The simple view of reading and evidence based practice Rhona Stainthorp Institute of Education, Reading University Morag Stuart Institute of Education, University of London
©Stainthorp & Stuart 2008
The Simple View of Reading
Language comprehension processes
Word recognition processes
GOOD
Good language Comprehension; poor word recognition
POOR
GOOD
POOR
Poor word recognition; Poor language comprehension
Good word recognition; good language comprehension
Language comprehension processes
Word recognition processes
Good word recognition; Poor language comprehension
©Stainthorp & Stuart 2008
Correcting a major misconception • “Between 1997 and 2006 the teaching of reading in the NLS Framework was underpinned by the searchlights model (DfES, 1998). • But by 2007 it had been replaced by the simple view of reading …. • Morag Stuart and Rhona Stainthorp were invited to develop this model as part of the Rose review of the teaching of reading.” Wyse & Jones, 2007 (pps 44-45)
©Stainthorp & Stuart 2008
Who did develop the simple view of reading? • The Rose Report was published in 2005. • Stuart & Stainthorp presented the simple view of reading in an annex to that report, and it was also referred to in the main body of the report. • Gough and Tunmer first proposed the simple view of reading in 1986, almost 20 years before Rose. • Stuart & Stainthorp clearly referenced this source in their presentation of the simple view of reading in the annex to the Rose Report. • We simply do not understand how any attentive reader of this report could form the impression that Stuart & Stainthorp developed this ‘model’. ©Stainthorp & Stuart 2008
Is the simple view of reading a ‘model’? • We made clear in the annex to the Rose Report that we consider the simple view of reading to be a useful conceptual framework. • When trying to understand something as complex and multifaceted as reading, it is helpful first to simplify – in this case, by delineating two major, essential, interacting but different components of reading. • In the annex to the Rose Report, we presented evidence which indicates that such a simplification is warranted, including: ©Stainthorp & Stuart 2008
– Evidence from factor analytic studies of reading, showing that typically two factors are identified, with word recognition loading heavily on one of these, and comprehension on the other – Evidence from studies showing that in various typical and atypical populations, good word recognition skills can be developed in the absence of good comprehension, and vice versa
• In other words, this simplification does not overly distort the object (reading) being simplified. ©Stainthorp & Stuart 2008
Is the simple view of reading a ‘model’? • It is certainly not a processing model: it tells us nothing at all about the complex processes that are involved in developing either of its two dimensions. • The Simple View is only the beginning!
• We need to understand the complex processes involved in skilled word recognition and its development if we are going to enable children to read the words on the page • We need to understand the even more complex processes involved in language comprehension and how language comprehension can be developed in children if we are going to enable children to understand what they read. • Of which, more later…….
©Stainthorp & Stuart 2008
A call for research evidence • Wyse & Jones (2007) go on to say that it is important that officially endorsed models such as the simple view of reading are informed by the widest possible research base. • They argue that the ‘model’ needs to be “subject to extensive critical evaluation so that we can be sure that its influence is justified”.
• They list the following three issues which they propose research will need to address: – Is there agreement that the simple view of reading is the most appropriate model? – Does the simple view appropriately reflect the evidence that is used to justify it? – Does the simple view reflect research on the teaching of reading?”
©Stainthorp & Stuart 2008
Is the simple view of reading the most appropriate model? • As we’ve already argued, it is not a model at all. It is a starting point for thinking about and investigating the complexity of reading. • We clearly think it is a good starting point, and so do an increasing number of researchers, if sheer volume of publications is taken as an indicator: – A Scholar Google search for the exact phrase “simple view of reading” brought up a list of 261 papers and book chapters published in the last five years
©Stainthorp & Stuart 2008
Does the simple view appropriately reflect the evidence that is used to justify it? • We confess to finding this a rather strange question: perhaps that is because we locate ourselves within the scientific community. • In science, evidence can never justify any particular proposition or theory. Consider the well known example of the proposition “All swans are white”. • Finding lots of white swans and no swans of any other colour might appear to ‘justify’ this proposition. It cannot. There always remains the possibility that somewhere a red swan, a black swan, a striped and spotted swan, might appear. • So, evidence, according to Popper, can only ever disprove a proposition or theory.
