UG and its contribution to methodology
Academically I have always been a generative linguist trained in the strict sense of formal linguistics and Universal Grammar. To some extent, it is not easy to look beyond your academic belief system. But once in a while a heated philosophical debate between the Unversalists and the Cognitivists would still catch my interest. I remember four or five years ago when I was still doing my dissertation in graduate school, one of my professors circulated a paper written by Andrew Nevins, David Pesetsky and Cilene Rodrigues to counter the arguments from Daniel Everett on the lack of recursion in the Pirahã language. It was quite an eye-opener. Before that I had never questioned the fundamental beliefs in the generative linguistic field. Thus I digged deeper into the debate and found that Michael Tomasello had written a paper saying that language is not an instinct at all and there is nothing to the program of Universal Grammar. Then I read the paper written by Hauser, Chomsky and Fitch on the core principle of UG, i.e. recursion. These papers really inspired me to look at the foundational issues in linguistics and particularly in generative linguistics. Although the cognitivist view on language is very interesting and probably would correct some biases in the formal field towards pure description and functional cognitive theories of language, such a view in general still seems to me to be lacking in methodological rigor and explanatory power. So I just continued my research as usual while keeping an eye on the cognitive side of story.
Last night, I was reading Kai von Fintel’s blog and came upon Lisa Matthewson’s talk slides on the scientific methodology on linguistic research as a counterargument against a new wave of attacks on generative linguistics from the cognitivist camp written by Evans and Levinson calling UG a myth. I guess the philosophical debate between the generative linguists and the functional and cognitivist researchers have never stopped ever since Chomsky defeated Skinner’s views in 1967. Once in a while a major paper would appear in the cognitivist field refuting the generative theory of UG and a heated debate would follow. This time again, the debate seems to be really heated, as can be seen from the responses that follow Evans and Levinson’s paper. Interestingly I have been reading Michael Tomasello’s book on the origins of human communication. Indeed, again such debates are very interesting and very illuminating in the sense that the fundamental issues and the ultimate research goals of generative linguistics and research in language(s) and human communication can be brought to the front and this will lead us to consider contributions from both sides.
All in all, I do think that in terms of the origin of human communication, Tomasello’s theory does sound very appealing. I am not buying into the whole “genetic mutation” view on the origin of language at all. There is indeed quite some truth in what the cognitivist views can teach us. But I do think to some extent the attack on UG as a research program is wrong. In addition to the responses that a lot of people have posted everywhere, I want to add two more points.
First, if there is no UG, then why are the majority of human languages so similar evan just on a superficial surface level of description? If indeed human communication arises from the need to cooperate, and is part of evolution, then why are most languages so similar, and can thus be analyzed with tools derived from the so-called “ethnocentric” “Euro-centric” linguists? Of course whatever can be used in language for communication should be constrained by the general physical and cognitive properties of humans. But if the general cognitive properties of humans constrain the design of language in such a way that most, if not all, human languages are built from a very similar draft, then these similar design principles could be reinforced genetically through evolution and passed down from one generation to the next. Thus to some extent we might still have some kind of UG. Even if we cannot prove the existence of genetic codes of UG, and even if the whole picture of UG might not be the same as it was described in the beginning, it can still function as a guideline in the research of language. I do not think there is anything wrong in assuming UG and trying to find commonalities among different languages at a deeper level. As Lisa Matthewson has already shown in her talk, pure description without a certain level of theoretical sophistication and assumption would in many cases lead to shallow inaccurate generalization.
Second, the scientific methodology as described in Lisa Matthewson’s talk slides is what makes linguistics a science. The advance that linguistics has made and also the contribution of modern generative linguistics to human knowledge are still greater than any negative effect that it might have, if any at all. In terms of methodology, the early structuralist linguists provided good formal tools. Then generative linguists added more robust formal tools and the fundamental shift to considering language as a computational system. Such methodological advances help many researcher to conduct their research with more rigor and logical precision. Although I agree that sometimes abstractionist theoretical inquiries might turn away from the facts of human language, the general soundness of this scientific methodology should not be denied completely.
In conclusion, such philosophical and methodological debates on the fundamental issues of research in human language are good for generative linguistics, in that many generative linguists might turn more of their attention to pure linguistic facts, rather than positing increasingly abstract theories. I have learned a great deal from reading papers on the cognitivist side, and they do remind me of the goal and methodology of linguistic research.