Você está na página 1de 16

The Analysis of Verbal Behavior 2000, 17, 89-104

Using Intraverbal Prompts to


Establish Tacts for
Children with Autism
Mark L. Sundberg, Katie Endicott,
and Peter Eigenheer
Behavior Analysts, Inc.
Some children with autism have difficulty acquiring tacts, despite their ability to mand, echo
words, and imitate actions. The current study focused on 2 nonvocal children who had ac-
quired a few mands using sign language, but had repeatedly failed to acquire signed tacts.
Two procedures were compared to determine the most effective approach for training tacts
to these participants. One procedure (the standard condition) used the general verbal prompt
"What is that?" The other procedure (the intraverbal condition) used a specific intraverbal
prompt "Sign [spoken word]." The results showed that both participants acquired nonimi-
tative verbal responses during the intraverbal condition but not during the standard condition.
One participant demonstrated complete transfer to pure tacts, and the other participant showed
a partial transfer to pure tacts. These results have implications for the design of language
intervention programs for children with autism who have difficulty acquiring tacts.

Children with autism frequently ers, but they repeatedly failed to ac-
show substantial linguistic gains as a quire signs as tacts (labels).
function of intensive behavioral inter- This language problem may be re-
vention (e.g., Lovaas, 1987; Maurice, lated to Skinner's (1957) point that the
Green, & Luce, 1996; Smith, 1993; mand benefits the speaker, whereas the
Sundberg & Partington, 1998). How- tact (and the other types of verbal be-
ever, some children are more difficult havior) benefits the listener (p. 36).
to teach than others in that they present Within the mand, establishing opera-
linguistic challenges not easily treated tions (EOs) directly relevant to the
by typical behavioral interventions. speaker evoke verbal behavior that re-
With these more difficult and often ceives specific reinforcement. The tact,
nonvocal children, special procedures however, is controlled by nonverbal
are required to move the children past discriminative stimuli and receives
certain linguistic barriers. The current only generalized conditioned reinforce-
study examined a language problem ment. Some types of mands (especially
experienced by 2 nonvocal children those involving food and drink) are of-
with autism. Both children had failed ten easier to teach than tacts, perhaps
to acquire functional speech and had because of the potent evocative effects
only limited success with sign lan- of the relevant EOs and the related rep-
guage. The children did acquire a few ertoire-altering effects of specific rein-
signs as mands (requests) for reinforc- forcement (Michael, 1993a).
Successful tacting requires attending
to nonverbal stimuli that may be irrel-
evant to a child or that are blocked by
We gratefully acknowledge Cindy Sundberg a child's ongoing EOs (e.g., EOs relat-
for her editorial comments on an earlier version
of this paper. Portions of the paper were pre- ed to self-stimulation). In addition, the
sented at the 25th annual convention of the As- consequences of the tact involve con-
sociation for Behavior Analysis, Chicago, May ditioned reinforcement, which may be
1999. ineffective for some children with au-
Reprints may be obtained from the first au-
thor, 3329 Vincent Rd., Pleasant Hill, California tism (Bijou & Ghezzi, 1999). Thus, the
94523. controlling variables of the tact may
89
90 MARK L. SUNDBERG et al.
make teaching this verbal operant more stimulus control transfer ... may occur
task-like and of less interest or value because the subject ignores the new
to children with autism. Therefore, it is stimulus and responds only to the de-
often the case that special procedures layed cue" (p. 703). These authors
are necessary to establish tacts for chil- found that a procedure that involved
dren who repeatedly fail to acquire this having the participants repeat the sam-
important type of verbal behavior with ple, and a progressive increase in the
the standard procedures. delay between the sample and the com-
When the participants in the current parison stimulus, resulted in the suc-
study were required to tact objects they cessful transfer of stimulus control.
would typically "guess" by emitting Given the data that show that an ex-
several signs in rapid succession. The isting verbal discriminative stimulus
participants emitted a correct sign only can block the establishment of nonver-
when given imitative prompts; thus, bal stimulus control, removing the ex-
