Accounting For Individual Differences When Comparing The Effectiveness of Remedial Language Teaching Methods

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Applied Psycholinguistics (1980) 1, 151-170

Printed in the United States of America

Accounting for individual differences


when comparing the effectiveness of
remedial language teaching methods
PHILIP FRIEDMAN KAREN A. FRIEDMAN
Howard University Howard University

ADDRESS FOR CORRESPONDENCE


Philip Friedman, Department of Psychoeducational Studies, School of Education, Howard
University, Washington, DC 20059

ABSTRACT
Two language intervention methods were compared on their effectiveness at improving syntactic
structures as measured by gain scores on the Developmental Sentence Scoring (DSS) procedure.
Forty-one expressive language delayed children were assigned to either an Interactive approach
{n = 25) or a Programmed approach (n = 16), following a series of 13 pre-training diagnostic
tests. An analysis of group mean DSS gain scores did not indicate a significant difference
between the two teaching methods. However, when individual differences on the pre-training
tests were incorporated into further group comparisons, disordinal Aptitude x Treatment
interactions occurred for measures of intelligence, initial syntax level, and visual-motor integra-
tion. The Interactive approach optimized syntax improvement for children with relatively high
pre-training scores on each of the three measures, whereas the Programmed approach resulted
in superior syntax performance for low-scoring children.

Different theoretical views of language acquisition among normally develop-


ing children have led to a variety of clinical methods for rehabilitative
language teaching. Some clinical researchers characterize language develop-
ment in terms of previously defined syntactic rule systems (e.g., Chomsky,
1957, 1965), and have organized their therapy procedures upon structured
grammatical components (Ingram & Eisenson, 1972; Lee, 1974; Zwitman &
Sonderman, 1979). Language acquisition models emphasizing the impor-
tance of meanings and concepts expressed by linguistic structures and lexical
referents (Bloom, 1970; Brown, 1973; McNeill, 1970; Schlesinger, 1974)
provided a framework for such techniques as the Language and Teaching
Strategy (Miller & Yoder, 1974) and the Environmental Language Inter-
vention Program (MacDonald & Blott, 1974). Behavioral or operant proce-
dures, as they apply to the fields of speech and language, have also been
thoroughly discussed (Brookshire, 1967; Girardeau & Spradlin, 1970;
Holland, 1967) and adapted for clinical intervention strategies (Gray &
Fygetakis, 1968; Kent, Klein, Falk, & Guenther, 1972; Risley & Wolf,
1967). Finally, the pragmatics of verbal interaction (Bates, 1976), such as an
ability to express intentions and a knowledge of standard conversational
© 1980 Cambridge University Press 0142-7146/80/020151-20 $02.50
Applied Psycholinguistics 1:2 152
Friedman & Friedman: Individual reactions to language training

rules, is just beginning to receive attention by clinicians (e.g., Muma, 1975).


When training for language competence, some investigators have either
controlled or considered additional subject variables such as cognition,
memory, and social awareness (Bricker & Bricker, 1974; Miller & Yoder,
1974), whereas others have ignored these concurrent developmental factors
(Carrier, 1974; Fokes, 1976; Monsees, 1970).
Despite reported successes beyond expected maturational gains when
using most of these intervention techniques any comparative analysis of their
results would be difficult. Because each procedure was designed and imple-
mented at a different location there was a wide divergence among popula-
tions treated and criteria used for determining success. In several of the
published results, clinical subjects were not identified beyond being language
retarded or disordered. A few investigators merely defined the minimum
levels of psycholinguistic behavior or language performance necessary for
acceptance into their programs (e.g., Lee, 1974; Miller & Yoder, 1974). In
general, there is a lack of available data to support the clinical benefits or
suitability of one program over another.
Language researchers have also avoided setting up their own contrasting
treatment groups to assess the relative efficiency of different methods for
handling similar problems. Obviously, the cost of simultaneously implement-
ing more than one intervention program is prohibitive. Perhaps clinical
researchers are also familiar with the "no significant difference" results that
have typically stemmed from such research. This general conclusion is
repeatedly found in studies on the differential treatment of problems such as
stuttering (Dalali & Sheehan, 1974; Martin & Haroldson, 1969; Peins,
McGough, & Lee, 1972), poor articulation (Sommers, 1964), and expressive
language defects in aphasics (Sarno, Silverman, & Sands, 1970) or preschool
children (Cazden, 1965). Progress was defined in each study as the differ-
ence between pre- and post-training test scores, and the gains were not
significantly affected by treatment type.
One consistent problem with past treatment studies is that main effects
have been emphasized in their statistical analyses. That is, measures of
central tendency were typically employed to evaluate the relative effective-
ness of contrasting procedures. Studies that use group averages to test for
treatment differences inevitably mask the effects of individual differences. It
is highly unusual that treatment Method A will be more effective for every
child than Method B. Typically some children will show more improvement
with a second method. However, use of the statistical average disguises this
fact, resulting in a persistent failure to find differences between groups.
In order to detect more easily differences in treatment methods, it appears
that researchers will have to incorporate into their analyses of language data
relevant individual difference measures. As Cronbach and Snow (1969, 1)
have pointed out, "the search for generally superior methods must be
supplemented by a search for ways of adapting instruction to the individual."
From an analysis standpoint, this means a search for significant interactions
between contrasting types of expressive language training and measures of
Applied Psycholinguistics 1:2 153
Friedman & Friedman: Individual reactions to language training

individual differences among children. In several studies (e.g., Bracht, 1970;


