Copyright is owned by the Author of the thesis. Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only. The thesis may not be reproduced elsewhere without 
the pennission of the Author. 



AN INVESTIGATION INTO CONTEXTUAL FACILITATION EFFECTS 
FROM A VERBAL-VISUAL FORMAT. 

by 

JUSTIN CHARLES FRANCIS HYDE 

A THESIS 

PRESENTED IN PARTIAL FULFILMENT OF 

THE REQUIREMENTS OF THE DEGREE OF 

MASTER 

IN 

EDUCATION 

EDUCATION DEPARTMENT 

MASSEY UNIVERSITY 

PALMERSTON NORTH, NEW ZEALAND 

NOVEMBER, 1992 



ABSTRACT 

This study investigated the use of contextual facilitation in 

children's beginning reading. A verbal-visual format was utilised to 

examine the way context influences children's ability to identify 

irregular words (words which do not follow normal spelling to sound 

relationships). The study also determined whether poor or good 

readers in each grade utilised context more or less according to a 

proposed stage pattern of word identification. 

The sample, comprised 113" children drawn from three grades from a 

Primary School in New Plymouth. Children were individually tested 

with the Peabody Picture Vocabulary Test-Revised, BURT Word 

Reading Test, Isolated Word Test and Contextual Facilitation Test. 

The BURT, Isolated Word Test and Context Facilitation Test were 

modified for easier presentation and active participation of the 

children to reduce confounding variables of earlier research. 

Two ANOVA's were applied to analyse data in this study. The first 

was used to assess the difference that exists in contextual 

facilitation across the grades tested. The second ANOVA assessed 

the interaction between grade by reading ability by word block 

difficulty for context facilitation and for movement through the 

proposed stage pattern in word recognition. A correlation and 

I I 



stepwise regression assessed the link between context facilitation 

and BURT scores for reading ability differences between the 

children in relation to contextual facilitation use. 

The first ANOVA revealed that utilisation of context with childrens' 

increasing age decreases for simple words, but increases with 

increasing word difficulty. The second ANOVA indicated a divergence 

in reading ability and contextual facilitation beginning at about 

Junior 2 grade, which possibly delayed childrens' development of 

automatic word decoding skills. The correlation and stepwise 

regression between contextual facilitation and BURT scores provides 

the range of the children's word identification ability. This showed a 

considerable range in ability from poor word decoding where context 

cannot be utilised, through an alliance between decoding and context 

to identify words; to highly efficient word decoding without the 

need for context. 

Evidence from this study indicates the existence of a stage pattern 

of sight word acquisition similar to that proposed by Adams and 

Huggins (1985). This consists of three stages: 1) non-recognition, 2) 

intermediate and 3) automatic word recognition. Results of. this 

study provide evidence that an alliance exists between children's 

decoding and context for words in the intermediate stage before 

automatic word recognition occurs. As a result, increasing word 

I I I 



identification ability into the child's non-recognition stage from 

intermediate stage expansion. Divergence in the stage pattern 

proposed begins in Junior 2 grade, where low ability readers lag 

behind in automatic word decoding and rely more on context. Further 

research beyond the generalised results from this study should 

include a longitudinal study to follow the reading development of 

individual students. 

IV 



This dissertation is dedicated to my parents, 

Jane and Gordon, whose understanding and 

support provided me with the encouragement 

to complete this work. 

V 



ACKNOWLEDGEMENTS 

There are several people whom I wish to acknowledge for their part 

in the construction of this thesis. First, my two supervisors Prof. 

William Tunmer and Dr. Ken Ryba who both helped in many ways. To 

Prof. Tunmer whose vast knowledge and experience in the reading 

field helped construct the analyses used in this investigation into 

contextual facilitation. To Dr. Ryba who helped me to become 

familiar with the computer system SPSSx to perform the statistical 

analyses in the thesis. Both Bill and Ken have provided much positive 

commentary on the drafts they read, and contributed much to my 

conceptual understanding of thesis work. Many thanks to both of you 

for always supplying me with positive encouragement and advice 

which will influence the way I conduct my future research. 

Special th.anks, of course, to the people who allowed me to become 

part of their school community for three months - Eddie Betts and 

the staff and children of Westown Primary School. We all seemed to 

work well together and ended the time still friends. Their positive 

encouragement would indicate that I gave them as much as they gave 

me. I hope this is the case. 

VI 



TABLE OF CONTENTS 

Abstract ................................................................................... i i 

. . 
Acknowledgements ................................................................. v 1 

List of Tables ......................................................................... x i 

List of Figures ........................................................................ xii 

1 Introduction and Overview ............................................ 1 

Intent of Present Study .......................................................................... ? 

2 Contextual Facilitation: 

A Review of the Literature ........................................... 9 

Semantic and Syntactic Redundancy ................................................ 9 

Phonological Recoding Reliance ......................................................... 1 4 

Research Investigations: Phonological Recoding ........................ 1 5 

1) Memory: Short-Term Memory (STM) ................................... 1 6 

2) Memory: STM and Context.. ..................................................... 1 8 

3) Automaticity and Attentional. Deficits ............................ 2 0 

4) Reading Level Match Designs 

for Contextual Effects .................................................................. 2 3 

5) Perfetti's (1985) Verbal Efficiency Theory .................. 2 7 

6) Interactive Compensatory Model ......................................... 2 8 

7) Short-Term Memory and the 

Acquisition of Reading .................................................................. 3 O 

Summary .............................................................................................. 3 2 

Vii 



Decoding Context Alliance .................................................................... 3 3 

Preliminary Research with Irregular Word Testing ................. .4 2 

Methodological lssues ........................................................ : ................... .4 9 

Summary ....................................................................................................... 51 

Research Questions .................................................................................. 5 7 

Tentative Hypothesis .............................................................................. 5 7 

3 Method ............................................................................... 5 9 

Subjects ........................................................................................................ 5 9 

Materials ....................................................................................................... 5 9 

(a) Peabody Picture Vocabulary Test-Revised 

(PPVT-R) .............................................................................................. 5 9 

(b) BURT Word Reading Test. ....................................................... 6 0 

(c) Isolated Word Test. .................................................................. 61 

(d) Contextual Facilitation Test ............................................... 6 2 

Procedure ...................................................................................................... 6 3 

Permission .......................................................................................... 6 3 

Establishing Rapport ...................................................................... 6 3 

Testing Situation ............................................................................. 6 4 

Testing Sessions .............................................................................. 6 4 

Tests Performed .............................................................................. 6 5 

(a) (PPVT-R) ............................................................................. 6 5 

(b) BURT Word Reading Test.. ............................................ 6 5 

(c) Isolated Word Test.. ....................................................... 6 8 

viii 



(d) Contextual Facilitation Test.. ............................................. 6 9 

4 Results and Discussion .................................. ~ ............... 7 2 

Correlation between Mean Context Effect 

and BURT Scores ........................................................................................ 8 7 

Stepwise Regression Analysis ............................................................ 9 2 

5 General Discussion and Conclusions ............................ 9 4 

Initial Concerns ......................................................................................... 9 5 

Analyses ........................................................................................................ 9 6 

Correlational Analysis ........................................................................... 1 O 3 

Stepwise Regression Analysis ............................................................ 1 O 5 

Limitations of the Study ....................................................................... 1 05 

Suggestions for Future Research ....................................................... 1 06 

Conclusions .................................................................................................. 1 O 8 

References ............................................................................... 1 1 O 

Appendices ............................................................................... 1 1 6 

Appendix A: Original Adams and Huggins Isolated Words 

and Sentence Context Tests ................................................................. 11 6 

Appendix B: Parental Permission ....................................................... 11 9 

Appendix C: Original BURT Test Format.. ........................................ 1 20 

Appendix D: Modified BURT Test Format Example Card ............ 1 22 

IX 



Appendix E: Modified BURT Test Record Sheet.. ........................... 1 24 

Appendix F: The Clay Test Words ....................................................... 1 26 

Appendix G: Isolated Word Test Example Card 

and Record Sheet. ...................................................................................... 1 2 7 

Appendix H: Practice Example Card 

for the Context facilitation Test.. .................................................... 1 30 

Appendix I: Contextual Facilitation Test Example Card 

and Record Sheet. ...................................................................................... 1 3 2 

Appendix J: Children's Raw Score Data ........................................... 1 35 

X 



LIST OF TABLES 

1 Means and (Standard Deviations) for Words 

Correctly Identified as a Function of Condition 

and Word Block for each Grade Level. .............................................. 7 3 

2 Means and (Standard Deviations) of Contextual 

facilitation for Grade X Reading Ability (High and Low) 

X Word Block .............................................................................................. 8 0 

3 Mean Context Effect versus BURT Score for all Grades ........... 8 8 

4 Correlations between Context Effect and BURT Raw Scores 

for all Grades at Intervals of Increasing 

and Decreasing Size ................................................................................. 8 9 

5 Summary of Multiple Regression Analysis Testing a 

Curvilinear Relationship between BURT Raw Score 

and Contextual Facilitation ................................................................... 93 

xi 



LIST OF FIGURES 

1 Mean Number of Words Correctly Identified as a Function 

of Grade Level and Condition ............................................................... 7 4 

2 Mean Number of Words Correctly Identified as a Function 

of Grade Level and Word Block ............................................................ 7 5 

3 Mean Number of Words Correctly Identified as a function 

of Condition and Word Block for all Grades Combined ............. 7 6 

4a The Mean Number of Words Correctly Identified as a 

Function of Condition and Word Block for the 

Junior 2 Grade ............................................................................................ 7 7 

