Traditional machine translation systems rely on large parallel corpora and perform adequately for resource-rich languages such as English, Chinese, and Spanish, but struggle with low-resource languages.