Beyond Monosyllables: Word Length and Spoken Word Recognition

Most empirical work in spoken word recognition has focused on monosyllabic words, and most theories and models are limited to monosyllables either in scope or in practice (in that the theories' predictions for polysyllabic words have largely been unexplored). We add to the small number of studies to date explicitly comparing recognition of monoand polysyllabic words with a study examining time course predictions that follow logically or by simulation from three theoretical perspectives. We tracked subjects’ eyes as they followed spoken instructions to click on pictures in a display (i.e., we used the visual world paradigm). Consistent with a computational model, we found a late-emerging advantage for long words over short words matched on several lexical dimensions, though the effect was weaker than expected. The results suggest new constraints on theories of spoken word recognition.