User:Mr. Stradivarius/Connectionism

Source: Wikipedia, the free encyclopedia.

Connectionist accounts of language acquisition are hypotheses and models of how humans learn language that are made within the framework of

emergent phenomenon; that is, that it can be explained by simple rules that work together to create a complex system of language when exposed to large amounts of natural language input. Connectionism is one of several constructivist
approaches to explaining language acquisition; along with the other constructivists, connectionists believe that language acquisition can be explained by general learning processes, and that there is no “special” process for language learning inside the brain.

Frequency

Connectionist models assume that language acquisition can be explained by the connections between synapses in the brain, and how they change over time. The basic assumption that connectionists make is that synapses that are fired at the same time will tend to be associated with one another. Thus the strongest connections are those which are fired together the most frequently. Furthermore, connectionists assume that this applies not just to single neurons, but to networks of neurons. These neural networks are hierarchical, meaning that connections are made at all different levels of language.

Chunking

Connectionists represent this through the concept of ‘’chunking’’. Chunking is the connection of different elements in short-term memory to make one ‘’chunk’’. This connection frees up mental resources to focus on other stimuli. The number of mental items a person can hold in short-term memory is limited to something between seven and nine. However, with chunking, the amount of information stored can be increased. According to Nick Ellis, this is the mechanism that makes the telephone number “0800-123777” easier to remember than the more random number “4957-632518”.[1] Chunking is also recursive; things which are chunked together may themselves be chunks, and so on.

Connectionists hold that native speakers of have learned to chunk a great deal of their language, from individual sounds and letters to entire phrases and clauses. As Nick Ellis puts it:

What’s next in a sentence beginning T…? Native English speakers know it is much more likely to be h or a vowel than it is z or other consonants, and that it could not be q. But they are never taught this. What is the first word in that sentence? We are likely to opt for the, or that, rather than thinks or theosophy. If The… begins the sentence, how does it continue? "With an adjective or noun," might be the reply. And, if the sentences (sic) starts with The cat…, then what? And then again, how should we complete The cat sat on the…?

— Nick Ellis, 2003[2]

Emergence of rule-like behavior

History

Reception

Notes

  1. ^ Ellis 2003, p. 76.
  2. ^ Ellis 2003, p. 75.

References

  • Doughty, Catherine; Long, Michael H., eds. (23 June 2003). The Handbook of Second Language Acquisition. Oxford: Blackwell. .