The study of errors allows researchers insight into the production of speech. Speech errors have been shown to accommodate in form to their erroneous environment, demonstrating that errors occur before the processing of the phonological rule component. That this configuration is a complete picture of the processing involved, however, has been called into question by the prevalence of non-accommodated errors that have been detected via instrumental analysis. This paper presents a model of speech production developed using Python ACT-R that uses a noisy recall system and explicit encoding of phonological rules. This system produces both accommodated and mismatch speech errors at the same rates as observed in the empirical study.