Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

J Behav Educ (2010) 19:62–75

DOI 10.1007/s10864-010-9097-2

ORIGINAL PAPER

An Examination of Two Techniques for Promoting


Response Generalization of Early Literacy Skills

Gary J. Duhon • Sara E. House • Brian C. Poncy •

Kim W. Hastings • Sally C. McClurg

Published online: 28 January 2010


Ó Springer Science+Business Media, LLC 2010

Abstract This study examined the application of two generalization procedures


designed to promote generalized responding across two early literacy skills. Letter
sound fluency was targeted using direct intervention for three subjects within a
multiple baseline design. After instruction was complete, two generalization pro-
cedures (cueing and providing sufficient response exemplars) were tested in an
increasingly intense application of the procedures to determine whether sufficient
levels of response generalization to letter sound blending could be programmed.
Results indicated that although some subjects demonstrated minor levels of
unprogrammed or spontaneous generalization, more complete levels of general-
ization were accomplished only once generalization techniques were employed.

Keywords Generalization  Early literacy  Cueing  Response generalization

Introduction

Generalization has been defined as ‘‘the occurrence of relevant behavior under


different nontraining conditions.’’ (Stokes and Baer 1977, p. 350) Generalization is
critical in education for the development of academic behaviors that students will
use in a variety of settings and across different tasks. The concept of generalization
has further been divided into generalization across environments, settings, or
materials (stimulus generalization) and generalization across skills (response
generalization). Stimulus generalization occurs when ‘‘the target behavior is
emitted in the presence of stimulus conditions other than those in which it was
trained.’’ (Cooper et al. 2007, p. 617) Response generalization is the ‘‘extent to

G. J. Duhon (&)  S. E. House  B. C. Poncy  K. W. Hastings  S. C. McClurg


School of Applied Health and Educational Psychology, Oklahoma State University,
423 Willard Hall, Stillwater, OK 74078, USA
e-mail: gary.duhon@okstate.edu

123
J Behav Educ (2010) 19:62–75 63

which a learner emits untrained responses that are functionally equivalent to the
trained target.’’ (Cooper et al. 2007, p. 620) Both forms of generalization are
desirable in academic responding, and examples of each are evident in research on
literacy instruction. For example, students are often taught to read words in isolation
and are then expected to generalize this skill to either novel situations or novel tasks.
When students respond accurately to words imbedded in text that they were taught
to read in isolation, this can be described as stimulus generalization (e.g., Nist and
Joseph 2008). However, when students spell words never instructed once they have
been taught to read these words, this is referred to as response generalization (e.g.,
Noell et al. 2006). When generalization does occur, it provides a savings of time and
effort in teaching, because students do not have to be explicitly taught every
instance in which a behavior should be emitted. However, research has shown that
to ensure generalization, explicit programming is often required (Berends and
Reitsma 2006; Mesmer et al. 2007; Thaler et al. 2004).
In most academic research where generalization is of interest, researchers employ
a ‘‘train and hope’’ method (e.g., Bonfiglio et al. 2004; Martens et al. 2007), in
which the occurrence of generalization is measured after intervention and without
explicitly programming for it. These ‘‘train and hope’’ methods have demonstrated
that some degree of unprogrammed generalization may occur between academic
skills (Bonfiglio et al. 2004; Martens et al. 2007; Noell et al. 2006). However, other
research has demonstrated that without explicit programming, students are unlikely
to fully generalize gains to novel academic stimuli (Daly et al. 1999; Eckert et al.
2002). Even when generalization does occur across skills without explicit
programming, it is often incomplete. In a study that examined generalization of
whole word instruction from reading to spelling and spelling to reading, Noell et al.
found that the students made gains on both of the generalization skills without
explicit programming. However, the students were not able to attain the same rate of
accuracy that they had with the explicitly taught reading and spelling words.
Overall, the academic research on generalization provides support for Stokes and
Baer’s (1977) conclusion that ‘‘train and hope’’ methods are insufficient for ensuring
that generalization will occur.
Research on generalization across academic skills has largely focused on oral
reading fluency (e.g., Ardoin et al. 2007, 2008; Daly et al. 1998; Weinstein and
Cooke 1992); however, oral reading fluency is only one component of reading
instruction (National Reading Panel 2000). Reading typically begins with phonemic
awareness and understanding of the alphabetic principle (National Reading Panel).
The skills of letter sound fluency and blending letter sounds have been shown to be
the reliable predictors of later oral reading fluency (Harn et al. 2008; Ritchey and
Speece 2006). Specifically, research has demonstrated that letter sound fluency at
the beginning of kindergarten accounts for a unique and significant proportion of
variance in end of kindergarten reading and spelling performance (Ritchey and
Speece). Also, blending of letter sounds, as measured by modified scoring
procedures for nonsense word fluency, has been shown to be an effective predictor
of later oral reading fluency achievement (Harn et al.).
Most of these studies have focused on the predictive nature of early literacy skills
for future oral reading fluency. Few studies have focused on how learning specific

