with assimilating final consonant in standard Finnish may appear in older forms, with a fixed final vowel. Thus, for example, standard Finnish has the word korte (horsetail), but the older form kortes appears in names like Kortesj酺vi and Kortesoja. For a survey of notations used in printed texts in the 19th century, refer to the Master’s thesis of Kari Blomster, Loppu虺ah苓en要uk貞en merkitseminen varhaisnykysuomessa, which also contains other detailed descriptions of the history of the phenomenon.

People who think that the assimilating final consonant should somehow be expressed in writing do not usually apply such ideas in their own texts. Various proposals have been made, more or less seriously.

I have thrown in the idea of simply writing according to the pronunciation. See Onko suomen kirjoitusj酺jestelm ihanteellinen?, subsection Loppukahdennus. I don’t regard the proposal as very realistic, though. However, it might one day be accepted for use in the situations where the presence of a final assimilating consonant may make a difference in meaning, as in some infinitives.

The glottal stop

In some forms of spoken Finnish, words that start with a vowel in standard Finnish may be pronounced so that they start with a glottal stop. This tendency is reflected, in some dialects, in situations where a consonant assimilation would appear. Thus, similarly to pronouncing mene pois as [menep pois], some people pronounce mene ulos as [meneʔ ʔulos], where the symbolʔ denotes a glottal stop.

Thus, there is some reality behind names and descriptions that refer to a glottal stop. However, the glottal stop is not counted as a sound in standard Finnish. The consonant assimilation at word boundaries is not a successor of any word-final glottal stop but different consonants, usually [k] or [h], sometimes [s] or[t].

Which words have an assimilating final consonant?

Finnish words (word forms) that have an assimilating final consonant can be divided into several groups. The phenomenon is an established part of standard Finnish in some of the groups, less common (or less obligatory) in many other groups, and very rare if not obsolete in a few groups, listed below as a separate table.

In the following groups of words, an assimilating final consonant is common (ranging from very common to rather common):

#Group Examples of word forms in the group
1Imperative, 2nd person singular (except )mene, kirjoita, vastaa
2Negative verb forms (except conditional) (en) ota, (ei) oteta, (鄟) ota, (鄟k礵n) ottako, (鄟k礵n) otettako, (en) ottane, (ei) otettane
3Infinitive (dictionary form)menn, kirjoittaa, vastata
4Words with 3rd person “possessive” suffix -nsa or -nstalonsa, p鳵ns
5Allative forms (-lle suffix) talolle, taloille, p鳵lle
6Adverbs with -nne suffix t鄚ne, minne
7Adverbs with -tse suffix (“prolative”) meritse, postitse
8Adverbs with -sti suffix kauniisti, pahasti
9Some other adverbs and postpositions kiinni, luo, taa
10The basic form of most nouns that end with -e hame, terve, askele
11The numeral for three, in base form kolme
12The reflexive pronoun itse
13Some (mostly archaic or dialectal) nouns and adjectives kiiru (= kiire)

In case 11, standard Finnish allows both gemination and lack of gemination. For example, the word kolmetoista (thirteen) may be read as [kolmettoista] or as [kolmetoista]. The latter is probably more common.

The second person singular of the imperative, such as mene (go!), is morphologically the same form as the one used in negative sentences like en mene (I don’t go), including the negative imperative like 鄟 mene (don’t go).

The 4th person, often misleadingly called “passiivi” in Finnish grammars, indicates an unspecified but personal actor. Thus, kirjoitetaan means that some people are writing, without specifying the people (or even whether that means one person or more). However, the actor is often somehow implied, and in widespread colloquial usage, the 4th person is used in the role of 1st person plural; for example, (me) menemme is more or less literary and formal, whereas as me menn鳵n is common especially in urban speach. The negative form of the 4th person, such as in ei anneta (or in me ei anneta, the colloquial version of (me) emme anna) has a final assimilating consonant. Quite often the negative form coincides with the infinitive, as e.g. in ei menn.

Among the adverbial suffixes, -sti is very common and can be attached to most adjectives. It roughly corresponds to English -ly in meaning and productivity. The other adverbial suffixes are mostly unproductive and appear only in a few words, some of which are rather common. The suffix -tse is productive to a limited extent.

Nouns that end with -e in the basic written form mostly belong to declinations that have the vowel lengthened (written as -ee-) in most singular forms. For example, terve: terveen: terveess: terveeksi etc. Such words have an assimilating final vowel in the nominative singular. However, this in practice optional in standard Finnish. Although tervetuloa is in principle pronounced as [tervettuloa], the alternative [tervetuloa] hardly sounds odd to most Finns.

The above-mentioned nouns have the partitive singular ending -tta or -tt, e.g. tervett, instead of the common -ta, -t, -a, or -. The first t actually belongs to the stem and stems from the original final vowel—which was, in fact, often t, but has otherwise been assimilated to the consonant that starts the case suffix.

There is a relative small number of nouns ending with -e that belong to declinations where the vowel is not lengthened, e.g. nalle: nallen : nallella etc. In such words, the nominative singular has no final consonant in pronunciation.

