Professional Documents
Culture Documents
Williams y Greer-A Comparison of Verbal-Behavior and Linguistic-Communication Curricula For Training Developmentally Delayed Adolescents To Acquire and Maintain Vocal Speech
Williams y Greer-A Comparison of Verbal-Behavior and Linguistic-Communication Curricula For Training Developmentally Delayed Adolescents To Acquire and Maintain Vocal Speech
Williams y Greer-A Comparison of Verbal-Behavior and Linguistic-Communication Curricula For Training Developmentally Delayed Adolescents To Acquire and Maintain Vocal Speech
1
The research reported herein was accomplished by the first author as a doctoral dissertation under the
sponsorship and within the programmatic research of the second author. The first author was a Keller
Fellow whose training was funded by a grant from the U.S. Department of Education. Reprints may be
obtained from the second author at Box 76, Teachers College, Columbia University, New York, NY 10027.
The first author is now at the Fred S. Keller School.
communication skills by applying operant procedures to train linguistic units (Guess,
Sailor, & Baer, 1976; Lovaas, 1977), these curricula have not systematically incorporated
Skinner’s (1957) conception of verbal behavior.
Method
Participants
Three adolescents with developmental disabilities participated in the experiment.
Student A was a 15 year old male with organic brain damage. He had limited vocabulary
which was echolalic in nature. He did not request objects or activities independently. He
had received several years of operant training in linguistic units (Guess et al., 1976) and
had a Stanford Binet I.Q. of 27. He read several survival words and engaged in
independent art activities.
Student B was a 16 year old nonverbal female who had been a resident of a non-
English speaking country until a few months before the onset of the study. She had no
verbal behavior in her native language and had a Stanford Binet I.Q. of 32. Within a few
months of arrival to the residential school, she had learned to brush her teeth, bath
herself, and dress and undress herself independently. Prior to the onset of the study she
acquired basic prelinguistic skills such as compliance to commands, imitation as a
general response class, and imitation of vocal sounds. She was trained linguistically to
label vocally sixteen items, to say “want” when presented with an item, and to say
“what’s that” when presented with a novel item (Steps 1 to 4 in Guess et al., 1976). All of
the above mentioned verbal skills were under the control of the verbal antecedent of the
instructor. At the onset of the study, she neither labeled nor requested items
independently.
Student C was a 16 year old male profoundly retarded (MA of 12.8 months on the
Cattell Infant Intelligence Test). After four years of instruction in the residential school,
he had learned to wash his face and hands, dress and undress himself, brush his teeth at a
gestural prompt level, match three dimensional objects, label the letters of the alphabet
and the numbers 1 to 10. He had progressed to step 12 in the linguistic curriculum (Guess
et al., 1976), but had not achieved criterion on step 7 (yes and no answers), nor could he
identify objects or activities when prompted by the instructor with questions such as “are
you eating a cookie?” He emitted the following words, but only under the verbal
antecedent of the instructor- “water,” “cookie,” “ball,” and “toy.”
Setting
The study took place in the classrooms of a residential school in a large
metropolitan area. Each student was in a separate classroom. The institution provided
intensive operant-based training procedures in the school and the residence. The school
included 48 children and adolescents with severe to profound retardation, many of them
who were also autistic. Each classroom had a teacher, two paraprofessionals, and six
students. The components of the school model are described elsewhere (Greer, 1992).
Curricula
The linguistic curriculum consisted of the curriculum and operant procedures
described in Guess et al. (1976). These authors based their program on five years of
research and had field tested it with over 100 individuals with severe handicapping
conditions. This curriculum consists of a series of individual training steps sequentially
designed to develop functional communication skills. There are 60 steps divided into four
parts. Part one (steps 1-9) deals with the student labeling novel items when asked by the
trainer “what’s that?” Part two (steps 10-29) deals with training the student to label
actions, persons, and things when asked by the trainer. Part three (steps 30-42) deals with
the student stating possession and color (“Is this my… Is this yours?” and “what color?”)
when asked by the trainer. Part four (steps 43-60) deals with the student stating size
(big/little), relationships (on/under), and location (inside/outside) when asked by the
trainer. The specific components of the linguistic curriculum are outlined in Table 1 with
step-by-step comparisons with the verbal behavior curriculum.
The verbal behavior curriculum was based on a pilot version of the “Scripted
Curriculum for Verbal Behavior” (Greer, 1986). One component of this curriculum
(nonverbal antecedent control of verbal responding) was relevant to the objectives of the
linguistic curriculum. The following verbal operants were trained with the verbal
behavior curriculum: echoics, mands, tacts, and autoclitics. Words or groups of words
(verbal operant units) were trained in a sequence consisting of echoic training, followed
respectively by either: a) mand training, b) tact training, and c) autoclitic training with
both tacts and mands.