123
64 J Behav Educ (2010) 19:62–75

early literacy skills may actually generalize to more advanced decoding skills. One
study that examined the generalization of early literacy principles to reading whole
words demonstrated that stimulus control over small units of phonemes in nonsense
words generalized to reading real words with the same phoneme units (Daly et al.
2004). In this study, two sets of nonsense words were learned by participants. The
first set of nonsense words was taught to students as a single unit, and students
responded to the nonsense words as a whole. Participants leaned the second set of
nonsense words by first learning the individual phonemes within the nonsense words
prior to blending these phonemes into the single unit response. Results indicated
that when the individual phonemes were learned, generalization occurred in the
presence of other nontrained real words consisting of the individual phonemes.
Ultimately, maintaining tight stimulus control over small aspects of a behavior
assisted in facilitating generalization to related stimuli. However, the study did not
examine the relationship between related early literacy skills.
Little is known about how learning basic skills, such as letter sound
correspondence, relates to improvements in more advanced skills such as letter
sound blending. One conceptualization is that this is a form of response
generalization, in which a more basic skill shares common elements with the more
advanced skill. As such, strengthening of the basic skill will also strengthen the
advanced behavior through its shared component behavior. Therefore, ‘‘training in
one area of skilled behavior may improve performance in another’’ (Skinner 1953,
p. 94). This appears to be an underlying assumption in education in that teaching
basic phonetic skills will lead to more advanced reading responses, but more
research is needed to understand the relationships between these skills. It is
unknown how strengthening which, if any, basic skills will lead to generalized
improvements in more advanced skill areas, or if in any given case, explicit
instruction is necessary.
The present study examined the relationship of increasing fluency in one skill,
letter sound fluency (LSF), with the development of a related but more advanced
early literacy skill, blending letter sounds. These two skills, at a minimum, share the
component of letter sound correspondence. However, it is unknown how improve-
ments in the basic skill of LSF will affect performance in blending of letter sounds.
The research on academic generalization demonstrates that some degree of
spontaneous generalization may occur, but that it is often incomplete (e.g., Daly
et al. 1999; Martens et al. 2007; Noell et al. 2006). This spontaneous generalization
may occur for a variety of reasons, but due to the nature of the concept, it is only
possible to speculate as to why this may occur. One possible reason some degree of
spontaneous generalization may occur is that the skill of sound blending contains
both letter sound correspondence and blending of these sounds. Once participants
reach proficiency on letter sound correspondence, the skill of blending those letters
may be more accurately or easily displayed without any apparent programming.
Therefore, the first goal of this study was to examine the degree to which
generalization occurred across the two related academic tasks, LSF to blending of
letter sounds without explicit programming. Based on the literature, it is unlikely
that LSF will produce a significant amount of generalization to blending of letter
sounds without some amount of explicit programming. Therefore, the second goal