the response was not a tact. Attempts isting verbal stimulus (e.g., "What is
to eliminate the participants' depen- that?") can eliminate an inappropriate
dence on imitative prompts by using verbal response, but additional mea-
delayed and partial prompting proce- sures are necessary to establish a new
dures repeatedly failed. This failure to form of nonverbal stimulus control.
acquire tacts has been observed in a The procedure of placing individual
number of children with autism. For objects in a box and opening the box
example, Partington, Sundberg, New- while giving, then fading, pointing
house, and Spengler (1994) observed prompts was an effective method for
that the verbal stimulus "What is establishing nonverbal control (Par-
that?" had inappropriately acquired in- tington et al., 1994). However, the pro-
traverbal control over a specific re- cedure had long intertrial intervals and
sponse (i.e., regardless of the object required the use of auxiliary objects
presented, the participant always (i.e., a box). The current study sought
signed "ball" when asked "What is to examine a faster way to eliminate
that?"). Partington et al. suggested that dependence on imitative prompts and
the participant's "failure to acquire establish nonverbal stimulus control.
tacts may have been due to the pres- Skinner (1957) suggested that the
ence of a verbal stimulus that blocked use of multiple sources of control can
the establishment of stimulus control help establish new forms of verbal be-
by a nonverbal stimulus" (p. 734). A havior. Multiple control occurs when
procedure that consisted of eliminating two separate sources of control com-
the verbal stimulus and increasing the bine to evoke a response. Skinner
salience of the nonverbal stimulus by points out that "separate sources of
placing it in a box and using pointing strength are additive. ... As a result,
prompts was effective in overcoming multiple causation produces many in-
the participant's failure to acquire tacts. teresting verbal effects" (p. 228). The
The results from Partington et al. common use of echoic and imitative
(1994) further support the findings of (motor) prompts along with a target
other researchers who have suggested stimulus is a well-established form of
that the blocking and overshadowing multiple control in language training.
of stimuli can impede the transfer of Less widely used, but sometimes more
stimulus control (e.g., Glat, Gould, effective, is the use of establishing op-
Stoddard, & Sidman, 1994; Singh & erations as additional sources of con-
Solman, 1990; Urcuioli, 1984). For ex- trol (e.g., Carroll & Hesse, 1987;
ample, Glat et al. found that in a de- Drash, High, & Tutor, 1999; Sundberg
layed-cue matching procedure involv- & Partington, 1998). For example, it is
ing printed-word comparison stimuli often easier to teach a child to say
(e.g., DOG) and dictated-name sample "book" when he sees a book (nonver-
stimuli (e.g., "dog"), the "failure of bal stimulus) and wants a book (estab-
INTRAVERBAL PROMPTS 91
lishing operation). The transfer to a ulus control to the target nonverbal
pure tact may occur because the rele- stimulus.
vant nonverbal stimulus is present in The current study examined the use
the original training condition and may of specific intraverbal prompts as a
gain some degree of stimulus control method to establish tacts for children
over the target response. with long histories of unsuccessful
Other types of verbal stimuli, such tacting. Specifically, the verbal stimu-
as textual (written stimuli) and intra- lus "What is that?" was dropped, and
verbal prompts, can also be used to the name of the object (e.g., "sign
teach new forms of verbal behavior. shoe") was vocally presented while the
However, textual stimuli, like echoic nonverbal object was displayed (e.g., a
and imitative stimuli (called formal shoe). This procedure was contrasted
prompts by Skinner, 1957), share the with the standard tact training proce-
common drawback of revealing to the dure (i.e., "What is that?" with imita-
participant the desired response form. tive prompts) suggested in many of the
Intraverbal prompts, like establishing behavioral language programs (e.g.,
operations, do not reveal the response Guess, Sailor, & Baer, 1976; Kent,
form but may still have the benefit of 1974; Lovaas, 1977; Sundberg & Par-
an added source of strength (these are tington, 1998).
called thematic prompts by Skinner,
1957). It is possible that thematic METHOD
prompts can facilitate the transfer of Participants and Setting
stimulus control more effectively than