Shavelson, Berliner, Ravitch, & Loeding, 1974; Tobias, 1973) this research
approach has resulted in systematic analyses of what have been termed
Aptitude x Treatment interactions (ATIs). An aptitude is defined as any
subject trait or ability that results in measurable individual differences.
If an ATI is present, the differential effectiveness of two or more
treatment approaches can be systematically related to some individual
difference characteristic of the children. Depending on the child's score on
this aptitude variable, one or the other method may lead to better results.
Statistically, this will be reflected in differential slopes for the regression
lines of Method A and Method B on this previously obtained child aptitude
measure. For example, Dowaliby and Schumer (1973) looked for the possi-
bility of an Aptitude x Treatment interaction between two modes of
classroom instruction and student differences in manifest anxiety. A differ-
ence in regression slopes revealed that although a teacher-centered mode
optimized learning for high-anxious students, a student-centered approach
resulted in superior exam performance for low-anxious students.
The present study was designed to compare the relative effectiveness of
two conceptually different language intervention programs in improving the
syntactic constructions of language-delayed preschool children. One program
controlled behavior changes through the use of specific operant conditioning
principles. Procedures were adapted from a variety of studies where they had
been successfully applied to modify language behaviors of normally develop-
ing (Harris & Hassemer, 1972; Liebert, Odom, Hill, & Huff, 1969; Rosen-
thai & Whitebrook, 1970) or disordered children (Lovaas, Berberich,
Perloff, & Schaeffer, 1966). A second remedial teaching program was based
on the research of psycholinguistic investigators who suggested that a child's
grammar at any stage of development is a legitimate, coherent set of
developing rules rather than an imperfect attempt at imitation of, and
selective reinforcing by, adults. Elaborate basic research studies of the
developmental stages of specific syntax structures, and the rules generated
during this development (e.g., Bloom, 1970; Brown, 1968; Cazden, 1968;
Klima & Bellugi, 1966), provided the guidelines for selecting goals and
procedures for this rehabilitation technique.
Prior to the different treatments, 13 standardized measures were given to
the children to help in diagnosing their specific language problems. These
aptitude variables were subsequently analyzed within the treatment groups
to demonstrate some of the procedures that may be used to examine
Aptitude x Treatment interactions.

Design
Two rehabilitative teaching programs were developed for use with small
groups of language-delayed children at the Northwestern University Speech
and Language Clinics. Twenty-five children were assigned to Interactive
Language Development Teaching (Lee, Koenigsknecht, & Mulhern, 1975)
Applied Psycholinguistics 1:2 154
Friedman & Friedman: Individual reactions to language training

and sixteen additional children to Programmed Language Presentation


(Koenigsknecht & Lee, 1974). The total teaching period for each child
lasted approximately eight months. During the course of study there were six
Interactive and four Programmed language groups. Most groups consisted of
four children, with a fifth in one Interactive group. The children met four
times a week in remedial teaching sessions that lasted one hour. Every child
was seen for three teaching blocks, each covering an eight-week period.
There was generally a vacation period between each teaching block. The
structure and presentation of the Programmed and Interactive lessons
followed a series of procedures described below.
Grammatical structures emphasized during these lessons depended on the
capabilities of children within a particular group and the rate of progress
evidenced over the course of training. However, measuring every individual
grammatical feature of a child's language would be so time-consuming as to
be clinically impractical. Therefore, the following syntax features were
selected for treatment within both programs based on their early appearance
in children's language and the ability to identify their developmental
progressions: 1) indefinite pronouns and/or noun modifiers, 2) personal
pronouns, 3) main verbs, 4) secondary verbs, 5) negatives, 6) conjunctions,
7) w/j-questions, and 8) interrogative reversals. Interrogative reversals were
those questions formed by reversing the subject and the first auxiliary verb of
a declarative statement (e.g., Is he coming?).
Based on previous basic research with each form, specific words or
structures within each classification have been grouped into a general
developmental order (Lee & Canter, 1971). The criterion measure used for
treatment comparisons was shown to be sufficiently sensitive to detect
changes in these features through a number of validation studies (Koenigs-
knecht & Friedman, 1976; Koenigsknecht, Friedman, & Hill, 1973;
Koenigsknecht & Lee, 1971).

Children and teachers

The 41 children receiving remedial training came from middle-class, mono-


lingual English-speaking homes within suburban Chicago. At the beginning
of their training period the children ranged in age from 38 to 69 months. The
mean age for children in the Interactive groups was 52.3 months and for
children in the Programmed groups, 51.5 months. Although no attempt was
made to control for sex of the children, boys outnumbered girls within each
treatment by almost a two-to-one margin.
Each of the children was given a complete hearing evaluation by audiology
graduate students and their supervisors within Northwestern's Department
of Communicative Disorders. If a hearing problem was detected, the child
was excluded from these experimental training groups. A judgment of
physiological status was generally made from case history information
obtained from referring doctors and other outside authorities. If these data
were not available, or if questions were raised, children were referred to
Applied Psycholinguistics 1:2 155
Friedman & Friedman: Individual reactions to language training

appropriate specialists who made the physiological determinations. None of


the children included in the treatment groups evidenced any general neuro-
muscular problems. Children assigned to both methods were generally found
to be functioning within a normal range of intelligence as measured on
individually administered Stanford-Binets (M = 99.88 for the Interactive
groups and M = 102.00 for the Programmed groups). However, four chil-
dren who were included in the sample did obtain scores in the 70s on this IQ
test. Based on several of the pre-treatment diagnostic measures all the
children were found to have severe language problems. Deficits in language
usage were most clearly shown on measures of expressive usage of grammar.
Comprehension of spoken syntax was also delayed. In summary, the difficul-
ties these preschool children would face in school would be massive if their
oral language skills were not improved. A more detailed analysis of these
pre-treatment deficiencies is given in the Results section.
Subjects were placed in treatment groups based on their availability
during scheduled time periods, their position on a clinic waiting list, and
their meeting the above physiological and cognitive requirements. It is
important to note that no child was assigned to a treatment based on the a
priori notion that it would be more appropriate or more effective. Although a
random assignment procedure was not feasible, later comparisons revealed
that the groups performed similarly on each diagnostic measure prior to
clinical intervention (see Results section).
Ninety-nine graduate students participated in the teaching programs as
part of their graduate training requirements in the area of speech and
language pathology. The vast majority of these teachers and all of their
supervisors were female. Student-teachers were rotated within each training
procedure on a prespecified schedule based on their classes and other clinic
responsibilities. Therefore, the children were being seen by different trainers
on each visit to the clinic. There were always two teachers within the
treatment room, one presenting or coordinating the predetermined lesson
procedures, and the other keeping order, providing lesson materials, and
handling special occurrences such as behavior problems. Other teachers
could observe the lesson from an adjoining room, through a one-way mirror,
and record information for the development of future lessons. Each student-
teacher was involved with the program during one eight-week teaching
block. During each eight-week period there were as many student-teachers
involved in the program as there were children in the group. All the teachers
had completed at least one course in child language development prior to
involvement in this study.