In the following groups of words, an assimilating final consonant occurs according to some descriptions of Finnish but rarely (if at all) in practice in modern language:

#Group Examples of word forms in the group
1Negative forms of the conditional, ending with -isi (ei) olisi, (emme) menisi, (ei) oltaisi
2Base form of participles ending with -tu or -ty saatu, tehty, menty
3Plural verb forms in the 1st and 2nd person, ending with -mme or -tte saamme, teette
4Words with a “possessive suffixes” of the 1st and 2nd person plural, ending with -mme or -nne maamme, kotinne
5Adverbs with -lti suffix paljolti, laajalti
6Adverbs with -ti suffix 鳵neti, kaiketi, olleti(kin)
7Some other adverbs alemma, edemm, irti, kahtia, kai, kolmia, kotia (= kotiin), taamma, tyk, tyynni, yh, ylemm, ymp酺i
8Some conjunctions eli, mik鄟i, sik鄟i, tai, vai
9Abessive forms, with -tta or -tt suffix rahatta, sanomatta
10Comitative forms without possessive suffix, i.e. ending with -ne (see below) molempine (poikineen), kaikkine (lisukkeineen)
11Translative forms, ending with -ksi taloksi, taloiksi
12The imperative of the negation verb (2nd person singular)

For example, in many dialects, the abessive ending is -ta or -t, i.e. with a single t instead of the double tt of standard Finnish. In such dialects, the ending often has an assimilating final consonant. Conceivably, speakers of such dialects may extend the feature to the abessive forms that they use when trying to speak standard Finnish.

Some of the cases in the table are from the booklet Suomen kielen 鳵nne- ja oikeinkirjoitusoppi by Aarni Penttil, published in 1948. It seems that its author tried to cover all the cases and got somewhat too anxious, so that he included purely dialectal cases, perhaps just mispronunciations.

The appearance of a final assimilating consonant in participles like saatu can hardly be regarded as acceptable standard Finnish, and it is not common. The reason for the phenomenon is probably analogue with participles ending with -nut or -nut (e.g., saanut). As described above, the final -t is very often assimilated (or lost) in pronunciation, creating a phenomenon very similar to the one discussed here, except for the orthography. Similar considerations apply to cases 3 and 4 in the table, except that no obvious reason for the phenomenon can be identified.

Conditions for omission of the consonant

Previously we have mentioned that that the assimilating final consonant is omitted

Although the assimilating final consonant is usually omitted before [h], it may affect the pronunciation. In Finnish, an intervocalic h is usually voiced, but probably more or less unvoiced (as at the start of a word) in words like olehan. This means that a written form like sy鐬鄚 is pronounced differently depending on whether the word sy in it is indicative (“eats”) or imperative (“eat!”).

Can the unwritten consonant be distinctive?

In most cases, the occurrence of an assimilating final consonant can be predicted and it makes no difference in meaning. However, it may make a difference, though usually ambiguities can be resolved by context. The most common situation where real ambiguity can occur is a sentence where a verb form such as kantaa (carry) could be parsed either as 3rd person singular (carries), or as the infinitive (to carry), and in standard speech, there is gemination in the latter case. In dialects where gemination does not occur, and in written language, this may allow two plausible interpretations for a sentence.

Here is a sample sentence with an ambiguity:
H鄚 halusi auttaa minua ja kantaa matkatavaroitani.
When kantaa is pronounced with gemination, [kɑntɑ:m], the interpretation is “He wanted to help me and to carry my luggage.” When kantaa is pronounced without gemination, [kɑntɑ:], the interpretation would be that it is a predicate and the meaning is “He wanted to help me and carries my luggage.” This is a subtle difference perhaps, but in other cases, there might be a substantial difference.

To take a real-life example, when the TV series on the Moomin characters was presented in Finland, I first misunderstood the start of its theme song. I heard K銛 Muumi laaksoon (AMoomin is walking into a valley). Later I realized that it is K銛 Muumilaaksoon! (Come to the Moomin valley!) pronounced without gemination, i.e. [k獠 mu:milɑ:kso:n] instead of standard [k獠m mu:milɑ:kso:n], and stressed so that it sounded more like three words, i.e. with three main stresses [k獠 mu:mi lɑ:kso:n].

Another example is the start of the national anthem, Maamme: Oi maamme, Suomi, synnyinmaa! Soi sana kultainen! The written verb form soi can be either indicative (sounds [loud]) or imperative (sound [loud]!), and there is nothing in the Finnish text that excludes either interpretation. The intended interpretation is imperative, as in the Swedish text, from which it has been translated. Yet, many Finns misinterpret the word as indicative and pronounce it as [soi] instead of the correct [sois]. (Of course, the [soi] pronunciation may also reflect dialect background.)

In some situations, written forms are unambiguous but the spoken form is ambiguous. The reason is that the assimilating final consonant may coincide with a consonant appearing as a suffix. The spoken sentence [ɑnnɑt t熡ne] can be interpreted as the indicative annat t鄚ne (you give [it] here, you are giving [it] here), or as the imperative anna t鄚ne (give [it] here!).

As you have seen, the gemination under discussion is rather irregular and subject to variation among speakers of Finnish. This is not surprising if you think that it has never been indicated in normal orthography (or even in Finnish dictionaries, as a rule) and that there is considerable dialectal variation.

Thus, in practice, if a foreigner fails to use the assimilating final consonant according to the rules of standard Finnish, this will probably not affect the understandability, except rarely. There is an affect on naturalness, but speakers of some Finnish dialects will make the same impression.