Language has been evolved, via minute changes, similar to the erosion process in landscapes over millions of years. It might be tempting to think of the development of language as the acquisition of function, specifically the acquisition of linguistic function by a physical type whose earlier functions were exclusively non-linguistic. It might also be tempting to think that theses changes occurred somewhat abruptly, over a fairly short time scale. The view here presented is almost diametrically opposed to that. Oversimplifying the matter, we can say that early linguistic function was the residuum of worn-away non-linguistic function, and later linguistic function the residuum of earlier. The wearing away is the inevitable consequence of use. Moreover, we cannot overstress the point that linguistic function is an ever-changing process that extends over a much larger timescale than the one we are usually tempted to consider. Also, these changes that occur are minute, unnoticeable yet incremental. The cumulative effect is felt (or observed) only over this large time scale. We suspect that some of these observable consequences of change may occur comparatively quite suddenly, maybe within a few generations of linguistic activity.
Language responds incrementally and adaptively to its applicability and effectiveness.
The respect in which this is an oversimplification is this: As with many or most evolutionary developments, it could not be predicted in what manner what residuum of what earlier feature may confer advantage, but we can confidently predict that the manner of its exploitation will self-select for some subfeatures, not select for others, and select against still others. Vocables may begin to emerge as the awkward exploitation of some relatively inefficient non-linguistic residuum, but the awkwardness itself exerts evolutionary pressures: what is not needed or actually impedes we can expect to be refined away.
A corresponding account applies to functionalization within language, and accounts, we presume for the morphological alterations that accompanies, for example, the logicalization of lexical vocabulary.
We assume many facts about early linguisticity because we think that structural changes in language, across the history of its evolution, are motivated by forces that the idiom of physics has successfully described. If the dynamics of change in language can be described using physics, we assume that observations based on a present sample of language will scale to past samples since these forces, as far as we know, do not change over time.