Title Page
 Table of Contents
 List of Tables
 List of Figures
 Review of the literature
 Appendix 1: Scoring key for Berry-Talbot...
 Appendix 2: Means, standard deviations,...
 Appendix 3: Summary of Bonferroni...
 Reference notes
 Biographical sketch

Title: Developmental changes in the linguistic performance correlates of reading disability
Internet Archive: http://www.archive.org/details/developmentalcha00flet
Full Citation
External Link: http://www.archive.org/details/developmentalcha00flet
 Material Information
Title: Developmental changes in the linguistic performance correlates of reading disability an evaluation of a theory
Physical Description: ix, 118 leaves : ill. ; 28 cm.
Language: English
Creator: Fletcher, Jack McFarlin, 1952-
Copyright Date: 1978
Subject: Reading disability   ( lcsh )
Children -- Language   ( lcsh )
Psychology thesis Ph. D   ( lcsh )
Dissertations, Academic -- Psychology -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 103-116.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by Jack McFarlin Fletcher.
 Record Information
Bibliographic ID: UF00098646
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000066142
oclc - 04394420
notis - AAH1357


This item has the following downloads:

PDF ( 5 MBs ) ( PDF )

Table of Contents
    Title Page
        Page i
        Page ii
    Table of Contents
        Page iii
        Page iv
    List of Tables
        Page v
    List of Figures
        Page vi
        Page vii
        Page viii
        Page ix
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
    Review of the literature
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
    Appendix 1: Scoring key for Berry-Talbot text of morphology
        Page 98
        Page 99
    Appendix 2: Means, standard deviations, and ranges for six language variales by age (5.5, 8.5, 11) and group (reading disabled vs. control)
        Page 100
    Appendix 3: Summary of Bonferroni T-tests for planned comparisons
        Page 101
    Reference notes
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
    Biographical sketch
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
Full Text








Special thanks go to the members of my committee: Drs.

Paul Satz, Hugh C. Davis, Wiley C. Rasbury, Warren J. Rice,

and Robert J. Scholes. Dr. Satz, Chairman, has been a friend

and colleague for whom I have diligently and mindlessly

slaved for several years. Dr. Davis and Dr. Rasbury were my

first real contacts in the Department and have been kind to

me for some time. Dr. Rice is a special friend with whom I

shared many secrets and stimulating conversations during my

V. A. ordeal. Dr. Scholes has supported this study from its

inception and freely gave of his technical and theoretical

expertise in planning and completing the study. Dr. Barry

Lester and Dr. Ira Fischler were also kind enough to discuss

the study and its results with me.

Two people helped with testing and I appreciate it:

Bill Greenberg and Bess Frye. Several people struggled with

the typing and should be acknowledged: Mrs. Virginia Walker,

Lynda, and Lois. My parents, Mr. and Mrs. H. Max Fletcher,

grandmother, Mrs. F. L. May, and in-laws, Mr. and Mrs.

William H. Russell, have provided several types of support

during my prolonged education and I am exceedingly grateful.

My siblings, Max, Elizabeth, and even Clara have helped in

various and sundry ways. Most importantly, my wife, Fanita,

has made it all worthwhile and it is to our love that this

work is dedicated.



LIST OF TABLES . . . . . .

LIST OF FIGURES . . . . . .

ABSTRACT . . . . . . . .



Phonological Component . . .


. . . . ii

. . . . vi

. . . . vii

. . . . 1

. . . . 7
. . . . 7

Syntax Component . . . . . . . . 25

Semantic Component . . . . . . . 41

Statement of the Problem and Hypotheses . . 59

CHAPTER III METHOD . . . . . . . . 66

CHAPTER IV RESULTS . . . . . . . . 74

CHAPTER V DISCUSSION . . . . . . . . 89

APPENDIX 1 Scoring Key for Berry-Talbot Test of
Morphology . . . . . . . . 98

APPENDIX 2 Means, Standard Deviations, and Ranges for
Six Language Variables by Age (5.5, 8.5,
11) and Group (Reading Disabled vs.
Control) . . . . . . . . 100

APPENDIX 3 Summary of Bonferroni T-Tests for
Planned Comparisons . . . . . 101



REFERENCE NOTES . . . . . . . . . 102

REFERENCES. . . . . . . . . . .. 103

BIOGRAPHICAL SKETCH .... .. ... ....... 117


Table Page

1. Descriptive Statistics by Age (5.5, 8.5,
11) and Group (Reading Disabled vs. Control). . 68

2. Examples of Stimulus Sentences for Syntax
Text . . . . . . . . . . 71

3. Summary of Univariate Analyses of Variance
Results for Four Language Measures by Age
(5.5, 8.5, and 11) and Reading Group (Reading
Disabled vs. Control) . . . . . . 76


Figure Page

1. Example of Picture Stimuli for Syntax Test . 72

2. Mean Scores (No. of Responses) on Verbal
Fluency Test by Age (5.5, 8.5, 11 years) and
Group (Reading Disabled vs. Control) . . 77

3. Mean Scores (Percent Correct) on Berry-Talbot
Test by Age (5.5, 8.5, 11 years) and Group
(Reading Disabled vs. Control). . . . 78

4. Mean Scores (Percent Correct) on ITPA Grammatic
Closure Subtest by Age (5.5, 8.5, 11 years)
and Group (Reading Disabled vs. Control) . 79

5. Mean Scores (Percent Correct) on Syntax Test
by Age (5.5, 8.5, 11 years) and Group
(Reading Disabled vs. Control) . . . 80

6. Mean Scores (Percent Corre-ct) on Reading
Levels I and II (Syntax Test) by Age (5.5,
8.5, 11 years) and Group (Reading Disabled vs.
Control) . . . . . . . . . 86

7. Preferred Interpretations (Reading Level I vs.
II) of Ambiguous Sentences (Syntax Test) by
Age (5.5, 8.5, 11 years) and Group (Reading
Disabled vs. Control) . . . . . 87

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Jack McFarlin Fletcher

June, 1978

Chairman: Paul Satz
Major Department: Psychology

A current theory postulates that developmental reading

disorders represent the behavioral manifestation of an

underlying immaturity in the central nervous system. This

immaturity is further hypothesized to delay those develop-

mental skills underlying the acquisition of reading according

to their ontogenetic sequence of development. Because the

early acquisition of reading is more dependent on sensorimotor

and preconceptual skills which develop ontogenetically early

(e.g., visual-perceptual integration), performance patterns in

younger disabled readers (age 5-7) are more likely to reflect

problems along this performance dimension (Hypothesis 1). In

contrast, because the later stages of reading are more depen-

dent on verbal-conceptual skills which develop ontogenetically

later, performance patterns of older disabled readers (age 10-

14) are more likely to reflect problems along this performance

dimension (Hypothesis 2).

A recent longitudinal study which involved retesting

the same children (good and poor readers) at three different

ages (5, 8, 11 years) provided some support for Hypotheses 1

and 2. However, the battery of tests failed to include

measures of language processing which assessed those morpho-

logical and syntactic skills for which developmental psycho-

linguistics has suggested an earlier development.

Recent work on the acquisition of reading emphasized the

importance of those linguistic skills for beginning and more

fluent readers. However, the nature and role of these skills

for reading acquisition changes with age, so that the possi-

bility of developmental changes in the linguistic performance

correlates of reading disability has important consequences

for Hypothesis 2.

The present study explored the general issue of develop-

mental changes in the linguistic correlates of reading dis-

abilitiy in an additional test of Hypothesis 2. Along with

a Peabody Picture Vocabulary Test and a Verbal Fluency Test,

two measures of morphological ability and a test of syntactic

comprehension were administered to disabled and nondisabled

readers at three age levels (5.5, 8.5, and 11). This study

represents one of the first studies of language function in

potential reading disabled children at the pre-school age, a

period when many linguistic skills undergo primary develop-

ment. Such an assessment was possible with a two year inter-

val for the youngest groups prior to collecting criterial



Results indicated (a) robust age effects across the

three ages on all measures; (b) differences between reading

groups on the two morphological tests and the highly corre-

lated PPVT at all ages; and (c) between-group differences on

the Syntax Test and Verbal Fluency Test only for 11-year-old

children, i.e., and age X group interaction. Multivariate

analyses of variance indicated a significant age X group

interaction suggesting that large differences between reading

groups were observed at the oldest age.

Although the results were largely compatible with Hy-

pothesis 2, variables in addition to developmental period of

acquisition (earlier vs. later) were invoked to explain the

results. It was argued on the basis of this study and

additional recent work that the nature of the skill (perceptu-

al vs. linguistic) may be less important for discriminating

reading level than the child's developmental readiness for

meeting the task demands of reading at different ages. In

this respect, the results of the present study were also com-

patible with models of the acquisition of reading emphasizing

the greater importance of linguistically based organizational

strategies for older, more fluent readers.



The relationship of linguistic factors and reading

achievement has been a topic of interest for some time.

Rabinovitch (1959) was among the first to suggest a major

role for language function in reading disability and argued

that linguistic problems might be observed both in the oral

reading and expressive language of the disabled reader.

More recently, Gibson and Levin (1975) have summarized con-

siderable research showing that language skills and their

development are an important component of the beginning and

more fluent stages of learning to read. However, the re-

lationship of these linguistic skills to reading acquisition

may be age-dependent. The relative importance of different

linguistic skills for learning to read may change as the

child develops ( Doehring, 1976; Gibson & Levin, 1975). This

emphasis on relating language and reading skills in normal

children has paralleled the greater prominence of linguis-

tically based explanations of reading disability (cf.

Vellutino, Note 1). Unfortunately, these explanations have

largely failed to address the possibility of age-dependent

relationships and the problem of developmental change as a

factor in reading disability..

One current theory of reading disability (Satz &

Sparrow, 1970; Satz & Van Nostrand, 1973; Satz, Taylor, Friel,

& Fletcher, Note 2; Fletcher & Satz, Note 3) specifically

addresses the issue of developmental change. As currently

stated, the theory recognizes that linguistic, perceptual,

and conceptual skills are underlying components of reading

disability at all ages. The relative importance of these

different skills, however, changes with age according to

their ontogenetic sequence of development and importance for

learning to read. Because the early acquisition of reading

is more dependent on sensorimotor and preconceptual skills

which develop ontogenetically early (e.g., visual-perceptual

integration), performance patterns in younger disabled readers

(age 5-7) are more likely to reflect problems along this

dimension (Hypothesis 1). In contrast, because the later

stages of reading are more dependent on verbal-conceptual

skills which develop ontogenetically later, performance pat-

terns in older disabled readers are more likely to reflect prob-

lems along this performance dimension (Hypothesis 2).

A recent longitudinal study (Fletcher & Satz, Note 3)

which involved retesting the same children (good and poor

readers) at three different ages (5, 8, and 11 years) pro-

vided some support for Hypotheses 1 and 2. Perhaps the most

serious limitation of this study stems from its choice of

language measures (cf. Jansky, Note 4). For example, assess-

ment of those syntactic and morphological skills for which

developmental psycholinguistics has suggested an earlier

development (Bloom, 1975; Dale, 1976) was not made. In view

of the role of cognitive organizational strategies based on

these linguistic skills for beginning and fluent readers

(Gibson & Levin, 1975; Smith, 1971), the acquisition of

earlier developing linguistic skills in younger disabled

readers has important consequences for Hypothesis 2.

The present review will focus primarily on the relation-

ship of developing linguistic skills to the reading process,

learning to read, and reading failure. Within the context

of Hypothesis 2, particular emphasis will be placed on

developmental changes in the linguistic performance corre-

lates of disabled readers. Evidence for age-dependent

relationships will be examined in each of three aspects of

linguistic performance: the phonological, syntactic, and

semantic components of language -(Langacker, 1968). Each

component will be viewed as a set of rules and strategies

for information processing in language, reading, and memory.

The phonological component, therefore, will be viewed as a

set of rules and strategies learned during language develop-

ment concerning linguistic sounds and their relationship to

obtaining meaning from written and spoken language. The

syntactic component will be viewed as a set of rules and

strategies for forming acceptable words and word sequences

(sentences). Syntactic skills are important for organizing

larger units of oral and written language. The semantic

component will be viewed as a set of rules and strategies

directly concerned with the meaning of oral and written

language. In the present context, word meaning, word access,

and the more general division of conceptual experience by

language are all pertinent to the semantic component

(Langacker, 1968).

For reading acquisition, Gibson (1971) and Gibson and

Levin (1975) have argued that the fundamental unit of pro-

cessing is the word. Words are composed of and constrained

by a variety of distinctive features: graphic, phonological,

syntactic, and semantic. Processing words in reading involves

simultaneous perception along all these dimensions. These

constraints form patterns of invariance (redundancy) and a

major portion of the development of perception in reading

consists of learning strategies for handling increasingly

large units of information (Gibson, 1970). However, develop-

mental changes in the acquisition and use of these process-

ing skills during reading acquisition are often observed.

These changes involve differences in rate of acquisition,

extent of maturation, and the relative importance of these

skills for different reading stages. Doehring (1976)

evaluated this theory in a study of the acquisition of a

number of the rapid visual processing skills involved in

learning to read along these dimensions. Graphic and phonolog-

ical constraints were more important for the rapid visual

processing of letters, syllables, and words, characteristic of

the earlier stages of reading acquisition. Later stages were

characterized by the use of strategies based on semantic and

syntactic constraints for processing units of information

larger than the word.

As this digression on reading theory shows, chronological

age, the characteristics of language, and language development

are all intrinsic to the prolonged process of learning to

read. For this review each of these language components will

be examined separately. Brief statements concerning the develop-

ment of these aspects of linguistic performance and their

general relationship to the reading process will introduce

each section. The major portion of each section will be

devoted to research findings with disabled readers. Whenever

possible, the possibility of age-dependent relationships

and developmental changes in the linguistic performance cor-

relates of reading disability (Hypothesis 2) will be explored.

However, as the present review will show, this question is

difficult to address on the basis of current research findings.