• Propositions and theories that have not yet been disproved remain conditional; they can never be shown to be true. • So, we would prefer to recast the question:
©Stainthorp & Stuart 2008
A recast of the question: • Is there to date any evidence that casts doubt on the proposition that reading comprises two dimensions, namely, word recognition and language comprehension? • The kind of evidence we would need here is, for example: – Demonstration that performance on tasks requiring word recognition skills and tasks requiring language comprehension abilities load on to a single factor in factor analytic studies • To our knowledge this has not yet been demonstrated – existing factor analytic studies have all identified at least two factors, with word recognition tasks loading on one, and language comprehension tasks loading on the other.
©Stainthorp & Stuart 2008
Is reading unidimensional? • If it is, then we should never be able to find children whose word recognition skills and language comprehension abilities are discrepant with each other. • But, discrepant patterns of performance like this have already been demonstrated. • This disproves the proposition that reading is unidimensional.
©Stainthorp & Stuart 2008
Does the simple view reflect research on the teaching of reading? • This question is sufficiently complex to require a presentation of its own. • It might be fruitful for a future UCET conference to consider the relations between research into the cognitive and linguistic processes involved in reading and learning to read, and research into the teaching of reading. • We would be pleased to participate in the exchange of views that such a conference would encourage. • However, in the interests of getting to where we want to get today, we will just reiterate that the simple view of reading is just a conceptual framework for thinking about the object ‘reading’.
©Stainthorp & Stuart 2008
Does the simple view reflect research on the teaching of reading? • There are certain inferences one can draw from the simple view of reading as to what might need to be taught. • Permissible inferences include: – Children need to be taught to recognise and understand the words on the page – Children need to be helped to understand the texts they read.
©Stainthorp & Stuart 2008
The inference we draw from the simple view of reading is that • The better we understand each of the two dimensions, the better placed we are to facilitate children’s reading development. • This means understanding at least four things: – what happens in the mind of a skilled and fluent reader as they recognise and understand the words on the page – how these word recognition processes develop as children learn to read – what happens in the mind of a skilled and fluent reader as they interact with texts to construct a valid personal interpretation of the meaning – how these comprehension processes develop as children learn to read.
©Stainthorp & Stuart 2008
Skilled word recognition: two major processing models
• The dual route cascade model • The ‘triangle’ model • Two sets of processes involved – Phonological recoding processes – Orthographic- semantic processes
©Stainthorp & Stuart 2008
Dual route cascade and triangle models printed word
printed word
letter identification orthographic store
orthographic store grapheme-phoneme
semantic store
conversion system
phonological store
semantic store
phonological store phoneme units
spoken word
spoken word ©Stainthorp & Stuart 2008
Developing word recognition skills • Predictors of success – Phonological awareness - especially phoneme awareness – Letter knowledge - both letter name and letter sound knowledge
• Both are essential to developing an understanding of the alphabetic principle ©Stainthorp & Stuart 2008
Knowledge and application of phonic rules facilitates development of phonological recoding processes : printed word
printed word
letter identification orthographic store grapheme-phoneme conversion system phonological store phoneme units spoken word
spoken word ©Stainthorp & Stuart 2008
Knowledge and application of phonic rules also facilitates development of orthographic/semantic processes printed word
WHY?
printed word
letter identification orthographic store
orthographic store semantic store
semantic store
phonological store
phonological store phoneme units
spoken word
spoken word ©Stainthorp & Stuart 2008
Knowledge and application of phonic rules also facilitates development of orthographic/semantic processes
WHY?
Two views: Share’s ‘self-teaching’ hypothesis Ehri’s ‘partial alphabetic’ phase
©Stainthorp & Stuart 2008
Self-teaching hypothesis •
Children who can apply phonic knowledge to read unfamiliar words will build a store of ‘sight vocabulary’* more quickly.
•
This is because left-to-right decoding of each grapheme forces attention sequentially on to each letter of the unfamiliar word.