123
J Behav Educ (2010) 19:62–75 65

of this study was to evaluate the efficacy of two techniques (cued responding and
exemplar training) designed to increase the degree of generalization of LSF to letter
sound blending. The first of the two techniques is simply to cue generalization or
‘‘instructing the learner to generalize’’ (Cooper et al. 2007, p. 648). A lack of
generalization may sometimes be the result of an inability to recognize the
opportunity to generalize. With cued generalization, the learner is simply informed
of this opportunity. In some cases, the cue was enhanced through the addition of
goal setting. The resulting combination of a cue with goal setting continued to serve
as a cue. However, goal setting was provided to determine whether a more effective
method of promoting generalization could be used while still being less intrusive
than the second technique. The second technique, provide sufficient response
exemplars (Stokes and Osnes 1989), consists of training that employs a sampling of
the total set of responses in the response class being taught. In this case, training
over a smaller sample of responses may be sufficient to produce generalization to
previously untaught behaviors. In other words, this technique provides models of the
generalized response and allows the participant to observe and practice the behavior
of interest. Based on the goals of this study, two research questions were examined.
First, will spontaneous generalization of letter sound blending occur when letter
sound fluency is increased to a criterion level of mastery? It was hypothesized that
spontaneous generalization would occur in some cases, but that it would not reach
that of the letter sound fluency criterion. The second question addressed in this
research was will the techniques of cued responding or exemplar training improve
the degree of generalization of letter sounds to letter sound blending? It was
hypothesized that these two techniques would improve the degree of generalization
of letter sound blending observed.

Method

Participants and Setting

Participants were three-first-grade students in general education who were referred


by their teachers for concerns with pre-literacy skills as part of the school district’s
pre-referral process. Samantha was a 7-year-old Caucasian girl. Alex was an 8-year-
old multi-racial boy who had been retained once in Kindergarten, and Lucy was a
7-year-old Caucasian girl currently repeating the first grade. All students were
receiving supplemental reading services for 30 min per day 5 days a week. The
supplemental reading services consisted of instruction in broad literacy constructs
unrelated to early literacy concepts. None of the participants had been diagnosed
with a learning disability at the time of the study.
Students were recruited from three different schools in a small Midwestern town.
After the start of the study, Alex was transferred to the same school as Lucy but was
in a different classroom. All sessions of the interventions took place in quiet areas,
either in the library or an unoccupied office, removed from classroom instruction
and were conducted by trained school psychology doctoral students. The
interventionists were in the process of completing an advanced school-based

123
66 J Behav Educ (2010) 19:62–75

practicum. Prior to the beginning of the study, the interventionist completed didactic
and supervised experiential training for implementation of DIBELS and AIMSweb
assessment systems. Each intervention session was 10–15 min long and was
implemented individually with each student one time a day, 5 times a week. Alex
received 42 intervention sessions, Lucy received 37 intervention sessions, and
Samantha received 45 intervention sessions.

Materials

Materials included AIMSweb (Shinn and Shinn 2002) first grade letter sound
fluency (LSF) progress monitoring probes. On each LSF probe, there were 10 rows
of 10 lowercase printed letters, for a total of 100 letters per probe. Reliability
estimates of letter sound fluency indicate that it has sufficient test–retest reliability
(.83) and equivalent form reliability (.82) (Elliott et al. 2001). Furthermore,
concurrent validity is moderate with the Test of Phonological awareness (.68) and
the Woodcock Johnson Psychoeducational Achievement Battery Letter-Word
Identification subtest (.62; Elliott et al. 2001). The criterion for success on the
LSF probes was set at 52 correct letter sounds per minute. This criterion was derived
from the AIMSweb System yearly norms for first grade performance on the LSF
Spring Benchmark for the 2007–2008 school year.
DIBELS (Good and Kaminski 2002) first grade nonsense word fluency (NWF)
probes were also utilized. Each probe consisted of 10 rows of five nonsense words,
for a total of 50 words per probe. Each nonsense word was two to three letters long
and included basic letter sounds, such as ‘‘sim,’’ ‘‘lat,’’ and ‘‘ab.’’ The 1-month,
alternate-form reliability for DIBELS NWF is .83 (Good et al. 2004). The norming
population for the NWF measure benchmarks consisted of 3,648 students from
schools that participated in the DIBELS Web data system as of May 20th, 2002
(Good et al. 2002). The benchmark level of performance for end of year first grade
performance on NWF is 49 Complete Letter Sounds. However, LSB is not the
standard measurement represented for this criterion; therefore, a specific criterion of
success on this measure could not be determined a priori.
A prize box was utilized, which contained small tangible items, such as pencils,
erasers, and stickers. The prizes included in the box were approved by each
student’s teacher and selected according to the student’s personal interest. For
example, Alex liked Spiderman, therefore Spiderman pencils and stickers were
included in the prize box for him. These prizes were employed as contingent
rewards when individuals exceeded daily goals during the intervention process.

