Statistical Identification of English Loanwords in Korean Using Automatically Generated Training Data