In both children and the human lineage, word learning predates rule making, but rule making eventually evolved as a way to relieve the stress on memorization, reported (or at least I think they reported) two members of the University of York, Dimitar Kazakov and George Tsoulas, to the “Ways to Protolanguage” conference in Torun, Poland.
Their presentation was titled “Applying recapitulation theory to language” and addressed the problem of whether the acquisition of language in children today offers any guide to how language evolved in the past. In other words, does the fact that modern children speak a kind of protolanguage before they speak syntactic language provide strong evidence that our ancestors used a protolanguage for many generations before they spoke a full, syntactic language?
The recapitulation theory is older than the theory of natural selection, and was put in its essential form as, “Ontogeny recapitulates phylogeny.” For those who don’t speak jargon: the development of an individual retraces the evolution of the species. The classic example is the development of a gill-like form into the human’s middle ear. Development takes that round-about route because it is as blind as evolution itself. Long ago our ancestors used gills. Over the eons the gills changed and eventually became part of the ear. But at no point did the DNA become aware of what it was doing and introduce a more direct route to the modern-day goal.
So that developmental process sounds fixed, although the process does lead to terminal addition (adding things to the end of the process that were not found in ancestral species) and condensation (acceleration of the development of the ancestral features). But genetics also allows for retardation of development. Humans, for example, develop much more slowly than chimpanzees. Our heads grow much larger and the skull seals up much later than the head of an ape. Thus, head development most definitely does not recapitulate the story of our ancestors. And thus, the famous proverb goes out the window as a handy rule of thumb. As Sportin’ Life says, “It ain’t necessarily so.”
But that observation needn’t always leave us scratching our heads. For example, language has certain computational constraints. Speakers have limited short-term memory, and very young speakers have even more limited short-term memory. The adult solution for the memory problem is to learn rules: thus work, worked, and working can be created by following rules and we don’t have to memorize each form individually. But for small children learning rules puts more stress on memory capacity. So words like go and went can be learned individually. Later a child begins using words and says go, goed making an error not made before. Ultimately the go-went pairing is learned as an irregular conjugation. Since our ancestral brains are known from fossils to have enlarged considerably, there seems good reason to believe the same stresses seen in children today were seen in ancestors. In this case then the development of language in a child does indicate the development of language in the lineage.