The Theory Behind Content-Based Instruction

This short paper has a nice concise statement of why content based instruction (CBI) is more effective than just teaching a language:

"One of the achievements of cognitive science is the confirmation of the dual nature of cognition given in the dictionary definition: all human intellectual activities, such as thinking, communicating, problem solving, and learning, require both processes and content (knowledge). This implies that attempting to raise people's cognitive abilities to high levels simply by improving processes such as "reading," "writing," "critical thinking" is nearly futile. To perform these processes well requires high levels of content knowledge on which the processes can operate."

Computer programs that understand natural language need databases of commonsense content knowledge like Cyc to function.

Humans like computers need these databases and they acquire them by osmosis through extensive reading and automatic word recognition:

"To efficiently read and comprehend, the decoding aspect of reading must become automatic, that is, performed without conscious attention. This can only be accomplished by hours and hours of practice in reading. This is one of the reasons why adults who leave literacy programs having completed just 50 to 100 or so hours of instruction do not make much improvement in general reading comprehension: they have not automated the decoding process. A second reason is that, to markedly improve reading comprehension, one must develop a large body of knowledge in long term memory relevant to what is being read. Like skills, the development of large bodies of knowledge takes a long time."

The article ends by suggesting that language training be combined with job training.