This paper examines the human mental lexicon and the associated word-recognition processes which together lie at the heart of our capacity to process spoken language as quickly and efficiently as we do. There is an emerging consensus that word recognition involves the activation of a set of lexical hypotheses and the selection of the target word from amongst this activated set. This view suggests itself quite naturally for speech given its continuous and sequential nature. More often than not listeners are in a position of having only partial sensory information about the target word — information that is insufficient to uniquely identify it. They are, nonetheless, continuously generating or activating lexical hypotheses on the basis of this partial information. Hence, it becomes an important research objective within this perspective to identify which words are activated during the recognition process — even if only temporarily or momentarily — and to determine how the appropriate word is ultimately selected and the inappropriate ones rejected.