Dependent Variables

The dependent variables assessed in this study were the number of correct letter
sounds identified in 1 min on AIMSweb first grade LSF probes henceforth referred
to as letter sound fluency (LSF) and the number of letter sounds blended (LSB)
correctly on DIBELS first grade NWF probes in 1 min. Sounds were counted as
accurate on the LSF measure when the student provided the most common letter
sound within 3 s. With LSB, sounds were counted as accurate when the student

123
J Behav Educ (2010) 19:62–75 67

correctly blended together at least two of the most common letter sounds within 3 s
for each nonsense word presented. The student received credit for each letter
included in the blend. For example, reading S-I-M as ‘‘sim’’ would count as three
accurately blended sounds, whereas reading it as ‘‘im’’ would count as two
accurately blended sounds.

Experimental Design

To establish experimental control of both the treatment skills and the generalization
skills, a multiple baseline design across participants was utilized (Harris and
Graham 1985). The multiple baseline design across participants was used to
establish the effects of the initial letter sound treatment implemented during the LSF
phase and the effects of generalization procedures on LSB during the generalization
phase. The use of a multiple baseline design across students allowed for the
evaluation of both the treatment effects across students and spontaneous general-
ization across skills.

Baseline

Prior to the beginning of the study, all three students were able to accurately identify
100% of the letter sounds. During baseline, the students were presented with one
LSF probe and one NWF probe for each session. The standard directions developed
by AIMSweb (Shinn and Shinn 2002) for LSF and by DIBELS (Good and Kaminski
2002) for NWF were used during the baseline administration.

Letter Sound Fluency Intervention

The purpose of the letter sound fluency intervention was to increase the number of
correct letter sounds per min on the LSF probes. In addition, letter sound blending
(LSB) would also be evaluated to determine the degree of LSB generalization
occurring in the absence of explicit training. During the LSF intervention, students
were given a repeated-practice with goal-setting intervention to increase their
fluency on letter sound recognition. The students practiced the letter sounds on three
different LSF probes each for 1 min using the standard AIMSweb directions (Shinn
and Shinn 2002). For each probe, the students were instructed to tell the sounds of as
many letters as they could, and if they came to a sound they did not know the
interventionists would tell it to them. After each practice session, students were
provided with error correction on sounds they missed. The students were then given
a fourth LSF probe with a goal statement. The goal was the mean of the previous
3 days’ scores on the goal or baseline probes. The students were told if they beat the
goal, they would receive a prize from the prize box. Students continued in the LSF
phase until they met the preset criterion of mean performance of 52 correct letter
sounds per min over three consecutive sessions. In addition, once the LSF
intervention was complete, a nonsense word probe was administered using the
standard DIBELS directions to asses for spontaneous generalization of LSB.

123
68 J Behav Educ (2010) 19:62–75

Generalization Interventions

Once the students met the criterion for the LSF phase, they began the generalization
procedures for LSB on the nonsense word probes. The purpose of the generalization
procedures was to increase the number of sound blends on the nonsense word probes
with a technique that required minimal amounts of training. The first generalization
procedure was an antecedent-based cueing procedure that simply provided a cue
that instructed the student to generalize. If students responded to the cueing
procedure, a goal-setting component was then added to maximize the amount of
generalization on LSB. If students did not respond to the cueing procedure, the
alternative and more intensive procedure of providing sufficient response exemplars
was utilized to directly target the LSB.

Cue Generalization

During this phase, the students were given one NWF probe that was administered
with modified DIBELS instructions. The standard instructions give the students the
option of either saying each letter sound or the whole word. These directions were
modified to cue the students to blend the words by instructing them to ‘‘read the
whole word’’ and omitting the option to say each letter sound. Therefore, students
were given the NWF probe and instructed that the words were make-believe words,
and they should read the whole words the best they could. If students misread a
word or did not say a word in 3 s, it was marked as an error, but students were not
provided with error correction.

