Finnish and Agglutinativeness

Every now and then - although I think it's getting less common now - people on conlanging communities online mention Finnish as a  typical "cartesian product" language - i.e. its morphological subsystems are very clear examples of systems that just line up morphemes in a perfectly orderly manner. So, the noun morphology consists of root-[number]-[case], and since you have two numbers and 14 cases*, you have 28 case-number forms, and once you add in the various discourse clitics you do get a huge number.

However, the case system is not as trivial as that. First, the purely odd gaps in it:

	singular	plural	pronouns
nominative	auto	autot	minä
accusative I			minut
accusative II	auton
genitive		autojen	minun

Secondarily, the local cases form a thing suggestive of cartesianness. First of all, yes, with these, we do have the full number * case thing going, (almost**) by simply putting -i- between the root and the case suffix.

	to	at	from
on	-lle	-llA	-ltA
in	-Vn***	-ssA	-stA

And we find a very clear implication here: -l- = on, -s- = in, doubled consonant = at or to, -tA = from, -A without t = at, -e = to. However, obviously, *-sse does not exist.

What makes this system less convincing as a cartesian product is the non-local usages of these forms, which do not really seem to form any "semantically cartesian" thing at all. E.g. -ltA and -lle both mark what something appears to be or what impression something gives, -llA marks instruments, -stA marks by whose opinion something is or does something, and -ssA weirdly enough also mark some relations that would sort of really rather more logically seem to desire the other noun to be marked (e.g. a ship is in a load of something, whereas one'd kinda expect that to be phrased as 'in the ship there is a load of something'), etc.

We further have two outliers, -ksi (to, role) and -na (at, role); they do not look alike at all, and in addition in almost no Finnish dialects does a (from, role)-case exist.

Furthermore, we have the comitative case - which signifies 'with'. It lacks morphologically singular forms altogether, although the forms can have singular meaning, so a form like 'vaimoineni' - which superficially looks like 'with my wives' - means 'with my wife'. A thing that makes this case peculiar, is that it cannot stand without a possessive suffix - meaning that the number of permissible forms is one less than half of the forms permissible for most nouns (i.e. for most numbers, you have two numbers times {five personal suffixes or no suffix at all}, this one only has plural times five personal suffixes.

The possessive suffixes introduce one more complication: for most cases, you just stick the possessive suffix at the end of the word, but for the whole of nominative (singular and plural) and accusative (singular and plural) and genitive singular, the possessed forms are indistinguishable - thus e.g. "my brother (nom)", "my brother (acc)", "my brothers (nom/acc)", and "my brother's" are identical - veljeni. Without a possessive suffix, these would be veli, veli/veljen, veljet, veljen - thus three surface forms of five underlying forms are conflated into one surface form once the possessive suffix is added.

From a more purely morphological point of view, it is well worth keeping in mind that Finnish has dozens of conjugations and declinations, and which conjugation or declination a word belongs to is not necessarily entirely obvious from its 'shape', and sometimes one form of a verb might coincide with another form of another.

In the verb system, the first obvious irregularity is the formation of the negative and of the negative past; the negative uses a negative particle inflected for person. The past tense negative uses the same negation particle, followed by a past participle. The past tense positive uses a suffix -i-. However, some verbs have an -i- in the stem, and thus have no distinct positive past tense.

Example:

kello soi = the bell rings
kello soi = the bell rang
kello ei soi = the bell doesn't ring
kello ei soinut = the bell didn't ring

 A peculiar situation indeed - a language that for a significant number of verbs only distinguishes their tense when they're negated. The passive provides a further oddity with regards to tense and negation:

syö[+ person] = [person] eats
e[+ person] syö = [person] ate

syödään noun = noun is being eaten
ei syödä noun = noun isn't being eaten

syötiin noun = noun was eaten
ei syöty noun = noun was not eaten

The passive in the present tense has a two-part suffix, of which the latter half is lost when negated. (Oftentimes, this will end up identical to the infinitive, but not always). In the past tense, it has a single suffix, -tiin, which, when negated, is replaced by the past passive participle.

The point of this post is to showcase some ways in which a morphology can be made somewhat non-cartesian; other ways exist - Chukchi, for instance, only distinguishes singular and plural in the absolutive case for most nouns. Russian conflates accusative and nominative for some nouns, accusative and genitive for some nouns, and has a separate accusative for feminine singulars.

* maybe more? who knows, man?

** In reality, what you do is take the partitive plural - which is by no means all that trivial to form, then cut off the -(t)a suffix, then maybe turn the -j- that might precede that into -i-, and voila, you have the plural root.

*** this one misbehaves a lot, being realized as V:n, hVn, sV:n, etc.