Pre-training individual difference measures


A variety of speech and language measures was obtained from the children
before the treatment period. These tests were conducted by a speech and
language pathologist who was not directly involved in the subsequent
language training procedures. A clinical psychologist with experience in
Applied Psycholinguistics 1:2 156
Friedman & Friedman: Individual reactions to language training

evaluating the functioning of language-impaired children assisted in pre-


treatment assessment. In addition to providing for a statistical analysis of
individual difference effects, a number of practical considerations were taken
into account when selecting the test battery. A series of short, easily
administered measures was desired, which would culminate in a general
picture of the many components of a child's language. Selection was limited
to a relatively small pool of available tests for which extensive reliability and
validity data had been reported and which could be employed with this
special preschool population. Since the clinic is located in a university
setting, tests were also chosen for didactic purposes. That is, students needed
to be provided with experience on a variety of testing procedures from some
of the most widely used measures. An inordinate number of the tests selected
had been developed at Northwestern; thus faculty and supervisors were
probably more familiar with and confident in these procedures, and local
norms were more readily available.
The 13 measures given to each child were Northwestern Syntax Screening
Test (NSST) - Receptive and Expressive sections (Lee, 1969; 1971);
Peabody Picture Vocabulary Test (PPVT) (Dunn, 1959); Stanford-Binet
Intelligence Scale Form L-M (Terman & Merrill, 1960); Preschool Attain-
ment Record (PAR) (Doll, 1966); Spencer-McGrady Sentence Repetition,
an unpublished test of ability to repeat sentences of increasing length with
generally increasing grammatical load (McGrady, 1964; Spencer, 1958);
Oral Commissions subtest of the Detroit Tests of Learning Aptitude
(DTLA) (Baker & Leland, 1967); Illinois Test of Psycholinguistic Abilities
(ITPA) - total and individual subtests (Kirk, McCarthy, & Kirk, 1968);
Draw-a-Man Test (Harris, 1963); Developmental Test of Visual-Motor
Integration (Beery & Buktenica, 1967); Memory Span for Digits - Forward,
a subtest of the Wechsler Intelligence Scale for Children (WISC) (Seashore,
Wesman & Doppelt, 1950); Northwestern University Articulation Profi-
ciency Test (Canter & Collins, unpublished research manual, 1965); Devel-
opmental Sentence Scoring (DSS) (Lee, 1974; Lee & Canter, 1971).

Language training procedures

1. Interactive Language Development Teaching (Lee et al., 1975) was


designed to approximate as closely as possible the conversational setting of
normal language development. The training rationale was based on a hy-
pothesis that normal children, during natural interaction with their parents,
spontaneously generate utterances that can be characterized by precise
grammatical rules. Several previous investigations (Bloom, 1970; Brown &
Bellugi, 1964; McNeill, 1971; Miller & Ervin, 1964) have suggested that
language acquisition often occurs as a result of parent-child interaction
about ongoing events of daily living. For example, McNeill emphasized the
importance of a "contiguity" between a child's semantic intentions and the
parent's production of appropriate grammatical structures. To capitalize on
this model of normal language acquisition a clinical setting was created
Applied Psycholinguistics 1:2 157
Friedman & Friedman: Individual reactions to language training

where interactions could be spontaneous, nonimitative, and meaningful, and


where both receptive and expressive skills could be practiced. The input
language of the teacher was simplified, repetitive, and confined to grammati-
cal structures at the child's developmental level. Teacher-child interchanges
were centered on a simple story presentation that duplicated the common,
everyday experiences of the children.
The teacher read a few lines of the story, then asked the child a question
designed to elicit a target response that incorporated one or more grammati-
cal goals of the lesson. These goals were determined from a pre-treatment
spontaneous language sample that was a requirement of the DSS screening
test. They generally involved identifying classes of pronouns, verb forms,
negatives, conjunctions, or questions that were being incorrectly used or were
one step beyond the child's current expressive language level. A sample was
selected from among the many possible structures that would be develop-
mentally appropriate for use in constructing the story and follow-up ques-
tions. Lee et al. (1975) have developed a text that provides sample stories
and suggested guidelines to highlight grammatical structures of the lesson, to
clarify meanings, and to increase the probability that a child's response will
approximate the selected target.
A different story theme was used in each lesson to present and teach
language structures. Although all of the children in the group heard the
same story, the teacher treated each child individually when asking questions
and responding to them. By carefully selecting and modifying questions
asked by each child, and by reformulating, remodeling, correcting, and
expanding each response, the teacher adapted the general procedure to
individual needs. The children were called upon in random order, and an
attempt was made to provide each child with an equal number of opportuni-
ties for response. The child was not merely asked to repeat selected lists of
sentences, but was stimulated to ask questions. It was natural for a child to
answer a question, and for someone to comment on his or her response. In
turn, it was just as natural for the child to feel free to elaborate upon, or
respond to, a question within the exchange. It was felt that this free
interchange encouraged the children to pay closer attention to grammatical
forms, to listen more closely, and to find meaning in what was being said in
the social context.
The final section of each lesson was devoted to a group activity in which
material related to the story was presented in a new form. There were
pencil-and-paper activities, which included drawing, tracing, coloring,
hiding, and finding objects, and there were imaginative activities such as
dressing up, playing store, and role playing. In all cases, the activity was
related to the theme, concepts, vocabulary, and grammatical structures
presented during the story period. Transcripts of the children's utterances
during this part of the lesson were used to judge their progress and to modify
the treatment goals in subsequent lessons. Thus, a continuity in the sequence
of lessons was achieved by an ongoing assessment of individual children
within the group.
Applied Psycholinguistics 1:2 1 58
Friedman & Friedman: Individual reactions to language training

2. Programmed Language Presentation (Koenigsknecht & Lee, 1974)


was the general term applied to a series of group oral language remediation
techniques for children demonstrating delayed syntax development. The
major teaching goal was to maximize the number of correct grammatical
constructions elicited from each child by applying information about oper-
ant, programming, and modeling procedures. The process through which
each child could achieve mastery of target syntax structures was carefully
outlined using a comprehensive organizational plan based upon a) results of
an analysis of the child's spontaneous language, b) information on normal
language acquisition, and c) carefully graded treatment steps of progressive
difficulty.
Initial program planning focused on orienting the children, within their
language performance abilities, to the programmed lesson routine. This
orientation consisted of establishing a positive attitude toward the treatment
setting and familiarizing the children with the program routines and proce-
dures. Controlled operant techniques were applied to reduce disruptive and
inappropriate behaviors and to encourage a "set" of attending behaviors.
Individualized language lessons were presented in a group format with the
teacher mediating between the instructional plan and the child's language
performance. Each program consisted of a pre-written lesson that detailed
both the stimulus and response sequences to be utilized. Simple visual
representations in the form of pictures or objects were an integral part of the
stimulus delivery. The lesson functioned on a turn-taking basis with two
20-minute periods of programming and a final 10-minute show-and-tell
period. This final segment provided a spontaneous speech activity and a time
for day-by-day assessment of carryover.
As a direct extension of the pre-treatment syntax analysis of the child's
spontaneous speech, error forms were selected as program targets. Treat-
ment was designed to reflect a normal pattern of language acquisition. Early
developing syntactic forms including indefinite pronouns, personal pronouns,
and main verbs were the initial focus of training. When these relatively
simple grammatical structures were consistently used in appropriate
contexts, the more complex secondary verbs, negatives, conjunctions, inter-
rogative reversals, and vWi-questions were emphasized.
The treatment construct not only provided a comprehensive guide to
language remediation planning, but also functioned as a hierarchy of stimu-
lus events. The target syntax was embedded in varying contexts to provide 10
different response trials. A response trial was defined as a set of interchanges
between teacher and child following the delivery of a designated stimulus.
No response trial was completed without a correct response being elicited. At
various intervals throughout the program children were requested to make
value judgments on the adequacy of their own responses and, when neces-
sary, self-correction or peer correction was encouraged.
Among the operant principles used during these programmed lessons were
a) shaping - reinforcing in small successive steps syntax productions that
resembled the desired form; b) imitation training - shifting the response to
Applied Psycholinguistics 1:2 159
Friedman & Friedman: Individual reactions to language training

