Morphological Complexity

The following is one example of morphological complexity (more specifically, or at least especially, complex exponence, a complex relationship between units and meaning and units of form). 

In Ket, verbs indicate both the subject and the object. For example, the word daŋtuŋ by itself means 'I see them', where d- means 'I', -aŋ- 'them', and -tuŋ 'see'. This always happens, even when you have nouns or pronouns, qīm bura dɛ'ŋ daŋtuŋ,  'the woman sees her friends', lit. 'woman her people she-sees-them'. The verb always comes at the end, but it's the markers on the verb, not the word order, that tells you who does what, i.e. bura dɛ'ŋ qīm daŋtuŋ, lit. 'her people woman she-seems-them' means the same thing. 

The information about the subject and the object on the verb is really important in Ket then, and we might expect it to be marked as clearly and efficiently as possible. What we find in Ket is quite the opposite: on a fixed but arbitrary basis, with no differences in meaning, different verbs select from one of at least seven different ways of indicating the subject, some of them redundant, usually showing up in different places in the verb (d-, -ba-, both d- and -ba-, di-, -də-, -d-,  both d- and -də- or both d- and -d-, all can mean 'I'), and at least three or more different ways of marking the object (-ij-, -i-, and -u-, can all mean 'her', while -b-, -m-, -i-, -u-, or the absence of a marker, can all mean 'it'). Some, but not all, object markers also indicate the tense (e.g. -ij- means 'her' for a verb in the present tense, -iru- or -dit- for a past tense verb), information that is in most cases already indicated elsewhere in the verb, and therefore also redundant. Some ways of indicating the subject overlap with ways of indicating the object, such that e.g. -ba- means 'I', the subject, in dbatavraq 'I pull it out' or baɣabda 'I hear it', but 'me', the object, in g=batuŋ 'you see me', and it is fixed but arbitrary which meaning it has in any given verb.  

Finally, the ordering of different pieces that can appear in the verb, of which there can be as many as nine at a time, is strictly fixed, but arbitrary, and each of the pieces above also has its own rules about its ordering in the verb, relative to other elements. For example, -də- 'I' always appears before the last part of the stem, i.e. the pieces that tell you the basic meaning of the verb, like 'hear', 'see', which in Ket (unlike most languages) often consists 

Compare a language like Turkish, where all verbs are conjugated the same way, and 'I' is always only marked with -Vm (-im, -ɨm, -ym, or -um depending on the vowel of the syllable before the suffix, but always fully predictable), which always only means 'I', and always shows up as the last piece of the verb, and we see just how needlessly complex this aspect of Ket grammar is. Nevertheless, we know systems like this can be maintained for thousands of years, across many generations of speakers, and sometimes grow even more opaque with time.