formal prompts when a participant is Participant 1 was a 5-year-old boy
dependent upon response-form with a diagnosis of autism. His speech
prompts. was very limited, but he could echo
Intraverbal prompts are common in several complete words and would oc-
our day-to-day verbal behavior. These casionally say random words that were
prompts can occur in the form of hints, understandable to the general commu-
clues, reminders, notes, lists, word as- nity. Despite intensive efforts to teach
sociations, and so on. They are verbal speech as a response form, he failed to
stimuli that evoke specific, nonechoic acquire any vocal mands, tacts, or in-
verbal responses. One of the benefits of traverbals or any correct receptive dis-
using sign language as a response form criminations. Because of his strong
is that it allows a type of intraverbal motor imitation skills, a sign language
prompt that is not available with vocal program was implemented, and he
or picture training. Specifically, there is quickly acquired several signs as
an intraverbal relation between a mands (e.g., candy, book, music, teeth-
signed word and a spoken word (as is er) but failed to acquire any receptive
the case for all translations between skills or tacts (labels), despite an inten-
languages). For example, when a sive daily program over several
teacher signs "shoe" and a child emits months. Most presentations of objects
the vocal response "shoe," this verbal to tact resulted in guessing by the par-
relation is intraverbal because both are ticipant (i.e., he would emit several in-
verbal stimuli and there is no point-to- correct signs in varying order).
point correspondence between the Participant 2 was a 4-year-old boy
stimulus and the response (Skinner, with a diagnosis of autism. His vocal
1957). This type of prompt is different abilities were very limited in that he
from an imitative prompt because it could only echo approximately 10 dif-
does not model the response form. The ferent sounds. Despite intensive efforts
child must emit the response form to teach speech as a response form, he
without a model. It is possible, then, failed to acquire any vocal mands,
that this type of prompting may result tacts, or intraverbals or any correct re-
in a more successful transfer of stim- ceptive discriminations. However, he
92 MARK L. SUNDBERG et al.
was able to imitate a wide variety of Table 1
motor movements, so a sign language
program was implemented. He quickly Objects assigned to each condition.
acquired several signs as mands (e.g., Note that the assignment of the same
book, candy, cookie, music, puzzle), object (e.g., bed) to the same condi-
but failed to acquire any receptive dis- tion for each participant was random.
criminations or tacts. Like Participant
1, most presentations of objects to tact Participant 1 Participant 2
resulted in guessing.
The study was conducted at the par- Intraverbal Scissors Table
ticipants' school, which was a private Bed Bed
school that primarily served children Standard Shoe Shoe
with autism. Typically, one session for Car Fish
each participant was conducted each Control Dog Dog
school morning at a table in a parti- Top Sock
tioned area (1.7 m by 2 m) in their reg-
ular classroom. The other children, the
teachers, and the aides were present in Procedure
the classroom during the study. Ses- Word selection and baseline. A pool
sions lasted 10 to 20 min for both of potential objects was selected based
training procedures. on their apparent lack of reinforcement
value and general developmental ap-
Dependent and Independent Variables propriateness (e.g., shoe, hat, car, bed,
top, cat, fish, dog, scissors, pen, paper,
The dependent variables consisted table, chair, cup, spoon). Each partici-
of the percentage of correct tacts for pant was assessed on (a) his ability to
the 10 training trials for each condition correctly imitate the sign for these ob-
(five trials per session for each object). jects, (b) his ability to correctly tact the
A correct tact was defined as emitting objects when presented with the object
the right sign for the object within 5 s, and asked "What is that?" and (c) his
without emitting any additional signs. ability to correctly provide the sign
Also, the cumulative number of correct given only the English word (e.g.,
tacts during a single pre- and postses- "sign shoe") with no object present.
sion pure tact probe was recorded For an object to qualify as a target tact,
across sessions. During this probe the the participant (a) must have been able
experimenter held up the target object to correctly imitate the sign for the ob-
