We study the linguistic differences between papers written by native and nonnative English speakers in our dataset of fifty physics papers. We obtained an accuracy of 84% for the classification algorithms. These differences can help improve the communication and understanding of scientific research.