4b The Mean Number of Words Correctly Identified as a 

Function of Condition and Word Block for the 

Standard 1 Grade ....................................................................................... 7 8 

4c The Mean Number of Words Correctly Identified as a 

Function of Condition and Word Block for the 

Standard 2 Grade ....................................................................................... 7 8 

5 Mean Number of Words Correctly Identified from 

Context Effect as a Function of Grade Level 

and Word Block ........................................................................................... 8 2 

6 Mean Number of Words Correctly Identified from 

Context Effect as a Function of Reading Level 

and Word Block ........................................................................................... 8 3 

Xii 



7 Mean Number of Words Correctly Identified from 

Context Effect as a Function of Grade Level 

and Reading Ability ............................................................. : .................... 84 

Ba Mean Contextual Facilitation Effect as a Function of 

Reading Ability and Word Block for the Junior 2 Grade ........... 8 5 

8b Mean Contextual Facilitation Effect as a Function of 

Reading Ability and Word Block for the Standard 1 Grade ..... 8 5 

8c Mean Contextual Facilitation Effect as a Function of 

Reading Ability and Word Block for the Standard 2 Grade ..... 8 6 

9 Context Effect with Raw BURT Scores for all Grades .............. 8 8 

1 O Scatterplot: Context Effect with Raw BURT Scores 

for each Subject from each Grade ..................................................... 9 2 

Xiii 



CHAPTER ONE 

INTRODUCTION AND OVERVIEW 

The specific aim of this study is to determine whether contextual 

facilitation increases the reading potential of good and poor readers. 

Two opposing views exist regarding children's reading potential and 

the issue of contextual facilitation. The first view put forward by 

Goodman (1967) and Smith (1971) proposes that skilled reading is an 

activity of relying on the syntactic and semantic redundancies of 

language to make predictions about text yet to be encountered. The 

process of learning to read, according to this "psycholinguistic" 

view, is one of picking up contextual cues. The second view put 

forward by Stanovich (1980, 1986) proposes that skilled reading 

relies on the rapid and accurate decoding of words, and is not a 

context reliant process. These views are briefly outlined below 

along with_ a description of the perspective adopted in this study. 

According to the psycholinguistic method of reading instruction, 

teachers should allow the child to make mistakes as the context 

will soon put them right. There must be no specific help with words; 

when a child finds difficulty with reading, the teacher's role is to 

encourage them to work out words in their own way. The children 

learn to read because the teachers prevent reading programmes from 

standing in the children's way. 



2 

Therefore, this psycholinguistic view considers reading acquisition 

to be a natural process in which exposure to reading materials 

is all that is necessary for acquisition. Interfering with the natural 

state of affairs may lead to failure, "because children will have to 

deal with concepts and structures that make little sense to them" 

(Calfee and Drum, 1986 p. 809). As a result, reading failure arises 

from efforts to "teach" about reading when children could easily 

learn about reading on their own. 

One apparent weakness in this view is that children can learn to 

read by relying on sentence context alone without phonic decoding. 

In this regard, it should be noted that even the most obvious 

contexts generate a number of alternative views. For example, take 

the sentence: "The boy climbs the fence." The choice for the target 

word fence is not only limited to climbable objects (e.g., ladder, 

wall, rope, bars, hill), but could also include adverbs like fastest or 

slowest. The delay generated by the child trying to work out the 

meaning of the target word from sentence context alone could be 

considerable. 

Advocates of the psycholinguistic method recommend that children 

be taught to skip unknown words, until the words are met repeatedly 

in various contexts to induce meaning. This method allows fluent 

readers to enlarge their vocabulary, but it presupposes only 



3 

occasional unknown words. The method is ineffective for the 

beginning reader who may be stumped by more than one word in the 

sentence. In the above sentence, a child who cannot read fence 

may also fail to read climbs. The contextual clues in the sentence 

now become uninformative: "The boy __ the __ ." Not having 

the means to progress further than this could lead to frustration and 

the avoidance of the area of reading. 

The alternative view, known as Phonological Decoding, states that 

reading by skilled readers is a process of rapid decoding of words. 

Stanovich (1980, 1986) suggests that the use of context is 

negligible by skilled readers but context usage by poor readers is 

significant. Reading according to Stanovich is hot a top-down 

context driven process, but rather a bottom-up automatic decoding 

process. This view encompasses fast automatic bottom-up 

processing, where spelling-to-sound mapping determines a child's 

ability to identify unknown words. 

When children have developed their knowledge about spelling-to­

sound correspondences, it is also important to develop their 

automaticity at word identification. Words must be recognised 

quickly to ensure that sufficient cognitive resources are available 

for comprehension processes to operate on the information in 

children's short-term memory. Stanovich (1986) suggests that 



4 

unrewarding early reading experiences will lead to less involvement 

in reading by the child. As a result, the poor reader's lack of practice 

prevents the development of speed and automaticity in word 

recognition. 

Research reviewed by Stanovich (1986) indicates that good readers 

use context less because of their good decoding skills in reading 

whereas poor readers utilise context more in reading because 

they have poor decoding skills. Stanovich (1986) also indicates that 

context facilitation is inversely related to a child's word 

recognition skill. Therefore, a child's word recognition skill 

determines the extent to which contextual information will be 

relied upon to complete lexical access. Although poor readers rely on 

context more than good readers for word identification in ongoing 

reading, r.esearch reviewed by Stanovich (1986) also suggests that 

good readers are better at producing missing words in visually or 

orally presented text. 

The view put forward in this study is that a cooperative mechanism 

exists between children's decoding and context usage while reading. 

It is proposed that children are at different levels of ability in 

terms of their utilisation of context to read words in text. Generally 

speaking there are three groups or levels. The first is the poor 

reading ability group who are both poor decoders and context users. 



5 

In this group the children's language prediction skills do not aid the 

decoding of words. At the opposite end of the scale, the good readers 

decoding skills are so good that context does not significantly 

increase their language prediction skills. The good readers use 

context only as a back-up system to check difficult or 

irregular words while reading. 

The majority of beginning readers fall between these two other 

groups, and utilise context and word decoding in a cooperative 

manner while reading. In this group language prediction ability by 

the use of context helps children become better at word decoding. 

The result is that they engage in more automatic processing of 

words, and are therefore, less reliant on language prediction for 

word identification. This intermediate group of children are, as 

Stanovich (1982) suggests, poor readers who are lliU deficient in 

their utilisation of context to faci I itate reading. However, the 

slower and less accurate decoding by the poorer readers in this 

intermediate group may deteriorate if the reading materials are too 

far beyond the child's reading ability. The consequence of this gap 

between the child's reading ability and the materials they read 

would increase each school year, and may as Stanovich (1986) 

suggests generate "Matthew effects". 

However, total reliance on either decoding or context for reading is 



6 

ineffective for the child. Relying on only context to read can limit 

the beginning reader's word specific knowledge. For instance, they 

will not be able to pronounce irregular words correctly, such as 

"stomach" which does not follow conventional letter-sound rules. 

Consequently, the beginning reader will also be restricted in their 

awareness of polyphonic letter sequences in words, such as, ~. or 

.filll. Total reliance by a child on context to identify unfamiliar words 

will also result in little progress. The words that can be predicted 

from context are typically high frequency words, and not 

meaningful content words. 

With these alternative views in mind, the aim of this study is to 

examine more closely the relationship between contextual 

facilitation effects and the development of word identification 

ability, with particular focus on irregular words. The view 

proposed in this study is that a child's ability to reflect on sentence 

context combined with emerging phonological recoding ability, 

influences reading by helping children develop further their decoding 

skills. That is, the ability to combine knowledge of the constraints 

of sentence context with incomplete graphophonemic information 

may help children identify unfamiliar words (including irregular 

ones), and thus, increase both their word specific knowledge and 

their knowledge of grapheme-phoneme correspondences. This 

suggests that beginning readers who are better able to use sentence 



7 

context will more rapidly progress to a point at which the ability 

to use context decreases in importance as a consequence of more 

automatic cognitive processes in decoding taking over. Consequently, 

context usage in poor readers may lag behind good readers but reach 

a similar point a few years later. 

Intent of the Present Study 

This study examines the development of contextual facilitation 

effects as a function of grade level and decoding ability. Previous 

research to investigate this issue was carried out by Adams and 

Huggins (1985) by using irregular words in a visual context for 

identification by the children. They found that accuracy of 

recognising irregular words of "intermediate familiarity" (which 

varied with age and ability) improved markedly with context for 

every age and ability group. 

Several important modifications were made in this study to 

overcome some shortcomings of the Adams and Huggins (1985) study 

regarding word presentation. The major alteration was the use of 

context in a verbal-visual format for the children to identify target 

words. This was aimed at reducing possible memory overload 

resulting from decoding the words preceding the target word. The 

words were also presented in a random order to the child, to avoid 



8 

possible order effects. The results from the context test for 

contextual facilitation provided by the children were compared to 

their isolated irregular word scores. The other modifications made 

to the Adams and Huggins (1985) method in this study are discussed 

in chapter three, (the method). 

The next chapter (chapter two) reviews the literature relating to the 

views put forward about context effect and decoding reliance. A 

case is then made based on a review of research to adopt a combined 

perspective involving elements from both the contextual and 

decoding approaches. Chapter three explains the procedures and 

tests used, and also modifications carried out on the materials used 

by Adams and Huggins (1985). The aim of the modifications is to 

focus more on contextual facilitation between ability groups and 

reduce the confounding variables of previous experiments. Chapter 

Four presents the results of the study and Chapter Five presents a 

discussion of the results. Finally, Chapter Six presents the 

conclusions of this study, and implications for teaching and further 

research about contextual facilitation. 