imitative control once the terminal syntax construction was obtained; c)


extension of stimulus control - gradual fading in of new environmental and
contextual stimuli and fading out of verbal prompts to transfer control of
desired syntax behavior from imitative to more appropriate stimuli; d)
generalization - establishing the appropriate syntax behaviors under other
than the original training conditions. Reinforcement contingencies were
critical for this approach to a) maintain high response rates, b) encourage
appropriate attending behaviors, and c) help the children discriminate and
stabilize adequate responses.

Criterion measure

The primary treatment goal of both language training programs was to


stimulate growth in the maturity and frequency of grammatical structures
used by the child during normal conversation. It was assumed that as a child
gained linguistic competence, the grammatical load of his or her speech
would increase along two dimensions: 1) new syntactic structures would be
learned in a developmental progression (maturity), and 2) more of these
forms would be combined into each sentence (frequency). Both of these
aspects of syntax development were measured through the Developmental
Sentence Scoring (DSS) procedure. This instrument was designed to assess
the level of a child's grammatical development from a tape-recorded sample
of spontaneous speech (Lee, 1974). Recordings were made of conversation
between the child and a remedial language teacher using toys, pictures, and
nursery stories as stimuli. Only the last 50 complete, consecutive, intelligible
sentences were transcribed from each taped interaction session. A combina-
tion of frequency and maturity of pronouns, verbs, negatives, conjunctions,
and questions within these sentences was assessed with the DSS. The
developmental order for each syntax category was hypothesized based on
considerable interaction with language-delayed and normal children, and
later developing forms were assigned higher scores (Lee, 1970; Lee &
Canter, 1971). An overall mean score per transcribed sentence was called the
child's DSS. The difference in DSS scores obtained before and after training
provided an objective record of the child's progress over the eight-month
training period.

RESULTS
Pre-training analyses
Means and standard deviations by group for each of the 13 pre-training
measures are shown in Table 1. A series of two-tailed t tests (Ferguson,
1971) was computed to examine whether children within the two training
groups were comparable on these attribute measures prior to intervention.
Results of these analyses, also shown in Table 1, did not reveal significant
differences on any of these measures. Of particular importance was the
finding that groups were comparable in mean syntactic development as
Applied Psycholinguistics 1:2 160
Friedman & Friedman: Individual reactions to language training

Table 1. Comparison statistics on 13 pre-training attribute variables

Programmed Interactive Group


Grouf.> (n - 16) Group (n = 25) Comparison

Test X SD X SD i"

NSST - Expressive 9.33 6.68 5.24 4.68 1.77


NSST - Receptive 22.50 5.89 21.76 5.71 0.28
Peabody Picture
Vocabulary Test 43.75 6.34 43.12 12.62 0.13
Stanford-Binet IQ 102.00 15.49 99.88 13.97 0.36
Preschool Attainment
Record 75.50 16.83 75.03 13.36 0.07
Sentence Repetition 3.00 1.60 1.96 1.43 1.74
Oral Commissions 12.43 2.64 11.76 4.70 0.36
ITPA - Total Score 116.38 41.80 114.92 35.21 0.10
Draw-a-Man 4.38 4.00 4.68 3.24 -0.22
Visual-Motor
Integration 4.13 2.85 4.71 2.09 -0.58
Memory Span for
Digits 3.13 0.35 2.79 0.98 0.94
NU Articulation 33.75 10.81 32.04 12.31 0.35
Developmental
Sentence Score 4.75 2.07 3.84 1.53 1.34

"None of the differences was significant at the .05 level.

measured on the pre-training DSS. The average syntax delay estimated for
those in the Interactive groups was 21 months and in the Programmed
groups, 17 months. These age-score delays were determined from norms
developed by Lee (1974).
Deficiencies in pre-training performance were apparent on two additional
expressive language tasks that placed some demand on the children's syntac-
tic skills (see Table 1). At the beginning of the treatment period the mean
expressive score on the NSST was more than two standard deviations below
expectation for a 3-year-old. The average Spencer-McGrady Sentence repe-
tition score was also substantially below reported norms. Two additional
measures indicating a general impairment in this group's language skills
were the Receptive section of the NSST and the Northwestern University
Articulation Proficiency Test. On the Receptive portion of the NSST, a
screening device for sampling comprehension of specific syntactic structures,
the mean age equivalent score for all subjects was more than one year below
the group's chronological age. On the general articulation proficiency test
the children obtained a mean of only 32.5 correct single-consonant produc-
tions out of a possible 67 items. Similarly, the mean pre-treatment perform-
Applied Psycholinguistics 1:2 161
Friedman & Friedman: Individual reactions to language training

ance on the Memory Span for Digits was less than the three-digit span
normally obtained by 3-year-olds.
It is also important to consider those measures on which these children
averaged at or above their age-level expectations as a group. Mean scores
reported in Table 1 for the Stanford-Binet intelligence scale, the total ITPA,
and the Preschool Attainment Record indicated that the general level of
functioning for these children was commensurate with their chronological
ages. However, individual subtests on the ITPA did show specific problems
in language learning and usage. Pre-treatment results from two receptive
language measures were also normal. These were the Pea body Picture
Vocabulary Test, a measure of receptive vocabulary for single words, and the
Oral Commissions subtest of the DTLA, a measure of comprehension of
sequenced basic commissions. Finally, the children scored within normal
limits on the two visual-motor skill measures: the Developmental Test of
Visual-Motor Integration and the Draw-a-Man test.