and said nothing. A correct tact was ject, but (b) must have failed to tact the
defined as the participant emitting the object, and (c) must have failed to emit
right sign for the object within 5 s, the sign given only the English word.
without emitting any additional signs. Six objects were selected from each
The independent variable consisted of participant's pool and were randomly
the manipulation of two types of verbal assigned to either the treatment condi-
prompts. tions or a control (untreated) group, re-
sulting in two objects in each condition
Experimental Design (see Table 1). A baseline was then con-
ducted on the six objects for each par-
A within-subject design with a be- ticipant. The baseline consisted of pre-
tween-subjects replication was used to senting each of the six objects and the
isolate the relevant independent vari- verbal discriminative stimulus (SD)
ables. The within-subject comparisons "What is that?" five times in varying
were achieved by the use of a multi- order for three sessions. The control
element design that included a reversal groups received an additional three
design (Ulman & Sulzer-Azaroff, baseline sessions prior to starting inter-
1975). vention on those objects.
INTRAVERBAL PROMPTS 93
Phase 1: Probes and intervention. was followed by the imitative and in-
Prior to the start of each daily training traverbal prompt procedure (but no re-
session, a presession single-trial pure turn to the original verbal prompt this
tact probe was conducted. Each of the time), and a correct response was fol-
four targeted objects from the two lowed by praise and a new trial on the
training conditions was held up in the other object.
participant's line of vision while the The second procedure (the intraver-
experimenter said nothing (i.e., no ver- bal condition) used a specific intra-
bal SD was presented). The order of the verbal vocal prompt (e.g., "sign bed")
objects presented was alternated be- along with the presentation of the ob-
tween the two conditions for each ses- ject (without the verbal prompt "What
sion. A correct response or a succes- is that?"). Correct responses and suc-
sive approximation to a correct re- cessive approximations were followed
sponse was followed by social praise by praise and a mand trial. An incor-
(e.g., "Good job!") and a mand trial in rect response or no response was fol-
which the experimenter asked the par- lowed by a correction procedure simi-
ticipant "What do you want?" Typi- lar to the one described above, except
cally, both participants asked for can- the return to the original prompt was a
dy, but they also asked for other ob- return to the specific intraverbal
jects such as a book, music, cookie, or prompt rather than the "What is it?"
puzzle. Following the delivery of the prompt.
requested item and either the consump- The order of training was alternated
tion of the item or approximately 30 s between the two conditions each ses-
of time with the manded item, the next sion. Within a condition, training was
object to be tacted was presented. An given on both of the objects from one
incorrect response or no response was condition, alternating between the two
ignored. Following the consequence objects each trial. There were five
for the fourth object, there was a brief training trials and correction proce-
pause (1 to 2 min) before starting the dures, if necessary, for each object.
tact training conditions. Following training on the two ob-
During training two types of tact jects from a condition, there was a
procedures were compared. One pro- brief pause (1 to 2 min) and a post-
cedure (the standard condition) used session single-trial pure tact probe on
the verbal prompt "What is that?" each object. This probe was conducted
along with the presentation of the ob- in exactly the same manner as the pre-
ject. A correct response or a successive session probe, except that the probe
approximation was followed by social was conducted on the two objects im-
praise (e.g., "Good job!") and a mand mediately after each condition rather
trial as described above. An incorrect than on all four objects successively, as
response or no response was followed was the case in the presession probe.
by a correction procedure that consist- Phase 2: Reversal. The verbal
ed of presenting an imitative and in- prompts were reversed for each set of
traverbal prompt (e.g., the experiment- objects during this condition. Thus, the
er would sign and say "bed") along two objects that were initially present-
with the "What is that?" prompt with ed with the standard format (i.e.,
