There is no shortage of special issues on language evolution in the current landscape of academic journals. However, probably none of the three upcoming special issues I know of (or the many more I don’t know of) will match Tecumseh Fitch’s special issue on “Empirical approaches in the study of Language Evolution” in “Psychonomic Bulletin and Review”, at least in terms of sheer size – by my count, the issue contains no less than 36 contributions by 39 mostly very well-known researchers.
The volume starts out with an impressive overview – which also serves as a review paper on recent advances in language evolution research – by Fitch himself. Like some of the other contributions, it is freely available with open access. As all contributions are available as “online first” papers at the moment and have not been assigned to an issue of the journal yet, the references section of the overview is also a good starting point for retrieving the other papers in the special issue.
Some of the papers are response articles to other contributions in the volume, which nicely highlights some key debates and open questions in the field. For example, both David Adger and Dan Bowling react to Simon Kirby’s paper on “Culture and biology in the emergence of linguistic structure”. Reviewing a large number of (both computational and behavioral) experiments using the Iterated Learning paradigm, including recent work on Bayesian Iterated Learning, Kirby argues that linguistic structure emerges as sets of behaviors (utterances) are transmitted through an informational bottleneck (the limited data available to the language learner) and the behaviors adapt to better pass through the bottleneck. According to Kirby, “[a]n overarching universal arising from this cultural process is that compressible sets of behaviours pass through the bottleneck more easily. If behaviours also need to be expressive then rich systematic structure appears to be the inevitable result.” Adger, however, argues that expressivity and compressibility are not sufficient to explain the emergence of structure. He points out that the systematicity of human languages is restricted in particular ways and that in the case of some grammatical phenomena, the simplest and most expressive option is logically possible but unattested in the world’s languages. He therefore argues that the human language capacity imposes strong constraints on language development, while the structures of particular languages arise in the way envisaged by the Iterated Learning model.
Kirby also discusses the relation between biological and cultural factors in language evolution. Probably the most far-reaching conclusion he draws from Iterated Learning models (in particular, from work by Bill Thompson et al.) is that the language faculty can only contain weak domain-specific constraints, while any hard constraints on the acquisition of language will almost certainly be domain-general. Bowling’s response is targeted at this aspect of Kirby’s theory. While being sympathetic with the emphasis on cultural evolution, he argues that it “fails to leave the nature-nurture dichotomy behind”, as constraints are identified as either cultural or biological. Unfortunately, Bowling doesn’t really have enough space to unfold this argument in more detail in this very short response paper.
A second paper in the special issue that is accompanied by a short commentary is Mark Johnson‘s “Marr’s levels and the minimalist program” (preprint). He discusses the question “what kind of simplicity is likely to be most related to the plausibility of an evolutionary event introducing a change to a cognitive system?” Obviously, this question bears important implications for Chomsky’s minimalist theory of language evolution, according to which a single mutation gave rise to the operation Merge, “a simple formal operation that yields the kinds of hierarchical structures found in human languages”. Johnson points out that just because a cognitive system is easy to describe does not necessarily mean that it is evolutionarily plausible. In order to approach the question “What kind of simplicity?”, he takes up David Marr’s levels of analysis of cognitive systems: the implementational level (the “hardware”), the algorithmic level (the representations and data structures involved), and the computational level (the goal(s) of the system; the information it manipulates; the constraints it must satisfy). He suggests that complexity of genomic encoding might be most closely related to complexity at the implementational level. The introduction of Merge, however, is complex at the computational level, while the changes on the other two levels could be quite complex. To strengthen the minimalist account of language evolution, then, one would have to either show systematic connections between the three levels, or demonstrate that a simple change to neural architecture can give rise to human language.
In her response paper, Amy Perfors (preprint) basically seconds Johnson’s position. However, she also points out that, from the perspective of Occam’s razor, computational simplicity might nevertheless be an important factor in model selection: “Because the more computationally complex a model or a theory is, the more difficult it is, plausibly, to represent or learn. For those reasons the simplicity of Merge is a theoretical asset when evaluating its cognitive plausibility.”
Kirby’s and Johnson’s papers and the respective responses can of course only give a glimpse of the thematic breadth of the special issue and the diversity of theoretical frameworks represented in the volume. Other topics include, e.g., the architecture of the “language-ready brain”, advances and missed opportunities in comparative research, and the role of different modalities in the evolution of language.