There are two basic possibilities – sudden or gradual – both for changes within languages and for changes to languages (language splitting). I discuss these in turn and then, in §5.3, consider the way in which language may have originated.

Changes within languages

I suggest that many types of change within a language are not gradual but rather happen fairly suddenly, often within the space of a generation or two. That is, change is more like a series of steps than it is like a steady incline.

If a new grammatical mechanism is innovated this is likely to happen all at once, rather than bit by bit. Bantu languages have about sixteen prefixes which combine marking of noun class and of number (singular or plural). It is unlikely that the ancestor language began with just one prefix, then added a second one a bit later, then another a bit later still, and so on, until eventually there were sixteen. I am not suggesting that this necessarily happened all at once (although it could have), but that there would not have been more than two or three stages involved. There might first have been a marking of ‘human’ versus ‘non-human’, for instance, and then more distinctions within ‘non-human’.

When postpositions develop into a case system, the whole system is likely to evolve more or less at the same time, rather than one postposition becoming one case (an affix to a noun rather than a free form in syntactic construction with it), then a hundred years later a second postposition developing into a second case, and so on, so that a system of seven cases developed in seven separate stages.