There is an argument to made against the proposition that language exists, or at least exists in the sense of being something that can be studied. It says that individual languages can be studied, and abstract generalizations can be made, but those generalizations don’t have predictive power about any particular language. They cannot because each language has its own constelation of unpredictable properties..
This argument is set forth by Morten Christiansen in a paper he has posted on the internet. In it he treats language as purely a cultural phenomenon. “Language has been shaped by cultural evolution to fit domain-general constraints deriving from the human brain” (p. 2). Domain-general constraints include “socio-pragmatic considerations, the nature of our thought processes, perception-motor factors, as well as cognitive limitations on learning, memory and processing” (4). In other words, there is no faculty of language in the narrow sense of some brain module that contributes only to linguistic production.
Furthermore, he rejects Chomsky’s distinction between internalized language (I-language) and externalized language (E-language). Christiansen argues that what is internalized comes from listening and whatever neural wiring there is governing language “cannot be understood independently of the input that gave rise to [it]” (7).
Fifty plus years after Chomsky sank Skinner the argument is still with us—is language instinctive or is it learned? Christiansen’s argument is far more sophisticated than Skinner’s and this sophistication is due very much to Chomsky’s intervention which forced everybody to grant that language is much more complex than can be allowed by learning by conditioning based solely on stimuli and rewards. In Skinner’s day, the language learner was considered a “black box,” an entity whose internal workings were unknown and irrelevant to the learning process. Today’s learning-based approach gets rid of the black box and puts heavy emphasis on what the brain has to contribute generally to learning.
So how do children learn to talk if they are not pre-wired for it? Christiansen proposes two answers. First, the language is adapted to the brain. This idea has been around since at least the early 1990s and was given wide prominence by Terrence Deacon. It states that people learn what they can learn and forget what they cannot; thus, by its very nature, anything passed down from generation to generation will be learnable. This idea has been discussed many times on this blog and I have also reported a variety of experiments in which “generations” begin with difficult-to-learn languages and turn them into much more regular, readily learned systems.
A second aid to learning proposed by Christinasen is a set of cues that help a listener, including a learner, follow spoken language. I was surprised to see this line of argumentation as it means Christiansen has given up on the idea that adapting to the brain is sufficient to explain language’s learnability.
A similar, yet different, idea was introduced some years ago back when Chomsky’s forces recognized that languages are too different for everything about them to be built into the brain. Chomskyans proposed a series of parameters that could define any language, but that system failed when parameters proved insufficient as predictors of the idiosyncrasies in an entire language. Now Christiansen proposes that cues such as the number of phonemes in a word or an initial th phoneme can be informative to English speakers.
I’m not going to discuss Christiansen's experiments and calculations because this is an idea that requires extensive examination and testing. At this point it is sufficient to note that he argues “every language today incorporates its own unique constellation of probabilistic cues to signal different aspects of structure” (8), e.g., some cues signal an English learner that a word is a noun and some a verb. I do wonder how those distinctions mean anything to a three-year-old.
In this approach “natural language is a culturally evolved multiple-cue integration system” (12). That description puts the theory at the polar opposite of the Chomskyan approach that sees natural language as an inborn biological faculty. Both parties leave room for the other side to play a role. Thus, Christiansen says is account “does not preclude that there may have been biological adaptations for aspects of language that would improve learning and use” (21). Meanwhile, Chomsky and his fellows grant that the language faculty may include domain-general capacities, and of course they grant that each language has a cultural history of its own. But in both cases their focus is on one side or t’other of the biology/culture divide. They are not focused on a co-evolutionary hypothesis.
My favorite argument against cultural evolution alone is the adaptation to vocalization shown by the human body. The control of lips and tongue is quite unlike the less precise control available to apes. The loss of air sacs in the throa, common to all apes,t widened the range of sounds available to speakers. The FOXP2 gene supports delicate motor control of vocalizations. Work with songbirds has shown that birds have special modules for learning their local song, and their foxp2 contributes to this process. That work makes for a very strong hypothesis that humans learn the phonology of their local language through a combination of cultural input and specific biological adaptations that promote the learning. The biological and cultural components are so mixed that neither leads, neither trails.
The co-evolution approach seems like the only way we shall ever escape the learning vs instinct battle that rages interminably in linguistics. However, this approach is not neutral in its implications. Co-evolution requires a long time, much longer than the 200 thousand years of Homo sapiens. In a co-evolutionary scenario, some biological adaptations are adaptations to existing speech.
For example, we know from work with Washoe and other signing chimps that apes were smart enough to make basic phrases, but no ape has come close to signing a sentence formed by subject + predicate. Thus, there is no reason to assume that the first speakers could unite two distinct foci of attention by means of a verb.
The move from phrases to sentences is hard to justify In either a culture-alone or a biology-alone scenario. Cultural selection should reduce an innovation as intellectually unusual as a true sentence because too many people will be unable to reproduce it accurately. Biological evolution has a similar problem. Why would an accidental mutation permitting the creation of a true sentence? You are not allowed to give a cultural rreason for selection, some sort of group use or benefit. Biolog-alone, remember! Some kind of survival or reproductive success is required. You can try to argue that better thinking leads to better survival (or even better sex) but it is very hard to prove such a thing. Meanwhile a co-evolutionary scenario can posit a human lineage speaking phrases (like a two-year old) and then undergoes biological changes that permit constructionn of subject + predicate sentences while culture finds a way to make use of the biological change. In co-evolution one side can produce a change while the other side can provides a basis for selection.
Besides language, there are many traits that seem distinctively and universally human (suggestive of biological support) and yet are highly variable within the human world (suggestive of cultural shaping). Take your pick—music, dance, humor, morals, language, mythology, family relationships—other animals don’t have them but every culture does. I’m betting on co-evolution as the key to understanding what mamkes humans tick..