Effects of remedial training

Both training groups did, on the average, show significant gains in overall
DSS scores during the training period, ?(24) = 5.59, p < .01 for the
Interactive group, and t{l5) = 2.51,/? < .05 for the Programmed group. The
mean gain per child was approximately 10.8 months over the eight-month
training period. However, a comparison of these average DSS gain scores for
the two groups did not indicate a significant effect of teaching type, f (39) =
1.30, p > .10 (two-tailed test). This main effect comparison ignored any
individual child differences that may have interacted with the two teaching
procedures. Further analyses incorporated information on the 13 pre-training
variables to assess the relative effectiveness of the two teaching approaches
at improving child syntax.

Aptitude x Treatment interactions


In order to test for Aptitude x Treatment interactions the DSS gain variable
was regressed separately within groups on each of the 13 pre-training
attribute variables. Most of the basic procedures for comparing regression
coefficients assume linearity of regression. That is, it must first be demon-
strated that the relationship between the pre-training variable and the DSS
gain score criterion was linear rather than curvilinear. A test of linearity
(Winer, 1971) was computed, by group, for each pre-training variable. A
significant departure from linearity was obtained for only one variable (NU
Articulation), and it was excluded from further analysis. For the remaining
variables a test of parallelism of slope for the two training groups was
performed (Walker & Lev, 1953). Results of these homogeneity tests, shown
in Table 2, indicated significantly different regression slopes for only three of
the aptitude measures: Stanford-Binet IQ, Initial DSS, and the Beery-
Buktenica Developmental Test of Visual-Motor Integration.
Applied Psycholinguistics 1:2 162
Friedman & Friedman: Individual reactions to language training

Table 2. Within-group correlations and tests of parallel slopes for DSS gain crite-
rion score regressed on 13 separate pre-training measures

Programmed Interactive Homogeneity


Group Group Test

Test (r) (r) (F)

NSST - Expressive .45 .02 1.11


NSST - Receptive .01 -.07 0.07
Peabody Picture Vocabulary Test -.05 -.10 0.01
Stanford-Binet IQ -.26 .50* 3.66*
Preschool Attainment Record -.23 -.12 0.11
Sentence Repetition -.49 .09 2.43
Oral Commissions -.23 .11 0.77
ITPA - Total Score -.28 -.01 0.52
Draw-a-Man -.32 .30 2.66
Visual-Motor Integration -.63** .46* 7.91**
Memory Span for Digits -.06 .01 0.04
NU Articulation -.24 -.17 a

Developmental Sentence Score -.76** -.11 4.21

"Homogeneity test was not computed because of nonlinear regression lines.


*P < .05.
*V<.oi.

Figure 1 shows that the interaction between the IQ measure and the two
teaching procedures was disordinal. That is, the two within-group regression
lines crossed over within the measured range of IQ scores. The Interactive
approach resulted in greater DSS improvement than did the Programmed
approach for children with IQs toward the top of the normal range.
Conversely, the Programmed method resulted in greater gains on the DSS
for those children with lower scores. The within-group Pearson product-
moment correlations, reported in Table 2, between IQ and DSS measures
also demonstrated the differential regression. A significant (p < .05) direct
relationship was observed for children within the Interactive group (r = .50),
and an inverse correlation within the Programmed group (r = -.26, p >
.05).
To further analyze this disordinal interaction the Johnson-Neyman tech-
nique was employed (Johnson & Neyman, 1936). This test defines a region
of homogeneity around the crossover point of two nonparallel regression
lines. In other words, a range of scores on the aptitude variable was specified
over which neither of the two teaching methods produced significantly more
improvement on the DSS criterion variable. Results of an application of this
Johnson-Neyman technique to the IQ data revealed a region of homogeneity
between 58 and 112 on the Stanford-Binet (p = .05). Although there were
nine children (22.0% of the entire sample) observed above this region of
Applied Psycholinguistics 1:2 163
Friedman & Friedman: Individual reactions to language training

3.6

3.0

S 2.4

S 1.8
a

1.2

0.6
I I I I I
70 80 90 100 110 120 1.0 2.0 3.0 4.0 5.0 6.0

STANFORD-BINET IQ INITIAL DSS VISUAL-MOTOR INTEGRATION

Figure 1. Significant disordinal interactions for DSS gain scores regressed on three
pre-treatment variables.

homogeneity, none was observed below. Because it is improper to interpret


regression lines extrapolated beyond the range of scores from which they
were developed, the lower boundary has no real meaning. Thus, any predic-
tion of an optimal treatment method for syntax improvement can be made
with less than 95% confidence for children with IQ scores below 112 on the
Stanford-Binet. However, the prediction that the Interactive approach is
optimal for children with IQ scores above 112 can be made with 95%
confidence.
When DSS gain scores were regressed on the pre-training DSS scores, the
resulting interaction of best fitting regression lines was again disordinal (see
Figure 1). For both training groups there was an inverse relationship
between initial DSS scores and DSS gain scores. That is, those children who
began training at lower syntactic levels tended to show more improvement
over the teaching period. However, this negative relationship was more
evident for those children within the Programmed group (r = —.76, p < .01)
than within the Interactive group (r = — .1 l , p > .05).
Results of the Johnson-Neyman procedure examining these regression
lines revealed a region of homogeneity between an initial DSS score of 1.6
and 5.5. Thus, any predictions of a "best method" at teaching for syntax
improvement could be made with less than 95% confidence for any child
beginning training at a DSS level between 1.6 and 5.5. However, based on
this analysis, two regions of significance were identified without extrapola-
tions from the actual data. For those children scoring below 1.6 on the DSS
screening test, the Programmed approach is predicted to optimize DSS
gains, as is the Interactive approach for those initially measuring above 5.5
(p = .05). There were two children within the sample observed below and six
children above the region of homogeneity.
The third variable showing a significant disordinal interaction with the
teaching procedures was visual-motor integration as measured on the Beery-
Buktenica test. The regression equation with the DSS gain criterion for the
Interactive group (K = A1X + 1.66) is illustrated in Figure 1. The
Applied Psycholinguistics 1:2 164
Friedman & Friedman: Individual reactions to language training

regression line for the group receiving Programmed training (Y = 31X +


2.97) appears from the figure to be strikingly different from the first one. A
significant (p < .05) direct correlation between the visual-motor and DSS
gain scores was observed with the Interactive group (/• = .46), whereas an
inverse correlation was observed with the Programmed group (r = - . 6 3 , p <
.05). Therefore, the two regression lines crossed at a point corresponding to a
visual-motor score of about 2.4. Only one region of significance was identi-
fied through the Johnson-Neyman procedure (p = .05). This region
extended beyond a visual-motor score of about 6.2. When children had
visual-motor scores above this point of significance (about 18% of the entire
sample), syntax improvement, as measured by DSS, was predicted with 95%
accuracy to be optimized by the Interactive approach.

