Prior research [Ranbom and Connine (2007)] showed that alternative variant forms of a word are represented in the lexicon as a function of variant frequency. Two experiments investigated whether these multiple representations conspire or compete during spoken word recognition of nasal flap variants. In experiment 1, nasal flap productions of high (i.e., “counter”) and low (i.e., “enter”) variant frequency words, as well as a set of nonword controls (i.e., “penter”), were presented in a phoneme monitoring task (detection of a /t/ sound). The results showed “higher‐t” detection rates for words than for nonwords, and higher‐t detection for high than for low‐variant frequency words. Experiment 2 used truncated versions of the stimuli from experiment 1 and showed comparable t‐detection rates across all conditions. The results suggest that coactivated multiple representations of phonological variants conspire during spoken word recognition.