The difficulty stems from multiple methodological and theo-

retical sources, not the least of which is the general lack of

concern for developmental phenomena in the area of childhood

reading disorders (cf. Fletcher & Satz, Note 3; Note 5). In

fact, age is often either a poorly manipulated or confounded

independent variable. Torgeson (1975) underscored the im-

portance of appropriately manipulating the age variable by

stating that "studies using subjects at one age may identify

deficits associated with reading disability which are different

from those found at another age (p. 421). Perhaps more

importantly, confounding the age variable by including children

from several age levels may obscure the relationship between

the dependent variable and reading achievement. The present

review will emphasize methodological factors only where they

are pertinent to the discussion. Major emphasis will be

placed on integrating a set of confusing and disparate find-

ings with general theories of language and reading acquisi-

tion in an effort to clarify the relationship between age

and linguistic skills in disabled readers. An attempt will

be made to show that there are age-dependent relationships

between language and reading failure. This relationship,

however, varies according to the nature of the linguistic

variable, its ontogenetic sequence of development (earlier

vs. later), and relative importance for reading acquisition.



Phonological Component

Phonological development. As Palermo and Molfese

(1972) noted, the literature on developmental phonology is in

stark contrast to the relatively advanced state of phono-

logical theory. What is known indicates that five year old

children have acquired and can distinguish the majority of the

sounds composing their language (Gibson & Levin, 1975). Some

advances, particularly in the articulation of certain sound

combinations, remain to be made, but even these relatively

difficult blends are generally mastered by age 8 (Palermo

& Molfese, 1972).

Phonology and reading. Phonological skills are quite

important for the initial phases of learning to read. Some-

how the child must bring to the task of learning to read the

knowledge of language accumulated over his first five years.

In this respect, the child must break the written code of

language and be able to pronounce and identify individual

words (Gibson & Levin, 1975; Golinkoff, 1975-1976). This

initial decoding process may be a necessary prerequisite for

future advances in reading comprehension.

Breaking the written code of language requires the

acquisition of a number of skills and strategies, including

visual-discrimination skills and different types of auditory-


mediational strategies (cf. Doehring, 1976; Gibson & Levin,

1975). These latter strategies pertain to the pronunciation

of written words, the more complex process of decomposing

sounds to their distinctive segments, and the use of intraword

redundancies for higher level processing (Gibson & Levin,

1975). Doehring (1976) showed that early in the reading

process, skills for processing strings of letters and words

seem to require speech mediation. The influence of these

mediational skills, however, diminished as the reader pro-

gressed. Mature readers appear to be able to directly

assess meaning without associating print and speech (Bradshaw,

1975). These stages may be mediated by the use of higher order

linguistic relationships (Doehring, 1976; Smith, 1971).

Early stages of reading require different types of

linguistic skills and strategies. In establishing the link

between oral and written language necessary for early decoding,

the child must acquire different auditory-mediational strate-

gies which reflect his understanding of the acoustic structure

of speech. However, neither the auditory-mediational skills

(nor the visual discrimination skills) underlying the early

decoding phases of reading are mastered by children when they

begin to read. Although the basic framework has been acquired,

the transfer of these skills to the task of reading requires

new learning which generally persists until age 8 (Gibson &

Levin, 1975). Problems with either visual discrimination or

acoustic mediational skills could hamper the child's early

ability to deal with the graphic features of written language

and master decoding.

It is interesting to note that linguistic skills which

the child has already acquired may transfer more directly to

the reading process. Both Weber (1970) and Biemuller (1970-

1971) compared the oral reading errors of first grade child-

ren. Weber found that these errors generally conformed to

grammatical constraints (syntactic and semantic) provided by

prior grammatical context. Thus, early readers applied their

knowledge of language to the new problem of word identifica-

tion. More importantly, Weber (1970) and Biemuller (1970-

1971) found no differences between good and poor readers

(first grade) in the number or type of these contextual er-

rors. In terms of correcting these errors, however, good

readers were distinctly superior (Weber, 1970). Biemuller

(1970-1971) described three stages in the initial process of

learning to read. The first phase was characterized by an

emphasis on use of contextual information. In the second

phase an increase in the number of graphically constrained

errors was observed, while the third phase was marked by a

co-occurrence of contextual and graphically constrained er-

rors. Most important, however, was Biemuller's finding that

the early reader's ability to use graphic information for

word identification differentiated good and poor achievers.

In other words, group differences emerged on the level of in-

dividual words and did not necessarily reflect the use of

higher order linguistic relationships.

Difficulties with visual discrimination skills and audi-

tory mediational phonetic segmentation skills concerning the

child's knowledge of the acoustic structure of speech could

hamper the early reader's ability to deal with the graphic

features processed during decoding. These auditory skills

may reflect both earlier developing phonetic segmentation

abilities and later developing skills for processing intraword

redundancies. Three types of studies regarding the disabled

reader's phonetic segmentation skills will be reviewed, along

with additional research on rapid naming skills and later

developing decoding skills.

Phonetic segmentation skills. Most of the research on

phonetic segmentation stems from work by Liberman and

Shankweiler (cf. Liberman, Shankweiler, Liberman, Fowler, &

Fischer, 1977). This work is based on the "assumption that

reading is somehow parasitic on speech.... In order to learn

to read, the child must map the written word to the spoken

word. .in order to do this, he must have some recognition

of the phonetic structure of his spoken language" (Liberman

& Shankweiler, 1976, p. 2). In this quotation the direct role

of auditory mediational factors for bridging the gap between

spoken and written language can be observed. Liberman,

Shankweiler, and associates have investigated three aspects

of this relationship as it pertains to reading achievement:

(1) awareness of phonetic segments; (2) phonetic representa-

tions in short-term memory; and (3) oral reading errors.

Awareness of phonetic segmentation. The importance of

phonetic segmentation is illustrated by the fact that because

the syllable, not the phoneme, is the minimal unit of articu-

lation, learning to read by sounding letters one by one is

impossible. Rather, reading analytically implies the discovery

of how letter segments (sounds) can be simultaneously blended

to arrive at the correct phonetic representation of each

syllable. Therefore, knowing how many phonemic segments form

a unit of articulation is vital to relating speech and written

language (Liberman et al., 1977).

These phonemic segmentation skills do not come naturally

to the child. Liberman, Shankweiler, Fischer, and Carter

(1974) explored the ability of four, five, and six year olds

prekindergartenn to Grade 1) to identify the number of

phonetic and syllabic segments in spoken utterances. The task

used required the child to tap a wooden mallet for each of

the segments in a list of test utterances. At age 4, no

children could identify phonemic segments, while half could

identify syllabic segments. At age 6, however, 70 percent

could identify phonemic segments and 90 percent could identify

syllabic segmentation. In summarizing this area, Gibson and

Levin (1975) suggested that phonemic segmentation skills were

not fully acquired until age 8.

Using their first grade sample, Liberman et al. (1977)

also contrasted good and poor readers on their segmentation

skills about four months after initial testing (beginning of

Grade 2). Half of the children in the lowest third of the

reading achievement distribution had failed the phonetic seg-

mentation task, while all of the top third of the distribution

had passed the task.

More rigorous investigations have strengthened the

relationship of phonetic segmentation skills and early read-

ing achievement. Helfgott (1976) measured segmentation and

blending skills in kindergarten children in an attempt to

predict first grade reading achievement. Segmentation of

spoken CVC words in kindergarten correlated at .75 with the

first grade word recognition subtest of the Wide Range Achieve-

ment Test. Zifcak (1976) found a significant correlational

relationship between phonetic segmentation on the dowel tapping

task (described earlier) and reading achievement in first

grade Ss. Treiman (Note 6), in a study of first and second

grade inner city children (largely blacks) also found a

high correlation between phonetic segmentation (measured by a

variation of the tapping task) and reading ability. These

results, on diverse samples with different criterion reading

measures, suggest that the child's ability to decompose

linguistic units into phonetic segments has a high relation-

ship with early reading achievement.

Phonetic recoding. A second set of studies concerns

phonetic recoding in short term memory. These studies are

predicated on the assumption that before longer segments of

speech (and reading) can be processed, a temporary acoustic

store must be established. This acoustic store takes the

form of a phonetic representation. Therefore, if poor readers

have trouble forming phonetic representations of language,as

indicated by the segmentation studies, differences between

good and poor readers should be observed in the amount of

phonetic coding used for short term memory tasks (Liberman &

Shankweiler, 1976).

Experiments pertaining to this hypothesis were sum-

marized by Liberman and Shankweiler (1976). Two experiments

(Liberman et al., 1977; Shankweiler & Liberman, 1976) com-

pared recall of phonetically confusable (rhyming) and non-

confusable (nonrhyming) strings of letters in second grade

reading groups. Results from both studies indicated that

regardless of presentation modality, the interference effect

of confusable stimuli was more apparent for good readers

than poor readers. No disruptive effects were observed on

nonconfusable stimuli for either reading group. Thus, tasks

requiring a phonetic strategy had a greater disruptive

effect on recall in good readers than poor readers, indicating

more consistent application of phonetic recoding strategies

in the good reader group.

A more rigorous experiment (Mark, 1977) gave beginning

readers (age unspecified) a list of 28 words to be read

aloud, followed by a recognition list containing the original

words and 28 new distractor words. Half the distractors were

phonetically confusable but visually dissimilar from an origi-

nal test word. The other 14 words were not confusable, i.e.,

didn't rhyme. Again, the inference effect of confusability

for recall was more pronounced in good than poor readers.

Oral reading errors. The third approach to early phonetic

encoding is based on research attempts attempting phonetic

analysis of reading errors made by second grade (and older)

disabled and nondisabled readers. Shankweiler and Liberman

(1972) found a distinct pattern of oral reading errors showing

that errors on the final consonant of a CVC syllable were

twice those on the initial consonant, while medial vowel

errors exceeded both possible consonant errors. In other

words, more errors were observed on medial vowel and final

position consonants. These results were interpreted as

revealing incomplete understanding of the phonological segments

of language in these Ss. When scanning from left to right

a child with incomplete phonological awareness will search

for any lexical word beginning with the initial sound. As

such, later errors later in the word would be more frequent

because the child failed to process the remainder of the word.

These interpretations were buttressed by a recent phonemic

analysis of the oral reading errors of second, third, and

fourth grade readers (Fowler, Liberman, & Shankweiler, 1976).

Although the number of errors decreased with increased age,

the error pattern remained the same.

These latter studies have led Liberman and Shankweiler

(1976) to argue against visual perceptual problems in ex-

plaining these observed error patterns, largely because of the

difference in vowel and consonant errors is difficult to ex-

plain in terms of distinctive features (physical shapes, etc.)

of letters. The absence of errors (e.g., reversals) presum-

ably reflecting perceptual anomalies is also cited as sub-

stantiating this argument. Liberman and Shankweiler (1976)

concluded that the problems of beginning and disabled readers

reflect phonetic segmentation difficulties as opposed to

visual-perceptual difficulties. It should be noted, however,

that both the oral reading error and the phonetic recoding

studies which support this hypothesis are based on null

(not negative) data obtained from relatively older (second

grade and above) readers, an age where visual-perceptual prob-

lems are expected to be less prominent (Benton, 1962; Fletcher

& Satz, Note 3; Note 5). Therefore, while the present series

of studies provides strong support for the role of phonologi-

cal factors in early reading, sweeping generalizations re-

garding other types of skills are inappropriate because of

the restricted age range.

Rapid automatized naming. Another source can be probed

for evidence of developmental changes in the linguistic skills

underlying decoding differences between reading groups. These

studies concern speed of digit and object naming in good and

poor readers at several age levels. Spring (1975) attempted

to predict first grade reading achievement (N = 44) with a

simple measure of digit naming speed. After a five month

follow-up,a correlation of .53 between digit naming speed and

reading achievement was found. In the large scale longitudi-

nal-predictive study of Jansky and deHirsch (1972), a picture

naming test was among the best kindergarten predictors of

Grade 2 reading ability, also correlating at .53. Similar

results were reported by Lindgren (1975) in a one year follow-


With relatively older readers, Spring and Capps (1974)

and Spring (1976) examined speech of digit, color, and picture

naming in small smaples of poor readers (N = 24) and good

readers (N = 24) from seven to 13 years of age. Overall results

indicated that disabled readers named more slowly than good

readers across the age range. Denckla and Rudel (1976a)

measured performance on several naming tasks in a large sample

of reading disabled, minimally brain damaged, and normal Ss at

four age levels: 7.5, 8.5, 9.5, and 11.5. Age related group

differences were found on all tasks, with a strong trend

towards an Age X Group interaction. The nature of this trend,

however, was not clear in this study. As opposed to indicating

that "dyslexic" children are "dysphasic" as implied by this

study and Denckla and Rudel (1976b), these findings are con-

sistent with a model which attributes reduced naming speed to

slower phonological coding in disabled readers (Spring &

Farmer, 1975). Except for the Denckla and Rudel (1976a) study,

these studies provide only moderate evidence for age-dependent

relationships in naming tasks. The high correlations with

early reading achievement (Jansky & deHirsch, 1972; Lindgren,

1975; Spring, 1975) are intriguing. Again, problems with

task definition make it difficult to specify the relationship

of these tasks to reading and its underlying components.

Later decoding skills. The preceding studies illustrat-

ed the role of speech perception, especially phonological me-

diation, for the decoding process in early reading. High cor-

relations with reading achievement were obtained, along with

group differences between good and poor readers. The importance

of these phonological factors is for the early learning of word

structures which is facilitated by the child's knowledge of the

acoustic structure of speech. Decoding in the intermediate

and older ages, however, reflects the use of some higher order

intraword relationships. In this respect, factors such as

pronounceability, orthographic structures (e.g., spelling

patterns), and other intraword structures are important

components of decoding strategies. These intraword struc-

tures compose different sources of redundancy which enable

the learning reader to efficiently process increasingly large

units of information. For earlier phases of reading lower

level analyses, e.g., sequential letter recognition and

spelling-sound correspondences may characterize decoding

strategies. With age and experience, higher levels of analy-

sis including the use of orthographic structure for identi-

fying clusters of letters, become more characteristic

(Doehring, 1976).