9 

CHAPTER TWO 

CONTEXTUAL FACILITATION: A REVIEW OF THE LITERATURE 

This chapter contains an examination of each of the views about 

contextual facilitation introduced in Chapter One. The first view, 

which primarily involves children using context to read by means of 

semantic and syntactic redundancy. The second view, primarily 

involving the utilisation of fast and accurate word decoding to read 

is referred to here as phonological recoding reliance. The third view 

put forward in this study proposes a cooperative mechanism 

between context use and word decoding. This view will be referred 

to in this study as the decoding-context alliance. Finally, in this 

chapter preliminary work into contextual facilitation, 

methodological issues and implications for the research 

methodology in children's beginning reading will be covered. 

SEMANTIC AND SYNTACTIC REDUNDANCY 

Goodman ("1973) suggests that reading is a "psycholinguistic 

guessing game," where children utilise context for semantic and 

syntactic cues to predict forthcoming words. The main proposition 

of th is view is that children who are efficient readers pay little 

attention to the majority of words in a text because the flow of 

language follows a predictable pattern. 



"1 0 

Smith ("197"1) believes that there is a limit to the amount of 

printed information that can be processed by the readers 

eyes while reading. As a result, the reader must rely on semantic 

and syntactic information to predict upcoming words and 

comprehend the text being read. Both Goodman ("1967, "1973) and 

Smith ("197"1) argued that higher levels of reading ability arise from 

children using more contextual information during word recognition. 

This implies that children who are poor readers rely more on graphic 

information while reading to recognise words, which in turn 

increases their comprehension of a text. 

The reading process Goodman and Smith describe is a top-down 

approach, where "an idea or concept generates perceptual 

hypotheses which are then checked against perceptual data" 

(Snowling, "1987 p. 62). For example, a child reading a book may 

expect the word table to be present on a page. This expectancy about 

the word table will decrease the perceptual processing required by 

the child. All that is required by the child is that the particular 

hypothesis about the word table is confirmed. For instance, if the 

word presented on the page began with a "t" and had an ascender 

"b", this may be adequate to confirm the child's hypothesis about 

the word table. 

Goodman and Smith argue that phonological, visual and orthographic 



1 1 

information only confirm or disconfirm the child's predictions 

about words in the text about to be read. The good readers are 

thought to pay little attention to the bulk of words in a text 

because the flow of language incorporated in the text follows a 

particular pattern. This would suggest that for good readers the 

reading process is only incidentally visual, and the flow of language 

is characterised by a series of guesses. Since adults can access 

word meanings directly without first deriving a phonological 

representation, there is no reason to teach children to read by a 

phonological method. Smith argues further that the fast reading 

speed of skilled readers proves that they cannot be attending to 

every letter in a word. Instead, they make use of syntactic, semantic 

and pragmatic information in recognising words. Therefore, 

beginning readers should be taught to utilise context to read in the 

same way. as adults. 

The poor readers lower level of reading ability is thought to result 

from "readers using increasing amounts of graphic information 

during word recognition" (Leu, DeGroff and Simons, 1986 p. 348). The 

result of an overdependence on graphic information according to 

Goodman and Smith is an obstruction in the poor reader's 

comprehension. Goodman and Smith propose that decoding words is 

difficult, unnatural and a harmful method of learning to read for 

children. Therefore, children should learn to read by reading, by 



"1 2 

deriving hypotheses from the context and prior knowledge of what 

the text is about. 

Smith is not explicit, however, about what methods should be used 

to teach children to read. Flaws appear in Smith's argument that 

guessing words from context is the all important factor in skilled 

reading. This was based on previous research experiments with 

adults reading words from difficult texts. Under the conditions of 

reading difficult texts, adult readers will make informed guesses 

to help them decipher words. However, there is no evidence to 

suggest that adults use guesswork to help them identify words in 

normal reading. The guesswork used by adults may help them 

identify a visually unfamiliar word or work out the meaning of a new 

word. But usually, contextual information is available too late to aid 

in word identification by adult readers. Normally adult readers have 

identified a word from its appearance before they have had time to 

decide what words are likely to appear in context. 

The assumption put forward by Goodman and Smith that less skilled 

readers cannot use context has little support in research on word 

recognition. Generally, the use of context will not be a problem for 

less skilled readers provided the text does not exceed their coding 

abilities. The view Goodman and Smith propose is that reading is a 

process of guesswork used by adults which can be applied to 



1 3 

children. However, "reading cannot be simply a guessing game" 

(Oakhill and Garnham, 1988 p. 97) for children. There· must also be an 

element of decoding of the printed text to either confirm or 

disconfirm guesses. Further, children should also be taught to 

decode text while reading. A child's decoding processes cannot be 

expected to arise as an offshoot of reading guesswork; word 

decoding instruction is necessary. However, some high ability 

children will be able to work out the rules for word decoding 

themselves without instruction. 

The most important point made by Goodman and Smith, is that 

success will only occur with reading when a child considers the 

materials are "worth" reading. Otherwise, if a child is only made 

to concentrate on visual information while reading, difficulty will 

be encou.ntered, and therefore, inability to make meaning for 

comprehension. 

It is not claimed here that abilities such as predicting words from 

context are unimportant in children's reading. This is not meant to 

discount the importance of such top-down reading skills. Rather, the 

view advanced here is that teaching only such top-down context 

skills is not enough for children to proceed in reading. Both decoding 

and context should have a place in a child's reading instruction and 

one should not be sacrificed for the other. The view presented next--



1 4 

Phonological Recoding Reliance--favours decoding over context to 

facilitate children's skilled reading. Within this view, context is 

only minimally used to facilitate children's skilled reading. 

PHONOLOGICAL RECODING RELIANCE 

The phonological recoding view put forward by Stanovich (1980, 

1986) assumes that reading operates in a bottom-up fashion, "where 

sensory input is analysed initially at the level of perceptual 

features" (Snowling, 1987 p. 61 ). Word perception for a child begins 

with detecting each letter's basic features, such as the ascenders 

"b" and descenders "p." When the letters in a word have been 

identified by the child letter combinations corresponding to sound 

units can be identified, for example "ch." Next, these sound units 

are synthesised and those which match stored representations in 

the child's memory will be accepted, signalling word perception. 

Therefore, perception for the reader in this decoding process moves 

from low level feature information, to high level conceptual 

information in word identification. 

A prerequisite for bottom-up processing is phonological awareness. 

This is the ability to reflect on and manipulate the phonemic 

segments of speech. This involves, for example, the ability to 

segment the spoken word .b.a1 into it's corresponding phonemes lbl, 



1 5 

lal, and ltl. Both phonological awareness and spelling-to-sound 

mapping should develop symbiotically for a beginning reader, with 

phonological awareness giving rise and causing the growth of 

spelling-to-sound mapping. As a result, the child learns that a 

phoneme relates to a lexical visual representation in a particular 

manner. Therefore, phonological awareness and spelling-to-sound 

mapping should be in place "early" in children's reading 

development, as they affect " the ease and accuracy with which we 

can pronounce English words" (Parkin and llett, 1986 p. 23). As 

Stanovich (1986) has noted, the absence of these prereading skills 

can set up a chain of escalating negative educational side effects. 

Stanovich (1986) calls these negative educational side effects 

"Matthew Effects," where small deficits in a child's reading ability 

at the beginning of school can become large deficits after a few 

years. 

Research Investigations: Phonological Recoding 

A great deal of research has been carried out on phonological 

awareness as a result of claims about the primary importance of 

automatic word decoding in reading. Many investigations have also 

been carried out by researchers into: (1) children's short-term 

memory and reading comprehension; (2) children's short-term 

memory and the variable of context in word identification; and 



1 6 

(3) children's automaticity and attentional deficits. More recently 

reading-level match designs have been used to investigate children's 

utilisation of context and visual word identification. 

Several theories and models about children's phonological recoding 

have evolved from the research, for example, Perfetti's Verbal 

Efficiency Theory and Stanovich's Interactive Compensatory 

Model. Many researchers agree with Tunmer (1990a) that children's 

phonological awareness significantly influences reading 

development in the early stages where focus is primarily on the 

acquisition of decoding skills. The level of a child's phonological 

awareness, however, does not decrease in the later stages of 

learning to read when the emphasis shifts towards text level 

processes for example, comprehension monitoring while reading. 

Phonological awareness still exists within the reader but moving 

through the stages of phonological recoding and automatic word 

recognition reduces the beginning reader's reliance on phonological 

awareness. 

1) Memory: Short-Term Memory (STM) 

Research performed by Stanovich, Cunningham and Feeman (1984a) 

tested first grade children's general intelligence, decoding speed, 

listening comprehension and phonological awareness in relation to 



1 7 

early reading acquisition. They found as previous researchers have 

discovered that phonological awareness partially · determines the 

child's success at breaking the spelling-to-sound code, and 

therefore, the child's ability to identify unknown words. The 

decoding speed variables of how quickly and accurately children 

could name pseudowords explained an additional amount of variance 

beyond that for phonological awareness. This supports previous 

research carried out on children's speed of decoding in reading, 

which in addition to accuracy, is also important, due to short-term 

memory limitations. That is, slow and inefficient decoding would 

adversely affect comprehension processes in working memory. 

In the case of children's reading "word coding and sentence 

comprehension processes must be fast enough to avoid 

desynchronization because of memory deactivation" (Lesgold and 

Perfetti, 1978 p. 326). Lesgold and Perfetti hypothesised that a 

major source of individual differences in children's reading 

performance is slow inefficient decoding resulting in reduced 

execution of higher level comprehension. Therefore, for poor readers 

less short-term memory is available because of problems with 

word decoding. 