the presentation of the object. A cor- "What is that?") were now presented
rect response (which always occurred with the intraverbal format (i.e., "Sign
under these conditions) was followed [object]"), and the two objects that
by praise and the re-presentation of the were initially presented with the intra-
original verbal prompt "What is that?" verbal format were now presented with
A correct response under these condi- the standard format.
tions was followed with praise only, Phase 3 (only for Participant 1). All
and a new trial on the other object in four objects were placed in the intra-
the set began. An incorrect response verbal condition.
94 MARK L. SUNDBERG et al.
Phase 4. The four control objects ther tact (M = 27.6%). Often he would
(two for each participant) were placed emit the same sign for both objects,
in the intraverbal condition. thus scoring 50% correct. During the
Phase 5: Reversal (only for Partic- intraverbal condition the participant
ipant 1). The verbal prompt was met the criterion of three consecutive
changed to "What is that?" during this sessions of 80% or better after 11 ses-
condition. sions. Following these sessions, his
performance stabilized at or above
Response Definition and Reliability 80% correct for the remainder of the
The experimenter recorded the oc- condition (M = 77.2% for the whole
currences of the signed responses and condition). Training was continued af-
scored them as correct, an approxima- ter the 11th session in an effort to see
tion, or incorrect. A correct response if acquisition would eventually occur
was scored if the participant emitted in the standard condition, and if a pure
the complete sign (e.g., for the sign tact would occur in either condition. It
"shoe" the closed fists coming togeth- is interesting to note that after a 13-day
er at the thumbs in front of the body). vacation (Session 26), performance in
An approximation was scored if the the standard condition dropped to 0%
participant emitted part of the sign but was maintained at or above 80% in
(e.g., for the sign "shoe" the closed the intraverbal condition.
fists without bringing the hands togeth- During the reversal phase, Partici-
er). Approximations were given half pant l's performance immediately de-
the point score of a fully correct re- teriorated when the objects from the in-
sponse. An incorrect response was traverbal condition were presented in
scored if the subject emitted a sign that the format of the standard condition (M
did not correspond with the object, = 10%). The verbal stimulus "What is
emitted more than one sign, or failed that?" consistently evoked a number of
to respond within 5 s. different signs in rapid succession, de-
A second observer independently re- spite a history of success on these ob-
corded the participants' responses dur- jects when they were in the intraverbal
ing 12.4% of the sessions (there were condition. On the other hand, when the
85 sessions for Participant 1 and 68 objects from the standard condition
sessions for Participant 2). Reliability were presented in the format of the in-
data were taken for the baselines, pre- traverbal condition there was an im-
and postsession probes, and the train- mediate increase in correct responding
ing sessions. A point-by-point reliabil- (M = 70%). When all four objects
ity method of dividing the total number were placed in the intraverbal condi-
of agreements by the number of agree- tion, performance stabilized at 80% or
ments plus disagreements and multi- better (M = 86.1 %).
plying by 100% was used. The mean The bottom panel of Figure 1 shows
percentage agreement score for the the results for Participant 2. His per-
baselines was 100%, the mean per- formance was very similar to that of
centage agreement score for the pre- Participant 1, except that there was
and postsession probes was 98.6% more variability in his scores. During
(range, 87.5% to 100%), and the mean the standard condition, Participant 2
percentage agreement score for the failed to acquire either tact (M = 22%).
training sessions was 93.6% (range, However, during the intraverbal con-
85% to 100%). dition he met the criterion of three con-
secutive sessions of 80% or better after
RESULTS eight sessions (M = 72.3% for the
Figure 1 shows the results of the ex- whole condition). Sessions were also
periment. During the standard condi- continued with this participant to see if
tion, Participant 1 failed to acquire ei- acquisition would eventually occur in
INTRAVERBAL PROMPTS 95
Sign Language Acquisition
Baseline Intervention Reversal All Intraverbal
100
Intraverbala
90

