Evolution is typically seen as change over time, involving all the individuals in a species, and involving long stretches of time.
For natural selection to work, attributes and behaviour across individuals in a species must be somewhat distributed, so that there is an inheritable differention across individuals, which affects the probabilities of reproduction and survival, and therefore provides a chance to optimize the opportunities for the species within the environment.
Research on evolution is normally based on a time sequence based on fossil records. Many hypotheses on evolution are verified through an analysis of such historical records.
There are also hypotheses based on analyses of intraspecies and interspecies comparisons. Much of this research is based on a comparison of presently existing species.
More contemporary research focuses on a search for similarities and differences in DNA structures.
We have experimented with plants for thousands of years, in selection and in producing hybrids. Historically there has been similar experimentation with animals, primarily for domestication. Currently there are relatively fewer experiments with animals, mostly with species where the intergenerational period is very short, and where such experimentation is ethically acceptable.
It is not clear how any of the above research methods can help us investigate the evolution of language. We need a new approach that can complement the methods above.