I've heard it mentioned that languages tend to evolve in a kind of merry-go-round pattern where a language that's agglutinative slowly turns fusional, that fusional language's inflections slowly break into separate helper words, and then those same helper words start to reattach themselves into word roots, and we end up with a synthetic language again.

The proposition makes sense as far as it goes, but when was it originally proposed, and more importantly, do we have any good evidence for it being the case? Do we eg. know of patterns in analytic languages where two separate-ish words get bound together more and more tightly and the word order starts to loosen?

  • 5
    It’s not that fusional inflections break into separate words, but that inflections in fusional languages tend to be lost over time, requiring circumlocutional constructions to express the same thing. That’s the stage of the proposed cycle that’s best attested, since it’s clearly evidenced in the historic development of most (Indo-)European languages. Apr 2, 2022 at 22:41
  • 4
    An example of patterns in analytic languages where two separate-ish words get bound together more and more tightly is the simple future tense in the Romance languages which is actually a composite form of infinitive + to have in the present tense: ‘we will love’ aimerons = aimer ‘to love’ + avons ‘we have’.
    – Yellow Sky
    Apr 2, 2022 at 23:05

1 Answer 1


You can find a lot of information on the process of turning from analytic to synthetic in the research of grammaticalization. One example is the development of Mandarin perfective marker le. It originally was a lexical word, pronounced as liǎo meaning to end, to finish. This pronunciation and meaning are still kept in modern Mandarin but rarely used alone. Note that when changing from a full lexical word to a perfective marker, the phonetic form has been reduced greatly: tone is lost and the vowel is reduced. In today's Mandarin it functions like a clitic, never appears alone and is always attached to other words to form a prosodic unit.

If you feel cliticization isn't synthetic enough, here is another example from Manchu: Manchu has a cislocative associated motion suffix -nji. It can be best translated into English as "come to do something" or "come and do something", but basically the idea is to describe the translational motion of the action's performer. It's a development from an analytic construction using converbs: V-me ji, which can be translated as "doing something, come". The imperfective converb marker -me is reduced to probably -m at some stage (Sibe, a language developed from originally a Manchu dialect, shows this development as well), and the nasal is assimilated in place of articulation of j, becoming n, thus giving you the suffix -nji. Here you have a development from a lexical verb (or at least, a light verb) to an inflectional suffix, phonetic form reduced in the process, and the meaning becomes more abstract. An analytic construction is turned into an inflectional affix. (Note that in Manchu, the analytic construction still lives on.)

  • 2
    Similarly the Estonian comitative case -ga, which is a fossilised and reduced form of kaasa ‘[along/together] with’, reduced from kansassa (Finnish form, since I’m not sure what the exact unreduced Estonian form would be), originally the inessive of kansa ‘people’, borrowed from Germanic *hansō (same word as in German Hansestadt). So from ‘among people’ to ‘together with’ to a case ending. Mar 2 at 20:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.