Diffusion models are recognized for their ability to consider both response time and accuracy as variants in the process. They provide predictions for the relationships between response time and the probability of correct versus error responses, as well as the shape of the response time distribution. For this project, response time data were used from a previously completed study of word recognition in bilinguals. The response time data (in msec) were collected from 94 participants given a lexical decision task. Participants were Italian native speakers who had acquired English as a second language later in life. Stimuli chosen were nonwords and words for both English and Italian. In both languages, stimuli were classified as having either a high frequency of occurrence within the language or a low frequency of occurrence. Stimuli were also classified as having either a large neighborhood size or a small neighborhood size. Amongst the words and nonwords from either language, the response times were separated based on whether or not the presented stimulus had a high or low frequency. Then for each category, correct responses were separated into five quantiles. Means response times across quantiles were found, and covariances were calculated. The response times from the human participants and the information derived from the quantiles will be used to determine parameters for the diffusion model that produce simulated response times closely matched to the data collected from the lexical decision task.