An echoic unit of verbal behavior consists of verbal responses having point-to-
point correspondence with a vocal verbal behavior unit of another speaker (Skinner,
1957). It is an exact imitation of the spoken word of anther speaker. (See also Vargas,
1986). The trainer says “ball” and the child says “ball.”
A mand unit is a verbal operant consisting of vocal sounds, a word, or words (and
gestural or textual equivalents) emitted in the presence of an item and its deprivation
(including lack of access to the item) or its deprivation alone (without the presence of the
object), which results in the speaker obtaining the item (reinforcer). Thus, a mand
specifies a known reinforcer, and it may consist of an utterance, a gesture, or other
behavior, but only when deprivation conditions are part of the controls. If a child is asked
“What do you want?” and responds with “I want ball”, the response is not a mand but
rather an intraverbal response. The latter response is primarily under the control of the
verbal antecedent “What do you want”, not the deprivation or the item’s presence. If the
same sequence is viewed from a linguistic perspective, the child is being taught the
linguistic structure and would presumably acquire the linguistic function of request.
A tact unit consists of verbal behavior under the control of nonverbal stimuli
(item, event, activity) and generalized reinforcers. It is verbal behavior that makes verbal
contact with the nonverbal world. For example, in the presence of a stapler, a child says
“stapler”, and a trainer or parent says “good” (generalized reinforcer) or presents the
child with a token. The word “stapler” could function as a mand if the reinforcer is access
to the stapler. Thus, to be a correct tact the response must be under the appropriate
antecedent stimulus control and a generalized reinforcer. If, however, the child is asked to
label the stapler, and the verbal antecedent (e.g., “What’s this?”) precedes the response
“stapler,” the response is not a tact according to a behaviorological analysis of verbal
behavior. From the linguists’ point of view, the child is learning linguistic labels for
stimuli. The specific antecedents and consequences of the behavior are not part of the
account. Thus, according to linguists, failure to label items spontaneously is a function of
lack of mental structures, not specific stimulus control.
Autoclitic units are collateral responses to mands or tacts which qualify, confirm,
negate, or affirm certain consequences on the listener’s part of other verbal behavior that
is being uttered. The word “cookie,” a whine, or a gesture (point) may function as a
mand. The following consists of a mand with autoclitics: “May I have one cookie,
please?” and is consequated with a cookie. The word “cookie” may also function as a
tact. The following consists of a tact with autoclitics: “That is a big cookie,” and it is
consequated with the generalized reinforcer “Yes, it is.”
Verbal units were selected for echoic training and assigned to mand training if the
words were ones that specified items or activities known to be reinforcers (e.g., high
preference items, events) to the student. If the verbal units were common items, activities,
or events in the child’s environment but were not high preference stimuli, they were
trained as tacts. If the verbal units modified either tacts or mands, they were trained as
autoclitics with the tacts and mands. In the normal sequence of the curriculum, words
were trained as both tacts and mands interchangeably. All the words trained in the verbal
behavior curriculum were different from those trained in the linguistic curriculum with
the exception of “Yes” and “No.” However, in the verbal behavior curriculum “Yes” and
“No” were trained functionally as qualifying autoclitics of affirmation and negation,
using establishing operations.
Training Procedures
Both curricula involved the use of operant procedures. Each had controlled
antecedent presentations and the use of reinforcement operations using items, activities,
or events known to function as reinforcement for the target students. There was no
difference in the number of correct responses per students between the two curriculum
conditions during training. This was an indication that operant procedures were applied
without systematic bias to the two curricula. That is, generic use of operant procedures
was controlled across curricula.
The student had to be able to sit still on command for 5 seconds and make eye
contact as a prerequisite to training in either curriculum. During the training of the verbal
behavior curriculum, the trainer called the student’s name to indicate that the trainer was
going to voice a unit (echoic), present a known reinforcer (mand), or present a common
item, or model and activity (tact). The antecedent conditions consisted of brief
deprivations of the item or activity (i.e., after sitting, walking was manded) and
accessibility to the item contingent on verbalizations. Trials that were not correct resulted
in countable instances of deprivation (Michael, 1982).
During echoic training, correct or successive approximations of mands were
reinforced by access to the item named and by nonvocal approval (smile, pat, nod). In the
case of tacts, correct or successive approximations were reinforced by nonvocal approval
followed by the opportunity to mand a generalized reinforcer. An echoic response
graduated to mand or tact training when either (a) the student echoed the response
correctly in five successive trials (first verbal behavior training phase), or 12 successive
trials (second verbal behavior phase) or (b) the student manded or tacted an item outside
the echoic discrete trial training session on two separate occasions or (c) the student
manded or tacted the item in five successive trials in the training setting before the trainer
presented the stimulus item to be echoed.