An earlier developing source of intraword redundancy

is letter-sound correspondence. Calfee, Venezky, and

Chapman (1969) explored the relationship of the child's

knowledge of letter-sound correspondences with reading

achievement. Snythetic words incorporating regular and ir-

regular letter-sound patterns were presented for pronoun-

ciation to good and poor readers in third grade, fifth grade,

high school, and college. Correlations between pronounci-

ation and reading achievement were highest in third graders,

decreasing substantially after that age as variables such

as IQ accounted for much more of the variability in read-

ing achievement. Differences between good and poor readers

were larger at the third grade level, decreasing with age

except on more complex patterns. A subsequent study (Venezky

& Johnson, 1973) gave similar synthetic words to first,


second,and third grade reading groups. Correlations with

reading comprehension were at .77 for first grade readers,

dropping to .63 for third grade readers.

In these studies the use of letter-sound correspondences

for pronouncing pseudowords is correlated with reading achieve-

ment in Grades 1-3, diminishing after this age. Whether this

relationship reflects phonological or orthographic sources of

redundancy is unclear. Tasks which don't require pronounci-

ation show that the use of spelling patterns for word identi-

fication emerges after Grade 2 ( Gibson & Levin, 1975).

Rosinski and Wheeler (1972) showed that third and fifth

graders could use spelling patterns to discriminate the

closeness of nonsense words to real words. First grader

performance was at the chance level. A reaction time study

(Santa, 1976-1977) compared first, second, and fifth grade

children and adults in the rapid visual recognition of dif-

ferent linguistic units: single letters, initial and final

letter clusters, and whole words. Fifth graders and adults

were faster processors for all these units. First graders

processed single letters most rapidly, while second graders

processed initial letter clusters as rapidly as single letters.

Developmental changes in the use of intraword struc-

tures can also be described for groups differing in reading

ability. Calfee et al. (1969) and Venezky and Johnson (1973)

found larger correlations with reading comprehension in younger

reading groups (Grades 1 and 3) than older reading groups


Grade 6) on the ability to use letter-sound correspondences

for word recognition. Santa (1976-1977) showed that poor

second grade readers were attempting to process different

units of information with a sequential letter strategy, while

achieving second graders were using strategies based on

higher order units of information.

Katz and Wicklund (1972) found no differences in

search time for single letters embedded in random letter

strings between good and poor readers in Grades 2 and 6.

However, good and poor first grade readers may differ

on the use of strategies based on single letter cues for

word identification. Rayner and Hagelberg (1975) compared

strategies based on letter shapes and whole word shapes in

kindergarten and good and poor first grade readers. Results

indicated that poor first grade readers employed strategies

based on initial letter shape less consistently than their

same aged counterparts, though both reading groups preferred

letter shape (and not word shape) as a word-recognition cue.

Expanding this study to the sixth grade, Rayner (1976) re-

vealed that the first letter strategy was important until

Grade 4, at which time whole word shape became the more im-

portant cue. Although good and poor readers were not employed

in this study, Rayner and Kaiser (1975) compared sixth grade

reading groups for reading text mutilated by altering first

and last target letters. Good readers were better at identi-

fying the mutilated word irrespective of the location of

the mutilation.

Some studies have explored the use of more complex

sources of intraword redundancy. In a variant of the Katz

and Wicklund (1972) study, Mason (1975) found differences

between good and poor sixth grade readers in single letter

search time when the letters were embedded in highly redun-

dant strings. Mason, Katz, and Wicklund (1975) showed that

this improvement also reflected differences between reader

groups for remembering spatial order. Katz and Wicklund

(1971) found no differences in the ability of fifth grade

reading groups to identify target words embedded within gram-

matical and nongrammatical sentences, with grammatically

facilitating identification for both groups. In contrast,

Samuels, Begy, and Chen (1975-1976), comparing fourth grade

reading groups, showed that good readers identified words

faster when the target words were embedded within contextual

(and orthographic) sources of redundancy. This finding was

apparent when the groups were equated on visual word recogni-

tion ability.

Two major points can be made on the basis of this re-

search on the use of intraword structure for decoding. The

first is that strategies based on higher order intraword

redundancies characterize later stages of decoding (Doehring,

1976; Mason, 1975; Rayner, 1976). This point shows that

the size, if not the nature, of the linguistic unit processed

changes as the child develops. The second point is that read-

ing group differences may be age-dependent. For example,

while Katz and Wicklund (1972) found no differences in the

ability of second and sixth grade reading groups to recog-


nize single letters, Rayner and Hagelberg (1975) found dif-

ferences in first grade reading groups use of initial letter

shape as a word-recognition cue. Although the tasks are

different, chronological age may be one variable underlying

the results in these studies.

It should be noted, however, that deficits on earlier

developing skills underlying decoding strategies seem to per-

sist in disabled readers throughout development. For ex-

ample, Calfee, Lindamood, and Lindamood (1973) measured per-

formance of kindergarten through twelfth grade Ss on a three

subtests of the Lindamood Auditory Conceptualization Test, a

variation on phonetic segmentation, as a function of reader

ability. On the easier task the largest differences emerged

between kindergarten achievement groups, with differences

persisting until Grade 4. The two more difficult tasks may

have been too difficult for Ss prior to Grade 2, at which

point large differences between achievement groups up to

Grade 10 were obtained.

Studies examining general decoding skills in relative-

ly older poor readers also found deficits. Perfetti and

Hogaboam (1975) measured oral response latencies to tachis-

toscopically presented pseudoword stimuli. Reaction time

differences were observed between both third and fifth grade

reading groups. Similar results were reported by Vellutino,

Smith, Steger, and Kaman (1975) for second and sixth grade

reader groups for the simple pronunciation of visually

Good readers in this study did not fully master these tasks
even at the high school level.


presented words. More specific studies of decoding ability

and access of single word meanings show that relatively older

reading groups defined on the basis of comprehension scores

still differ on decoding skills. Golinkoff and Rosinski (1976)

found both third and fifth grade poor readers to have longer

latencies than their respective controls for identifying

visually presented trigrams. Pace and Golinkoff (1976) ex-

tended this finding to more difficult real words. Interest-

ingly, in this latter study, the difference was larger between

third grade than fifth grade reading groups.

One common aspect of all the studies cited as evidence

for later decoding problems concerns task requirements for

pronunciation of visually presented stimuli. This may re-

flect a confounding because the pronunciation responses also

required "reading" by the disabled readers. A variety of fac-

tors underlie reading disorders, so that conclusions are

confounded bythese task requirements. Age-dependent relation-

ships between reading groups on the use of intraword struc-

ture are more apparent when a pronunciation response is not

required (Rosinski & Wheeler, 1972; Santa, 1976-1977). More-

over, deficits in the use of later developing higher order

intraword sources of redundancy can be described, which could

also explain the persistence of these deficits. These ex-

amples illustrate some of the problems with task definition

that make interpretation of many of these studies difficult

(Doehring, Note 7).

Summary. The studies reviewed under the phonological

section provide several sources of evidence for developmental

changes in the linguistic performance correlates of reading

achievement. The first source concerns the high correlation-

al relationship of performance on phonetic segmentation tasks

(Liberman et al., 1977) and naming speed tasks (Jansky &

deHirsch, 1972; Spring, 1975) with reading achievement. The

second source reflects the tendency for group differences to

diminish with increased age on phonetic segmentation tasks

(Liberman et al., 1977) and oral reading phonemic error pat-

terns (Liberman & Shankweiler, 1976). The third source con-

cerns larger group differences between relatively older read-

ing groups in the ability to use later developing higher-

order intraword redundancies (Mason, 1975; Rayner, 1976;

Rosinski & Wheeler, 1972). In general, group differences

appear to reflect (a) the size of the linguistic unit processed

(b) the importance of the linguistic unit for the reading pro-

cess, (c) the extent to which auditory mediational factors

are involved in processing the linguistic unit, and (d) the

unit's ontogenetic sequence of development.

It should be noted that decoding problems, no matter

what the source, may characterize poor readers throughout

development. This factor may help to explain the persistence

of reading problems into adolescence and adulthood (cf. Satz

et al., Note 2). As normal children progress in learning to

read, many decoding processes become automatized (LaBerge &

Samuels, 1974). When processing reaches an automatic level,

attention can be devoted to other elements of the reading

process. In fluent readers, comprehension is hindered by

over-reliance on decoding strategies. Many poor readers,


however, never seem to progress beyond the decoding level of

reading acquisition into a higher level comprehension stage

(Golinkoff, 1975-1976). They seem less able to minimize their

reliance on visual and auditory cues, the "mechanics" of

reading, so that more attention can be focuesd on the direct

extraction of meaning, i.e., comprehension. Perhaps more

importantly, these strategies may not constitute a useful

reading strategy for poor readers. Even relatively older

readers don't seem to master the basic strategies underlying

decoding. Kolers (1975) compared the ability of seventh

grade (age 10-14.5) good and poor readers on tasks permitting

separation of graphemic and semantic constituents of text.

The results indicated that even at this later age good readers

were more sensitive to variations in complex sources of graph-

emic structure (typography) than poor readers.

Several hypotheses concerning the inability of at least

some poor readers to master decoding can be provided. The

first hypothesis postulates a basic deficit in acquiring the

skills underlying decoding (Kolers, 1975; Vellutino, Note 1).

However, the evidence for developmental changes in the skills

underlying decoding strategies is inconsistent with this hy-

pothesis. A second hypothesis suggests that later comprehen-

sion processes are dependent on the acquisition of earlier

developing decoding skills (Farnham-Diggory & Gregg, 1976;

Golinkoff, 1975-1976). If decoding skills do not reach an

automatic level, then higher order comprehension processing

cannot be obtained and poor readers have no alternative to

their poorly learned decoding skills. Both hypotheses are

supported by the persistence of decoding problems throughout

development in poor readers. A third and different hypothesis

suggests that different factors underlie the failure to ac-

quire early decoding and later comprehension processes.

Under this hypothesis, poor readers continue to rely on in-

adequate decoding skills because they fail to acquire later

developing skills necessary for comprehension. Additional in-

formation on this latter hypothesis will accrue from the next

two sections, which concern more directly comprehension

strategies and processes.

Syntactic Component

Acquisition of syntax. Syntactic development is the most

thoroughly analyzed aspect of language acquisition. The acqui-

sition of the different structures and rules composing the

syntactic component of language begins at an early age (Bloom,

1975; Brown, 1973). Once begun, development is quite rapid.

Earlier studies of syntactic development employed longitudinal-

naturalistic observation of very young (ages 1.5 to 3) children

and analyzed oral samples of speech for the presence of dif-

ferent syntactic structures (cf. Bloom, 1975; Brown, 1973).

It was apparent in these earlier studies that children knew

quite a bit about the grammar of their language, which led

some authors (McNeill, 1970) to argue that syntactic develop-

ment was complete prior to age 5. Subsequent experimental

investigations (Chomsky, 1969; Kessell, 1970; Scholes, Tanis,

& Turner, 1977) showed that the acquisition of some more

complex syntactic structures proceeds much later into develop-

ment. Palermo and Molfese (1972) interpreted these data as

indicating major advances in language development between

ages 5-7 and 11-14. Despite this interpretation, it is evi-

dent that the major portion of the acquisition of syntax is

complete around the child's fifth year.

Organizational constraints and reading. The role of

higher order linguistic skills for reading (and language and

memory) is for the organization and extraction of written

information in units larger than the single word (Gibson &

Levin, 1975; Smith, 1971). For example, syntactic rules can

be viewed as a system of constraints which increase the redun-

dancy in a string of words. By developing strategies for pro-

cessing these redundant units, fluent readers can handle larg-

er units of information. This processing reduces the load on

short term memory and enables more efficient extraction of

information from text (Gibson & Levin, 1975; Smith, 1971).

The greater organization of input provided by the application

of these higher order linguistic strategies greatly facilitates

comprehension. The present section of this review focuses

almost exclusively on comprehension, not decoding.

The influence of grammatical constraints on the extraction

of information can be demonstrated in a variety of contexts. The

large number of studies on eye-voice span (summarized by Gibson

& Levin, 1975) show that older readers are progressively sensi-

tive to syntactically constrained units of text. Moreover,

older good readers have larger eye-voice spans than younger

and/or poor readers. Rode (1974-1975) explored the relation-

ship of syntactic structure to the eye-voice spans of third,

fourth, and fifth grade readers. Although younger readers

had shorter eye-voice spans, they also attempted to use syn-

tactic constraints for reading. Older readers, however,

tended to use the clause as the basic unit of processing,

while younger readers used a smaller unit, the phrase.

Previously cited studies by Weber (1970) and Biemuller

(1970-1971) revealed that first grade beginning readers were

highly sensitive to syntactic (and semantic) sources of

grammatical context. Sawyer (1976) evaluated the ability of

adults to recognize target phrases in tachistoscopically pro-

duced sentences differing in the amount of grammatical redun-

dancy. Recognition was facilitated when the grammatical

structure of the sentence was highly constrained. Wisher

(1976) examined the role of syntactic expectations during adult

reading. He showed that more rehearsal for recalling a string

of numbers occurred while reading a sentence if the reader knew

the syntactic structure prior to reading the sentence. Sum-

marizing this interesting area, Gibson & Levin (1975) concluded

that by the fourth grade, children regularly used grammatical

constraints for reading comprehension.

Organizational constraints and memory. In an important

review of developmental changes in the cognitive organization

of memory, Hagen, Jongeward, and Kail (1975) suggested that

children may use different stimulus attributes for memory at

different ages. Younger children (age 5-7) focus on the

acoustic and spatial attributes of the items to be remembered,

while the strategies of older children resembled those of

adults in their emphasis on syntactic,semantic, and other

higher order organizational constraints. The use of these

strategies, however, made for better and more efficient


Additional research exists on this developmental theory

of attribute dominance (Underwood, 1969). In a false recogni-

tion memory task, Bach & Underwood (1970) found that recall

error was characterized by acoustic features in second graders

and by associative features in sixth graders. Perlmutter and

Myers (1976) found that the recognition memory of 3 and 5 year

olds was not improved by providing verbal labels for the

children. McCarver (1972) showed that providing different

types of organizational cues (visual and verbal) improved

recall in Grade 4 and college level Ss, but not in Grade 1 or

kindergarten level Ss. Paris and Lindauer (1976) demonstrated

that the use of extralinguistic information was more effective

for facilitating sentence comprehension and recall in older

(age 11-12) but not younger children (age 6-7). Cramer (1975)

used a false recognition paradigm to show that verbal labeling

was more characteristic of Grade 4 and 6 children than Grade 2

children. In a similar experiment Cramer (1976) showed that

encoding on the basis of visual attributes characterized first

graders, while visual and verbal encoding characterized fourth


These studies tend to show developmental changes in the

attributes used for information processing. Similar changes

may be apparent for learning to read. The new problem pre-

sented by learning to read concerns the graphic and phonological

features (visual and acoustic) of written language. Older

readers, however, can rely on higher-order linguistic con-

straints (syntactic, semantic, and extra-linguistic) for

processing larger units. In terms of memory organization,

younger children seem to rely more on acoustic and spatial

attributes for remembering information, while older children

and adults use linguistically based skills to organize larger

units of information for more efficient processing.

