Top-down processing can be of benefit in comprehending speech. Linguistic knowledge of syntax, semantics and phonology helps us to understand speech, yet our process of understanding does not always occur in an orderly structure beginning with decoding and working bottom-up through to word interpretation. Our knowledge and expectations of the world mean that it is quite possible to understand the meaning of a word prior to actually decoding the sounds. This is illustrated in the phonemic restoration effect (Warren & Warren, 1970; Samuel, 1997, as cited in Eysenck & Keane 2010), an illusory effect in which a missing or ambiguous phoneme is perceived as normal by the listener. This illustrates a benefit of top-down processing in speech comprehension when noisy environments can impair perception. Additionally, listeners will frequently have hypotheses about what is coming next, for example, the sentence 'she was so angry, she picked up the gun, aimed and ____' (adapted from Grosjean, 1980 as cited in Buck 2001). Background knowledge of what occurs when anger and guns are present means we do not need to pay much attention to the following acoustic information. On expecting the word 'fired' or 'shot' we would stop paying attention after hearing the first letter. Top-down processing is also advantageous in everyday situations such as greeting and saying goodbye to friends and colleagues. Knowledge of what happens coupled with visual information such as body language and gesture determine what is being said rather than decoding of the acoustic information, which may not even be processed (Buck, 2001).
Buck, G. (2001). Assessing Listening Cambridge University Press pp. 1-10.
Eysenck, M.W. & Kean M.T. (2010). Cognitive Psychology A Student's Handbook Psychology Press.