•
This sequential attention to each grapheme increases the likelihood that the child will form an accurate memory of the spelling pattern. * by ‘sight vocabulary’ we mean, storage of spelling patterns of familiar words in an orthographic lexicon, and linkage of these spelling patterns to the meanings of the words they represent. ©Stainthorp & Stuart 2008
There are at least five research studies which have tested this hypothesis: • Share (1999) – 2nd graders; read short texts containing nonwords (preef) – Tested 3 days later: • preef / preaf ? – chose preef • ‘preef’ read more rapidly than ‘preaf’ • /prif/ more likely to be spelled ‘preef’ than ‘preaf’
©Stainthorp & Stuart 2008
There are at least five research studies which have tested this hypothesis: • Cunningham, Perry, Stanovich & Share (2002). – 32 2nd graders; nonwords in stories – Tested learning by • homophone choice (which is the coldest town in the world? Yait / Yate • Spelling • Naming
• Advantage for seen nonwords in all conditions
©Stainthorp & Stuart 2008
There are at least five research studies which have tested this hypothesis: • Bowey & Muller (2005) – 3rd grade children; read series of short stories; contained 4 or 8 nonwords – Tested on lists of nonwords, spelt as seen in stories (‘lork’) or new, homophonic, spelling (‘lawk’) – ‘as seen’ nonwords (lork) read faster than new spellings (lawk) – Supports Share’s self-teaching hypothesis ©Stainthorp & Stuart 2008
There are at least five research studies which have tested this hypothesis: • Nation, Angell & Castles (2006) – 8 and 9 year olds; read nonwords in stories or isolation; 1, 2 or 4 repetitions – Asked to select nonwords from visually similar and phonologically similar foils • ferd furd ferp furp
– Learning increased with number of repetitions, but some observed with single exposure – No advantage for presentation in story context
©Stainthorp & Stuart 2008
The general conclusion from these five studies is • Ability to decode unfamiliar items leads to orthographic learning – i.e. use of phonic skills to read unfamiliar items facilitates storage of those items in sight vocabulary
©Stainthorp & Stuart 2008
Partial alphabetic phase rain
orthography r
n
phonology /reI n/ drops of water falling from the sky ©Stainthorp & Stuart 2008
Evidence consistent with Ehri’s hypothesis: Storage of ‘boundary’ letters • Stuart & Coltheart (1988) • Longitudinal study • Assessed phonological awareness in final term in nursery, and July year R • Assessed letter name and letter sound knowledge in January, March and July of Year R • Recorded children’s single word reading errors every two months throughout Year R and first term of Year 1
©Stainthorp & Stuart 2008
• Grouped children into four groups, according to PA and letter sound knowledge – Group A = above mean PA in nursery; knew 15+ letter sounds in January Yr R – Group B = above mean PA in nursery; knew 15+ letter sounds in March Year R – Group C = above mean PA either in nursery or by July Year R; knew 15+ letter sounds in July Year R – Group D = still failed to meet PA and/or letter sound knowledge criteria in July Year R.
©Stainthorp & Stuart 2008
• Asked children to read single words on 7 occasions: Nov, Jan, March, May and July of Year R, and Sept and Nov of Year 1. • Categorised each child’s errors on each test into five groups: – 1 = error shared no letters with target word – 2 = error shared one or more letters with target word, but not in same order or position in word – 4 = error shared one or more final letters with target word.
