Understanding language is a funny thing. We start listening before we have even heard anything, with a preset expectation depending on our surroundings; then we hear a stream of sounds which we immediately connect with a set of words in our heads. We know they are significant, especially if they are uttered in the cadences of our own language, or a familiar one. How does that happen? And reading is equally strange – how do we make sense of so many peculiar signs?
Psychologists think that listening to speech is a much older use of language than reading words; evolutionary psychologists suggest that man was making complex noises well before he invented written language, with the corresponding necessity to read.
How do we divide up what we hear into the right words? If you have ever learned another language, you will know that a difficult task for the novice is to learn to tell when one word ends and another begins. It seems to be a constant stream when spoken quickly, without breaks which could give clues as to the word boundaries. Some cues, such as rhythm, are ‘pre-lexical’, meaning that they don’t depend on vocabulary knowledge and recognition. In English, for instance, we often stress the first syllable of each important word, giving the listener a strong clue that a new word has started. Different languages have different rhythmic habits, which give different clues as to word boundaries. Even babies have the ability to pick up patterns in their own native language.
Where syllable stress doesn’t work, however, we may have to rely on ‘lexical knowledge’ – a kind of dictionary in our heads which picks up parts of words and tried to make them into complete ones. Some models of language comprehension suggest that when we hear a part-word, we ‘activate’ in our brains a whole set of words which might fit – for instance, when we hear ‘wh-’, we may instantly trigger ‘what’, ‘why’, ‘where’, and other possible words in our heads. Then, when we hear more of the word, we may get to a point when there is only one possible interpretation. The important thing in these models, is that we are constantly coming up with theories and possibilities, all the way until the point where a word is certain. Some psychologists have come up with computer models to simulate how we activate some words, and inhibit others, in an interactive way (e.g. the TRACE model of McLelland and Elman, 1986).
When we read, it seems that we focus on the early part of a sequence – we often fixate at a point around 4 letters into our usual 18-letter reading span. Our vision is only detailed in the middle of our sight (the ‘fovea’), so we prioritize, and choose the earlier information, as it is likely to be more important.
When we read words, we can do it in two ways: all-at-once, or by dividing them up into sections. For example, some children learn to read in whole-word chunks, just by familiarity; and some children learn in word-sections, by trying to build up a rule-based understanding of words. Both ways are equally valid. Coltheart and others incorporated both routes in their ‘dual route’ theory of how we read. It explains why some words are very slow to process – irregular words, for instance, have no rules to hang on to; if, furthermore, a word is rare as well as irregular, then there will be no quick way of either recognizing it or working it out, and we will be slow to respond. Other factors affect how quickly we can respond to words – for instance, we respond faster to words which have a lot of similarly-rhyming words in the same language!
You will notice that words split up into recognizable bits, called ‘morphemes’. These are the basic sounds of a language. Theories vary as to whether we just recognize whole words, or decompose what we see. In 1958, Berko found that children were quite good at adding bits to new words (e.g. ‘s’ for plural), as if they understand rules and can apply them to ‘bits’ of words – this suggests that we all use some ‘decompositional’ functions, not just recognition.
But, in addition to recognition and rules, also influential is whether we expect a particular meaning. There is evidence that we can be ‘prepared’ to interpret words in particular ways, and do not just read with an open mind. Underlying conceptual thinking seems to be always influencing how we interpret what we read. Even so, in the early stages of seeing a word, it may be that ALL possible meanings get triggered initially (autonomous processing). Although there is evidence for this, some psychologists have also found evidence that the context of a sentence, and any dominant meaning, can influence very early word perception.
Finally, we read words in sentences, which are amazingly creative things, full of variety. Here we cannot use recognition all the time, because most sentences are unique. We must use our deeper brain to construct meanings and interpretations. A sentence begins as an unknown, until we begin to be able to assign words to particular roles (i.e. syntax). Some psychologists have assumed that we do this in a carefully sequential way, putting a sentence together in a single ‘best fit so far’ interpretation. But some (e.g. MacDonald et al., 1994) take a different view, and feel that we constantly play with many different possible meanings, treating a sentence a bit like a jigsaw.
What seems clear in all cases, is that, even where there are established rules and understandings, extra clues such as rhythm, pitch and timing make a difference. A word’s frequency of use in a particular context makes a difference. And even the non-verbal environment – the contextual setting outside the words – has been shown to have an effect.
So next time you speak or listen, read or write – just remember how much processing is going into the interpretation of what you produce. With all that energetic computation and theorizing going on, it’s a truly wonderful thing that we understand anyone’s thoughts at all!