Syntactic function in good and poor readers. These

parallels between the acquisition of reading and development

of memory attributes are informative. In the preceding section

on phonological factors and decoding it was shown that reading

group differences tend to emerge as a function of the size of

the linguistic unit processed, the importance of the unit

for the reading process, and its'ontogenetic sequence of

development. The present section focuses on units of informa-

tion larger than the word, the use of which characterizes more

advanced readers (Gibson & Levin, 1975). However, a major

portion of the syntactic component of language is acquired

by the time the child begins reading, a clear contrast to the

new learning presented by the graphic porperties of written

language. Therefore, it is not clear whether possible age-

dependent relationships will reflect mere development (Satz

& Van Nostrand, 1973) or a more complex relationship between

ontogenetic development and the role of the skill in the read-

ing process. This section will review research on syntactic

function in good and poor readers to see what relationships

emerge. Several different types of studies have been attempt-

ed, based on (a) morphological knowledge; (b) oral reading

errors; (c) oral language performance; and (d) syntactic


Morphological knowledge. Morphemes are the smallest

meaningful grammatical units. In essence, they are phonemes

which have meaning (e.g., -ed, -s, etc.). Just as syntactic

rules can refer to the set of formation rules used to generate

meaningful sentences, morphemes can be considered a set of

word formation rules for creating meaningful words. Linguists

generally separate morphemes into two types: inflections (e.g.,

pluralization) which do not change the grammatical class of a

word, and derivations (e.g., adverbalization) which do change

word class (Dale, 1976).

Numerous studies have used the Grammatic Closure subtest

of the Illinois Test of Psycholinquistic Abilities (ITPA;

Kirk, McCarthy, & Kirk, 1968) to explore potential differ-

ences in morphology between good and poor readers. Grammatic

Closure is a test of morphology using real words as response

eliciting stimuli. Studies in this area are generally poorly

designed, fail to control for chronological age, and seldom

employ children younger than Grade 2. Nonetheless, a simple

generalization can be made for good and poor readers at and

above the second grade. In summarizing the many studies of

theITPA and academic achievement, Hammill, Parker, and Newcomer

(1975) concluded that "Only Grammatic Closure, the most lin-

guistic of the ITPA subtests, consistently predicted academic

achievement in that it evidenced significant predictive and

diagnostic relationships in this study" (p. 251). Summarily,

Grammatic Closure almost always separates second grade and

older good and poor readers.

A similar approach to studying morphology in good and

poor readers is based on one of the most important early

studies in developmental psycholinguistics (Berko, 1958).

Berko measured first and second grader's knowledge of a varie-

ty of morphemes using nonsense word stimuli. Berko found that

the second graders in her study had acquired most of the

morphophomenic forms measured, though other studies have

suggested some caution regarding this conclusion (Palermo &

Molfese, 1972).

In looking at reading group differences, Brittain (1970-

1971) correlated performance on the Berko Test with reading

achievement in Grades 1 and 2. A. partial correlation (correct-

ing for IQ) of only.36 was obtained with first grade achieve-

ment, while the correlation with second grade achievement was

.70. Swallow (1972) gave multiple measures of linguistic and

nonlinguistic function, including the ITPA and Berko Test, to

good (N =30) and poor (N =30) second grade achievers. Group

differences were obtained on the Berko Test, Grammatic Closure,

and other measures (including the Bender-Gestalt). Wiig,

Semel, and Crouse (1973) gave the Berko Test to groups of "High

Risk" children and controls and learning disable children and

controls (aqe 9.5). The "High Risk" Ss (age 4.4) were defined

in terms of a history of neonatal trauma with subsequent neuro-

logical disability. Differences were found between experimental

and control groups at their respective age levels. Vogel

(1974, 1975, 1977) gave a variant of the Berko Test, the

Berry-Talbot Test of English Morphology (Berry, 1969) and the

ITPA Grammatic Closure subtest, to groups of second grade good

and poor readers. Robust differences were found on both

measures between these well-defined groups. A different sort

of study (Barganz,1971) required Grade 5 good and poor readers

to inflect and derive different target words embedded within

sentence context. Performance by poor readers was poor in all

three presentation conditions (oral-oral, oral-visual, and


It is evident that second grade and older poor readers

perform more poorly on measures of morphological knowledge.

The basis for this difference, however, is not clear, for the

construct characteristics of the morphological tests employed

have not been clarified. For example, Grammatic Closure

almost always loads on the general Verbal Comprehension factor

often observed in factor analyses of the WISC (Newcomer, Hare,

Hammill, & McGettigan, 1975). Correlations with vocabulary

tests are uniformly high. The net effect of these findings

is to question the suggestions of some authors (Vogel, 1975;

Wiig & Semel, 1976) concerning the specificity of morphological

knowledge as an underlying factor in reading disability. De-

spite these shortcomings, it is important to note that differ-

ences in performance are observed at the Grade 5 level, especi-

ally in view of the fact that these skills undergo primary

development between 5 and 8 years of age (Berko, 1958; Gibson

& Levin, 1975). It would be interesting to know the performance

characteristics of potentially disabled readers (age 5) prior

to the measurable onset of reading disability (age 8), es-

pecially in view of the lower correlations with first grade

than second grade reading achievement (Brittain, 1970-1971).

Oral reading errors. Two oral reading studies which

failed to find differences in the linguistic error patterns

of good and poor first grade readers were described earlier

( Biemuller, 1970-1971; Weber, 1970). In both studies, 90

percent of the oral reading errors invariably "made sense"

given preceding grammatical context. Somewhat different re-

sults were reported by Little (1975), who analyzed oral read-

ing errors of third grade average and disabled readers. These

errors were also compared with performance on the Development-

al Sentence Scoring Test (Lee & Canter, 1971). Although no re-

lationship between oral reading errors and syntactic develop-

ment was evident, the errors of average readers did conform

more to grammatical constraints within stimulus sentences

than for poor readers. Isakson and Miller (1976) defined

groups of fourth grade children equivalent on word recognition

skills but differing in comprehension ability. Results, based

on oral reading errors at the verb position, revealed that poor

comprehenders were less disturbed by snytactic (and semantic)

violations of sentence structure than good comprehenders, in

whom error rates increased. This study replicated similar

finding by Clay and Imlach (1971) and Weinstein and Rabino-

vitch (1971) showing that (relatively older) poor readers

seemed insensitive to grammatical constraints, processing words

one at a time. Less use was made of syntactic (and semantic)

context cues necessary for processing groups of words. Oakan,

Weiner, and Cromer (1971) and Steiner, Weiner, and Cromer

(1971) provided different types of comprehension training for

good and poor fifth grade comprehenders. Again, oral reading

errors indicated that performance was disrupted only in good


These studies show that older poor readers make less

use of the organizational facilitation provided by syntactic

constraints for reading comprehension. They also show that

even when poor and good readers are matched for word recog-

nition skills (i.e., decoding), comprehension difficulties

still arise. Cromer (1970) distinguished between two types

of poor readers. The "deficit group" was characterized by

poor decoding skills, while the other "difference group"

possessed adequate decoding skills and poor comprehension

skills. Even if the child learned to read word by word (i.e.,

learned decoding), major comprehension problems may still be

evident because of a failure to acquire or use linguistic skills

underlying fluent reading. It is unfortunate that this inter-

esting hypothesis has not been examined longitudinally or

cross-sectionally across broader (well-defined) age ranges.

If inadequate decoding skills characterized poor readers at

one age, while poor comprehension skills characterized poor

readers at a contrasting age, considerable insight into the

developmental course of reading disability would be the result.

Oral language. Several studies have examined the re-

lationship of syntactic characteristics of oral language and

reading achievement. From a correlation paradigm, Bougere

(1969-1970), Ribovitch (1975), and Mahaffey (1975) all cor-

related different measures of oral syntax (sentence length,

number of kernal phrases, transformational complexity, etc.)

with reading achievement in first grade samples. Only

Ribovitch (1975) obtained even a moderate correlation with

reading achievement, which was lower than the correlation of

intelligence and other linguistic and non-linguistic measures.

Mahaffey (1975) found letter naming skills to be more predic-

tive, while Bougere (1969-1970) found Metropolitan Readiness

Test subtests to be highly correlated with reading achieve-

ment. Bougere (1969-1970) concluded that while different

measures of oral syntax were not-correlated with Grade 1

reading achievement, they may be more important in later grades.

Hensley (1974) provided similar results by finding no oral

syntax differences between poor, average, and superior first

grade readers.

Correlational studies of older readers provide different

results. Second grade (Harris, 1975) and seventh grade (Kuntz,

1975) correlations of reading achievement and oral syntax

(using the Falk Sentence Construction Test) were quite high

(.68 to .70). Experimental studies of oral syntax between

relatively older reading groups provide similar results. Fry,

Johnson, and Muehl (1970) found differences on a variety of

oral language measures between second grade good and poor read-

ers. Similar findings were reported by Dumas (1976) with third

grade reading groups. Calvert (1973) found that the oral

language of good fifth and sixth grade readers was syntactical-

ly more complex and mature than comparably aged poor readers.

Summarily, these oral language studies show that oral syntax

is less related to the discrimination of first grade reading

ability, but more related to the discrimination of reading

abilities in second grade and older children.

Syntactic comprehension. The age effects described for

oral reading and oral syntax are largely based on measures of

language production. The present section will examine studies

based on the comprehension of sentences to see if a similar

generalization is possible.

For younger reading groups, two studies are quite per-

tinent. Falk (1977) compared the ability of good and poor

first and second grade readers to answer questions about 23

spoken sentences varying in syntactic complexity. A signifi-

cant age effect was observed, but reading group differences

were apparent only for second grade (not first grade) readers.

Taylor (1977) dichotomized first and second grade readers on

the basis of Metropolitan Achievement Test scores. He then

asked these Ss to judge the grammatical acceptability of

disrupted sentences. Correct judgements increased with age,

with semantic disruptions more easily identified and correlated

with achievement at both grade.levels. Syntactic disruptions,

however, were correlated with reading achievement only at the

second grade level (not first grade).

With older reading groups Vogel (1975) failed to find

differences in second grade on the Northwestern Syntax Screen-

ing Test (Lee, 1971), a task requiring picture selection on

the basis of different sentence types. In contrast, Semel and

Wiig (1975) found differences on this measure between reading

groups sampled across a broader age range (7-11.5) and there-

fore slightly older (about one year).

Dawson (1974) attempted a replication of Chomsky (1969)

with good and poor third and fourth graders from lower SES

backgrounds. Chomsky (1969) studied the acquisition of more

complex syntactic forms in children aged 5-9 and found con-

tinued development up to age 9. Dawson (1974) obtained an age

effect and a significant correlation with reading achievement

was reported. However, the order of acquisition was different

from that found by Chomsky (1969). Chomsky (1972) examined

Ss from her 1969 sample with regard to reading achievement.

She found that Ss with greater syntactic competence were also

better readers. This conclusion, however, was confounded by

chronological age and intelligence differences within her


Berger (1975) and Wiig and Semel (1973, 1975, 1976)

examined comprehension of different sentence types in good and

poor readers about 11 years old using sentence repetition tasks.

Differences in sentence comprehension were found in both

studies. Finally, Guthrie (1973) examined the relationship

between sentence comprehension and the use of syntactic cues

during silent reading. Disabled readers, about 10 years old,

were selected across a broad age range, while younger (7.5

years)and older (10 years) control groups were employed. The

task required Ss to read silently and select different words

from contrasting syntactic classes which would make the sen-

tence passages acceptable. Results indicated that while com-

prehension was much lower in disabled readers, the pattern of

errors was quite similar. These results are similar to those

obtained by Oakan et al. (1971) and Steiner et al. (1971) for

oder readers. Rabinovitch and Strassberg (1968) also showed

that syntactic cues did not facilitate comprehension in fourth

grade poor readers using sentence repetition and sentence

learning tasks.

Summary. The syntactic comprehension studies highlight

a variety of methodological and substantive issues which

plague this research area. For example, many studies confound

chronological age either by using disabled readers from a broad

age range (e.g., Guthrie, 1973; Semel & Wiig, 1975) or by fail-

ing to sample groups across the entire age range (age 5-14).

These studies, especially in the comprehension area, uniformly

fail to manipulate variables clearly described as syntactic.

Identifying target words (Guthrie, 1973), repeating sentences

of different syntactic complexity (Wiig & Semel, 1973;

Weinstein & Rabinovitch, 1971), or selecting pictures on the

basis of two sentences differing in single word meanings

(e.g., Northwestern Syntax Screening Test), manipulate a num-

ber of lexical, semantic, conceptual, and syntactic variables.

These variables, all important for comprehension, may develop

at different rates (cf. Section III) and may have different re-

lationships with other skills and abilities (e.g., intelli-

gence). Furthermore, because a variety of linguistic and

nonlinguistic deficits affect reading disability, it is not

clear that tasks requiring reading of the child (e.g.,

Guthrie, 1973; Oakan et al., 1971; Steiner et al., 1971)

yield conclusions specific to linguistic skills. Finally,

few of the studies cited relate substantially to current

knowledge concerning language development and the acquisition

of reading. Taken together, however, the four types of

studies reviewed in this section do provide support for

developmental changes in the linguistic performance corre-

lates of reading disability. This evidence is strongest for

the oral reading and oral language studies, which show that

syntactic variables contribute more to the discrimination of

reading level in older children than younger children. The

morphological and syntactic comprehension studies also show

that syntactic variables separate older reader groups. How-

ever, few of these studies have employed younger age groups

(before Grade 2), so that conclusions regarding age-dependence

from these studies are limited by this empirical shortcoming.