During mand training, the items to be manded were displayed on a plate or table
(at least two at a time) in front of the student. The student had to mand the object which
was then used as an immediate natural reinforcer (eating a bit of food, or 5 seconds of
contact with an item). If the item was not manded within 5 seconds, the item was moved
closer. If necessary, the student’s name and a gesture (point) was used. When the student
completed 5 consecutive responses (criterion) without prompt, the stimulus was removed
and a new one added. Subsequently, the acquired mand was to be used in other settings.
The mand became part of the list of responses used as generalized reinforcers during the
tact training. Each mand was “incidental” in that access to the consequence was
contingent on vocal responses.
During tact training, the item to be tacted was displayed on the table in front of
the student. If the student did not tact independently, a lip prompt (the topography of the
word presented with the lips, without emitting any sound) was used. If, with the lip
prompt the student did not emit the correct response, then a vocal prompt was used.
Vocal prompts were faded as soon as possible. When the student emitted the correct
response independently (without any prompts), he could mand a reinforcer. For each tact
session, there were at least two previously learned mand items or activities available
(these were changed as new mands were acquired).
The autoclitic training began after a series of mands and tacts were in the
student’s verbal repertoire. Echoic, mand, and tact sessions were conducted in rotated
order. That is, a session devoted to echoic training might be followed by tact or mand
training sessions. As soon as possible, autoclitics were trained echoically then added to
mands or tacts already in the student’s repertoire (i.e., “This is a _____.” “May I have a
(the) _____, please.” or “I want some _____, please.”).
The learning of “yes” and “no” responses was an objective in each
curriculum. The specific components of each curriculum taught to each student were
based on the individual response history of each student. Two of the students did not have
the “yes” and “no” responses in their repertoires. These responses were trained first
linguistically and then as autoclitics of affirmation and negation in the verbal behavior
curriculum. For example, if a student had not achieved criterion on responding “yes” or
“no” to questions in the linguistic curriculum and the “yes” and “no” responses were next
in the verbal behavior sequence, the student was trained to emit “yes” and “no” using
verbal behavior procedures.
In the verbal behavior curriculum, students were presented with previously
established reinforcers, one of which was made less desirable to the student (mustard on a
cookie versus no mustard, or soda with cola syrup or other flavoring versus soda with no
flavoring). These procedures functioned as establishing operations (Michael, 1982)
consistent with Skinner’s conception of the mand and the qualifying function of
autoclitics (Skinner, 1957). In one case (Table 2), the student required one trial in which
an adult modeled “yes” and “no” to the desirable and undesirable item respectively. The
student subsequently acquired the “yes” and “no” criterion. The general characteristics of
each curriculum are outlined in Table 1. The specific curriculum for each student is
shown in Tables 2 and 3 which also provide a step-by-step scenario for each student. See
Greer, Dorow, McCorkle, Williams, and Asnes (1991) for additional research on
establishing operations.
Interobserver Agreement
A second trained observer independently collected data on correct and incorrect
responses during some of the 20-trial training sessions and on the number of words
emitted in training sessions and maintenance probes. Independent observations were
conducted for 20 sessions (25%) of the training sessions received by Student A, 19
sessions (24%) for Student B, and 25 sessions (31%) for Student C. Percentage
Agreement was calculated for correct/incorrect responses by dividing point by point
agreement on words emitted correctly only when the trial was correct (all words correct)
by agreement plus disagreement and multiplying times 100%. Training sessions mean
agreements were 99% for Student A (range 97% to 100%), 98% for Student B (range 92
to 100%) and 92% for Student C (range 90% to 100%).
Second observations were conducted on all of the eight maintenance probes
received by all three students. The mean and range percentages of agreement were 95%
for Student A (94% to 100%), 99% for Student B (98% to 100%), and 97% for Student C
(96% to 100%).
Experimental Design
The design for all three students consisted of a BABA reversal treatment design
(where B represented the verbal behavior curriculum and A the linguistic curriculum)
with maintenance probes (one for each curriculum) at the conclusion of the first A phase,
second B phase, and second A phase. The design was characterized as a reversal because
the curricula were reversed from phase to phase. Each phase consisted of 20 sessions,
established a priori, to maintain equivalent sessions under each phase. Maintenance
probes were conducted at the end of the first A phase, the end of the second B phase, and
the end of the second A phase. Multiple probes were conducted after the entire
instruction for all responses was completed. The steps and repertoires trained to each
student are shown in Tables 2 and 3.
