A study of grammar transfer in a second order recurrent network

It has been known that people, after being exposed to sentences generated by an artificial grammar, acquire implicit grammatical knowledge and are able to transfer the knowledge to inputs that are generated by a similar grammar. In the current paper, the ability of a second order recurrent neural network to transfer grammatical knowledge from one language (generated by a Finite State Machine) to another language is investigated, where the latter language differs in the syntax from the former language but uses the same vocabulary. We sought the measure of syntactic differences that affects the amount of transfer. The result shows that the effect is sensitive to the frequency of subsequences of words in the both languages.