The way in which we process (and therefore understand) written and spoken texts is often described as either top-down or bottom-up.
Top-down processing is the idea that we start with “higher-level” features – background knowledge, context, overall meaning - and proceed through a series of steps “down” to “lower-level” semantic, syntactical and phonological features.
In bottom-up processing, on the other hand, we start by recognising phonemes, combining these into syllables, syllables into words, words into clauses, and so on “up” to contextual and background information.
This contextual information at the top can come from knowledge about the world or the speaker/writer, from a mental image or expectation set up before or during listening or reading (often called a schema), or from predictions based on the probability of one word following another.
- How language is used in texts
|↑ ↓||↑ ↓|
|Knowledge of the situation and context||Context|
|↑ ↓||↑ ↓|
Watching the TV news in our first language, for example, we may make use primarily of top-down processing. Our previous experience of watching TV news gives us some knowledge and expectations from which to make predictions about the likely content, as well as the style of language that will likely be used by presenters and journalists. As a news item starts, we may recognise it as an ongoing story, and call upon our knowledge of the story’s context. From there, we progress to “lower level” features to understand the finer details of the story.
L2 learners may use use a combination of bottom-up and top-down processing to understand a text, for example compensating for a lack of vocabulary by employing top-down cues about context.