Cue Generalization with Goal Setting

During this phase, a goal, which was based on the mean LSB score of the previous
three NWF probes, was added to the cueing procedure. The students were informed
of their goal at the beginning of each session and told that if they beat their goal,
they would be allowed to select a reward from the prize box. At the end of the
sessions, if the students beat the mean score of their previous three NWF probes,
they were allowed to pick one prize from the prize box. If the students did not beat
their goal, they were told that they would have the opportunity to get a prize the next
day.

Sufficient Response Exemplars

In this procedure, blending nonsense words was directly taught using modeling and
guided practice with corrective feedback. During the modeling and practice
procedure, the student was given a NWF probe. On the first five words of the probe,
the interventionist provided exemplars of correct blending by modeling the
nonsense word. The student was then instructed to repeat the five words previously
modeled. If the student incorrectly read the nonsense word, the word was modeled
again, and the student was asked to repeat the word. The student was then asked to
read the next two words independently without modeling. If they were not able to

123
J Behav Educ (2010) 19:62–75 69

correctly read the two words, the interventionist repeated the modeling and guided
practice procedures for the next five words. If the student correctly read the two
words, the interventionist administered a new NWF probe using directions identical
to the cueing phase.

Treatment Integrity and Inter-Scorer Agreement

Procedural integrity data were collected for a total of 33% of all phases of
intervention by an independent observer. For each of these phases, an independent
observer watched the intervention sessions and checked off steps on an intervention
protocol as the examiner completed each step. There were 18 steps for the LSF
intervention, eight steps for the blending and blending with goal interventions, and
16 steps for the sufficient response exemplar sessions. Based on step-by-step
agreement, the procedural integrity was 100% for each intervention phase.
Furthermore, inter-scorer agreement was calculated. An independent scorer
re-scored 33% of the LSF probes and 35% of the NWF probes. There was 100%
agreement on the scores for the LSF probes and 91% agreement for the NWF
probes.

Results

Baseline

Participant responses to both the LSF intervention and the generalization


interventions are displayed in Fig. 1, and Table 1 provides the means and standard
deviations of the baseline, treatment, and final generalization phase for each
participant. During baseline, all three participants demonstrated LSF that was well
below the goal of 52 letter sounds per min. Mean performance for Alex was 21 letter
sounds per min with only three LSB correctly per min. Lucy’s mean performance
during baseline was 20 letter sounds per min and one LSB. Mean letter sounds per
min and LSB for Samantha were 23 and 2, respectively.

Letter Sound Fluency Intervention

As a result of a stable baseline, the letter sound fluency intervention was initiated
first for Alex. In response to the letter sound fluency intervention, Alex’s LSF
reached the goal of 52 letter sounds per min (mean 52 LS per min in last three
sessions) in approximately 27 sessions. Lucy was the next participant to enter the
sound fluency intervention phase of the multiple baseline design. As a result of
exposure to the letter sound fluency intervention, Lucy performed above the goal for
LSF (mean 53 LS per min in last three sessions). Samantha was the final participant
to begin the letter sound fluency intervention. In response to the letter sound fluency
intervention, Samantha improved performance from 23 letter sounds per min during
baseline to 36 letter sounds per min during the last three sessions of the letter sound
fluency intervention. Although Samantha did not reach the criterion performance of

123
70 J Behav Educ (2010) 19:62–75

Correct Letter Sounds per Min

Sessions

Fig. 1 LSF and LSB performance for each subject across phases

123
J Behav Educ (2010) 19:62–75 71

Table 1 Mean performance across treatment phases


Baseline LSF Final generalization
training phase

Alex
LSF 21 44.22
(2) (8.68)
LSB 3.3 19.19 30.67
(3.06) (7.46) (5.39)
Lucy
LSF 20 42.81
(2) (8.48)
LSB 1.20 9.44 30
(2.68) (5.76) (5.52)
Samantha
LSF 23.67 33.35
(3.67) (4.45)
LSB 1.89 1 18.36
(3.55) (2.77) (5.59)

Note. Standard deviations appear in parentheses below means

52 letter sounds per min, we still decided to evaluate the degree of spontaneous
generalization that occurred.
Once participants had completed the letter sound fluency phase of the study, an
examination of letter sound blending (LSB) with nonsense words was conducted to
determine the level of spontaneous generalization that had occurred. Both Lucy and
Alex demonstrated a level of spontaneous generalization to LSB during the letter
sound intervention. Initial performance during the baseline of three LSB for Alex
improved to mean of 25 LSB during the last three sessions of letter sound
intervention. Lucy demonstrated a small degree of spontaneous generalization
during the letter sound intervention phase. Prior to any instruction with letter sound
fluency, Lucy averaged one LSB. However, after instruction, her performance
increased to six LSB. Finally, Samantha did not demonstrate any spontaneous
generalization during the letter sound intervention. Performance during the letter
sound fluency intervention was similar to that of baseline levels for LSB, averaging
about one letter sound blended under both conditions.

