Language learning as language use: A cross-linguistic model of child language development.
Last modified on May 16, 2022
Notes
In this model, comprehension and production are two sides of the same coin.
Comprehension task
The model “chunks” the input utterance, incrementing the count of that chunk in the chunkatory.
chunkatory = dict() # for each sentence... # TODO: write the pseudocode for chunkify chunks = chunkify(sentence) # uses running avg. of BTPs for chunk in chunks: chunkatory[chunk] += 1
Production task
Given a “bag of words” derived from a child utterance, model has to figure out how to put it in the right order.
Of the longest chunks, select that with the highest BTP given the previous item.
# first phase: chunkify the input bag_of_words_chunks = # ... # second phase: pull out the chunks output = '' prev_item = '<start>' bag_of_words = # { set, of, all, input, words } while bag_of_words is not empty: output += get_highest_BTP_chunk(bag_of_words_chunks)
Link
Bibliography
[1]
S. M. McCauley and M. H. Christiansen, “Language learning as language use: A cross-linguistic model of child language development,” Psychological review, vol. 126, no. 1, pp. 1–51, 2019, doi: 10.1037/rev0000126.