READING: DECODING
المؤلف:
John Field
المصدر:
Psycholinguistics
الجزء والصفحة:
P234
2025-10-01
327
READING: DECODING
The perceptual process of identifying the letters and words in a text. Decoding is regarded as a distinct process from lexical access, since we can process the form of an unfamiliar word (even attributing a possible pronunciation) without knowing its meaning.
The process of decoding is highly automatic in a skilled reader. This means that it makes few demands upon working memory, leaving capacity spare for higher-level processes such as constructing a meaning representation. It also means that our recognition of written words is so little subject to our control that we find it difficult to suppress the process in order to perform a task such as the Stroop test.
Decoding must be linked to a phonological representation of the words that are being read. It is this which enables us to map unfamiliar strings of letters on to words that we know in speech; to assign pronunciations to words that we have never heard (e.g. technical terms or Proper Nouns); and to read rhyming poetry, puns etc.
We need to distinguish two types of phonological operation:
Pre-lexical, in the form of letter-to-sound mappings which enable us to work out the pronunciation of unfamiliar words.
Post-lexical, in the form of the inner speech which readers sometimes report hearing in their heads. This seems to be a means of retaining word-level information in working memory. The advantage of storing information phonologically is that it does not interfere with the ongoing visual process of reading. However, it must be stored in an abbreviated form. Speaking is much slower than silent reading, so we cannot store words exactly as we say them.
A dual route model of decoding suggests that we process written words in two ways. A lexical route enables us to match whole words while a sub-lexical route permits us to identify words by means of grapheme-phoneme correspondence (GPC) rules which specify the relationships between spelling and sound. Though the sub-lexical route is slower, it permits us to attribute a pronunciation to unfamiliar words.
An alternative interactive account of decoding represents the reader as processing a text simultaneously at several different levels of representation: taking account of letter features (curves, vertical lines etc.), letters, letter sequences and whole words. Evidence feeds up from one level to another, and also feeds down. Thus, when the letters WORareidentified at letter level and the order 1W 2O 3R is recorded at word sequence level, words such as WORK, WORD, WORM are activated at lexical level. But there is also a ‘downwards’ flow of information, with the reader’s lexical knowledge confirming that there are indeed words in the lexicon beginning with WOR, so that the evidence from letter recognition is to be trusted. Lexical knowledge also tells us that the word is more likely to be WORD or WORK than the less frequent WORM.
A very different view, espoused by proponents of whole word teaching methods, holds that efficient decoding is heavily dependent upon rapid word shape recognition. Experimental evidence does not support this hypothesis. Both children and adults have little difficulty in identifying words presented in unfamiliar zigzag or vertical forms, and text presented in AlTeRnAtInG cAsE only delays word recognition by about 10 per cent. Furthermore, eye movement studies show that longer words cause longer fixations– which should not occur if words are being processed as wholes. These findings suggest that sub-word units doindeed play a part in decoding. When words are confused in reading, it may not be because they resemble each other in overall shape but simply because they contain letters which resemble each other.
Sometimes cited in support of the ‘whole word’ view is a finding that subjects identify target letters more quickly when they occur later in a word. One would expect the opposite if words were processed letter by letter from left to right. However, the interactive model neatly accounts for this effect by suggesting that lexical knowledge of the existence of a word becomes available as more and more of the word is seen, and assists the process of decoding.
A reader processes highly frequent letter strings such as-ATION more quickly than others; this may mean that there is a level of representation for unanalysed chunks. The gradual recognition of chunked sequences is an experience reported by many who have spent time in a country with a different alphabetic system from their own.
Technology allows us to track a reader’s eye movements across the page; this has provided valuable insights into the decoding process. It has also dispelled a number of myths. It appears that skilled readers do not, as once supposed, make longer eye sweeps (saccades) than less skilled. Nor do skilled readers skip important words. The majority of content words (about 80 per cent) are fixated by all readers, though readers seem to have the ability to distinguish and by-pass function words, of which they only fixate about 40 per cent. What distinguishes a skilled reader from an unskilled is the degree of back-tracking that occurs. In a reader with good decoding skills, only about 10 per cent of saccades are regressive, and these often occur because a word is unexpected in its context rather than because it has been inaccurately processed.
See also: Eye movements, Reading aloud, Reading: bottom-up vs top-down, Reading: skilled
Further reading: Oakhill and Beard (1999); Oakhill and Garnham (1988); Rayner and Pollatsek (1989)
الاكثر قراءة في Linguistics fields
اخر الاخبار
اخبار العتبة العباسية المقدسة