Generalization Interventions

Cue Generalization

During this phase of the study, the first generalization intervention implemented was
the cueing intervention. We decided to implement this procedure with Lucy first,
because LSB performance was trending downward, and the criterion for LSF had
been reached. As a result of the cueing procedure, there was an immediate level

123
72 J Behav Educ (2010) 19:62–75

change in LSB and an upward trend in performance. As a result of the intervention,


Lucy improved LSB from a mean of six sounds blended to a mean of 24 sounds
blended in the last three sessions of intervention resulting in performance which
approximated 45% of letter sound fluency performance. Cueing was next
implemented with Samantha. Although Samantha did not reach the pre-established
criterion of 52 for LSF, her LSB baseline during the letter sound intervention phase
was more stable relative to Alex’s more variable baseline. As a result, we decided
that the cueing intervention would begin. In response to the cueing intervention
Samantha demonstrated a slight increase in level for LSB increasing from one LSB
to an average of eight LSB per min across the entire phase. This was a level of
generalization in LSB that approximated 22% of performance in LSF. Finally, the
cueing procedure was implemented for Alex. Upon implementation of the cueing
procedure, the level of LSB increased from 25 letter sounds blended to 34 letter
sounds blended in the last three sessions of intervention resulting in a performance
level of sound blending that approximated 65% of LSF performance.

Cue Generalization with Goal Setting

As the level of LSB generalization for Alex most closely approximated LSF, no
further intervention was implemented. However, generalization strategies for Lucy
and Samantha were continued in an attempt to produce greater degrees of
generalization. Lucy generalized to a large extent with cueing alone; therefore, goal
setting with reward was added in an attempt to increase the overall level of
generalization. Cueing with goal setting for Lucy resulted in an upward trend with
performance in the last three sessions averaging 34 LSB per min which
approximated 66% of the overall LSF performance.

Sufficient Response Exemplars

Samantha’s response to cueing alone was minimal, so the modeling intervention


was implemented in an attempt to promote generalization. Results of the modeling
intervention resulted in a strong upward trend with performance in the last three
sessions of 24 LSB which was 66% of total LSF performance.

Discussion

The current study contained two main goals. The first was to examine the degree to
which generalization occurred across two related academic task, LSF to blending of
nonsense words, without explicit programming. Results indicate that although all
three participants responded to the letter sound fluency intervention, unprogrammed
generalization to sound blending occurred at moderate rates for only one student. In
the absence of any generalization programming, Alex did demonstrate a level of
generalization to sound blending. However, Lucy and Samantha demonstrated
minimal to a complete absence of generalization. These results are consistent with