1 8 

2) Memorv: STM and Context 

Later research by Perfetti, Goldman and Hogaboam (1979) added the 

variable of context in word identification for good and poor readers. 

Perfetti et al. (1979) presented their test words in isolation, 

unrelated word lists and a story context for the children to identify. 

In the isolated word condition each word was presented to the 

children by a slide projector. In the list condition, a list of unrelated 

words was presented through headphones to the children prior to 

each target word. In the story context, the words were presented 

through headphones and the target words were presented at 

intervals visually for identification by the children. 

The results of Perfetti et al. (1979) research indicated that the 

story context facilitated word identification for both good and poor 

readers. A two-way interaction for reading skill by context 

indicated that context had a greater effect for poor readers' 

latencies, compared to good readers' vocalisation latencies. In the 

research by Perfetti et al. (1979) it seemed unlikely that the 

children's anticipation of specific words is the context mechanism 

that generates skilled reading. While context supports and aids word 

recognition, this decreases with increasing reading ability 

depending on the beginning reader's skill. However, if a beginning 

reader can make anticipations about words to come in a text, word 



1 9 

preactivation will occur. As a result, word identification reduces to 

word verification for the beginning reader in the context situation. 

Sometimes, a different word may be anticipated by the beginning 

reader and the verification procedure fails, which means that the 

word must be retrieved from memory. 

As a consequence, the subsequent drain on short-term memory 

capacity can cause a loss of comprehension for the poor reader. If 

the good readers are good at context-free word decoding as Perfetti 

et al. suggests, they should not be adversely affected by the 

misanticipation. Poor readers on the other hand should be affected 

by misanticipation because it is their "basic" word decoding 

processes which are not developed, and therefore, short-term 

memory becomes overloaded with word decoding and not used for 

comprehension. The good readers, according to Perfetti et al (1979) 

are so good at word identification while reading that context is of 

little consequence. The word recognition processes of the good 

readers are relatively "attention free," and executed more rapidly 

than contextual processes to eliminate lexical alternatives. 

Therefore, poor readers may fail to read well because too much 

attention is devoted to decoding words, which exceeds short-term 

memory limitations. 



20 

3) Automaticitv and Attentional deficits 

Curtis (1980) builds upon Lesgold et al's. (1978), and Perfetti et 

al's. (1979) research, by proposing that children who are poor 

readers may fail in part, due to a failure to develop automaticity in 

word recognition. The consequence of this lack of automaticity 

is a deficit in the amount of attention available by the poor 

reader for comprehension. In most word identification models, 

children's access to word meaning is achieved from graphemic 

encoding or via phonological coding. When a child's access to the 

word codes occurs automatically there is no competition with other 

reading processes for processing capacity. Instead, these word code 

sequences can be integrated into the child's memory. However, 

when decoding is not automatic, "competition among reading 

processes_ may result, and the probability that there will be enough 

capacity to complete all necessary processing decreases" (Curtis, 

1980 p. 657). 

Mutual facilitation exists between children's word decoding and 

phonological awareness and spelling-to-sound code breaking, as 

indicated by Stanovich, Cunningham and Feeman (1984) to develop 

automaticity. Even with children as young as 5 to 6 years of 

age evidence exists for the importance of verbal comprehension 

ability, a general ability that will increase in importance as the 



21 

child's reading skill develops. It is likely that phonological 

awareness underlies a child's ability to segment and· analyse speech, 

and this determines early success at decoding unknown words from 

breaking the spelling-to-sound code. It is important for children to 

develop automaticity in their reading ability because inefficient 

word decoding absorbs too much cognitive capacity and reduces 

reading comprehension. 

Research by Stanovich et al. (1984) supports Lesgold and Perfetti 

(1978) and Stanovich's (1980) reading theories, all of which place 

emphasis on the limited capacity of a child's short-term memory. 

Children must be able to recognise words rapidly, "in order to 

provide sufficient word meanings for comprehension processes 

operating on activated information in short-term memory" 

(Stanovich et al, 1984 p. 296). The child, as a result, has more 

cognitive capacity for reading comprehension processes, as 

automaticity develops in word recognition. The poor reader, 

unfortunately, allocates more cognitive capacity to word decoding, 

because of inadequate low level word decoding. 

The inadequate word decoding skills of poor readers according to 

Stanovich et al. (1984) and Stanovich (1986) generate "Matthew 

effects," leading to unrewarding early reading experiences. As a 

result, the poor reader becomes less involved in reading related 



22 

activities, and therefore, fail to fully develop their vocabulary. The 

lack of reading exposure and practice by the poor reader leads to a 

failure to develop automaticity and speed at the word recognition 

level. Slow and capacity draining word recognition processes ensue 

for the poor reader, requiring cognitive capacity which should have 

been allocated to comprehension. 

As a result, the poor readers reading for meaning is thwarted, 

unrewarding reading experiences multiply, and reading practice is 

avoided or tolerated without cognitive involvement. Therefore, 

according to Stanovich (1986), the poor reader becomes one of the 

reading "poor getting poorer" because of negative reading 

experiences, and consequently move into a downward literary spiral. 

To summarise so far, it has been documented previously that the 

speed and accuracy of a reader's word decoding and recognition is 

related to reading comprehension. The result is that the good readers 

have more cognitive capacity available for the higher level 

comprehension processes, due to automaticity in word recognition. 

The problem, as Stanovich et al (1984) suggests is ascertaining how 

context-free word decoding and contextual facilitation interact to 

provide rapid word recognition associated with skilled reading. 

Perfetti et al (1979), and West, Stanovich, Feeman and Cunningham 

(1983) found that when the reading contexts are understood by good 



23 

younger and poor older readers, the poor older readers showed as 

much context facilitation as the good readers. However, the poor 

readers may be compensating for their poor automatic word decoding 

skills by their longer experiences with reading. In order to reduce 

this confounding reading experience variable between children of 

different age groups, reading level match designs were applied. 

4) Reading Level Match Designs for Contextual Effects 

Stanovich et al. (1984a) proposed that there is a level of difficulty 

reached in a text where poor readers cannot identify enough words 

to sustain contextual facilitation. Good readers on the other hand 

continue to display facilitation effects, on account of their more 

efficient context-free word recognition processes. In a difficult 

text the poor readers contextual mechanisms have been rendered 

inoperative because of inadequate decoding processes for word 

identification. The question put forward by Stanovich was "when 

less skilled readers reach a certain level of context-free decoding 

efficiency, do they display as much contextual facilitation as 

skilled readers did when the latter were at the same level?" 

(Stanovich et al, 1984a p. 669). To answer this question a reading 

level match design was applied, where children from the middle 

school grades of 3 to 6 were selected for context effect in word 

recognition. In the reading level match design the older poor readers 



24 

were matched with good younger readers who were reading at the 

same developmental level. 

Stanovich et al's. (1984a) results indicated that when both age 

groups were at the same reading level, the poor readers derived 

just as large context facilitation effects on word recognition as the 

good readers. The results were derived from both good and poor 

readers use of coherent paragraphs, and when their decoding skills 

were similar for words out of context. 

Stanovich et al (1984a) suggests that a developmental lag exists 

between the good and poor readers developing subskills of reading. 

And that at a given point in time, a child is using contextual 

information to a degree commensurate with their decoding skills. 

The context use deficiencies, which occur for some poor readers is 

due to, inadequate context-free word decoding. The result is that 

poor word decoding degrades the proceeding contextual information 

available to the poor reader, and therefore, reading comprehension 

suffers. 

Further research by Stanovich, Nathan and Vala-Rossi (1985) applied 

a reading level match design to measure children's ability to use 

context to speed word recognition. Stanovich et al. (1985) examined 

the developmental lag model between third and fifth grade 



25 

children's individual differences in reading skill. The third grade 

children's data, and the scores for the fifth grade children on word 

naming in related and neutral contexts were highly correlated. 

However, unlike the data from the younger group, the fifth grade 

reading ability data was negatively correlated with contextual 

facilitation. The results indicated that the better readers displayed 

smaller context effects than the poor readers. This converges with 

other findings from Curtis (1980), and Stanovich et al (1984), where 

Stanovich suggests that a compensatory pattern exists for context 

use for the reading skill groups in each grade. Within the fifth grade 

group, the readers with poorer decoding ability were probably 

relying more on context to compensate for their less efficient 

bottom-up decoding skills. This compensation breaks down for poor 

third grade readers as their processing of contextual information is 

relatively inefficient and cannot compensate for poor decoding. 

The above evidence indicates that children's decoding skills are 

essential for skilled reading comes back to the point of phonemic 

awareness. Indirectly, if a child's phonemic awareness is low, .their 

decoding skills for words in a text will also be low, and as a result, 

their reading is slow and laborious. 

To summarise, research evidence is accumulating that phonemic 



26 

awareness "aids" children's early reading successes by enabling the 

child to discover how phonemes are related to graphemes (Stanovich 

et al., "1984, "1984a; Tun mer and Nesdale "1985). Unfortunately, for 

children who have poor phonemic awareness, they will have 

difficulty breaking the spelling-to-sound code and this delay may 

cause less exposure to text. The delay in spelling-to-sound code 

breaking may exacerbate the children's reading problems as 

they move into more difficult reading materials. The lack of 

exposure and practice with materials poor readers can read delays 

their attainment of automatic word-recognition. As a result, slow 

cognitive capacity draining processes are used by the poor reader to 

utilise context which should have gone to gaining comprehension 

from the text. 