The age-dependent relationships that do emerge in this

section are consistent with the third hypothesis outlined

in the Phonological section (cf. p. 25 ). If syntactic skills

contribute more to the discrimination of reading level at

later years, then the persistence of decoding problems in

disabled readers could reflect-difficulties with the later

acquisition of skills underlying the development of compre-

hension strategies.

This possibility does not rule out the other two hy-

potheses, which postulated a basic defect in decoding skills

or attributed later comprehension problems to inadequate

development of earlier hierarchically related decoding skills

(Farnham-Diggory & Gregg, 1975; Golinkoff, 1975-1976; Perfetti,

Note 8). The age-dependent relationships that were described

do suggest that even if poor readers acquired earlier decoding,

later comprehension problems could still arise.

It should be noted that many syntactic skills are ac-

quired at an age (before Grade 2) which precedes the period

when they become more important for the reading process

(Grade 4). Of course, having these skills does not neces-

sarily imply that they can be applied in all the appropriate

situations (Flavell, Beach & Chinsky, 1966; Torgeson, 1977).

Although children may be able to apply their knowledge of

syntactic rules to verbal utterances, using these skills for

reading (and memory) may require additional development. The

acquisition of language has a much broader meaning than simp-

ly possessing certain skills and concepts. For the applica-

tion of these skills to reading, many factors could influence

possible age-dependent relationships, including the fact that

new skills must be acquired early in the reading process

(Gibson & Levin, 1975) and other types of developmental hier-

archies (White, 1965). Similar problems were considered

by Doehring (1976; Note 7), who emphasized the need for ad-

ditional knowledge concerning language, reading, and their

development. These interpretive problems have been evident

throughout this review and will continue to present diffi-

culties in the section on semantics.

Semantic Component

Semantic system and its development. Theoretical gen-

eralizations regarding the nature of the semantic system are

difficult to support (Dale, 1976). The semantic component

of language is generally defined as having to do with mean-

ing, i.e., the "ideas" expressed by language. Ordinarily it

includes a lexicon, or word dictionary, and a set of rules re-

lating lexical entries to one another (internally) and to

other aspects of language and cognition (externally). The

size of the lexicon and its organization concerns the internal

aspects of the semantic component. How the system relates to

other components of language and to more general extralin-

guistic and conceptual aspects of cognition are external

semantic concerns. These latter aspects of semantics pertain

to the organization of our conceptual experience provided by

language (Dale, 1976; Langacker, 1968).

Developmental generalizations regarding semantic pro-

cesses are even more difficult to maintain. Anglin (1970),

McNeil (1970), and Slobin (1971) all contended that the

semantic component develops more slowly in childhood than the

syntactic and phonological components of language. Dale (1976),

however, showed that this generalization is quite dependent

on the method of study employed and the specific words and

word types employed. Development proceeds rather early for

some aspects of semantics. Frasure and Entwisle (1973)

compared kindergarten, first and third grade children of

different social class backgrounds in their ability to use

syntactic and semantic cues to facilitate sentence recall.

Contrary to the idea of later, more gradual semantic develop-

ment, semantic cues facilitated performance at every grade

level. Also contrary to the idea of earlier and more rapid

syntactic acquisition was the finding that syntactic cues

were facilitative only at later grades, especially for lower

class children. These findings are consistent with studies

of certain types of semantic cues, e.g., taxonomic clustering,

which children may use for recall at relatively early ages

(Lange, 1973).

It may be that generalizations which relate semantic

development to more general aspects of conceptual development

will prove more useful in the long run. In this respect,

These studies do show early use (Age 5) of semantic
strategies by children and may on the surface conflict with
the review of organizational constraints in memory (cf. p.27 )
However, it is not clear that the use of these cues in young-
er children reflects more than lower order associative skills,
such that different strategies may underlie the facilitative
effects of these cues at different ages (Lange, 1973; Hagen
et al., 1975). Also, it is not clear that these attributes
are more important than acoustic and spatial attributes at an
earlier age.

Slobin (1971) has argued that syntactic and semantic pro-

cesses do not develop in unison. Semantic development must

be preceded by more general cognitive development--an idea

must exist before it can be expressed. Palermo and Molfese

(1972) argued that this notion applies to all aspects of

language, such that a general theory of language must be

embedded in a general theory of cognition. More specific-

ally, Dale (1976) noted that the issue of what idea is ex-

pressed semanticallyy) is difficult to separate from the for-

mal aspects of how ideas are expressed. Conceptual develop-

ment, at least in terms of comprehension, seems to develop

well into adolescence, so that it may be more reasonable to

expect semantic development to proceed in a similar fashion.

However, separating semantic and conceptual variables will be

difficult when attempts are made to "measure" these constructs,

so that this conclusion may not be entirely accurate. There

is less evidence suggesting rapid phases of acquisition for

semantic and conceptual skills, in contrast to common descrip-

tions of syntactic development (McNeill, 1970; Palermo &

Molfese, 1972).

Semantic function in good and poor readers. This di-

gression may be of little assistance in conceptualizing stu-

dies of semantic function in disabled readers. One suspicion,

however, is that developmental changes will be more difficult

to describe. Semantic aspects of language are important for

reading throughout development. Even younger children under-

stand the need to read for meaning (Gibson & Levin, 1975).

Unfortunately, semantic theory is not sufficiently developed

to provide any clear resolution of those problems. Studies

with disabled readers reflect more of the traditional verbal

learning emphases instead of the recent structural linguistic

emphases. Therefore, the studies reviewed under this section

largely concern associative and mediational processes in four

areas: (a) paired-associative learning; (b) auditory-visual

integration; (c) memory processes and (d) conceptual-linguis-

tic function. At the end, a few studies will be reviewed in

a fifth area pertaining more specifically to a theory of

semantics in disabled readers.

Paired associate learning. As Gibson and Levin (1975,

pp. 285-287) have noted, several conventional theories of

early reading acquisition viewed paired associate learning

(PAL) as the basic process involved. This view highlighted

the need for verbal responses to printed verbal stimuli, in

essence a "visual-verbal" associative task (Vellutino, Note

1). While this view is simplistic, it does provide a ration-

ale for examining PAL in different age groups of good and poor


Several studies have examined the predictiveness of rate

of learning various PAL for early reading achievement. Lam-

bert (1970) found that rate of PAL correlated significantly

with first grade reading achievement even after demographic

patterns were incorporated into prediction. Stevenson, Parker,

Wilkinson, Hegion, and Fish (1976) found that a verbal recall

PAL task administered in kindergarten reliably contributed to


reading achievement in Grades 1-3.

These correlational studies, of course, do not neces-

sarily imply differences between younger good and poor read-

ers. Several studies have examined PAL performance in experi-

mentally defined reading groups. These studies have used a

variety of PAL tasks in different modalities (auditory, vis-

ual, and tactical) and have used PAL tasks to investigate

both learning efficiency variables and the possibility of a

general verbal associative deficit in poor readers. Al-

worth (1974) compared performance of good and poor readers

in Grades 1 and 2 on a variety of visual-verbal PAL tasks.

Differences between groups in learning rate were not reliab-

ly obtained. Otto (1961) found differences in PAL learning

rates between good and poor second, fourth, and sixth grade

reading groups. Although differences were significant at

each age, an Age X Group interaction was also obtained. This

interaction revealed larger differences between second grade

than sixth grade reading groups. However, the task may have

been too easy for the sixth grade controls. In a subsequent

experiment, Otto, Koenke, and Cooper (1968) found a similar

Age X Group interaction for both visual-visual and visual-

verbal PAL learning rates. While second grade reading groups

made more errors and were slower on both types of PAL tasks,

no differences between fifth grade groups were found because

the task was too easy.

Differences between older reading groups on different

types of PAL tasks are reliably obtained. Vellutino, Steger,

Harding, and Phillips (1975) compared reading group perform-

ance (Grades 4-6) on visual-visual (nonverbal) and visual-

verbal PAL tasks. The groups differed only on the visual-

verbal PAL tasks (score = absolute errors). Similarly,

Rudel, Denckla, and Spalten (1976) compared 10 year old read-

ing groups on tactual-verbal (Braille letters) and auditory-

verbal (Morse code) PAL learning. Poor readers learned fewer

letters regardless of modality, leading Rudel et al. (1976)

and Vellutino (Note 1) to argue that these differences re-

flected general verbal encoding and retrieval difficulties.

Reinforcing this conclusion is a study by Gascon and Good-

glass (1970) with fifth grade reading groups. These authors

compared PAL performance using nonsense forms containing

different degrees of visual and auditory enrichment. Dif-

ferences were found regardless of modality and enrichment.

Task ceilings and floors are apparently very important

factors in these PAL studies. Samuels and Anderson (1973)

found differences between second grade reading groups on

an easy PAL task, but not on a more difficult PAL task, sug-

gesting the possibility of a floor effect. Camp and Dahlem

(1975) performed a study (confounding the age variable) which

compared PAL and serial learning in older poor readers (no

controls). Performance on the tasks were highly correlated,

with task difficulty forming the major determinant of per-


The influence of task difficulty on PAL tasks suggests

that group differences may reflect more general learning

efficiency variables as opposed to specific verbal associative

deficits. Differences on these PAL tasks emerge regardless

of presentation modality (Gascon & Goodglass, 1970; Otto, 1961;

Rudel et al., 1976) and for verbal and nonverbal material

(Otto, 1961; Samuels & Anderson, 1973). Moreover, group

differences often appear to be largely quantitative in nature.

Camp (1973) examined learning curve differences on PAL tasks

between good and poor readers aged 8 to 18. Quantitative

patterns were observed which tended to attenuate, while

qualitative differences in learning curves were not apparent.

Bartel, Grill, and Bartel (1973) looked for qualitative

differences in the achievement of the syntagmatic-paradigmatic

shift, which is usually complete between ages seven to nine

(Riley & Fite, 1974). No differences in paradigmatic word

association responses were found in 8 year old good and poor


Interpretation of these studies in terms of development-

al change is difficult. In general, group differences are

obtained between reading groups at all ages from Grades 2 to

6 (Otto, 1961). Prior to Grade 2, Stevenson et al. (1976)

found a verbal PAL task to be an important predictor of read-

ing achievement, while Alworth (1974) was not able to obtain

reliable group differences on PAL tasks in Grades 1 and 2.

Bartel et al. (1973) found no qualitative differences between

The study by Veliutino, Steger et al. (1975) is an
exception to this general finding.


second grade reading groups in the achievement of an impor-

tant development change, the syntagmatic-paradigmatic shift.

Group differences beyond Grade 2 were reliably obtained, but

with several PAL tasks under different presentation condi-

tions. Searching for developmental changes in PAL learning

tasks illustrates many of the problems with task definition

and surplus construct variance also evident in the next

section concerning auditory-visual integration.

Auditory-visual integration. A traditional research

topic in reading disabilities concerns auditory-visual inte-

gration (AVI), matching visual and auditory sequential pat-

terns. Initial experiments (Birch & Belmont, 1964; 1965)

were based on the idea that decoding graphic symbols to

sounds involved integrating auditory and visual input. These

experiments required matching spatially arranged dot patterns

with patterns tapped by a pencil. Results revealed group

differences mainly in Grades 1 and 2. After Grade 2 perform-

ance was more related to IQ, leading Birch and Belmont (1965)

to argue that early reading disability was more related to

perceptual deficits, while later reading disabilities were

more related to the measurement characteristics of IQ.

The major reason for including these experiments in

this section concerns subsequent AVI studies by Blank and

Bridger (1966) and Blank, Weider, and Bridger (1968). These

authors argued that the AVI performance of 7 and 9 year old

disabled readers reflects difficulty in conceptually labeling

sequential patterns, such that performance differences reflect

verbal mediational problems instead of perceptual factors.

This argument is based largely on the observation that while

both reading groups attempted to label the patterns in their

study, the disabled reader group was much poorer at labeling.

This argument has received considerable popular but

only moderate empirical support. The major reason for dis-

cussing it under semantic processing is that these experiments

are so often cited as indicating "verbal coding" problems in

disabled readers (cf. Vellutino, Note 1). Several factors,

however, provide shortcomings for this hypothesis. Most basic

to the point that children at the age when AVI was related to

overall reading achievement (especially first grade) in Birch

and Belmont (1965) do not seem to spontaneously apply verbal

coding and rehearsal strategies for processing information

(Hagen et al., 1975). Although these strategies do come into

prominence at age 9, a recent experiment by Drader (1975) with

older (11-12 years) reading groups questioned the assumption

that reading groups differ in verbally encoding sensory infor-

mation. Drader required good and poor readers to verbally

label temporally presented patterns of lights, sounds, and

finger taps. No group differences were found for labelling

temporal patterns in auditory, visual, or tactual modalities.

However, the labelling task was very difficult for all Ss,

suggesting the possibility of a floor effect. Even so, if

verbal labelling of sequential-patterns is too difficult for


11 year olds, why invoke it as an explanation of performance

differences in 9 year olds (Blank & Bridger, 1966; Bridger et al.,

1975)? Moreover, two factor analytic studies with younger

children show that measures of AVI tend to load with measures

of sensorimotor-perceptual skills as opposed to verbal-concept-

ual skills. Beran (1971) factored four AVI measures, two IQ

tests,two reading achievement measures, and several visuo-

spatial and auditory-temporal tasks in a small sample (N = 96)

of first graders. Three factors emerged: a general sensory

factor, an intelligence factor, and a temporal relations fac-

tor. Performance on AVI tests was related to Factor I and

III tests, not the intelligence factor (II). In a much larger

study (N = 425 second graders), Fletcher and Satz (Note 3)

factor analyzed a measure of AVI (Birch & Belmont, 1965) and

several measures involving somatosensory, visual-perceptual,

visual-motor, and verbal-conceptual abilities. The AVI meas-

ure loaded on a general sensori-motor-perceptual factor and

not on either of two additional verbal factors. Finally, if

the sum of this evidence does not provide some reservations

regarding the AVI verbal mediation hypothesis, Blank, Berenz-

weig, and Bridger (1975) performed a series of AVI studies on

9 year old good and poor readers. Not once was the verbal

encoding hypothesis offered, with the authors suggesting that

group differences reflected selective attentional factors and

would probably be obtained for processing any sort of complex,

meaningful information.

