Often in linguistic texts or introductions to language, it is stated that we don’t know how languages change. Often cited is Sapir’s hypothesis of language drift. But why does the drift start? There is also the push-pull chain described by Aitchison.
To me, two factors loom largest, well before borrowing or substrate, and they are: running out of breath and the need for emphasis and intensity.
The former accounts for the well-known phenomenon of the erosion of endings. The tones of Chinese resulted from the loss of final consonants, the long declensional and conjugational endings of Proto-Indo-European gave us the paradigms found in Latin, which in turn gave way to the reduced paradigmatic endings of the Romance languages, esp in the substantives.
In everyday language we can hear the way speakers try to highlight, demarcate, elevate, diminish, intensify, or make more colorful their utterances. Aitchison, again, gives excellent examples of how a chink in the grammar is provided by such little nudges in the language that result in what was originally a “marked” or unusualy form becomes the norm through overuse; we become used to it and it looses its “punch”, and so it is then the norm and we go off in search of other ways to give our speech punch.
Borrowing can dig into the grammar of a language, sometimes direct borrowing as when English borrowed third person plural forms from Old Norse (they/them/their) or indirectly through a kind of substratum influence as when Celtic structures are suggested as the model for “do”-support in English questions and negatives.
John McWhorter has suggested in Language Interrupted that extensive intertwining of lanauges is possible but there is a particular vehicle for change which is intimately tied to the history of the speakers of the languge in quesiton. This is when large numbers of foreigners acquire the language but incompletely, usually by paring off the complexities that accrete to a language spoken over time by a closed speech community. For English, this happend when the Vikings invaded and settled in large numbers and acquired English. He shows how complex forms like noun plurals were smoothed or planed down to one: -s. He does this for Arabic, Malay, Mandarin Chinese, and Persian. He compares their sister languages – in the case of English, all the Germanic languages extant – a demmonstrates their retention of these features e.g. irregular (so-called) plurals in German. This often happens with languaes of empire when large numbers of people must acquire the language for administrative and social reasons.
Wm. Labov has conducted research and shown that key figures in a community seem to be a nucleus of change. So not everyone need be given to running out of breath at the ends of words or driven to innovate in order to emphasize and highlight, but if these nuclear people do, then the innovations may spread to them. There was a study of Martha’s Vineyard speakers and how social distinctions determined how people spoke. Social factors probably loom as large as the physical and psychological elements I posit as the main reasons.