According. to Stanovich ("1986), the good reader more rapidly attains 

a stage of proficiency where decoding skill no longer determines 

their reading level. The good readers word recognition becomes more 

automatic and less resource demanding. The only limiting factor for 

these children is their general language skills. As a result, good 

readers expend less cognitive capacity, not because they rely on 

context, but because their automatic word decoding processes are 

more efficient and powerful. This has been referred to by Perfetti 

("1985) as Verbal Efficiency Theory, which gives speed and 

automaticity of decoding and semantic access central roles in 



27 

comprehension failure for poor readers. 

5) Perfetti's (1985) Verbal Efficiency Theory 

Perfetti (1985) argues that individual differences in comprehension 

between readers' ability arise because of differences in the 

efficiency of low level processes. The Verbal Efficiency Theory 

assumes that the processes of decoding and comprehension compete 

for a limited amount of processing capacity in a reader's short-

term memory. Therefore, the less efficient a reader's decoding 

process is, the smaller the amount of information it can 

work on. The result is that fewer other comprehension processes 

can occur. One way a reader's word decoding and comprehension 

processes can be made more efficient is by being faster. Fast 

processes_ only need to remain briefly in short-term memory. Slow 

processes incur another disadvantage because the contents of short­

term memory has a limited time-span. The information that a slow 

process is working on may decay before the process has analysed it 

fully, and therefore reading comprehension suffers. 

The essential point here is that processing and storage in short­

term memory have to compete against one another. These two 

functions are a likely source of difference in reading comprehension. 

The more automatic decoding and semantic access processes of good 



28 

readers allows greater efficiency of short-term memory use. This in 

turn leaves more processing capacity for · comprehension 

processes."The fact that processing and storage in short-term 

memory have to be traded-off against each other is a potential 

source of difference in reading comprehension" (Oakhill and 

Garnham, 1988 p. 122). 

If poor readers fail to develop automatic word decoding skills to 

read efficiently, they will have less short-term memory capacity 

for reading comprehension processes. As a result, when more 

cognitive capacity is required for decoding by the poor reader, less 

memory space is available for comprehension processes. 

Furthermore, the slow decoding processes of poor readers may mean 

that words identified at the beginning of a sentence may be lost 

because of more reliance on context absorbing more short-term 

memory capacity. Therefore, the meaning of the sentence is lost 

before the child completes the sentence. This has been found by 

Stanovich (1986), where context facilitation is inversely related to 

the automatic word recognition skill of the reader. Stanovich (1986) 

has referred to this as the Interactive Compensatory Model. 

6) Interactive Compensatory Model 

Stanovich's (1986) Interactive Compensatory Model proposes that 



29 

when the bottom-up decoding processes for word recognition are 

deficient, the reader compensates by relying on context. As a result, 

"Reading skill is not determined by skill at contextual prediction but 

rather that the level of word-recognition skill determines the 

extent to which contextual information will be relied on to complete 

the process of lexical access" (Stanovich, 1986 p. 370). 

When poor readers encounter difficult materials, their slow and 

inaccurate word decoding processes may in fact degrade the 

contextual information they receive. Thus, comprehension fails not 

because of the readers over-reliance on decoding, but because their 

decoding skill is not developed enough. The child's word decoding 

skill weakness may "intensify" as they progress through each school 

year, with the reading materials becoming more difficult each year. 

If the development of a child's vocabulary knowledge facilitates 

reading comprehension and if reading itself leads to vocabulary 

growth, which in turn, allows more efficient reading, then a 

reciprocal relationship exists. This reciprocal relationship between 

vocabulary growth and reading should continue to develop further 

growth in reading throughout a reader's development. Unfortunately, 

the problem for the poor reader is that a good deal of knowledge 

acquisition occurs in the vocabulary development process. As a 

result, the poor readers vocabulary knowledge base would be less 



30 

developed because of less reading experience. Conversely, the 

effect for good readers who are reading well and have good 

vocabularies will read more, learn more word meanings, and as a 

result read even better. The growth in vocabulary from increased 

reading volume by the good readers, plus their skill differences in 

reading could mean that they are the "rich-getting-richer" which 

Stanovich (1986) describes. 

The skill difference indicated by Stanovich (1986) could arise from 

a cumulative advantage cycle embedded in the course of the reading 

process for the good readers. Unfortunately, the reverse process 

applies to poor readers with poor vocabularies who read slowly and 

without enjoyment leading to the child reading less. As a result, 

inhibition occurs on the poor reader's vocabulary growth, generating 

a cumulative disadvantage cycle. Ellis and Large (1988) link together 

the utilisation of short-term memory and the acquisition of reading 

skills and provide a time scale with which to observe positive or 

negative cycles of achievement as Stanovich (1986) suggests exist. 

7) Short-term memory and the Acquisition of Reading 

Ellis and Large's (1988) research tested forty children from the age 

of five to seven years old, as they learned to read. Each year the 

children were tested for ability on 44 variables measuring: reading, 



31 

spelling, vocabulary, short-term memory, visual perception and 

discrimination, auditory-visual integration, language knowledge, 

phonological awareness, grammatical knowledge, rote knowledge, 

ordering ability and performance on the Weschler intelligence scale. 

Over the first three years of a child's reading acquisition a blending 

of knowledge occurs. The child's acquisition process begins with a 

blend of knowledge about the visual characteristics of letters of the 

alphabet, phonological awareness and visual short-term memory 

processes. Thereafter, the child's reading acquisition develops in 

association with knowledge of the sounds of language, the 

correspondences between sounds and visual patterns, skills for the 

analysis of visual patterns and syntactic skills. 

The correlational analyses performed by Ellis and Large for the early 

stages of. children's letter acquisition found that the predictors are 

phonological skills, such as, phoneme segmentation, sound blending, 

and the part-whole relationship; and that these previous skills in 

letter recognition predict later levels. However, phonological 

skills predict children's letter recognition, and not the reverse. As 

a result, phonological awareness facilitates the child's acquisition 

of letter-recognition and breaking the spelling-to-sound code. This 

view by Ellis and Large supports Stanovich's (1986) claims 

concerning children's progress in acquiring the basic skills for 

beginning reading. 



32 

Ellis and Large found that reading ability is related to children's 

short-term memory span. From the ages of five to six years, the 

child's short-term memory skills of word and sentence spans 

develop symbiotically with reading. From six to seven years the 

interdependence has increased, and reading is a better predictor of 

children's short-term memory skills. The development of basic 

reading skills allows the child to attend to, manipulate and 

remember the sounds of words and sentences. The child continues to 

build up the ability of making sense out of sentences by the age of 

seven, "where word order and syntax are paramount" (Ellis and 

Large, 1988 p. 67). Ellis and Large propose that phonological coding 

and short-term memory are important components when children 

begin to learn to read. 

Summary_ 

The Phonological Recoding View of children's beginning reading 

includes several important facets of reading. The first, is that the 

development of phonological awareness is important in helping 

children break the spelling-to-sound code. Without this basic ability 

the child will not be able to advance in reading and thus acquire the 

spin-off skills of reading, which aid in the effective utilisation of 

short-term memory skills in reading. Without the basic word 

recognition skills in place early, automaticity in children's reading 



33 

is not possible. 

As a result, the child's lack of automaticity may overload short­

term memory capacity in trying to decipher words. This takes up 

greater cognitive capacity that should be applied to comprehension. 

The poor readers, according to the Phonological Recoding View, 

use context to identify words. If the poor reader is slow at word 

decoding, the reader loses cognitive capacity in word recognition, 

which in effect causes the child to lose the first words of the 

sentence from memory, and therefore, comprehension. As a result of 

a lack of automaticity in word recognition, a negative cycle of 

achievement effects can occur for the poor reader, called "Matthew 

effects." 

However, context can facilitate both good and poor readers, when the 

materials being read are within the readers experience level. The 

view put forward next is that a cooperative mechanism exists 

between contextual use and phonological decoding in children's 

beginning reading. 

DECODING-CONTEXT ALLIANCE 

The essential view of this alliance is that both context and decoding 

are required for a child's success in reading, but context use 



34 

decreases as decoding becomes more automatic for the child. As a 

consequence of the development of word automaticity from context 

use to derive words of intermediate familiarity, comprehension of 

text increases for the child. 

Reading, whether by good or poor readers, is a combination of 

ongoing decoding and context usage. However, the ongoing 

development of a child's reading ability to become a proficient 

reader is a "complex process involving the integration of 

grammatical, phonological and short-term memory skills" (Siegel 

and Ryan, 1988 p. 28). Fluid and rapid reading involves, not 

necessarily at a conscious level by the reader, the identification of 

the class of word that is likely to appear in the text. The beginning 

reader needs to understand the basic letter-sound correspondences 

for decoding words effectively. Memory capacity is also required by 

the reader to gain comprehension from the words and sentences 

which have been read. 

The total reliance on either decoding or context for efficient reading 

will be ineffective for the beginning reader. Teaching children to 

only phonologically recode words may restrict their progress in 

reading. As a result of an over reliance on decoding, the child may 

be restricted to pronouncing homographic spelling patterns (patterns 

that have different pronunciations in different words) in the same 



35 

way, for example, the words ~ like ~ and ~ like b.Qn.e.. If this 

is left unchecked incorrect word pronunciations will result, which 

will retard the reading development of the beginning reader. Unlike 

beginning reader's who possess knowledge that some letter 

sequences have different pronunciations by using sentence context, 

depending on the word in which the sequence exists. The poor reader 

will persist in pronouncing polyphonic letter sequences in the same 

way, for example, own in the words flown and clown. 

