Restricted Access Thesis
The purpose of this study is to examine how a computer model learns a second language with a different language typology than its native language. Testing two artificial SOV languages and two artificial SVO languages using back propagation connectionism, acquisition is rated based upon correct answers to randomized exposure of a second language. This is explained by using simplified neural network terms so that computational SLA research can be introduced to K-12 practitioners and other interested parties. This study discovered that language typologies are easily learned. More specific analysis of the network reveals that plural articles might be more difficult to learn than singular articles. This study establishes a precedent for linguistic typology studies and calls for more computational work to be completed in SLA research.
Mumford, Brian W, "Computational acquisition of a second language: a test of back propagation connectionism" (2012). School of Education Student Capstones and Dissertations. 496.