Semantic factors in memory. A few studies have examined

rehearsal strategies in memory for good and poor readers.

Leslie (1974) compared the effect of rehearsal instructions

on the spatial order short term memory of second and fifth

grade reading groups. One group of good and poor readers was

instructed to name and cumulatively rehearse the names of

sequentially presented common objects while the control groups

were told to simply reproduce the sequence of objects. Results

indicated no effect of rehearsal at Grade 2, with good readers

having better overall recall than poor readers. However, at

Grade 5, rehearsal hindered performance of both reading groups,

with no overall group differences in recall! These latter

findings are inconsistent with normal developmental research

on rehearsal (Hagen et al., 1975). A better designed study

(Tarver, Hallahan, Kaufman, and Ball, 1976) showed age-linked

performance differences between good and poor readers at 8.5

and 10 years of age, and an additional 13.5 year old poor

reader group. Using the Hagen Central-Incidental learning

task, a serial learning recall task, results revealed the pres-

ence of a primacy effect in controls, but not younger dis-

abled readers. Furthermore, the primacy effect exhibited by

10 year old disabled readers resembled that of eight year

old normals, with performance by the 13.5 year old disabled

group resembling normal 11 year old performance. Tarver et

al. (1976) summarized their results by noting that "the

evidence suggests that a developmental lag of about 2 years

is characteristic of the learning disabled" (p. 383). It

should be noted that the task was not employed with younger

(age 5-7) or older groups (age 14-16) so that the possibility

of earlier differences or of catching up cannot be attested to.

Earlier differences seem unlikely in as much as the verbal

rehearsal strategy underlying performance does not spontaneous-

ly emerge until Grade 2 (Hagen et al., 1975), though the age

related improvements and possibility of catching up are in-

triguing. Kastner and Rickards (1974) also reported differ-

ences in verbal rehearsal strategies between Grade 3 reading

groups. In this study, good third grade readers consistently

used verbal rehearsal, while poor readers seemed to waver,

alternating between verbal and visual strategies for repeating

serial sequences of novel and familiar tapped stimuli.

Several studies have investigated the role of conceptual-

semantic organization in facilitating short term retention of

disabled and nondisabled readers. These studies are reminis-

cent of Clay and Imlach (1973), Weinstein and Rabinovitch

(1971), and Guthrie (1973), each of which manipulated variables

of a more syntactic nature. Freston and Drew (1974) provided

a group of disabled readers (mean age = 11) with free recall

lists of paradigmatic associations. These lists were either

unorganized or organized via retrieval cues (conceptual cate-

gories) embedded within the list. Material organization had

no effect on recall. Parker, Freston, and Drew (1975) ex-

panded this experiment by including a control group. Free

recall lists were again provided to reading groups (mean age

= 10) and two variables were manipulated: material organiza-

tion and difficulty level. Although difficulty level influ-

enced performance for both groups, material organization

facilitated recall only for the control group. These findings

do seem related to research on temporal ordering (Bakker,

1972) and serial memory deficits (Corkin, 1974) in disabled


A well conceived study by Waller (1976) employed a se-

mantic integration task (Paris & Carter, 1973; Paris & Lind-

auer, 1975) to tease out differences in sentential recogni-

tion memory between fifth grade reading groups. Each subject

read a series of "acquisition" sentences. After an inter-

ference task, a recognition test composing either original

sentences or different transformations of the acquisition

sentences was presented. The child was required to respond

"yes" or "no" if he recognized the sentence as being identical

to the original acquisition sentence. No group differences

were observed if the new sentence was a true or false premise,

or a false inference. If the sentence was a true inference,

i.e., derived directly from the acquisition list, or if it

changed number or tense, poor readers performed more poorly.

An alternative approach to studying memory functions in

good and poor readers was reported by Torgeson and Goldman

(1977). These authors investigated the possibility that memory

deficits often observed in poor readers reflect "a lack of

ability or inclination to use efficient task strategies"

(p. 56). A memory task permitting observation of the use of

verbal rehearsal was employed with second grade reading groups

(cf. Flavell et al., 1966). Differences were obtained for

total recall and the actual use of rehearsal strategies. In

a second phase of the experiment, Ss performed a task which

facilitated use of rehearsal. After completion of this task,

group differences were not observed,with concommitant signifi-

cant increases for amount of rehearsal and total recall in poor

readers. Showing that an apparent deficit reflected a failure

to apply a previously acquired strategy is consistent with an

intriguing hypothesis expressed by Torgeson (1977). This

hypothesis suggests that problems with executive and organiza-

tional functions underlie apparent poor reader deficits on

many tasks. Poor readers may be deficient in applying strate-

gies, but not in the actual acquisition of the strategy.

Conceptual-semantic function. The Torgeson hypothesis

is quite interesting, but may fail to explain findings re-

sulting from more global non-mnemonic studies of conceptual-

semantic function in good and poor readers. LaPointe (1976)

compared performance of learning disabled and nondisabled

adolescents on the Token Test, a general measure of receptive

language. Differences were obtained and appeared to relate

to the retention of critical sentential elements across sub-

tests. The relationship of this test to age and general

intelligence was explored by Silverstein, Raskin, Davidson,

and Bloom (1977). These authors gave the Token Test and

WISC to 46 reading disabled children (mean age = 9.5 years).

Age was moderately correlated with performance on the Token

Test (r = .42). Partial correlations (holding age constant),

were reported for UISC subtests and scaled scores, including

Verbal IQ (r = .61) and Full Scale IQ (r = .54).

Several cross-sectional studies comparing disabled and

nondisabled readers on a variety of conceptual-linguistic

("semantic") have been conducted. Satz, Rardin,and Ross

(1971) and Satz and Van Nostrand (1973) compared performance

of good and poor readers at two age groups (7 and 11) on a

variety of perceptual and linguistic tests. The linguistic

measures largely involved measures of verbal fluency, vocabu-

lary, abstract reasoning, and dichotic listening. Group

differences on these conceptual-linguistic measures were ob-

tained between older but not younger groups. Sabatino and

Hayden (1970) performed separate principle component analyses

on a variety of sensorimotor-perceptual and conceptual-

linguistic measures in two large samples of disabled readers.

For younger readers (age 7.5) the major proportion of the

variance was explained by sensorimotor-perceptual measures,

while for older readers (age 9-11) "psycholinguistic" measures

were more explanatory. Gruen (1972) compared the predictive

accuracy of a battery of perceptual-motor and cognitive-intel-

lectual tasks administered to a large group of first grade

(N = 204) and third grade Ss (N = 202) against end of year

reading achievement (1 year follow-up). The multiple re-

gression analyses showed that the perceptual-motor tests

accounted for more of the explained variation in reading achieve-

ment scores (vocabulary and comprehension) for first grade boys

and girls. In contrast, the cognitive-intellectual tests ac-

counted for more of the explained variation in reading achieve-

ment scores for third grade boys and girls. Finally, in a

longitudinal study involving repeat testing of the same group

of children at three ages 5, 8, 11 years), Fletcher and Satz

(Note 3) showed that measures of conceptual-linguistic skill

(word comprehension, verbal fluency, vocabulary, etc.) are

more related to reading achievement in older reading groups

(age 10-14) than younger reading groups (age 5-7).

It should be apparent that the definition of "semantics"

in this section has now expanded to include more general stu-

dies of conceptual-linguistic function, i.e., thinking with

language. In this section on semantics the first two topics,

PAL and AVI were criticized in terms of the extent of their

relationship to semantic and verbal processing. Furthermore,

early deficits on these tasks could be related to perceptual

as well as language functioning. This section explored the

more general area of thinking wi-th language and regularly ob-

served a greater contribution of conceptual-linguistic vari-

ables in older (age 8-14), but not younger (age 5-7) reading

groups. These studies are consistent with experimental and

clinical observations of general verbal comprehension deficits

in intermediate and older disabled readers on standardized

language (ITPA; Newcomer & Hammill, 1975) and intelligence

tests (WAIS; Belmont & Birch, 1966; Sattler, 1973). With re-

gard to semantics, the relationship broached concerns the use

of language to organize conceptual experience (Dale, 1976;

Langacker, 1968). To the extent that conceptual factors must

be included in a theory of semantics, this latter section

does address developmental changes in semantics between dis-

abled and nondisabled readers.

Specific semantic skills. Whether these age linked con-

ceptual-linguistic differences extend to more specific semantic

skills (e.g., lexical access), is not clear. A trio of stu-

dies (Golinkoff & Rosinski, 1976; Pace & Golinkoff, 1976;

Rosinski, Golinkoff, & Kukish, 1975), summarized in Golinkoff

(1975-1976) are quite pertinent here. These studies employed

good and poor comprehenders in Grades 3 and 5. Single word ac-

cess was measured with a series of picture-word interference

tasks and timed sets of decoding tests. Interference tasks

required Ss to label common pictures as rapidly as possible

while ignoring printed words and trigrams superimposed on the

pictures. Group differences were obtained on the decoding

tasks, but the effects of semantic interference were equal for

the two groups. Rosinski et al. (1975) interpreted these re-

sults as indicating that decoding and semantic access skills

are independent, with minimal decoding skills required for

accessing single word meanings. Golinkoff (1975-1976)

summarized these findings by noting that "poor comprehenders

may readily obtain the meaning of common printed words" (p. 639).

On the other hand, Perfetti (Note 8) found fifth grade poor

readers to be deficient on other types of semantic skills, e.g.,

semantic categorization. This suggests that along with de-

ficiencies in semantic knowledge, poor comprehenders are less

proficient in the use of semantic cues than good comprehenders.

Nonetheless, as Golinkoff (1975-1976) notes, not all aspects

of semantic access separate good and poor readers, an intrigu-

ing hypothesis.


Contrasting results using a rather different paradigm

were reported by Denner (1970), who required first grade

good and poor readers and third to fifth grade poor readers

to perform tasks based on an experiment by Farnham-Diggory

(1967). This experiment used pictures of different movements

and objects to represent contrasting activities. Denner

(1970) found differences between reading groups (both ages)

only when Ss were presented cards depicting several actions

and asked to perform. In this case poor readers apparently

could not synthesize the individual cards into an integrated

activity despite their ability to perform activities repre-

sented by single cards.

Kolers (1975), using a task permitting separation of

graphemic and semantic aspects of word processing (cf. p 24),

found no differences in the semantic processing of fifth

grade good and poor readers. These two studies show age-

dependent relationships in the opposite direction of those

described for conceptual-semantic skills. Again, problems of

task definition makes interpretation of these results diffi-


Summary. No ready generalizations emerge in closing this

section. It should be noted that a number of the oral reading

and language studies cited earlier (e.g., Biemuller, 1970-1971;

Frey, Johnson, & Muehl, 1970; Weber, 1970;) also pertain to

semantic issues. These studies explored linguistic con-

straints in terms of whether the errors made sense, there-

fore conforming to either syntactic or semantic acceptibility.

More generally, it appears that semantic-conceptual skills

(vocabulary, verbal fluency, etc.) develop into adolescence

and contribute more to the separation of reading groups at

older ages. Other semantic variables, including verbal re-

hearsal and reconstructive processes in memory have been

studied, but exclusively in children beyond Grade 2. These

studies tend to show differences between whatever age groups

were examined. Studies using PAL and AVI tasks presented sub-

stantial interpretative problems. Differences were obtained

at several ages, but the interpretation of these differences

for verbal processing was not clear. The tentative nature of

these conclusions may reflect the unintegrated status of seman-

tic theory at the present time (Dale, 1976) and the lag in

applying these concepts to reading theory.

Statement of the Problem and Hypotheses

Overall summary. The present review has described re-

search on language function in disabled and nondisabled read-

ers within the phonological, syntactic, and semantic components

of language. Particular emphasis was given to the possibility

of developmental changes in the linguistic correlates of

reading disability. This emphasis stemmed from the hypothesis

(Satz, et al. Note 2) that larger differences in language

function would be observed between older (age 10-14) than

younger (age 5-7) reading groups.

Literature reviewed in phonology was not consistent

with this hypothesis, though developmental changes were ap-

parent. Phonological processes are more important for the


decoding process characteristic of early reading. In this

respect, different skills based on the child's understanding

of the phonetic structure of language (e.g., phonetic segmen-

tation) are important for relating oral language to written

language. Because these skills are largely acquired by age 8,

it would seem unlikely that disabled readers would have more

difficulty with phonetic representation at older ages.

Developmental changes, however, were apparent and may relate

to the need to process increasingly large intraword units of

information (e.g., orthographic structures) for progress in

reading. As such, the age-dependent relationships that

emerged seemed to reflect the size of the linguistic unit

processed, its ontogenetic development (earlier vs. later),

and its role in the reading process. Skills with a phono-

logical basis appear to diminish in importance with age and

reader experience ( Calfee et al., 1969; Doehring, 1976; Gibson

& Levin, 1975) and meaning may be directly accessed in mature

readers as decoding processes become automatized (Bradshaw,

1975; Golinkoff, 1975-1976; LaBerge & Samuels, 1974). It was

noted that decoding problems seemed to persist at least in

some disabled readers throughout development. The persistence

of this problem did not seem to reflect a complete failure

to acquire earlier-developing decoding skills, though this

hypothesis could not be ruled out. Alternative hypotheses

concerning disabled readers continued use of an inefficient

and poorly learned reading strategy (decoding) suggested that

later developing comprehension skills are dependent on the

acquisition of earlier developing decoding skills. Similarly,

a third hypothesis attributed the persistence of decoding,

difficulties to age-related problems in the later acquisition

of comprehension skills.