Likewise, the total reliance on context by children to identify 

unfamiliar words will also result in little progress. "The words 

which can be predicted from context by a reader are typically high 

frequency function words, e.g. Qll, ~. iQ (Tunmer, 1990b p.199) and 

not meaningful content words. Also, the "over-stressing of context 

as a cueing system for reading is, in fact, unhelpful for beginning 

readers who do not have the wide background understanding 

and word knowledge of fluent readers" (Wray, 1989 p. 5). The 

poor beginning reader who is made to rely on context to identify 

words will make little progress, as the words they can predict from 

context will be words they already recognise. Another problem 

suggested by Konopak (1981), especially for poor readers, is the 

failure of some beginning reader texts to provide sufficient clues 

for meaning. 



36 

The view put forward by this study is that children who are in the 

process of acquiring grapheme-phoneme correspondences cannot 

recode all of the unfamiliar words they encounter completely. 

Contextual facilitation may influence children's phonological 

recoding development by combining the knowledge of the constraints 

of sentential context with incomplete phonological knowledge which 

aids in the identification of unfamiliar irregular words and 

polyphonic sequences. As a result, each new word correctly 

identified by a beginning reader would increase their knowledge of 

letter-sound correspondences. 

Tunmer, Nesdale and Wright (1987) follow a similar line of 

reasoning by applying syntactical awareness tasks, specifically oral 

cloze to help beginning reader's identify unfamiliar words. It is 

important, however, at this point to distinguish syntactical 

awareness from contextual facilitation. Contextual facilitation has 

the target word present in the text for the reader to identify. In 

syntactical awareness, the reader produces the word that fits the 

text. In applying syntactical awareness tasks, such as oral cloze to 

words with homographic spelling patterns can aid the reader in word 

identification. For example, in the identification of polyphonic 

sounding words like couQh (which can be pronounced like bouQh, 

rough or dough) by the beginning reader in a sentence, such as: "The 

man had a bad ..... " "When beginning readers apply their limited 



37 

knowledge of the letter-sound correspondences to such words, the 

result may often be close enough to the correct form that context 

can be used to arrive at a correct identification" (Tunmer, Nesdale 

and Wright, 1987 p. 26). But also, when arriving either syntactically 

or contextually to pronounce a target word the beginning reader may 

learn that some spelling patterns have two or more pronunciations. 

Children should also be encouraged to use a combination of 

strategies to identify new words they do not know when reading. 

For example, in the sentence "He put the cake on the ..... " (Garton 

and Pratt, 1990 p. 207). In this sentence the context does not 

provide enough information for the child to be sure of the correct 

word to complete the sentence. The sentence context may suggest to 

the child that the word "table" is a possibility. The child may then 

look at the word for clues to see if it starts with a "t", and as a 

result, decide that the word is table. The use of this letter­

clue technique may be useful when children encounter irregular 

words in context. Usually, irregular words, such as "stomach" do 

contain some clues to their phonetic representation for the reader. 

Tunmer (1990's) provides the example of the word "yacht" as an 

irregular spelling. The first and last letters of "yacht" provide 

clues to the phonemic representation of the word for the reader. 

When confronted with an irregular word, if the "beginning readers 



38 

apply their knowledge of grapheme-phoneme correspondences to 

such words, the result will be close enough to the correct form that 

sentence context can be used to arrive at a correct identification" 

(Tun mer, 1990a p. 101 ). If a child has difficulty identifying the 

word "yacht" in isolation, a context provided by other words in a 

sentence, such as "The man was sailing his ..... " (Garton and Pratt, 

1990 p. 208) will enable the child to decode it. The child could then 

put together his or her knowledge about the world (people sail boats 

that are often called yachts), with their knowledge of grapheme­

phoneme correspondences, for example, knowing what the first 

letter in the irregular word "yacht" corresponds to. 

Therefore, learning to read is a two-way process for children, in 

which sentence context and word decoding apply. The child should be 

able to combine the rules of grapheme-phoneme correspondence to 

extract sound cues, and using their knowledge of language and the 

world to find words that match these cues. This two-way process 

between context and decoding cannot occur for the child if the 

words are encountered in isolation. Accordingly, reading materials 

provided for children must be contextualised and meaningful, while 

at the same time generating opportunities to learn grapheme­

phoneme correspondence rules. 

As a consequence of a child's learning to read process, decoding 



39 

helps access meaning, but equally, consideration of meaning assists 

the decoding process. This two-way interaction is consistent with 

Tunmer's (1990b) hypothesis in the learning to read process, where 

syntactic awareness plays an important role for children in learning 

to recode words. To illustrate this point, for example, consider 

again the sentence "The man sailed his yacht." The process of 

decoding the last word "yacht" may proceed two ways. One child may 

know the correspondence rule for the letter "y" and utilise the sound 

this letter makes to search through words that the child knows to 

find one that fits the sentence context. In carrying out this lexical 

process the child finds the word yacht. A second child having 

difficulty with the rules of letter correspondence may use the 

context of the sentence to look for a word that fits. The child may 

come up with the word "boat," but realises this does not start with 

a "y," as_ "y" does not sound like "b." This may trigger another 

search in the child's memory, until the child finds the word "yacht." 

Yet, even this process will not yield correct answers every 

time as children will invariably make mistakes. Usually, 

mistakes occur from a child's over-reliance on the use of contextual 

strategies and a lack of consideration for the grapheme-phoneme 

structure of the word for identification. In other instances, it 

may be that combinations of phonological recoding skills and 

contextual cues will still lead to an incorrect word identification by 



40 

the child. For example, in the sentence "After school Steven went 

back to David's house," a child could read the final word as "home," 

from using context cues and their knowledge of the sound 

corresponding to the letter h or letters .b,Q.. It is important to 

distinguish that this type of contextual facilitation and letter-sound 

knowledge differs in this study to that proposed by Goodman and 

Smith. 

Goodman and Smith argue that the use of context to predict words is 

the primary factor in "ongoing" sentence processing for children. In 

contrast, the view proposed in this study is that the ability to 

reflect on sentence context combined with ongoing phonological 

recoding is essential for children to acquire word recognition skills 

and extend their automatic word recognition range. This decoding­

context al.liance is supported by Tunmer (1990a) who stresses that 

total reliance on context to identify unfamiliar words by children 

will result in little progress in reading. As noted earlier, words that 

can be correctly predicted from context are high frequency function 

words, which can be easily recognised by children. 

The errors children make when reading can also provide useful 

information about context or decoding strategies used for 

word identification. Consistent error patterns may indicate that 

a child is relying too much on one type of reading strategy. On the 



41 

one hand, the child who frequently reads words which fit the context 

but have no relationship to the graphemic structure will only be 

using context. On the other hand, a child who "reads" words that 

contain a sound that relates to one of the graphemes, usually the 

first, but has no relationship to the context, may be relying on 

limited-knowledge of correspondence rules, basing word 

identification on the initial sound, and ignoring the context. 

It is advanced in this study that total reliance on either phonological 

recoding or context strategies to identify words for a child will 

result in little progress. As a consequence of this over reliance on a 

particular strategy, the child who is a poor reader may be greatly 

disadvantaged in further education. Stanovich (1986) suggests that 

good readers recognise unfamiliar words through more efficient 

bottom-up_ word decoding processes in order to comprehend text, and 

not by using top-down context. Poor readers according to Stanovich 

(1986) take longer to read, due to poor decoding skills, and rely 

more on context for help. Consequently, the poor reader who does 

rely on context to read will require more cognitive effort, and as a 

result, comprehension of the text will decrease. Stanovich (1986) 

stresses that reading skill is not determined by skill at contextual 

prediction, "but rather the level of word recognition skill 

determines the extent to which contextual information will be 

relied on" (Stanovich, 1986 p. 370), and complete the process of 



42 

lexical access. 

The view proposed in this study is that the use of both context and 

decoding by children facilitates children's beginning reading. As a 

consequence, good readers may use context to improve their 

decoding and recognition of words of intermediate familiarity. These 

intermediate words then move into the category of automatic 

recognition, and as a result, leaves more cognitive processing 

capacity for comprehension. For good readers, the use of context to 

aid reading should decrease as more automatic cognitive processes 

take over. In contrast, poor readers who can use context less 

efficiently to recognise words of intermediate familiarity, may lag 

behind the good readers. However, eventually the poor readers may 

reach a similar point of context use to identify words, perhaps two 

or three y_ears later. 

Preliminary Research with Irregular Word Testing 

The research design used in this study is an experimental type, 

modifying contextual facilitation research performed by Adams and 

Huggins (1985) in the following ways: 

(1) The order of words from the Adams and Huggins Isolated Word 

Test were randomised in order to break up the easy to difficult order 



43 

and thus hold the child's attention. (2) Cards were used with only six 

words per card, instead of a list of fifty words in the presentation 

to the children. Also, the children actively participated by moving a 

blank card down each test card to reveal one word at a time to 

focus their attention. (3) High Frequency words words were included 

in the Isolated Word Test, in order to provide the children with 

words they could readily identify, and thus hold their attention. (4) 

The contextual Facilitation Test was presented in a similar fashion 

to that of the Isolated Word Test except that sentences were used. 

Adams and Huggins compared the abilities of good and poor readers 

to read a frequency-graduated series of irregularly spelled words. 

Two word identification test format presentations were used to 

test the children's sight vocabularies. The first test format 

consisted of irregular words presented to the children in isolation 

while the second test format consisted of the same irregular 

words presented in a sentence context to the children. 

Adams and Huggins assumed that a reader's knowledge is 

organised hierarchically such that the output of any level of 

processing is input for the next. In this way the reader extracts 

information from the written page. For the mature readers 

information is moved progressively upwards from a words visual 

detail through comprehensive levels of interpretation which 



44 

corresponds to the information flow in bottom-up word processing. 