• 1, 2 and 4 = ‘bad’ errors. Children with large proportions of these in their corpus did worse on standardised reading test. – 3 = error shared one or more initial letters with target word – 5 = error shared one or more initial and final letters with target word
• 3 and 5 = ‘good’ errors. Children with large proportions of these in their corpus did better on standardised reading test. Good errors include those which represent boundary letters. • In a later study, children whose single word reading errors at age 6 mostly included boundary letters were significantly better readers 2 years later (Savage, Stuart & Hill, 2001)
©Stainthorp & Stuart 2008
Relation between change over time in proportion of ‘good’ and ‘bad’ errors and change over time in PA and letter sound knowledge
Group A = above mean PA in nursery; knew 15+ letter sounds in January Yr R
©Stainthorp & Stuart 2008
Relation between change over time in proportion of ‘good’ and ‘bad’ errors and change over time in PA and letter sound knowledge
Group B = above mean PA in nursery; knew 15+ letter sounds in March Year R ©Stainthorp & Stuart 2008
Relation between change over time in proportion of ‘good’ and ‘bad’ errors and change over time in PA and letter sound knowledge
Group C = above mean PA either in nursery or by July Year R; knew 15+ letter sounds ©Stainthorp & Stuart 2008 in July
Relation between change over time in proportion of ‘good’ and ‘bad’ errors and change over time in PA and letter sound knowledge
Group D = still failed to meet PA and/or letter sound knowledge criteria in July Year R ©Stainthorp & Stuart 2008
A different kind of self-teaching hypothesis 5-year-olds who can identify initial phonemes and recognise letters for those phonemes learn ‘sight vocabulary’ more quickly (Stuart, Masterson & Dixon, 2000) Their sight vocabulary representations are more detailed and include boundary consonants (Dixon, Masterson & Stuart, 2002) These representations serve as a database from which further GPCs can be inferred (Stuart, Masterson, Dixon & Quinlan,1999) In this proposition, partial use of the alphabetic principle facilitates development of representations in sight vocabulary which include boundary letters. Having fixed the links between boundary letters and the sounds they represent, the remaining (medial) letters must map on to the remaining (medial) sounds. This allows the child, for example, to start inferring the links between medial vowel digraph spellings and sounds This proposition has the power to explain why some children are able to develop word recognition skills with very little teaching: because they have the necessary knowledge, skills and understanding to develop a self-sustaining learning system. ©Stainthorp & Stuart 2008
PA and letter-sound knowledge facilitate development of sight vocabulary • Taught new words to 5year-olds through shared book reading • PA lets + knew letter sounds; could identify initial phonemes • PA lets – Did not know letter sounds; could not identify initial phonemes
16 14 12 10 PA lets + PA lets -
8 6 4 2 0 12
24
36
Stuart, Masterson, & Dixon. (2000). •
©Stainthorp & Stuart 2008
Boundary letters are included in the representation of words stored • • •
Dixon, Masterson & Stuart (2002) Taught 10 new words 3 groups of children – – –
•
Identify initial and final phonemes; know letter sounds Identify initial phonemes; know letter sounds Don’t identify phonemes; don’t know letter sounds
First group only accepted variants of words presented if they contained the correct initial and final letters – – –
e.g. for ‘sandal’, most likely to accept SANDAL SARDAL SANCAL SADNAL Rejected NASDAL, SANLAD, etc.
©Stainthorp & Stuart 2008
Do children infer grapheme-phoneme correspondences from sight vocabulary? • Stuart, Masterson, Dixon & Quinlan (1999) – 6 to 7 year olds who had not been taught vowel digraphs – Asked to read four-letter nonwords – Half contained vowel digraphs they’d seen frequently in their reading material – Half contained digraphs seen much less frequently – Nonwords with frequently experienced digraphs more likely to be read correctly – Children who were most successful on this task could also identify medial phonemes in words ©Stainthorp & Stuart 2008
Summary so far: • The simple view of reading is not a model of reading and we did not develop it • However, the proposition it makes, that reading consists of two major interacting dimensions, word recognition and language comprehension, has not to date been falsified • Much is now known and agreed about the cognitive processes by which skilled readers recognise and understand written words • Much is also known about how children develop these cognitive processes, and about the foundations upon which they are developed. • We have shown you a tiny fraction of some of this knowledge here. • What about the comprehension dimension and its interaction with word reading as reading develops? ©Stainthorp & Stuart 2008
Comprehension • Gough and Tunmer are clear that they are referring to linguistic comprehension because they see the same processes underlying comprehension whether by ear or by eye. • They define it as ‘the process by which, given lexical (i.e. word) information, sentences and discourse are interpreted’.
©Stainthorp & Stuart 2008
Framework not theory • Gough and Tunmer proposed the SVR as a way of capturing in a transparent way the interactive relationship between the two complex sets of processes that are required if texts are to be read efficiently and independently with understanding.
©Stainthorp & Stuart 2008
Gough and Tunmer further make clear that word recognition is necessary but not sufficient for reading because ability to pronounce printed words does not guarantee understanding of the text so represented. Furthermore, linguistic comprehension is likewise necessary, but not sufficient, for reading: if you cannot recognise the words that comprise the written text, you cannot recover the lexical information necessary for the application of linguistic processes that lead to comprehension. – Rose Review, p.76
©Stainthorp & Stuart 2008
• They originally set out the simple view as an equation: R=DxC This draws attention to the notion that: 1. If the print cannot be translated into word then reading cannot be taking place because there is no way of accessing the meaning of the text. 2. If the text as decoded cannot be understood then likewise reading cannot be said to be taking place.