Like phonological skills, syntactic skills appeared to

develop primarily in the younger years (Dale, 1976) though

acquisition of more complex syntactic structures proceeds much

later in development (Palermo & Molfese, 1972). Unlike

phonological skills, syntactic skills are important for the

comprehension processes characteristic of more advanced

stages of reading (Gibson & Levin, 1975; Smith, 1971). Thus,

it is unclear as to what aspect of development would pro-

duce age-linked performance change between reading groups:

early acquisition or importance of the skill for the reading

process. Caution should be exercised in formulating these

problems since language development takes place in a context

much broader than simple verbal utterances. Applying syn-

tactic skills in all the appropriate situations may require

later development (Flavell et al., 1966).

In reviewing the literature concerning syntactic skills,

age-dependent relationships could be described which suggested

that tasks requiring the use of grammatical skills were more

likely to produce differences between older reading groups.

This possibility was not fully resolved because the chrono-

logical age variable was often confounded or insufficiently

manipulated. Also, the measurement characteristics of tasks

employed often confounded a variety of linguistic and reading

variables. These variables may develop at different rates and

be differentially involved in the reading process at different


Semantic development appears highly related to conceptual

development (Slobin, 1971). Although semantic development

appears to develop slowly and proceed late into adolescence,

such a conclusion appears to depend on what aspect of seman-

tics is measured (Dale, 1976). With advances in the theory of

semantics, such a conclusion will undoubtedly require modifi-


In a sense, semantic processes are always involved in

reading, for even beginning children understand the need to

obtain meaning from reading. From an organizational point of

view, however, semantic aspects of language are more involved

in the advanced comprehension stages of reading when decoding

skills are less important. The fluent reader uses a variety

of syntactic and semantic skills to organize written language

so that larger units of information can be efficiently pro-

cessed. Such a conclusion is consistent with research on the

role of cognitive strategies in the development of memory

(Hagen et al., 1975).

Literature concerning reading group differences on

semantic processing tasks presented considerable interpretive

difficulty. Most studies were based on relatively older age

groups and differences were generally observed across the age

range employed (Grade 2 and above). On semantic tasks with a

strong conceptual component, age-dependent relationships did

emerge, and suggested larger differences between older (age

10-14) than younger (aqe 7-8) reading groups. More specific

conclusions regarding semantic processes in disabled readers

awaits progress in both reading and semantic theory.

Problem and hypothesis. The problem to be explored for

this dissertation concerns possible age-dependent relation-

ships for syntactic comprehension skills in disabled readers.

This problem is particularly interesting because of the

relatively early development of those syntactic skills used

in language (Dale, 1976). Despite this earlier acquisition,

organizational strategies based on syntactic comprehension are

more important for later developing stages of reading acquisi-

tion. In this respect, it is not clear whether the crucial

postulate of the Satz et al., prediction (Hypothesis 2) con-

cerns the ontogenetic sequence of development or the relative

importance of the skill for learning to read. Earlier versions

of this theory (Satz and Van Nostrand, 1973) predicted group

differences largely on the basis of rate of ontogenetic develop-

ment. This version would predict larger differences on syn-

tactic comprehension tasks between younger reading groups

(age 7-8), despite the greater importance of this skill for

later reading stages.

In this study careful consideration will be given to

measuring linguistic variables clearly described as syntactic

and not confounded by lexical processing requirements. The

chronological age variable will be carefully controlled.

Furthermore, this study will assess earlier developing lan-

guage skills at an age (5.5) prior to the onset of reading

disability, but at a period when many of these skills undergo


primary development (age 5 to 8). By assessing these skills

across a broad age range (5.5 11), the issue of ontogenetic

sequence of development versus importance of the skill for

reading can be addressed. Finally, an attempt will be made to

measure language skills using tasks important to other studies

(e.g., morphology, vocabulary, and verbal fluency) so that de-

velopmental interpretations of these skills can be made for

reference comparison. Verbal fluency measures are particularly

important, inasmuch as it is a later developing skill

(Thurstone, 1955), with differences more often found between

older than younger reading groups (Satz et al. 1971; 1973).

One specific hypothesis can be stated, which is based

on the Satz et al. theory (Hypothesis 2):

1. Measures of linguistic function will contribute

more to the discrimination of reading level at older ages

(10-14) than younger ages (5-7).

Confirmation of this hypothesis would suggest that

factors in addition to ontogenetic sequence of development

must underlie developmental changes in the linguistic per-

formance correlates of reading disabilities (Satz & Van

Nostrand, 1973). This finding would also be compatible with

models of the acquisition of reading highlighting the impor-

tance of linguistically based organizational strategies for

fluent reading. If this hypothesis is not confirmed, then ear-

lier versions of the Satz et al. theory (Satz & Van Nostrand,

1973) which predicted on the basis of ontogenetic sequence of

development would be confirmed. This finding is less likely

in view of previous research (cf. Syntax section) suggesting

that syntactic skills in oral reading and language contribute

more to the discrimination of reading level at older ages

than younger ages. Whether this finding extends to syntactic

comprehension skills was not clear because previous research

has not been employed sufficiently for young age groups

(5-7 years). It should be noted that this view of syntactic

development derives from studies based on the application of

these skills for verbal utterances. To reiterate, language

development takes place in a much broader context and the

application of these skills in appropriate situations (e.g.,

reading, memory) may require later development (Flavell et

al., 1966).

No differences between reading groups at any age would be

contrary to the theory of developmental changes in the per-

formance correlates of reading disability (Fletcher & Satz,

Note 3). Over and above this hypothesis, however, the present

study represents an attempt to clarify the relationship be-

tween chronological age and language development in disabled

readers. If linguistic skills figure prominently in reading

disabilities, then language problem difficulties should also

be observed in nonreading language performance of disabled

readers (Rabinovitch, 1959).



Subjects. The children for this study were selected

at three age levels (5.5, 8.5, and 11 years) from three dif-

ferent longitudinal samples followed since kindergarten

(1970, 1971, and 1974). The 11 year old Ss were selected

during Grade 5 from the initial kindergarten population

of white males (N-497) obtained in 1970 (Satz et al., Note

2). The 8.5 year old Ss were selected during Grade 3 from

a subsequent cross-validational sample of white males

followed since kindergarten year (1971). The 11 and 8.5

year old Ss both came from the three schools participating

in the project with the largest number of Ss. The 5.5 year

old Ss were white males selected during kindergarten from

a third sample (N-120) obtained in 1974.

To obtain reading criteria on the 5.5 year old Ss,

a two year time interval was allowed to lapse. At the end

of Grade 1, criterion information based on the actual class-

room reading level of the Ss was collected. Children were

assigned to the reading disabled group (RD) if their classroom

reading level was below the Primer level and to the control

group (C) if reading was at or above the First Grade reader.

A similar age appropriate criterion, along with scores

on standardized reading tests, was previously available for

the 8.5 and 11 year old Ss.

To ensure the objectivity of group formation, four

judges classified Ss on the basis of criterion group infor-

mation for each project year available and the initial (KG)

prediction of the Satz-Friel Screening Battery (cf. Satz

et al., Note 2). Initial inclusion criteria constituted

agreement of (1) reading level criterion during the year Ss

received the language tests and (2) judges' classification.

Because less criterial information was available for the 5.5

year group, potential Ss were excluded if the judges' classi-

fication (RD vs. C) was different from the prediction of the

Satz-Friel Screening Battery. This additional requirement

resulted in the exclusion of five potential controls who were

predicted RD subjects (i.e., false positives). Sufficient

information was available on the 8.5 and 11 year old groups

so that group membership could be determined with consider-

able certainty. Additionally, a Peabody Picture Vocabulary

Test (PPVT) IQ of at least 85 was required for the two older

groups to screen out Ss with gross intellectual handicaps.

Such a criterion was not used with the 5.5 year old group

because of the unreliability of the PPVT IQ score at that

age and possible bias in the youngest age group by screening

out Ss with severe language handicaps. Table 1 presents the

results of the selection process, including sample sizes, ages,

PPVT IQ scores, SES levels, and mean criterion reading group.

It should be noted that no significant Age effects or Age X

Group interactions were obtained on PPVT, SES, and reading


Descriptive Statistics by Age (5.5, 8.5, 11) and
Group (Reading Disabled vs. Control)

Age Group N AGE SES PPVT Group

5.5 RD 11 65.09 1.45 92.91 1.27

C 23 66.43 1.95 113.30 3.61

8.5 RD 12 105.25 1.67 101.33 1.42

C 12 104.75 1.92 110.08 3.25

11 RD 12 130.75 1.75 101.58 1.25

C 15 131.67 1.93 121.67 3.40

aBased on the following scale:

1 = Severe Reading Disability

2 Mild Reading Disability

3 = Average Reading Ability

4 = Superior Reading Ability

group (F > .05). However, groups at each age level differed

on reading group (t < .05), while the youngest reading groups

(5.5) were significantly different on SES (t < .05). Overall,

the effect of these sampling differences does not appear to

bias significance tests in the direction of rejecting the

null hypothesis.

Tests. The tests included a verbal fluency measure,

two tests of morphological ability, and a measure of syntactic

comprehension. None of these tests required the children to

read. Verbal Fluency (Spreen, 1965) requires the child to

encode (orally) under timed conditions as many words as

possible denoting objects in different categories (e.g.,

rooms in the house, foods). Alternative forms were used at

different ages to control for possible floor and ceiling

effects. Verbal Fluency represented a reference point with

previous cross-sectional studies (e.g., Satz, Rardin & Ross,


The ITPA Grammatic Closure subtest (Kirk, McCarthy,

& Kirk, 1968) requires oral inflections of a variety of

real words, thus assessing the child's ability to use

plurals, possessives, and other grammatical morphemes. The

Berry-Talbot Test (Berry, 1969) assesses the child's knowl-

edge of English morphology by requiring inflection and

derivation of nonsense words from various stimuli. Thus,

for pluralization the child is shown a picture of a bird-

like creature and told it is a "nad." Then, two of the

pictured creatures are presented and the child is asked to

finish the sentence "There are two (i.e., nads).

Items and acceptable response for the Berry-Talbot Test are

presented in Appendix 1. These morphological tests have

been shown to tap linguistic skills developing primarily

between ages 5 to 8 (Berko, 1958). Both tests have been ad-

ministered to relatively older reading disabled children

(Vogel, 1975; Wiig & Semel, 1976).

The Syntax Test (Scholes, Tanist & Turner, 1977)

assesses the child's ability to use syntactic clues based

on the location and presence of the article "the" in the

resolution of direct-indirect object constituents. As Table

2 indicates, these constituents are presented in several

different ways. Two levels of Reading (I and II) are given,

representing different locations of the direct and indirect

objects. Within each level of Reading, two clues are given

as to location, either presence of the Article (A) or substi-

tution of a pause or Disjuncture (D). Three levels of deri-

vational complexity are also presented under each Reading

level. The Base form (B) is the basic unambiguous form of

these sentences. The second set of linguistic forms (A and

D) are more complex and present the ambiguity of direct-in-

direct object constituents. An additional pseudo-cleft trans-

formation (AT and DT) was applied to these forms (A and D),

making the Base form even more complex. This third structure

is designed to ensure appropriateness of the measure for the

older Ss.


Examples of Stimulus Sentences for Syntax Test

Reading Clue Sentence

I B He showed pictures to the girls' baby.
I A He showed the girl's baby the pictures.
I D He showed the girl's baby / pictures.
I AT It's the girl's baby the pictures
were shown to.
I DT It's the girl's baby / pictures were
shown to.
II B He showed baby pictures to the girls.
II A He showed the girls the baby pictures.
II D He showed the girls / baby pictures.
II AT It's the girls the baby pictures
were shown to.
II DT It's the girls / baby pictures were
shown to.
Ambiguous He showed the girls baby pictures.

1 Ifts IVy '"*1 ^-


1 3

f^r ( i

2 4
Fig. 1. Examples of Picture Stimuli for Syntax Test

Four pictures were given for each of the three sets

of sentences (N=33). Two pictures represented the opposing

types of Reading. To show how this test works, consider

Figure 1 and sentences 2 and 7 in Table 2. Should a child

fail to process the syntactic information by sentence 2

(picture 1), then the wrong type of Reading, sentence 7,

would be selected (picture 3). The other two pictures (2

and 4) represent lexically inappropriate forms of the sen-

tence stimulus. In the event of a failure to process the

major lexical items or lack of attention, an equal probabili-

ty for selection of an inappropriate sentence representation

is embedded within the test. Additionally, an Ambiguous form

(Amb) of these sentences with no clearly depicted interpre-

tation was included to assess preference for one or the other

levels of Reading (sentence 2 vs. 7).

Testing Conditions. All Ss were tested individually in

a quiet setting (usually in a mobile laboratory parked out-

side the school) by experienced examiners trained by the

author. The order of testing was constant, as follows:

Grammatic Closure, Berry-Talbot, Syntax, and Verbal Fluency.

The 5.5 and 8.5 year old Ss were tested in the Fall and

Winter of 1974-1975. The 11 year old Ss were tested in the

Spring of 1976. Testing was performed without knowledge of

the Ss reading group membership (RD vs. C).

This test has been used to measure the acquisition of
syntax in a variety of normal and deaf samples (cf. Scholes,



Test of overall significance. Because performance on

the four dependent variables is correlated, a 3 x 2 multi-

variate analysis of variance was conducted on the four

measures simultaneously. The MANOVA takes advantage of the

dependent structure among correlated measures, making for

greater power and control over Type I error rates (Hummell

& Sligo, 1971; Timm, 1975). Results of this test indicated

a significant Age effect (F=138.20; df=2,79; p < .0001),

significant Group effect (F=156.68; df=2,79; p <.0001), and

a significant Age X Group interaction (F=9.42; df=2,79;

p < .01). Following the significant MANOVA, the relative

contribution of each dependent variable to the obtained

effect must be determined. Although both univariate F-tests

(Hummell & Sligo, 1971) and Bonferroni t-tests for planned

comparisons were used (Timm, 1975), the present analysis

sought the linear composite of dependent variables maximiz-

ing these effects. Following a significant MANOVA, each de-

pendent variable can be correlated with the derived discrimi-

nant function maximizing the effect. Higher correlations re-

veal a greater contribution to the observed effect. For the

Age effect, approximately equal contribution of each measure

was obtained. The two morphological measures, Berry-Talbot

(r=.88) and Grammatic Closure (r=.81) had the greater contri-

bution to the Group effect, while Verbal Fluency (r-.72)

and Syntax (r=.62) maximized the Age X Group interaction.