Top-down word processing occurs as the readers processing system 

searches for information to satisfy partially activated 

higher level word knowledge complexes. In the case of the mature 

reader this top-down word processing generates automatic 

priming of lower level word identification complexes. 

In order to manage these automatic word processing· complexes, 

schema-theory models have adopted the concept of a central 

limited-capacity processor from human information processing 

theories. The central processor is responsible for setting the 

interpretive goals of the system, and "the proportion of attentional 

capacity allocated to higher order dimensions determines whether 

and how text will be understood. The proportion allocated to problem 

areas in . the system determines whether and how they will be 

overcome." (Adams and Huggins, 1985 p. 263). 

However, this view of attentional capacity by Adams and Huggins 

(1985), depends on how one imagines the interplay between the 

reader's relevant contextual knowledge and decoding skills occurs. 

This interplay between contextual and word decoding skills in turn 

determines how much attention the reader allocates to these sub­

tasks of reading. The prediction arises that either good readers 

profit more from context to read text or poor readers profit more 



45 

from context to read text. 

The first prediction arises from the good readers having more 

elaborate ingrained knowledge and skills compared to those of the 

poor readers. Specifically, it is suggested that because the good 

readers have higher levels of sophistication in the syntactic and 

semantic relationships of the text they should have greater 

sensitivity to contextual cues in texts than the poor readers. In 

addition, the good readers should be more adept at letter and 

word recognition, and 

capacity should be 

comprehension. 

as a result, more cognitive 

available for context clues 

processing 

for text 

The second prediction is that poor readers should gain more from 

context to. comprehend text. This arises from the premise that poor 

readers are such poor decoders they should gain more from context 

while reading. The poor readers, however, can use top-down 

syntactic and semantic support for their uncertain bottom-up 

encoding of a text's visual dimensions. As a result, this type of 

context use by the poor reader may compensate for their decoding 

difficulties. Conversely, it may also be argued that the automatic 

word recognition skills of the good readers are so efficient in 

reading without context, there is little room for improvement with 

context. 



46 

Adams and Huggins (1985) research design used children from grade 

two to grade five to test the two predictions· about context 

facilitation. All of the children were native speakers of English, and 

none were classified as dyslexic by the schools. Adams and Huggins 

(1985) found it impossible to equate IQ scores across reading 

abilities, they therefore, excluded children whose IQ scores were 

below 100, or above 125 to improve the matching of good and poor 

readers. To verify the IQ score Adams and Huggins (1985) 

administered the information, vocabulary, picture arrangement of 

the WISC to each child. If the WISC fell below 80 or above 130, the 

child was dropped from the sample. Stanford and Gates-McGinnitie 

reading comprehension scores were also obtained for each child. 

Children who scored below the 4th Stanine on both tests were 

classified as poor readers. Those who scored within or above the 

5th stanine on both tests were classified as good readers. 

In the study by Adams and Huggins (1985), using the frequency 

graduated list of 50 irregular words (formatted from easy to 

difficult in Appendix A.), it was found that the children's 

performance declined with increasing word difficulty. The o~der 

good readers penetrated further into the isolated irregular word test 

than the younger and poorer readers. An ANOVA for the main effects 

of both grade and children's reading ability was highly significant 

and the interaction between grade and reading ability was also 



47 

significant. 

The identification of the isolated irregular words in a sentence 

context specific for that word was the next test for the children. 

The sentences were designed to provide moderate priming for the 

children, but not deterministic priming for the target word. The 

principle difference between the isolated irregular word test and 

the context facilitation test was quantitative. The children were, as 

a rule, able to read more of the list of irregular words with context. 

An ANOVA was performed on the children's results for grade by 

ability by test for the context condition. The effects of the test or, 

equivalently context, were significant, as were those of grade and 

children's reading ability. 

The results of the tests performed by Adams and Huggins (1985) 

indicate that: ·(a) There are large differences in the depth of 

children's sight vocabularies associated with age, or amount of 

schooling and with reading ability. (b) The presence of meaningful 

context is a potent aid to word recognition regardless of age and 

ability. 

The data Adams and Huggins obtained would support the latter part 

of the prediction, in which, poor reader's should profit more from 

context to read text. This suggests congruous context facilitates 



48 

word recognition and that word recognition is inversely related to 

children's reading ability. One must take into account that the 

irregular words were chosen by Adams and Huggins (1985) to be 

within the expected familiarity range of the poorest readers in the 

sample. Consequently, the results obtained for the good readers 

would suggest that more of the irregular words would be more 

familiar and easier to identify without context. As a resu It, the 

good readers' results may be interpreted as a decrease in 

sensitivity to context with increasing reading ability. This may in 

fact, reflect an ability-related increase in subjective familiarity 

with the target words. The data reported by Adams and Huggins 

(1985) does not permit the conclusion that either good readers 

should gain more from context or the converse. This is possibly due 

to, contextual facilitation test presentation biases which may have 

facilitated the good readers' performance because of their better 

word decoding skills to identify the target irregular word. 

However, Adams and Huggins (1985) do propose that a stage of 

sight word acquisition exists where three stages occur for a word to 

attain sight word status for children. The first stage is where the 

words observed lacked any useful internal representation of their 

orthography for the child even when context was supplied. The 

second stage is transitional, where children cannot recognise the 

target irregular words in isolation but do so when context is 



49 

supplied. The most sophisticated stage occurs last, when the child 

can recognise the isolated words without the aid of context. In this 

stage word recognition is automatic and it can be assumed that the 

word is securely represented in the child's visual lexicon. 

Overall, the results Adams and Huggins obtained would tend to 

support the Lesgold and Perfetti (1978) view, which emphasises 

that the processes involved in word recognition are the most 

troublesome for beginning readers. Specifically, Lesgold and Perfetti 

argue that the processes involved in word recognition are slow and 

effortful, and may themselves, place limits on a child's short-term 

memory. The possible short-term memory overload for a child may 

result in the disruption and displacement of context for 

identifying target words. The short-term memory overload factor is 

an important confounding variable which the present study hopes to 

reduce, by using a verbal-visual context test format for presenting 

the sentences (context facilitation test) to the children. 

Methodological Issues 

Several important methodological issues arise from previous 

experimentation in contextual facilitation and syntactical 

awareness. However, an important distinction must be made between 

both research area's: (1) Contextual facilitation deals with the 



50 

actual target words being present in the texts the subject reads. (2) 

Syntactical awareness requires the subject "to reflect upon and 

manipulate aspects of the internal grammatical structure of 

sentences" (Tunmer, Nesdale and Wright, 1987 p. 25). Although, 

syntactical awareness follows a different aim to the one followed 

in this study, research into syntactical awareness provides 

important guidelines. 

An important confounding test factor of the Adams and Huggins 

(1985) research which also applies to syntactical awareness tests, 

has arisen from the presentation of materials to the subjects in a 

written mode rather than an oral mode. As a result, "differences in 

measures of syntactic awareness presented in a written mode maybe 

due to decoding abilities of good and poor reader's" (Tunmer, 

Nesdale and Wright, 1987 p. i 80). Therefore, beginning reader's 

who are already having trouble identifying words in a text for a 

contextual facilitation or syntactical awareness test, will also have 

trouble organising these words into larger structural units. 

Another confounding factor is the use of sentences in 

contextual facilitation tests which are too long for the beginning 

reader's to retain in memory. This procedure would potentially 

reduce the poor reader's ability to segment and decode words in 

order to reach the target word. As Tunmer, Bowey and Grieve (1983) 



51 

suggest in their research into young children's awareness of the 

word as a unit of spoken language (which also applies to this study) 

where the use of long sentences may confound some children's 

memory ability to identify words. In effect, the poor readers efforts 

to decode preceding words to the target word would degrade the 

overall context effect from short-term memory overload. As a 

result, "the magnitude of semantic priming effects in visual word 

recognition may be a rather poor indicator of real ability to use 

contextual information" (Bentin, Deutsch and Liberman, 1990 p. 

148). 

A relatively unbiased method for comparing good and poor readers' 

contextual facilitation is to reduce biased context effects, for 

example, eliminating the need to decode print, as in verbal 

presentation, and forcing all readers regardless of skill to use 

context to the same extent. Therefore, it was decided in this present 

study that all the readers would have the context test presented in a 

verbal-visual fashion. The purpose for this context test modification 

is to help reduce the children's decoding deficits and short­

term memory overload. 

Summary 

The review began by considering the two opposing views regarding 



52 

contextual facilitation and children's successful reading: Semantic 

and Syntactic Redundancy; and Phonological Recoding Reliance. A 

third view was proposed from a synthesis of the two preceding 

views. This was referred to as the Decoding-Context Alliance in 

children's successful beginning reading. 

According to the psycholinguistic view, children skilled in reading 

mainly utilise context and rely less on the phonological, visual or 

orthographic cues provided by a text. Children read by a process 

which, according to Goodman and Smith, is interactive between 

language and thought. The good readers are thought to pay little 

attention to the bulk of words of a text because the flow of language 

follows a predictable pattern. Therefore, reading is a natural 

process for children which is incidental and characterised by a 

series of. guesses. Goodman and Smith indicate that too great a 

dependence on graphic information by children when reading can 

impede their comprehension in reading. As a result, the primary goal 

of making meaning from the words in context while the child reads 

is defeated. 

The alternative view proposed by Stanovich (1980, 1986) suggests 

that the processes of reading for children operate in a "bottom-up" 

fashion, instead of a "top-down" fashion as in the 

"psycholinguistic guessing game." In the "bottom-up" process, 



53 

sensory input is analysed by the reader at the level of a word's 

perceptual features. 