123
J Behav Educ (2010) 19:62–75 73

previous research which has established that without explicit programming,


generalization is not a guarantee (Stokes and Baer 1977).
The second goal was to evaluate the efficacy of two generalization techniques on
the generalization of letter sound blending. These two techniques were cued
responding and response exemplar training. The cued responding represents an
antecedent-based manipulation to facilitate generalization that is similar by
incorporating common salient stimuli procedures described by Stokes and Osnes
(1989). Results indicate that for two of the participants (Alex and Lucy), the less
intensive cueing procedure produced significant improvements in generalization
over previous phases. For Alex, the implementation of the cueing procedure
elevated sound blending to levels that approximated his previous LSF performance.
Altering the administration of probe instructions to cue generalization also resulted
in improved sound blending for Lucy. These improvements were not as significant
when compared to Alex. However, the addition of a goal to the cueing procedure did
result in improvements in performance that, similar to Alex, approximated her
previous LSF performance. Finally, Samantha only demonstrated minimal improve-
ments in sound blending as a result of the cueing procedure. As a result of this lack
of progress, response exemplar training was provided. Once exposed to response
exemplar training, Samantha’s letter sound blending improved to levels that
approximated her previous LSF performance. The response exemplar training
reduced the narrow training effects of the LSF skills for Samantha. This effect of the
response exemplar training is similar to the generalization research on diverse
training to produce generalized responses (Stokes and Osnes 1989). The results of
these procedures on the performance of letter sound blending are that all three
subjects improved letter sound blending to levels that approximated about 66% of
overall LSF. To accomplish this, three different (two conceptually distinct)
procedures were utilized. This indicates once again that although generalization is a
desired outcome of instruction, achieving generalization is an outcome that is not
always arrived at through the same path. It is important to note that for the two
participants who responded to the cueing procedures, no instruction or modeling of
blending was required for generalization to occur. This implies that the skill
necessary to blend may have been present prior to training of LSF. If this was the
case, the lack of generalization prior to cueing may have been the result of poor
stimulus control over the behavior of sound blending. Providing the cue in
combination with advanced LSF may have served to bring the behavior of sound
blending under stimulus control thus resulting in increased sound blending for Alex
and Lucy. It is also possible that the generalization that occurred was a result of
increased skill proficiency with letter sound fluency. This increased skill proficiency
likely provided the keystone skill necessary to perform the more advanced blending
behavior.
If attempts at understanding the variables that promote generalization are to
continue, it is important that we not only identify and validate methods to produce
generalization, but we need to attempt to develop an understanding of the conditions
under which the different techniques are effective. Through this understanding,
assessment techniques can be identified that are able to predict which generalization
technique will likely produce the greatest degree of generalization for the least

123
74 J Behav Educ (2010) 19:62–75

effort. Future research should utilize current classification systems such as the one
forwarded by Stokes and Osnes (1989) as a guide for the development of new
generalization techniques as well as a means for linking assessments to treatment
recommendations. In the current research, techniques classified by Stokes and Osnes
as training diversely and incorporating functional mediators were differentially
successful across participants. Future research should not only expand on methods
within these categories but evaluate the interchangeability of these methods within
and between these categories.
As with all research, the current study contains limitations that should be noted
when evaluating the results. First, this study incorporated a single-subject design.
Although the use of single-subject designs has strong internal validity, the external
validity is unknown. In order to determine whether these results are consistent
outside of the current participants, replication should be conducted. Second, one
participant (Samantha) did not reach the criterion for LSF prior to the generalization
procedures being tested. It is possible that if the criterion was met that the less
intensive forms of the generalization procedures would have been effective in
producing higher rates of sound blending. Samantha had been receiving intervention
for LSF for over a month without any improvements in sound blending. The
decision was made to change phases simply because the researchers and her current
teacher believed that improvements in her basic reading skills (i.e., sound blending)
were essential, and delaying more intense efforts at remediation any longer may be
detrimental to Samantha.

References

Ardoin, S., Eckert, T., & Cole, C. (2008). Promoting generalization of reading: A comparison of two
fluency-based interventions for improving general education student’s oral reading rate. Journal of
Behavioral Education, 17(3), 237–252.
Ardoin, S., McCall, M., & Klubnik, C. (2007). Promoting generalization of oral reading fluency:
Providing drill versus practice opportunities. Journal of Behavioral Education, 16(1), 54–69.
Berends, I., & Reitsma, P. (2006). Remediation of fluency: Word specific or generalized training effects?
Reading and Writing, 19, 221–234.
Bonfiglio, C., Daly, E., Martens, B., Lin, L., & Corsaut, S. (2004). An experimental analysis of reading
interventions: Generalization across instructional strategies, time, and passages. Journal of Applied
Behavior Analysis, 37(1), 111–114.
Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis (2nd ed.). Upper Saddle
River, NJ: Merrill/Prentice Hall.
Daly, E., Chafouleas, S., Persampieri, M., Bonfiglio, C., & LaFleur, K. (2004). Teaching phoneme
segmenting and blending as critical early literacy skills: An experimental analysis of minimal
textual repertoires. Journal of Behavioral Education, 13(3), 165–178.
Daly, E., Martens, B., Dool, E., & Hintze, J. (1998). Using brief functional analysis to select interventions
for oral reading. Journal of Behavioral Education, 8(2), 203–218.
Daly, E., III., Martens, B., Hamler, K., Dool, E., & Eckert, T. (1999). A brief experimental analysis for
identifying instructional components needed to improve oral reading fluency. Journal of Applied
Behavior Analysis, 32(1), 83–94.
Eckert, T., Ardoin, S., Daly, E., & Martens, B. (2002). Improving oral reading fluency: A brief
experimental analysis of combining an antecedent intervention with consequences. Journal of
Applied Behavior Analysis, 35(3), 271–282.

