[jesuslop]

Some of the context of this can be found in the 'Poverty of stimulus' wikipedia entry. Chomsky argument was tried to be cast in theorem form in a paper of E. M. Gold to apparently settle the question (in favour of Chomsky), but despite that result the computational linguists have been busy making useful work in NLP with machine learning methods, and now the debate has shifted from armchair discussion to the question of the interest of the emerging new language technologies and the role of unsupervised learning. Less famous, Zellig Harris (Chomsky thesis advisor), is now cited in deep learning language modelling literature as a precursor of the distributional model pushed by Bengio that is now used in word2vec and what not nlp.