Commentary: Beyond tone and climate: broadening the framework

In this commentary, we address some general questions regarding the use of statistical methods for the purposes of correlating linguistic with nonlinguistic features, as well as engage some specific claims in the position article by Everett et al. (2016). In the process, we attempt to arrive at possible causes for misunderstandings among researchers coming from different disciplines. This short piece is not meant to be a thorough review of the article, but simply to focus on those areas where we can provide some productive feedback. First, we do believe that it is an empirical question whether or not certain linguistic features are influenced by external forces, and one should indeed pursue such research questions using statistical methods. As pointed out by Ladd et al. (2015), correlational studies of this kind can help one choose among competing hypotheses. While it is true that finding a correlation does not necessarily establish causation, it is also true that having discovered a correlation is better than having none. In addition, understanding how human language evolved will require such approaches. Perhaps part of the reason why some linguists seem skeptical about the findings of this kind has to do with what we can call a problem of proportion. Isolated studies like this, without being embedded in a comprehensive framework, may suggest to the reader, even when this is not claimed by the author(s), that a linguistic phenomenon can be completely reduced to a nonlinguistic factor. What would help would be to embed these findings into a larger, more comprehensive framework about change in phonological systems, as well as how these systems evolve. Against the background of such a comprehensive framework, one can then try to determine how much of a role is played by climate as opposed to other factors, such as language-internal …