123
J Behav Educ (2010) 19:62–75 75

Elliott, J., Lee, S. W., & Tollefson, N. (2001). A reliability and validity study of the Dynamic Indicators
of Basic Early Literacy Skills—Modified. School Psychology Review, 30, 33–49.
Good, R. H., & Kaminski, R. A. (Eds.). (2002). Dynamic indicators of basic early literacy skills (6th ed.).
Eugene, OR: Institute for Development of Educational Achievement.
Good, R. H., Kaminski, R. A., Shinn, M., Bratten, J., Shinn, M., Laimon, L., et al. (2004). Technical
adequacy and decision making utility of DIBELS. Eugene, OR: University of Oregon. (Technical
Report No. 7).
Good, R. H., Wallin, J., Simmons, D. C., Kame’enui, E. J., & Kaminski, R. A. (2002). System-wide
percentile ranks for DIBELS benchmark assessment. Eugene, OR: University of Oregon. (Technical
Report 9).
Harn, B., Stoolmiller, M., & Chard, D. (2008). Measuring the dimensions of alphabetic principle on the
reading development of first graders: The role of automaticity and unitization. Journal of Learning
Disabilities, 41(2), 143.
Harris, K. R., & Graham, S. (1985). Improving learning disabled students’ composition skills: Self-
control. Learning Disability Quarterly, 8(1), 27–36.
Martens, B., Eckert, T., Begeny, J., Lewandowski, L., DiGennaro, F., Montarello, S., et al. (2007). Effects
of a fluency-building program on the reading performance of low-achieving second and third grade
students. Journal of Behavioral Education, 16(1), 38–53.
Mesmer, E., Duhon, G., & Dodson, K. (2007). The effects of programming common stimuli for
enhancing stimulus generalization of academic behavior. Journal of Applied Behavior Analysis, 40,
553–557.
National Reading Panel. (2000). Teaching children to read: An evidence-base of scientific research
literature on reading and its implications for reading instructions. Bethesda, MD: National
Institutes of Health.
Nist, L., & Joseph, L. M. (2008). Effectiveness and efficiency of flashcard drill instructional method on
urban first-graders’ word recognition, maintenance, and generalization. School Psychology Review,
37, 294–308.
Noell, G., Connell, J., & Duhon, G. (2006). Spontaneous response generalization during whole word
instruction: Reading to spell and spelling to read. Journal of Behavioral Education, 15(3), 121–130.
Ritchey, K., & Speece, D. (2006). From letter names to word reading: The nascent role of sublexical
fluency. Contemporary Educational Psychology, 31(3), 301–327.
Shinn, M. M., & Shinn, M. R. (2002). AIMSweb training workbook: Administration and scoring of early
literacy measures for use with AIMSweb. Eden Prairie, MN: Edformation.
Skinner, B. F. (1953). Science and human behavior. New York: MacMillan.
Stokes, T., & Baer, D. (1977). An implicit technology of generalization. Journal of Applied Behavior
Analysis, 10(2), 349.
Stokes, T., & Osnes, P. G. (1989). An operant pursuit of generalization. Behavior Therapy, 20, 337–355.
Thaler, V., Ebner, E., Wimmer, H., & Landerl, K. (2004). Training reading fluency in dysfluent readers
with high reading accuracy: Word specific effects but low transfer to untrained words. Annals of
Dyslexia, 54, 89–113.
Weinstein, G., & Cooke, N. (1992). The effects of two repeated reading interventions on generalization of
fluency. Learning Disability Quarterly, 15(1), 21–28.

123

You might also like