The Learning and Usage of Second Language Speech Sounds: A Computational and Neural Approach