Evolang, the major biannual conference about the origins of language, has gotten underway in Utrecht, the Netherlands. Most of the day has been given to registration, and handing name tags and orientation material. This evening the first plenary session talk has just ended and people are making their way to the buffet. The opening presentation was given by Maggie Tallerman from Newcastle University. She defended the concept of a "language faculty." Although a conference like this one is sure to have surprises, the biggest contest is expected to be about whether the brain includes a module dedicated to processing language.
Tallerman offered her presentation as a rebuttal to a recent paper by Nicholas Evans and Stephen Levinson reported on this blog last December. (See: Is Anything Universal in Language?) But in the end, her defense seemed less energetic than expected and seemed to close on a surprisingly defensive note.
The old Chomskyan position was that there is a universal grammar common to all languages, it is built into our brains, and that it is specifically concerned with language. The Evans and Levinson paper argued that any commonalities of language arise from general cognitive powers and there are enormous differences between languages.
Tallerman dismissed these differences as "superficial" and overstated. She said there is evidence for a language faculty, specifically linguistic cognitive faculties. She presented the interesting case of an autistic person with an unusually strong ability to learn new languages. "Christopher," as he is called in case studies, did well with natural languages, but could not learn invented languages that have unknown or impossible features. For example, an invented language called Epun uses subject-verb order to make an affirmative statement (e.g.: I ran; She flew; They bought). Negative statements have no special verbal markers, like not. To form a negative statement, use verb-subject word order. In other words, "Ran I" means I did not run; "Flew she," means she did not fly. Epun had other rules as well that changed word order when we would expect some other kind of marker. Christopher was unable to Epunthem; however, normal people were able to master it.
Tallerman explains this data as showing that Christopher has a healthy language faculty which enables him to learn natural languages, but, because of his autism, has impaired normal cognitive powers. The normal people were said to master Epun because of their ability to switch from the language faculty to more general cognitive powers.
A more technical argument concerned "constituency." In formal syntax, words or other elements that form a coherent phrase are put within brackets. The elements between the start and close of a bracket are called constituents and can be replaced by some other element. For example, the phrase "All Gaul" can be replaced with the single word, it.
Suppose, however, somebody said, "Gaul is all divided...." The presence of the verb obstructs the effort to remove a constituent part. You cannot replace "Gaul is all" with it. Evans and Levinson said that there are some languages without constituency. They mentioned Latin as an example. Their point is that interpreting a language without any fixed word-order requires a different cognitive system from one that interprets English. If we have different interpretive systems, interpretation must be learned. There cannot be one innate faculty for both interpretive systems.
Tallerman gave an extended examination of a language without constituency, an Australian language, Wambaya. She concluded that despite appearances there are fixed rules of order after all and that there is little justification for the claim that different interpretive systems exist for different languages.
She also argued, almost as an aside, was that if there is no language faculty specific to our species, it is hard to explain why only human primates can learn a natural language grammar.
Many at the conference disagreed and will have plenty of opportunities to present the other side, but I was most surprised by the defensiveness and caution of the presentation. It had none of the old cockey swagger.
She conceded that there is far more linguistic diversity than the proponents of a universal grammar once suspected. Oops. And how honorable of her to mention it.
She also did her integrity proud by saying that finding a concrete universal is anything far from a trivial challenge. Universals tend to exist at a very abstract level.
Most surprising was her conclusion, which seemed to look for a compromise with Evans and Levinson. She called it a rapprochement. She says that Chomsky has some recent ideas striking similar to the people she has been rebutting. He now suggests that the development of language depends on genes, experience, and "principles not specific to the faculty of language." He has also said, "We need no longer assume that the means of generating structure expressions are highly articulated and specific to language."
When the previous Evolang conference ended two years ago in Barcelona this blog carried a summary saying that Chomsky's generative grammar was no longer the dominant approach (see: Paradigm Lost). Tallerman's lead-off lecture (made possible by a sudden change in speaking arrangements) gave a loyal Chomskyite the opportunity to come out swinging and prove me wrong. Instead she made a series of rebuttals of various strengths but none decisive. Mixed in were a series of tactical retreats and concessions. And this was just the first day.