Jemma, an 11-year-old who until last August had severe difficulties with reading, is now enjoying William Horwood’s The Willows in Winter. She is sailing through page 92, on which Toad and Rat are up in a flying machine, when she comes to the phrase “Toad’s wild cries”. She reads it as “Toad’s wild creatures”, partly from the long-term context of the book, partly from the immediate context of wild, and partly from the first two letters, which are consistent with the mistaken guess.

Tina, who is nearly eight and still in the very early stages of reading, makes this error in her first lesson on The Cat in the Hat: Text: The sun did not shine.

Tina: The sun did not rise.

This guess ignores completely the information conveyed by letters. It makes sense, but not the sense the author intended - in the book it is raining, but not dark. Moreover, Tina could not reasonably be expected to deduce from the picture of a rainy day the less direct message that the sun was not shining.

These examples show children using reading strategies from Professor Kenneth Goodman’s “psycholinguistic guessing game”, a theory first proposed in 1967, which holds that fluent readers and beginners alike piece together the meaning of a text on the basis of predictions of what it will say, which they check from a series of “cues” from the text.

While the flaws in this model of fluent reading are now widely recognised - Dr Colin Harrison of Nottingham University has recently argued that it lacks detail, that good readers recognise words automatically rather than relying on context, and that they fix their eyes on nearly every word as they read - parts of it are so close to the ways in which teachers see children thinking as they learn to read that it remains plausible. We need, therefore, to begin to clarify the role of guessing in learning to read, and to identify its place in the development of fluent reading.

The guessing game basically involves intelligent improvisation in which children use any means at their disposal to try to make sense of a word or series of words they do not immediately recognise, including pictures and context as well as their knowledge of the alphabetic system. Recording and analysing their errors as they do this can, as Goodman has demonstrated, provide an insight into their thinking, but he uses his “miscue analysis” chiefly to identify children’s use of the cueing system that is at the heart of his model of fluent reading.

If fluent readers do not operate as he suggests, the technique needs to be modified to show how far a pupil has developed the strategies which successful readers use, and how far less effective strategies have been discarded. Once this happens, some of the features of intelligent behaviour which are noted for praise in miscue analysis become indications of reading problems. It is not that children thinking in these ways are stupid, but that they need to adjust the way they use their intelligence in order to read successfully.

As there is usually no sure way of telling which part of a picture is supposed to provide the relevant cue, pictures in an alphabetic system are a haphazard source of information on individual words.

Context is a more abstract source of information, establishing and maintaining continuity of meaning and enabling the fluent reader to distinguish the meaning of words which share the same spelling (for example “a lead weight” and “lead on”). Once again, however, it is unreliable as a cue to an unknown word. Some errors based on context, such as a misreading of “convinced” as “confident”, may make no difference to meaning, but others do, and simply checking for meaning does not always enable the learner to know whether he or she is mistaken or not.

Rachel, an older beginner who has cerebral palsy, read the newspaper headline “Harrods Boss in Sleaze Storm” as “Harrods Boss in Sale”. Her prediction is confirmed by the first letter of the misread word, and remains plausible in the later context, as turmoil is often reported during sales. However, language is not always predictable, and much of the art of fluent reading is in following the course of the text as the author changes tack, sometimes deliberately avoiding the obvious. Using all, and not only part, of the information contained in the letters is central to this process, and encouraging prediction coupled with light sampling of letters makes this more difficult.

These examples show that beginning readers, particularly those who have difficulty, cannot be relied on to work out unknown words for themselves, even if they do use the additional information available to them. We also know, from Professor Katharine Perera’s research, that fluency in more able readers develops in conjunction with fast, accurate word-recognition. From the saga of the guessing game we have learned that the intelligent but immature behaviour of beginning readers does not always lead to progress. If we are to make an impact on reading difficulties, we must make the alphabetic system and its relation to spoken language as clear and accessible as possible, and continue to develop systematic teaching which ensures that children of all abilities steadily expand their capacity to recognise words at sight.