This analysis indicates that only two measures, Verbal

Fluency and Syntax, contributed to the interaction predicted

by the hypothesis tested.

Univariate analyses of variance. To extend interpre-

tation of the MANOVA, separate 3 x 2 univariate ANOVA's

were conducted on each dependent measure separately and are
contained in Table 3. Specific planned comparisons (q=9)

were also completed using Bonferroni t-tests on each

dependent and are presented in Appendix 3. In each case

comparisons between each level of age within RD and C groups

were made (q=6), along with comparisons between groups at

each age (q=3). Comparisons were considered significant if

mean differences were greater than the critical value of

Bonferroni's t at the .05 level. It should be noted that

this comparison procedure always keeps the critical value

of alpha at or below the experiment-wise alpha level, thus

lowering the probability of Type 1 errors (Timm, 1975).

These results are not presented in detail and the reader is

referred to Appendix 3 for specifics.

For Verbal Fluency, results from the univariate ANOVA

indicated a significant Age effect (F=72.49; df=2,79; p < .0001),

significant Group effect (F=9.87; df=l,79; p < .002) and a

Basic descriptive data are presented in Appendix 2
which should be consulted regarding variability and the ade-
quacy of test floors and ceilings. These data are not specific-
ally presented because of the detail, except where this infor-
mation is essential.


Summary of Univariate Analyses of Variance Results for Four Language
Measures by Age (5.5, 8.5, and 11) and Reading Group (Read-
ing Disabled vs. Control)

Verbal Fluency Berry-Talbot Grammatic Closure Syntax

Source df F Prob > F F Prob > F F Prob > F F Prob > F

AGE 2,79 72.49 .0001 49.64 .0001 70.17 .0001 51.51 .0001

GROUP 1,79 9.87 .002 46.43 .0001 39.43 .0001 5.38 .02

AGE*GROUP 2,79 4.94 .01 1.10 .34 <1 -- 4.56 .01



5 1/2 8 1/2 II

Fig. 2. Mean Scores
by Age (5.5,
Disabled vs.

(No.of Responses) on Verbal Fluency Test
8.5, 11 years) and Group (Reading





5 1/2 8 1/2

Fig. 3. Mean Scores (Percent
by Age (5.5, 8.5, 11
Disabled vs. Control)

Correct) on Berry-Talbot Test
years) and Group (Reading

95 C

80- RD


o 65-



5 1/2 81/2 1I
Fig. 4. Mean Scores (Percent Correct) on ITPA Grammatic
Closure Subtest by Age (5.5, 8.5, 11 years) and
Group (Reading Disabled vs. Control.










Fig. 5. Mean Scores (Percent Correct) on Syntax Test by
Age (5.5, 8.5, 11 years) and Group (Reading
Disabled vs. Control).

significant Age X Group interaction (F=4.94; df-2,79;

p < .01). Figure 2 shows that differences across all three

ages were significant. However, despite the trend towards

increased between-group differences across age, only at age

11 were the differences between groups statistically signifi-

cant (cf. Appendix 3). Performance at age 11 was poorest for

the RD group.

Results for the two morphological measures were some-

what different. As Table 3 indicates, significant differ-

ences across ages (p < .0001) and between groups (p .0001)

were apparent. However, there was no Age X Group inter-

action (F< 1), indicating a parallel performance between

groups across age. Figure 3 (Grammatic Closure) and Figure

4 (Berry-Talbot) summarize these findings. The figures show

that differences across all ages and between groups at each

age were significant, with lower performance by the RD group

at each age. However, the rate of development was parallel

across age, hence the absence of an Age X Group interaction.

The variable of major theoretical interest in this

study is the Syntax Test. Figure 5 displays the direction

of the overall results. As Table 3 indicates, there was a

significant Age effect (F=51.51; df=2,79; p < .0001), sig-

nificant Group effect (F=5.38; df-1,79; p < .02), and sig-

nificant Age X Group interaction (F=4.56; df-2,79; p < .01).

Figure 5 shows the direction of this interaction: Increased

differences between groups across age. Indeed, although

differences across all ages were significant, reading groups

differed only at age 11, with the RD group performing more

poorly (cf. Appendix 3). This finding did not emerge as a

function of floor effects. Both groups at the younger age

were correct in 60 percent of their responses. Furthermore,

group differences were not apparent at the 8.5 year level,

when both groups were correct on 85 percent of their responses.

Additional analyses: Berry-Talbot. The purpose of

these analyses was to explore some of the linguistic deter-

minants underlying performance. For this analysis Berry-

Talbot scores were divided into Inflected and Derived sub-

scores. Stimuli were scored as Inflected when the morpho-

logical operation required did not change the word class of

the stimulus (e.g., as in noun pluralization). Derived scores

reflect a change in word class, e.g., changing the noun "spot"

to the adverb "spotty." Derivation is a more complex linguis-

tic operation and may have a slower rate of development. The

major question addressed was the possibility of different

performance between groups as a function of the morphological

operation required (Inflected vs. Derived). The question was

treated as a multivariate problem with two correlated measures

in a 3 x 2 design. A MANOVA across the two measures indicated

a significant Age effect (F=63.77; df=2,79; p < .0001) and

Group effect (F=45.89; df-1,79; p <.0001). The Age X Group

For further information concerning the floors and
ceilings of the testsused, consult Appendix 2.

interaction was not significant (F=2.84; p < .10). A uni-

variate ANOVA on Inflected scores revealed only significant

Age (F-33.25; df=2,79; p < .0001) and Group (F=46.20; df-1,79;

p < .0001) effects, with between group performance parallel

across ages. For Derived scores, significant univariate Age

(F=39.96, df=2,79; p < .0001) and Group F=8.77; df=1,79;

p < .004) effects were observed. Although there was no Age

X Group interaction (<1), between group differences were

significant only at the 8.5 and 11 year old ages. This

finding probably reflects a floor effect for Derived stimuli

at the 5.5 year age, thus documenting the complexity of the

Derivation task.

Additional analyses: Syntax Test. A second set of

analyses explored the underlying dimensional performance on

the Syntax Test. The major question concerned the role of

later acquisition of structural forms embedded within the

Snytax Test (cf. Table 2) in producing the between group dif-

ferences at the 11 year age. The first analysis compared

performance as a function of the two levels of Reading.

Reading I is acquired and later in childhood than Reading II

(Scholes, Tanis, & Turner, 1977; Suarez, Note 9 ). A MANOVA

across Readings revealed a significant Age effect (F=63.54; df

-2,79; p <.0001), a significant Group effect (F=4.88; df=l,79;

p < .05) and an Age X Group interaction (F-4.23; df=2,79; p <

.05). Subsequent univariate ANOVA's for Reading I produced

an Age effect (F=11.93; df=l,79; p < .0001), marginally

significant Group effect (F=3.59; df=1l,79; p < .05) and an

Age X Group interaction (F=3.18; df=2,79; p < .05). For

Reading II, only the Age effect was significant (F-=37.89;

df-2,79; p < .0001). Thus, as Figure 6 shows, the interaction

between Age and Group was apparent largely on the later de-

veloping Reading I stimuli.

The next question concerned the development of the

different structural forms within Reading I and II and their

contributions to the Age X Group interaction observed on the

Syntax Test. A separate study (Suarez, Note 9 ) explored

performance on the Syntax Test in a much larger sample

(N=192) of normal children unselected for reading ability.

Three age groups (5.5, 8.5, and 11.5) were used, with per-

formance on the Syntax Test largely comparable with that of

the control Ss in the present study. Suarez (Note 9 ) also

showed that the Base forms (Reading I and II) were acquired

earliest, while the Disjuncture forms (ID and IDT) were ac-

quired latest.

To explore this question, a MANOVA was conducted across

the 10 sentence types. Then each dependent variable was

correlated with the discriminant function derived for the

overall effect. Because the discriminant function represents

the linear combination of variables maximizing an effect,

higher correlations indicate a greater relative contribution

to the observed effect (Timm, 1975). This approach circum-

vented problems associated with high Type I error rates with

the number of dependent F-tests and simultaneous comparisons

otherwise involved in interpreting the MANOVA.

The MANOVA across the 10 sentence types revealed a

significant Age effect (F-81.25; df=2,79; p < .0001), Group

effect (F-11.54; df=l,79; p < .01) and a significant Age X

Group interaction (F=11.07; df-2,79; p < .01). The highest

correlations with the discriminant function could be ranked

as follows: ID (r=.62) IDT (r=.52), IAT (r=.37) and IA

(r=.33). This ranking shows that the major contribution to

the Age X Group interaction observed on the Syntax Test was

made by Reading I sentence structures. Also, the degree of

contribution corresponded exactly to the order of acquisition

observed by Suarez (Note 9 ).

A final analysis concerned the Ambiguous forms of the

sentences and is important in understanding preference for

Reading I vs. Reading II. Because Reading II stimuli are

probably more frequently heard in conversation, preference

for this Reading is likely to be observed (Scholes, Tanis &

Turner, 1977). Figure 7 shows the direction of changes in

preference across age and by group. It is apparent that these

changes were non-linear, which was also apparent in Suarez

(Note 9 ), represented in Figure 7 as the Standardization (S)

group. The major finding here is that when Reading I was

preferred, marginally significant univariate Group effects

were observed (F=3.65; df=l,79; p < .06), but no Age effect

or Age X Group interaction. Inspection of Figure 7 shows that

although choices on the Ambiguous forms were non-linear and

parallel across the younger groups, differences emerged

between the 11 year old groups. Here the control group indi-

- ----RD


--------RL I


Fig. 6. Mean Scores (Percent Correct) on Reading Levels I
and II (Syntax Test) by Age (5.5, 8.5, 11 years)
and Group (Reading Disabled vs. Control).















51/2 81/2 1
Fig. 7. Preferred Interpretations (Reading Level I vs.
II) of Ambiguous Sentences (Syntax Test) by
Age (5.5, 8.5, 11 years) and Group (Reading
Disabled vs. Control).


cates a continued increase in preference for Reading I

interpretations, while the reading disabled group shows no

such increase, preferring Reading I interpretations at about

the same frequency as their 5.5 year old counterparts.


The present study provides some additional evidence

for developmental changes in the linguistic performance

correlates of reading disabilities. Although reading group

differences in language processing were obtained across

several ages, the magnitude of these differences varied with

chronological age. In this respect, the results were con-

sistent with a hypothesis advanced by a theory of develop-

mental reading disorders (Fletcher & Satz, Note 3; Satz et al.

Note 2; Satz & Sparrow, 1970; Satz & Van Nostrand, 1973)

which predicts larger differences in linguistic function

between older disabled and nondisabled readers. These results

were also compatible with several longitudinal-predictive

follow-up studies (cf. Fletcher and Satz, Note 3; Satz et al.,

Note 2) and cross-sectional studies (Satz, Rardin & Ross, 1971;

Satz & Van Nostrand, 1973; Sparrow & Satz, 1970) which em-

ployed language tests primarily sensitive to semantic-con-

ceptual skills (e.g., verbal fluency, dichotic listening,

vocabulary, and abstract reasoning). These studies uniformly

showed that measures of language processing contributed more

to the discrimination of reading level at older ages (10-14

years) than younger ages (5-7 years).

Perhaps equally intriguing is the indication from

the present study that performance differences seemed age-

dependent only on certain types of language measures. Verbal

Fluency and the Syntax Test made the major contribution to

the multivariate Age X Group interaction, while group dif-

ferences on the two morphology measures were essentially

parallel across age. Some insight into this latter finding

is afforded when the relationship of these two tests and the

PPVT is examined. Correlating the morphological measures and

the PPVT in the 1974 kindergarten sample (N = 120) showed that

these measures shared similar variances (r > .70). In view

of the large differences in PPVT IQ between reading groups

across the three ages (cf. Table 1), group differences on the

morphology measures may reflect more of a general verbal com-

prehension and intellectual deficit often noted in reading

disabled groups (Eisenberg, 1966). Vocabulary tests are the

best single measures of general intelligence ("g") and the

PPVT IQ score does correlate substantially with, for example,

WISC Verbal IQ scores (about .75; Buros, 1973).

Factor analyses of vocabulary tests place them on a

general Verbal Comprehension factor. Similar findings are

apparent for factor analyses of the Grammatic Closure sub-

test, which also tends to load on a general Verbal Compre-

hension factor, along with measures of vocabulary and reading

comprehension (Newcomer & Hammill, 1975). Recent studies of

the relationship of a factor composed of vocabulary, abstract

reasoning, verbal fluency, and dichotic listening tests to


reading achievement (Satz et al., Note 2; Fletcher & Satz,

Note 3) have shown that this construct dimension contributes

more to the discrimination of reading level at older ages

than younger ages. These findings emerged despite the

presence of univariate group differences on these measures

at all ages (5-11 years) considered. Therefore, even though

group differences are readily observed at all ages on

measures such as the PPVT and the morphological measures,

these skills contributed less to the discrimination of read-

ing level at younger ages than other types of linguistic and

nonlinguistic skills.

The age-dependent relationship observed for Verbal

Fluency and the Syntax Test appears to be multiply determined.

Thurstone (1955), in a series of prospective and cross-

sectional studies, showed that a.Verbal Fluency factor peaked

some eight years after a Perceptual Speed factor. In this

respect, the age dependent relationship observed for the

Verbal Fluency measure used in this study and other studies

(Satz et al., 1971 ; Satz & Van Nostrand, 1973) may reflect the

later ontogenetic development of this skill. Similar inter-

pretations can be provided for the Syntax Test. On linguistic

forms which are acquired earlier (i.e., Reading II stimuli),

no group differences were observed at any age. This was

apparent even when Ss (age 8.5) were correct on 85 percent of

their responses. However, group differences and an Age X

Group interaction suggesting poorer performance by older

reading disabled Ss were apparent on the later developing

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs