Current understanding of speech perception suggests that the routines that listeners use to segment the speech signal for word recognition are language specific. English speakers exploit stress patterns for lexical segmentation whereas Spanish speakers use syllable structure. This study investigated L2 lexical segmentation and specifically examined whether L2 learners with a syllable-based L1 (Spanish) exhibit stress-based segmentation in L2 English. 81 English-Spanish learners and 72 Spanish-English learners completed a word-spotting task that manipulated lexical stress and syllable structure. Participants aurally identified English monosyllabic words embedded in disyllabic pseudowords. Results showed that Spanish-English learners transferred L1 segmentation routines to L2 English. Spanish-English learners exhibited syllable-based segmentation of English regardless of differences in lexical stress, which was more prominent as L2 proficiency increased. English-Spanish learners showed stress-based segmentation regardless of differences in syllable structure. This study suggests that listeners develop only the segmentation routines that correspond to the phonological composition of their L1.