n contrast, when listening to directions to a friend’s house, comprehension is achieved by dividing and decoding the sound signal bit by bit. The ability to separate the stream of speech into individual words becomes more important here, if we are to recognise, for example, the name of a street or an instruction to take a particular bus.
In reality, fluent listening normally depends on the use of both processes operating simultaneously. Think about talking to your friends (in your first language) in a noisy bar. It is likely that you ‘guess’ the content of large sections of the conversation, based on your knowledge of the topic and what has already been said. In this way, you rely more on top-down processing to make up for unreliability in the sound signal, which forms an obstacle to bottom-up processing. Similarly, second-language listeners often revert to their knowledge of the topic and situation when faced with unfamiliar vocabulary or structures, so using top-down processing to compensate for difficulties in bottom-up processing. On the other hand, if a listener is unable to understand anything of what she hears, she will not even be able to establish the topic of conversation, so top-down processing will also be very limited.