Category Archives: parallel processing

Language Acquisition

Apparently these things start to learn words at an accelerated rate around the two year mark. That is to say that they seem to abruptly begin to amass and employ a wide variety of words. This phenomenon is referred to as “vocabulary explosion.”

Convincing generalized theories of language acquisition have been around for around 30 years, the most successful of which is probably Noam Chomsky’s theory of Universal Grammar. But such theories do not attempt to describe the dynamics of communication mastery introduced above. That is one of the tasks of the ingenious minds at work on the subject today.

A recent submission to Science Magazine by the Psychologist Bob McMurray (ref. 1) attempts to computationally model the observed acceleration of the word uptake process. Although past explanations of this phenomenon have invoked specialized and well timed brain mechanisms, Dr. McMurray’s work attempts a more parsimonious description.

He concludes:

“Acceleration is guaranteed in any system in which (i) words are acquired in parallel, that is, the system builds representations for multiple words simultaneously, and (ii) the difficulty of learning words is distributed such that there are few words that can be acquired quickly and a greater number that take longer. This distribution of difficulty derives from many factors, including frequency, phonology, syntax, the child’s capabilities, and the contexts where words appear.”

He goes on to demonstrate that languages seem to display such a distribution of word difficulty, and to show that his model captures the accelerating behavior well.

The real beauty of the work however, is the posited inherent parallelism. Such ability in the human brain has long been suggested by a wide variety of scientists and philosophers. Indeed, I scarcely need use the word suggested, as we know that certain things happen in parallel, the processing of visual information, for instance, does not happen one pixel at a time but rather proceeds by working on the entire pattern of light that falls on the retina at once.

Dr. McMurray has thus figured out an elegant way to apply what should be thought of as a basic property of the brain to explain what seemed an exceedingly difficult problem, something everybody who works on complex systems hopes to be able to do.

1. McMurray B., (2007) Defusing the childhood vocabulary explosion.
Science 317 (5838):631.