This study investigated the effectiveness of two methods of accent modification instruction. Thirty nonnative English speakers received an intonational-based instruction method for the pronunciation of American English. Half of those speakers received additional instruction and activities that were based on theories of cognitive processing of language. Three expert listeners evaluated pre- and post-instruction recordings of each speaker. Listeners were asked to rate the use of speech characteristics determined to be instrumental for intelligible speech on a 5-point Likert scale and three yes/no responses. As a whole, the intonational-based instruction method resulted in significant increases in the use of positive speech characteristics. A modification of that method did not result in significant differences in any of the speech characteristics. Findings of this study support the focus on suprasegmentals in pronunciation training of English as a second language and may lead to more diverse methods and designs for accent modification research.