Neither is sufficient; both are necessary. ©Stainthorp & Stuart 2008
Hoover and Gough then attempted to encapsulate the relationship between the two sets of processes in a diagrammatic form
©Stainthorp & Stuart 2008
From Hoover and Gough, 1990
©Stainthorp & Stuart 2008
The diagram alerts us to the fact that the balance of learning needs across the two dimensions changes as children become more fluent and automatic readers of words.
©Stainthorp & Stuart 2008
• Children come to school with considerably developed linguistic comprehension skills. • They have much to learn as far as content is concerned; there is still considerable development in vocabulary, syntax and semantics to take place But • they have the linguistic comprehension skills necessary to understand the spoken language around them. • Language development is not complete but they are not starting ab initio. – Language, unlike reading is a first order activity. ©Stainthorp & Stuart 2008
The two dimensions: Word reading accuracy and comprehension
©Stainthorp & Stuart 2008
Word reading accuracy and comprehension • Oakhill, Cain & Bryant, 2003 • 102 children assessed on • oral vocabulary • single word reading accuracy • phonological awareness • working memory • syntax • ability to draw inferences • understanding of story structure, • comprehension monitoring ability • This was a longitudinal study with the children being tested at time 1 - age 7 years, and time 2 - age 8 years ©Stainthorp & Stuart 2008
• The study was designed to investigate the contribution of these measures to individual differences in word reading accuracy and reading comprehension as measured by the Neale Analysis of Reading Ability (NARA).
©Stainthorp & Stuart 2008
Predictors of NARA reading accuracy – single word reading accuracy – phonological awareness
Predictors of NARA reading comprehension – ability to draw inferences – understanding of story structure – comprehension monitoring ability
The pattern of performance was the same at both ages ©Stainthorp & Stuart 2008
• This shows a clear dissociation between predictors of comprehension and predictors of word recognition. • This finding could have been predicted from the Simple View of Reading
©Stainthorp & Stuart 2008
Word reading accuracy and comprehension • Muter, Hulme, Snowling and Stevenson (2004) • 90 children assessed on – – – –
phonological skills letter knowledge, grammatical skills, vocabulary knowledge
• This was a longitudinal study with children being tested on entry to school (mean age 4y 9m) and followed for 2 years ©Stainthorp & Stuart 2008
The study was designed to investigate the power of these early skills to predict later word recognition and reading comprehension.
©Stainthorp & Stuart 2008
Predictors of word recognition skills – letter knowledge – phoneme sensitivity
Predictors of reading comprehension – prior word recognition skills – vocabulary knowledge – grammatical skills
Again, different skills and knowledge were shown to underlie performance on each of the two dimensions of reading ©Stainthorp & Stuart 2008
Word reading accuracy and comprehension • Nation and Snowling (1997) • 107 7 – 9 year olds assessed on measures of: – listening comprehension – word reading accuracy with and without context – nonword reading – text comprehension (NARA) – sentence comprehension (Suffolk reading scale)
©Stainthorp & Stuart 2008
Factor analysis Factor 1: Decoding (Word recognition)
Factor 2: Comprehension
• word reading accuracy with and without context • nonword reading • Suffolk test
• narrative listening • Neale comprehension
? Sentence completion tasks maybe more dependent on basic word recognition skills than on comprehension?
Listening and reading comprehension loaded onto the same factor This is consistent with the view that oral and written language comprehension depend to some extent on the same underlying language comprehension system. ©Stainthorp & Stuart 2008
• We have established that there is evidence consistent with the proposition made in the SVR that there are two dimensions of reading.
• Let us return to the diagram.
