Understanding spoken language is an exceptional computational achievement o
f the human cognitive apparatus. Theories of how humans recognize spoken wo
rds fall into two categories: Some theories assume a fully bottom-up flow o
f information, in which successively more abstract representations are comp
uted. Other theories, in contrast, assert that activation of a more abstrac
t representation (e.g., a word) can affect the activation of smaller units
(e.g., phonemes or syllables). The two experimental conditions reported her
e demonstrate the top-down influence of word representations on the activat
ion of smaller perceptual units. The results show that perceptual processes
are not strictly bottom-up: Computations at logically lower levels of proc
essing are affected by computations at logically more abstract levels. Thes
e results constrain and inform theories of the architecture of human percep
tual processing of speech.