The necessary prerequisites for children's reading according to 

"bottom-up" processing are: (a) phonological awareness, and (b) 

spelling-to-sound mapping. Both of these factors provide the basis 

with which beginning readers can accurately pronounce words. 

This literature review outlined the extensive research that has been 

carried out regarding good and poor readers automatic word 

decoding in reading. This research indicates that it is important to 

develop children's word automaticity according to the phonological 

recoding view, for two reasons: (1) Speed in decoding words is 

important because of the notion of limited capacity of a reader's 

short-term_ memory; (2) the result of automatic decoding is that 

cognitive capacity becomes free for allocation to comprehension 

processes in a child's reading. 

Inadequate word decoding skills can lead to unrewarding reading 

experiences for children and a lack of attainment in reading 

according to Stanovich ("1986). The lack of reading practice which 

can occur for poor readers can lead to failure to develop 

automaticity and speed of word recognition. Instead, the poor reader 

uses slow context processes for word identification, thus draining 



54 

cognitive capacity which should have been allocated to 

comprehension in reading. 

Stanovich (1986) suggests that a developmental lag may exist for 

poor readers, whereby they do reach a similar level of context free 

word decoding but are delayed by two to three years. These poor 

readers however, would have some basic word decoding skills and 

would not necessarily fall into the negative cycle of achievement 

where "Matthew effects" occur in reading. 

The view put forward by this study represents an alliance of both 

the decoding and context views of reading. Good and poor readers use 

a combination of ongoing context and word decoding in reading 

depending on the children's reading ability. Total reliance on either 

phonological recoding or context will be ineffective for the 

beginning reader. (a) Total reliance on phonological recoding to read 

would restrict children to pronouncing polyphonic words in the same 

way causing incorrect word pronunciations and a restriction in the 

beginning reader's development. (b) Total reliance on context to 

identify words will also restrict children's progress in reading as 

the words that can be predicted from context are high frequency 

words, and not content words. 

The alliance view suggests that children who are in the process of 



55 

acquiring grapheme-phoneme correspondences in reading will not 

be able to recode all unfamiliar words comple·tely. Contextual 

facilitation may influence the children's phonological recoding 

development by allowing them to combine knowledge of the 

constraints of sentential context with incomplete phonological 

knowledge, to identify words of intermediate familiarity. Words of 

automatic familiarity can be identified without context because the 

child's automatic word decoding is so good. Words which are beyond 

the stage of intermediate familiarity are not represented in the 

child's memory for automatic recognition, and context will not aid 

the words identification. 

When a beginning reader applies their limited knowledge of letter­

sound correspondences, the result may be close enough for context 

to provide the correct word identification for words of intermediate 

familiarity. As a result, the intermediate word moves into the 

child's automatic recognition stage which leaves more cognitive 

capacity to go to comprehension and the identification of difficult 

words beyond the intermediate stage. 

The consideration of whether children rely more on the context 

strategy or the word decoding strategy in beginning reading is 

problematic. These problems have been indicated in Adams and 

Huggins (1985) research study, where irregular word identification 



56 

was used to test whether good readers used context more than poor 

readers (Irregular words were used because they dci not conform to 

normal spelling-to-sound patterns, and their correct reading in 

isolation would indicate that they were represented in the readers' 

memory.) Adams and Huggins results suggest that the reverse may 

apply, where poor readers derive more context effect to identify 

irregular words than good readers. At this point, the important 

methodological issue of only using the context in a visual format for 

children to identify irregular words applies. The confounding 

element of preceding word decoding overload in children's short­

term memory capacity to identify target irregular words in 

sentences applies in only the visual context. 

Overall, the intent of this study is to reduce: (1) Children's short­

term mem_ory overload by applying context in a verbal-visual format 

to identify irregular words. (2) Preceding decoding of sentence 

words before the target irregular word from the verbal-visual 

format, with the aim of providing a more accurate context 

facilitation measure between good and poor readers. The other 

modifications to the Adams and Huggins (1985) research method by 

this study will be discussed in chapter three (method). 



57 

Research Questions 

The following research questions arise from the review of the 

literature: (i) What is the relationship between contextual 

facilitation effects and word decoding ability? 

(2) How does accuracy of recognising irregular words of varying 

difficulty change as a function of age and ability when the words are 

presented in isolation and when they are preceded by context 

presented in a verbal-visual format? 

Tentative Hypothesis 

It was hypothesised that prior context can increase children's word 

recognition accuracy but only under certain circumstances. The level 

of context usage decreases from a high degree in children's 

beginning reading to a lower level as more automatic word 

recognition processes take over. In effect, the good reader's 

vocabulary moves through a stage pattern of word recognition, 

specifically in the intermediate word difficulty range. As these 

intermediate words move into the automatic word recognition stage 

of the child's vocabulary, context is relied on less. The lower 

reliance on context for word recognition allows cognitive capacity 

to become available for comprehension. 



58 

On the other hand poor reader's appear to use context less to read 

because of their poorer decoding skills in word identification. The 

poor readers context use may increase at a slower rate to a peak 

similar to that achieved by the good readers for words of 

intermediate difficulty and then decrease in a developmental lag 

two to three years later. 



SUBJECTS 

CHAPTER THREE 

METI-DD 

59 

The study was carried out on a urban Primary school with a role of 

approximately 250 children, consisting primarily of middle class 

Caucasians. The total number of children in the study was 113 (mean 

age= 7.56 years, S.D= .83 years). Three grades were selected from 

the school, and as many children tested as possible from each grade 

complement consisting of: 38 children from Junior 2 (mean age= 6.65 

years, S.D= .29 years); 39 children from Standard 1 (mean age= 7.35 

years, S.D= .36 years); and 36 children from Standard 2 (mean age= 

8.54 years, S.D= .35 years). The overall sample of children included 

only those who had permission from their parents, and did not 

include any children who did not choose to participate. As a result, 

the children who took part in the study were selected in a non­

random fashion. All of the children who participated in the study 

took part in all of the tests for word recognition. 

MATERIALS 

(a) Peabody Picture Vocabulary Test-Revised (PPVT-R) 

Form M of the PPVT-R by Dunn and Dunn (1981) was used to 



60 

measure the children's "receptive vocabulary," and provides a 

verbal estimate of their overall intelligence. The· child's date of 

birth determines the starting position on the PPVT-R test. The test 

consists of 175 pictorial plates, each containing four pictures. As 

each plate is presented the examiner verbally provides a stimulus 

word, and the child responds by indicating the picture that best 

illustrates the meaning of the word. The PPVT-R test is designed for 

any individual who can hear stimulus words, see drawings and 

communicate a response. 

(b) BURT Word Reading Test 

The BURT Word Reading Test, New Zealand revision by Gilmore, 

Croft and Reid (1981) was the first word test administered to the 

children. The BURT test is an individually administered measure of 

an aspect of a child's word recognition skills. In its original form 

the BURT test consists of 11 O words printed in differing sizes of 

type, and graded in difficulty from easy to difficult. The child is 

asked to read as many words from the test as possible, until 1 O 

consecutive words are read incorrectly. At this point, the child is 

given the opportunity to look at the remaining words to see if they 

can identify any others. Consequently, the test provides a broad 

estimate of the child's reading achievement. A copy of the BURT test 

sheet in its original form is provided in Appendix C. 



61 

The BURT test was modified in this study from its original format 

for testing the children in the following ways. The presentation 

format was altered from all the words being presented at one time 

to the child, to five words at a time on separate cards. The test 

cards were presented to the child one at a time, beginning with the 

easy words and finishing with the difficult words. An example of one 

of the modified test cards is provided in Appendix D. Each test card 

had all of the words presented in the same type and size of print. 

The reason for this modification was to prevent a drop off in the 

child's attention to the words in the test. The child's lack of 

attention to the test words can arise from being confronted with 

words of increasing difficulty and decreasing size. 

(c) Isolated Word Test 

The original form of the Adams and Huggins (1985) Isolated Word 

Test, consisted of a list of 50 words with irregular spelling-to­

sound correspondences. The test provides a measure of a child's 

knowledge of irregular words of decreasing frequency. The 50 words 

were grouped in blocks of 1 0 words each of decreasing frequency 

(and therefore increasing difficulty). 

The format of the original Isolated Word Test was modified in 

several subtle ways. First, 26 high frequency words from the 



62 

"Clay Test" (Clay, 1979 p. 33) were randomly selected by computer 

and ordered for inclusion in the Isolated Word Test. (A copy of the 

high frequency word list is provided in Appendix F). However, unlike 

Adams and Huggins (1985) the order of the irregular words was 

randomised and interspersed at two irregular word intervals with a 

high frequency word. These words were included to provide words 

that the children could readily identify. The format for the Isolated 

Word Test began and finished with a high frequency word, 

providing a total of 76 words for the children to identify. 

(d) Contextual Facilitation Test 

The Adams and Huggins (1985) Contextual Facilitation Test 

consisted of the same 50 irregular word from the Isolated Word 

Test, but. each was presented as the last content word of a 

meaningful sentence. The sentences in the Contextual Facilitation 

test, unlike those in the Adams and Huggins (1985) study were 

presented in a verbal-visual context to the children. Again, cards 

were used to present the test sentences to the children. Five 

sentences were presented on each card, with the irregular target 

word as the last or second to last word. The irregular word order 

followed the same pattern as in the Isolated Word Test. 

Each sentence was presented verbally to the children by the 



63 

experimenter, who pointed to each word that