©Stainthorp & Stuart 2008
The Simple View of Reading
Language comprehension processes
Word recognition processes
GOOD
Good language comprehension; poor word recognition
POOR
GOOD
POOR
Poor word recognition; Poor language comprehension
Good word recognition; good language comprehension
Language comprehension processes
Word recognition processes
Good word recognition; Poor language comprehension
©Stainthorp & Stuart 2008
Good word recognition; poor comprehension • This pattern of performance is sometimes observed in children with developmental disorders: – Specific language impairment (Bishop & Adams, 1990) – Autistic spectrum disorders (Snowling & Frith, 1986) – Hyperlexic children (Newman et al., 2007)
©Stainthorp & Stuart 2008
• It is a pattern also found in some typically developing children who are precocious readers (Jackson, Donaldson & Cleland, 1988; Pennington, Johnson & Welch, 1987; Stainthorp and Hughes, 1998, 1999, 2005) – These children show exceptional word reading skills, way above chronological age with lower, but acceptable comprehension skills
©Stainthorp & Stuart 2008
The Simple View of Reading
Language comprehension processes
Word recognition processes
GOOD
Good language comprehension; poor word recognition
POOR
GOOD
POOR
Poor word recognition; Poor language comprehension
Good word recognition; good language comprehension
Language comprehension processes
Word recognition processes
Good word recognition; Poor language comprehension
©Stainthorp & Stuart 2008
Poor word reading skills with good comprehension • Spooner, Baddeley & Gathercole (2004) • 80 7- to 8-year-old children divided into two groups. 1. Normal listening comprehension, normal word recognition. 2. Normal listening comprehension; poor word recognition. • The group with poor word recognition skills did worse on a reading comprehension test. ©Stainthorp & Stuart 2008
This study furnishes evidence that poor word reading skills can impact negatively on comprehension of written texts in the face of normal comprehension when listening.
©Stainthorp & Stuart 2008
Development and the balance of skills • Catts, Adlof & Weismer (2007) – Evidence from a longitudinal study
• Reading comprehension in the early grades is more heavily dependent on word recognition skills than on language comprehension ability. – This is when word reading skills are still developing.
©Stainthorp & Stuart 2008
• Grades 2 and 4 (age 7 years and 9 years) • They compared 2 groups: – Children with poor word recognition skills and age appropriate language comprehension; – children with age appropriate word recognition skills and poor language comprehension
• Both had reading comprehension scores significantly lower than those of typically developing readers. • This shows that in these early grade levels, poor word recognition skills impaired reading comprehension even when language comprehension abilities were similar to those of typically developing readers. ©Stainthorp & Stuart 2008
• By 8th grade, reading comprehension in the group with poor word recognition skills was no longer significantly impaired relative to that of the typically developing readers. – Presumably by 8th grade, although they were still less skilled in terms of word recognition than the typically developing 14 year olds, they had reached a minimal level of competence in word recognition that allowed their good language comprehension ability to have an effect on reading comprehension. ©Stainthorp & Stuart 2008
Balance • The balance of the contribution from the two different processes shifts. • Once children achieve a degree of proficiency in word reading there is more processing capacity available for comprehending the texts. • Decoding is time limited and can be automated. • Comprehension may always be effortful to a degree.
©Stainthorp & Stuart 2008
The components of the comprehension system
Adapted from Perfetti, 1999 ©Stainthorp & Stuart 2008
The shaded boxes are the components required for comprehending spoken language. The knowledge sources of general knowledge, language and vocabulary are used to support the comprehension processes to generate a mental representation of the spoken message.
When texts are read there are further processes involved in order that the words can be identified. This is represented by the clear boxes in the diagram. The visual word identification processes have to be incorporated into the system. However, the comprehension processes themselves remain the same. • ©Stainthorp & Stuart 2008
• In order to generate a mental representation of a spoken message the comprehension processes use information from the language system, vocabulary and the level of general knowledge of the listener. • Improvements in all these areas underlie developmental improvements in language comprehension. • This improves whether or not the listener becomes literate.
©Stainthorp & Stuart 2008
• However, becoming literate has the potential to extend both vocabulary and general knowledge and thus to accelerate improvements in language comprehension.
©Stainthorp & Stuart 2008
When the ‘listener’ becomes a reader, the unique development that takes place is that the ‘reader’ is able to identify the written words and encode their meanings so that the mental representation is constructed using additional output from the written word identification system. ©Stainthorp & Stuart 2008
Implications for ITE • This evidence, though complex, is necessary knowledge for student teachers so that they understand – why they have to ensure children are taught to use a phonic strategy for initial word reading – why they must also focus on language comprehension
• This will enable them to ensure that children make adequate progress on both dimensions. ©Stainthorp & Stuart 2008