Results
Figure 1 shows the number of words emitted per student during training sessions
for each phase. Student A emitted a mean of 168 words (range 0 to 300) during the first B
phase (verbal behavior phase) and a mean of 59 words (range 8 to 158) during the first A
phase (subsequent language phase). During the second B phase (verbal behavior phase)
he emitted a mean of 103 words per session (range 15 to 280), while during the second A
phase (language phase) his mean was 26 (range 5 to 72). Student B emitted a mean of 36
words (range 3 to 100) during the first B phase (verbal behavior phase) and a mean of 13
words (range 2 to 18) during the first A phase (subsequent language phase). During the
second B phase (verbal behavior phase) she emitted a mean of 30 words per session
(range 6 to 73), while during the second A phase (language phase) her mean was 40
(range 0 to 96). Student C emitted a mean of 29 words (range 2 to 68) during the first B
phase (verbal behavior phase) and a mean of 5 words (range 1 to 11) during the first A
phase (subsequent language phase). During the second B phase (verbal behavior phase)
he emitted a mean of 56 words per session (range 12 to 78), while during the second
A phase (language phase) his mean was 11 (range 5 to 31). For all students, sharp
decreases in words correct (see Figure 1) signaled attainment of criterion and the
introduction of new words.
Students B and C emitted words that had been trained in pervious phases but were
not being trained during that particular training phase (Figure 1). During the second B
phase, Student B emitted 72 words that had been trained during the first B phase (verbal
behavior phase). During the second A phase, she emitted 135 words trained in the prior
two verbal behavior phases and 2 words trained during the prior language phase.
During the first A phase, second B phase, and second A phase, student C emitted 209,
157, and 72 words, respectively, that had been trained during the training of the verbal
behavior curriculum. During the second B phase and a second A phase, he emitted a total
of 20 and 7 words, respectively, that had been trained during the training of the language
based curriculum (Figure 1).
Figure 2 shows the number of correct trials (single or multiple words) in the
maintenance probes. The figure shows the correct trials and total units trained during the
most recent prior phases for each curriculum and all trained responses for the final
maintenance probe conducted two days after the end of the study. Student A emitted 37
correct responses to 56 possibilities from the language curriculum after a latency of four
days. He emitted 42 correct responses out of 58 possibilities from the verbal behavior
after a latency of 8 months. For the second probe (following the second verbal behavior
phase), he emitted 25 correct responses to 56 possibilities from the language based
curriculum after a latency of nine weeks. He emitted 39 correct responses of 54
possibilities in verbal behavior after a latency of three days. For the third probe, student
A emitted 10 correct responses out of 18 possibilities from the language curriculum after
a latency of five days, and 48 correct responses out of 54 possibilities from the verbal
behavior curriculum after a latency of four weeks. For the final probe (which tested all
words trained under each curriculum), Student A emitted 42 correct responses out of 73
words trained in the language curriculum (58%) and 65 correct responses out of 80
possibilities (81%) from the verbal behavior curriculum.
Student B emitted 11 correct responses to 16 possibilities from the language
curriculum after a latency of two days. She emitted 19 correct responses out of 39
possibilities from the verbal behavior after a latency of 8 months. For the second probe
(following the second verbal behavior phase), she emitted 12 correct responses to 16
possibilities from the language based curriculum after a latency of three weeks. She
emitted 19 correct responses of 22 possibilities in verbal behavior after a time of four
days. For the third probe, Student B emitted 11 correct responses out of 36 possibilities
from the language curriculum after a latency of five days, and 15 correct responses out of
22 possibilities from the verbal behavior curriculum after a latency of two months. For
the final probe (which tested all words trained under each curriculum), Student B emitted
34 correct responses out of 52 words trained in the language curriculum (65%) and 58
correct responses out of 67 possibilities (87%) from the verbal behavior curriculum.
Student C emitted 3 correct responses to 6 possibilities from the language
curriculum after a latency of three days. He emitted 6 correct responses out of 23
possibilities from the verbal behavior after a latency of 10 months. For the second probe
(following the second verbal behavior phase), he emitted 3 correct responses to 6
possibilities from the language based curriculum after a latency of two months. He
emitted 13 correct responses of 14 possibilities in verbal behavior after a latency of three
days. For the third probe, Student C emitted 10 correct responses out of 16 possibilities
from the language curriculum after a latency of five days, and 14 correct responses out of
14 possibilities from the verbal behavior curriculum after a latency of four weeks. For the
final probe (which tested all words trained under each curriculum), Student C emitted 6
correct responses out of 16 words trained in the language curriculum (37%) and 43
correct responses out of 54 possibilities (80%) from the verbal behavior curriculum.
At the final probe for all the students, substantially more words learned under the
verbal behavior curriculum were maintained than the words learned under the language
curriculum. The number of correct responses maintained also differed markedly by
curriculum for Student C on all probes, and for Student A on the third and final probes.
Student B showed marked differences by curriculum only at the final maintenance probe.
All three students emitted more words per session during the verbal behavior
training than during the language training with one exception. Student B emitted more
words during the second A phase (last language phase) than during the preceding second
B phase (verbal behavior training). Both students B and C emitted verbal behavior units
appropriately during language training (generalization), and Student C emitted some
language units trained in prior phases.
Discussion
References