DISCUSSION

The findings of this study indicated that research efforts examining the
relative merits of different remedial language training procedures may be
enhanced by the inclusion of individual difference variables in their designs.
Rather than accounting for a child's characteristics, which may differen-
tially interact with specific training procedures, most language training
comparison studies have simply discussed them as error variance. This
unexplained variability within groups could serve only to statistically mask
treatment differences that might otherwise have proven important. Perhaps
there are very few main effects to be found, and the relative effectiveness of
different intervention procedures will be observed only when successful
attempts are made to account for individual differences.
Although some experimenters of remedial language training procedures
have incorporated individual characteristic variables into their research, they
have done so by simply treating these variables as additional factors in an
analysis of variance design. For example, to examine the relationship
between intelligence and DSS gain scores, or its interaction with the two
language training procedures, the entire subject pool might have initially
been stratified into high- and low-IQ groups. Children within each intelli-
gence group would then be randomly assigned to the two language interven-
tion techniques. The resulting Treatments x IQ Levels factorial analysis of
variance would have been a less powerful analysis procedure than a compari-
son of within-group regression lines. Creating artificial levels of a continuous
individual difference variable, such as IQ, results in the loss of considerable
information. By lumping high intelligent children into a single group, any
individual difference effects within that group have been lost. There is also
some difficulty in deciding on the exact point of dichotomy between groups
of high and low intelligent children. Should the data be examined for a
natural split in the score distribution, or should a measure of central
tendency be used as a cutoff point? No matter how a group division is
determined, it will inevitably lead to placing subjects who differ by only a
few points on some attribute measure into different classification groups. In
Applied Psycholinguistics 1:2 165
Friedman & Friedman: Individual reactions to language training

addition, when starting out with typically small treatment samples, dividing
subjects into even smaller subgroups makes it more difficult to satisfy the
underlying distribution assumptions for appropriately using most parametric
statistics.
A variety of previous methods studies have demonstrated that global,
multifactor measures are less likely to show significant Aptitude x Treat-
ment interactions than unifactor measures of specific subject characteristics
(see Bracht, 1970). Therefore, the significant disordinal interaction observed
between intelligence scores and the language procedures was not expected.
However, for children who scored high within this group on the intelligence
measure, the Interactive approach does, on the surface, seem to be more
effective than the Programmed approach. It is certainly not as repetitive or
drill-like, and appears less likely to bore higher IQ children. In addition, it is
possible that to be effective the Interactive story procedure requires children
to come into training with a higher level of conceptual skill. Some of the
lower IQ children may have been confused by the relatively complicated
discussion procedures of the Interactive approach.
A major finding of this study was the inverse relationship for both training
groups between initial syntax level and amount of syntactic growth ^s
measured by the DSS. That is, children who scored lowest on the DSS
procedure prior to either teaching approach showed the largest gains during
training. The statistical artifact of regression toward the mean, based on
imperfect test-retest correlations, might at least partially explain these
negative coefficients. One criterion for determining that a child was delayed
in expressive language and in need of remedial training was an abnormally
low score on the DSS screening task. Since the DSS is not a perfectly reliable
instrument, any errors of measurement during this initial screening would
tend to be consistently biased against those who scored poorly on the test.
This was particularly true for children measuring at the very lowest levels,
who were then selected for remedial training. Their scores would have the
greatest tendency to regress upward when retested with the same instrument.
It is also possible that children who initially scored at the lowest syntax
levels simply demonstrated a spontaneous spurt in language development
that happened to coincide with the training period. This is the well-known
phenomenon of the "late bloomer." However, since children were arbitrarily
assigned to the two teaching procedures, such a spontaneous language jump
would not explain the stronger inverse relationship observed within the
Programmed group.
It seems more likely that the nature of the DSS instrument itself may have
prevented the more advanced children from demonstrating syntactic growth
that had actually occurred. The hierarchical maturity scales developed for
each of the DSS syntax categories may be interpreted only as ordinal
measures. That is, a jump in score from 1 to 3 on any category might not
reflect the same magnitude of syntactic development as a jump from 5 to 6.
It is entirely possible that numerical DSS gains at the higher category stages
were considerably more difficult to make than gains at lower levels. This
Applied Psycholinguistics 1:2 166
Friedman & Friedman: Individual reactions to language training

would allow children who were beginning their training at less mature stages
of grammatical development to more easily show quantitative improvement
than those at a more mature stage. In addition, if one intervention procedure
was more effective at high levels of syntax construction, and the other at
lower levels, this would be reflected in differential correlation coefficients.
Such confounding effects as regression artifacts and maturational gains
could be more easily explained if a control group of untrained language-
disordered children were administered the pre- and post-tests. However, in
the area of child language disabilities, clinical researchers attempting to
establish the effectiveness of their procedures face an important limitation:
They cannot ethically use control groups. It would be unconscionable to
withhold clinical training for any period of time from a child who needs it
and might gain from it. This preoludes designs in which one group of
children receives treatment while treatment is withheld from a comparable
group in order to show that the method produced results.
That a visual-motor aptitude measure significantly interacted with the
teaching methods was not at all surprising. The Beery-Buktenica, through a
copying task, measures the child's ability to actively demonstrate visual-
motor skills. It must be pointed out that the Interactive approach lessons
were designed to employ a combination of media to get the training message
across. In fact, the use of visual-motor skills was specifically incorporated at
the end of each lesson. Children with more advanced visual-motor skills
might have looked forward to this portion of the lesson, and were, therefore,
more motivated to participate in the preceding story telling. On the other
hand, the Programmed approach required a passive reaction to verbal and
pictorial stimuli, rather than an active demonstration of motor skills. Chil-
dren who were inclined to more active visual-motor behaviors may have lost
interest in this procedure, resulting in an inverse correlation between visual-
motor integration scores and DSS gains.
Although several interpretations have been presented for the three signifi-
cant disordinal interactions, some discussion seems necessary of why so many
of the individual difference measures did not interact with the treatments.
This is particularly critical for the other measures of language abilities. Why
wouldn't high scorers on the NSST Expressive or Receptive, or any of the
other developmental language measures, behave similarly to high scorers on
the DSS? It must be emphasized that significantly different training-group
slopes were obtained for only 3 of the 13 pre-treatment variables.
One explanation is that, when the two contrasting types of language
training were designed, not enough thought was given to analyzing the data
for Aptitude x Treatment interactions. An effort should have been made to
avoid overlapping procedures in the contrasting methods. For example,
although teacher reinforcements were not carefully controlled during the
Interactive training, they were extensively used to stimulate "natural"
interaction. It was also inefficient first to identify alternative training
approaches and then through trial and error to find individual difference
variables to interact with these teaching methods. The analysis of an
Applied Psycholinguistics 1:2 167
Friedman & Friedman: Individual reactions to language training

interaction effect should be a carefully planned part of the development of


intervention programs, rather than an afterthought. To be differentially
effective for children with divergent histories and aptitudes the alternative
teaching methods should demand different child attributes for successful
performance. Unfortunately, language comprehension, memory, good articu-
lation, and other skills were required, to some extent, within both language
treatment approaches. For greater success, an attempt has to be made to
develop contrasting treatments in which different child processes are relevant
to the same training objectives.
Grouping the different tests according to the function being measured
provided further insight about the results. As can be seen in Table 2, other
tests within the battery that measured expressive language functions showed
differential within-treatment correlation patterns similar to the pre-
treatment DSS. The same was true for the Draw-a-Man test relative to the
test for visual-motor integration. Perhaps such factors as the small sample
size, smaller within-group variabilities, and specific scoring idiosyncrasies
made it more difficult to locate statistically significant ATIs for certain tests.
Although measures of expressive language, visual-motor skills, and general
intelligence all seemed good candidates for ATIs, those tapping receptive
language functions, memory, articulation, or other more global characteris-
tics did not differentiate the effectiveness of these language treatment
approaches. Further research is necessary to determine the relative impor-
tance of the abilities tapped by the measures, and the specific scoring
characteristics of the measures themselves, for obtaining significant Aptitude
x Treatment interactions.

ACKNOWLEDGMENTS
Some of the observed training groups were part of the Modified Language Input
project supported by Grant #OEG-0-71-0949 from the Office of Education. The
authors wish to thank Laura L. Lee and Roy A. Koenigsknecht for suggestions about
the manuscript, and Susan T. Mulhern for aid in collection of the data. Requests for
reprints should be sent to Philip Friedman, Department of Psychoeducational Stud-
ies, Howard University, Washington, D.C. 20059.

REFERENCES
Baker, H. J., & Leland, B. Detroit tests of learning aptitude. Indianapolis, Ind.: Bobbs-Merrill,
1967.
Bates, E. Pragmatics and sociolinguistics in child language. In D. Morehead and A. Morehead
(Eds.), Normal and deficient child language. Baltimore: University Park Press, 1976.
Beery, K. E., & Buktenica, N. A. Developmental test of visual-motor integration. Chicago:
Follett, 1967.
Bloom, L. Language development: Form and function in emerging grammars. Cambridge,
Mass.: MIT Press, 1970.
Bracht, G. H. Experimental factors related to aptitude-treatment interactions. Review of
Educational Research, 1970,40, 627-645.
Bricker, W., & Bricker, D. Early language training strategy. In R. Schiefelbusch and L. Lloyd
Applied Psycholinguistics 1:2 168
Friedman & Friedman: Individual reactions to language training

(Eds.), Language perspectives - acquisition, retardation and intervention. Baltimore:


University Park Press, 1974.
Brookshire, R. H. Speech pathology and the experimental analysis of behavior. Journal of
Speech and Hearing Disorders. 1967, 32, 215-277.
Brown, R. The development of wA-questions in child speech. Journal of Verbal Learning and
Verbal Behavior, 1968, 7, 279-290.
A first language: The early stages. Cambridge, Mass.: Harvard University Press, 1973.
Brown, R., & Bellugi, U. Three processes in the child's acquisition of syntax. Harvard
Educational Review, 1964,54, 133-151.
Canter, G. J., & Collins, N. H. Northwestern University articulation proficiency test: Conso-
nant singles. Unpublished research, 1965.
Carrier, J. Application of functional analysis and non-speech response mode to teaching
language. In L. McReynolds (Ed.), Developing systematic procedures for teaching
children's language. Washington: American Speech and Hearing Association Mono-
graph No. 18, 1974,47-95.
Cazden, C. Environmental assistance to the child's acquisition of grammar. Unpublished
doctoral dissertation. Harvard University, 1965.
The acquisition of noun and verb inflections. Child Development. 1968, 39, 433-448.
Chomsky, N. Syntactic structures. The Hague: Mouton, 1957.
Aspects of the theory of syntax. Cambridge, Mass.: MIT Press, 1965.
Cronbach, L. J., & Snow, R. E. Individual differences in learning ability as a function of
instructional variables. United States Office of Education final report. Palo Alto:
Stanford University, 1969.
Dalali, I. D., & Sheehan, J. G. Stuttering and assertion training. Journal of Communication
Disorders, 1974,7,97-113.
Doll, E. A. Preschool attainment record. Circle Pines, Minn.: American Guidance Services,
1966.
Dowaliby, F. J., & Schumer, H. Teacher-centered versus student-centered mode of college
classroom instruction as related to manifest anxiety. Journal of Educational Psychology,
1973,64, 125-132.
Dunn, L. M. Peabody picture vocabulary test. Circle Pines, Minn.: American Guidance
Services, 1959.
Ferguson, G. A. Statistical analysis in psychology and education. New York: McGraw-Hill,
1971.
Fokes, J. Fok.es sentence builders. New York: New York Times Teaching Resources, 1976.
Girardeau, F. L., & Spradlin, J. E. A functional analysis approach to speech and language.
American Speech and Hearing Association Monograph, No. 14, 1970.
Gray, B. B., & Fygetakis, L. The development of language as a function of programmed
conditioning. Behavioral Research and Therapy, 1968, 6, 455-460.
Harris, D. B. Goodenough-Harris drawing test manual. New York: Harcourt, Brace and
World, 1963.
Harris, D. B., & Hassemer, W. G. Some factors affecting the complexity of children's
sentences: The effects of modeling, age, sex, and bilingualism. Journal of Experimental
Child Psychology, 1972,13, 447-455.
Holland, A. L. Some applications of behavioral principles to clinical speech problems. Journal
of Speech and Hearing Disorders, 1967,52, 11-18.
Ingram, D., & Eisenson, J. Therapeutic approaches III: Establishing and developing language
in congenitally aphasic children. In J. Eisenson (Ed.), Aphasia in children. New York:
Harper & Row, 1972.
Johnson, P. O., & Neyman, J. Tests of certain linear hypotheses and their applications to some
educational problems. Statistical Research Memoirs, 1936, /, 57-93.
Kent, L., Klein, D., Falk, A., & Guenther, H. A language acquisition program for the retarded.
In J. McLean, D. Yoder, and R. Schiefelbusch (Eds.), Language intervention with the
retarded. Baltimore: University Park Press, 1972.
Kirk, S., McCarthy, J., & Kirk, W. The Illinois test of psycholinguistic abilities. Urbana, III.:
University of Illinois Press, 1968.
Applied Psycholinguistics 1:2 169
Friedman & Friedman: Individual reactions to language training

Klima, E. S., & Bellugi, U. Syntactic regularities in the speech of children. In J. Tyons and R. J.
Wales, (Eds.), Psycholinguistics papers. Edinburgh University Press, 1966.
Koenigsknecht, R. A., & Friedman, P. Syntax development in boys and girls. Child Develop-
ment, 1916,4, 1109-1115.
Koenigsknecht, R. A., Friedman, P., & Hill, D. Developmental sentence scoring procedures
used in repeated measurement of spontaneous syntax usage of preschool children: A
longitudinal study. Paper presented at a meeting of the American Speech and Hearing
Association, Detroit, 1973.
Koenigsknecht, R. A., & Lee, L. L. Validity and reliability of developmental sentence scoring:
A method for measuring syntactic development in children's spontaneous speech. Paper
presented at a meeting of the American Speech and Hearing Association, Chicago, 1971.
The assessment of grammatical development in children and the clinical presentation of
grammatical structure to children with language problems. United States Office of
Education final report. Evanston, 111.: Northwestern University Press, 1974.
Lee, L. L. The Northwestern syntax screening test. Evanston, 111.: Northwestern University
Press, 1969, 1971.
A screening test for syntax development. Journal of Speech and Hearing Disorders, 1970,
55,103-112.
Developmental sentence analysis: A grammatical assessment procedure for speech and
language clinicians. Evanston, 111.: Northwestern University Press, 1974.
Lee, L. L., & Canter, S. M. Developmental sentence scoring: A clinical procedure for
estimating syntactic development in children's spontaneous speech. Journal of Speech
and Hearing Disorders, 1971, 36. 315-340.
Lee, L. L., Koenigsknecht, R. A., & Mulhern, S. Interactive language development teaching.
Evanston, III.: Northwestern University Press, 1975.
Liebert, R. M., Odom, R. D., Hill, J. H., & Hun", R. L. Effects of age and rule familiarity on
the production of modeled language constructions. Developmental Psychology, 1969, /,
108-112.
Lovaas, O. I., Berberich, J. P., Perloff, B. F., & Schaeffer, B. Acquisition of imitative speech by
schizophrenic children. Science, 1966, 151, 705-707.
MacDonald, J., & Blott, J. Environmental language intervention: The rationale for a diagnostic
and training strategy through rules, context, and generalization. Journal of Speech and
Hearing Disorders, 1974, 39, 244-256.
Martin, R., & Haroldson, S. The effects of two treatment procedures on stuttering. Journal of
Communication Disorders, 1969, 2, 115-125.
McGrady, H. Verbal and nonverbal functions of school children with speech and language
disorders. Ph.D. dissertation, Northwestern University, 1964.
McNeill, D. The acquisition of language: The study of developmental psycholinguistics. New
York: Harper & Row, 1970.
The capacity for the ontogenesis of grammar. In D. Slobin (Ed.), The ontogenesis of
grammar. New York: Academic Press, 1971.
Miller, J., & Yoder, D. An ontogenetic language teaching strategy for retarded children. In R.
Schiefelbusch and L. Lloyd (Eds.), Language perspectives - acquisition, retardation,
and intervention. Baltimore: University Park Press, 1974.
Miller, W., & Ervin, S. The development of grammar in child language. Child Development
Monographs, 1964, 29, 9-34.
Monsees, E. Structured language. Washington: Children's Speech and Hearing Center, 1970.
Muma, J. The communication game: Dump and play. Journal of Speech and Hearing
Disorders. 1975, 40, 296-300.
Peins, M., McGough, W. E., & Lee, B. S. Evaluation of a taperecorded method of stuttering
therapy: Improvement in speaking task. Journal of Speech and Hearing Disorders, 1972,
15. 364-371.
Risley, T., & Wolf, M. Establishing functional speech in echolalic children. Behavioral
Research and Therapy, 1967, 5, 73-88.
Rosenthal, T. L., and Whitebrook, J. S. Incentives versus instructions in transmitting grammati-
cal parameters with experimenter as model. Behavioral Research and Therapy, 1970, 8,
189-196.
Applied Psycholinguistics 1:2 170
Friedman & Friedman: Individual reactions to language training

Sarno, M. T., Silverman, M., & Sands, E. Speech therapy and language recovery in severe
aphasia. Journal of Speech and Hearing Disorders, 1970, 13, 607-623.
Schlesinger, I. Relational concepts underlying language. In R. Schiefelbusch and L. Lloyd
(Eds.), Language perspectives — acquisition, retardation and intervention. Baltimore:
University Park Press, 1974.
Seashore, H. G., Wesman, A. G., & Doppelt, J. E. The standardization of the Wechsler
Intelligence Scale for Children. Journal of Consulting Psychology, 1950, 14, 99-110.
Shavelson, R. J., Berliner, D. C , Ravitch, M. M., & Loeding, D. Effects of position and type of
question on learning from prose material: Interaction of treatments with individual
differences. Journal of Educational Psychology, 1974, 66, 40-48.
Sommers, R. K. Effectiveness of individual versus group articulation therapy. Paper presented
at a meeting of the American Speech and Hearing Association, San Francisco, 1964.
Spencer, E. M. An investigation of the maturation of various factors of auditory perception in
preschool children. Ph.D. dissertation, Northwestern University, 1958.
Terman, L. M., & Merrill, M. A. Stanford-Binet Intelligence Scale. Boston: Houghton Mifflin,
1960.
Tobias, S. Sequence, familiarity, and attribute by treatment interaction in programmed instruc-
tion. Journal of Educational Psychology, 1973, 64, 133-141.
Walker, H. M., & Lev, J. Statistical inference. New York: Holt, 1953.
Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1971.
Zwitman, D. H., & Sonderman, J. C. A syntax program design to present base linguistic
structures to language-disordered children. Journal of Communication Disorders, 1979,
12. 323-335.

Revision received: 28 January 1980

You might also like