80-

70 -
* Intraverbal
U)
L60-

50
CD0 Standard:

5 1-0 1 20 25 30 35 40 45 50

Baseline Intervention Reversal

80

90
50
20 -

Vacations
80
ernte Irvticantraverbal
nt a sa
70-
30
Th 20pncice epeetth bet "sh ewn Exper"mentfor Priiat,ad"he"ad"ih
10 -
atcpn

20-

0 5 10 15 20 25 30 35 40 45 50 55 60
Sessions
Fig. 1. Percentage of correct tacts for Participants 1 and 2 in the intraverbal and standard condition.
The open circles represent the objects "shoe" and "car" for Participant 1, and "shoe" and "fish"
for Participant 2. The closed circles represent "scissors" and "bed" for Participant 1, and "table"
and "bed" for Participant 2.

the standard condition, or if a pure tact (who had previously worked with this
would occur. participant on other skills), there was
It is interesting to note how a change no substantial change in his perfor-
in the experimenter affected this partic- mance, but by the next session his per-
ipant's performance. During the first formance deteriorated. Observations of
session with the new experimenter his behavior during these sessions
96 MARK L. SUNDBERG et al.
showed an increase in his task-avoid- 2, top panel), Participant 1 performed
ance behaviors (e.g., falling on the slightly better with the standard pro-
floor, climbing on the table), suggest- cedure. On this probe he correctly tact-
ing the instructional stimulus control ed an object from the standard condi-
established by the first experimenter tion seven times in 56 trials (12.5%),
did not generalize to the second exper- whereas in the intraverbal condition he
imenter. However, the split in perfor- correctly tacted an object five times in
mance between the two conditions was 56 trials (8.9%). During the reversal
maintained, and his performance grad- phase, however, his performance was
ually returned to the previous level, substantially better in the intraverbal
and then improved further. condition. He correctly tacted an object
During the reversal phase, Partici- in the intraverbal condition four times
pant 2's performance deteriorated in 12 trials (33%), whereas in the stan-
somewhat when the objects from the dard condition he never successfully
intraverbal condition were presented in tacted an object in 12 trials (0%).
the format of the standard condition, When the conditions were the same, he
but his performance recovered after showed better performance in one in-
three sessions (M = 79.4% for the traverbal condition, with five correct
whole condition). However, his perfor- tacts out of 24 trials (20.8%), than he
mance never returned to the level did in the other, with one correct tact
achieved in the final five sessions of out of 24 trials (4.2%).
the previous condition. When the ob- On the postsession probe (Figure 2,
jects that had previously been in the bottom panel) Participant 1 correctly
standard condition were presented in tacted an object from the intraverbal
the intraverbal format, performance condition 17.5 times in 54 trials
improved (M = 70%), but not to the (32.4%), whereas in the standard con-
level that had been achieved with the dition he correctly tacted an object
other objects during the intraverbal 12.5 times in 54 trials (23.1%). He was
condition in the previous phase. How- successful in tacting all four objects at
ever, it was observed that during the least three times (but never all four in
standard condition the verbal stimulus the same session). The separation be-
"What is that?" began to function ap- tween the two conditions appears to
propriately as an SD for the participant widen in the later sessions of this
to look at the object, rather than as a phase. During the reversal phase he
verbal stimulus that immediately correctly tacted an object in the intra-
evoked a verbal response without re- verbal condition six times in 12 trials
gard to the presence of a nonverbal (50%), whereas in the standard condi-
stimulus. In addition, this participant tion he correctly tacted an object four
correctly tacted the objects in the pre- times in 12 trials (33.3%). When the
and postsession probes; thus, these ob- conditions were the same (all intra-
jects were not returned to the intra- verbal), the slight separation between
verbal condition. the two sets of objects was lost. The
Figures 2 and 3 show performances total number of correct tacts in this
on the pre- and postsession pure tact condition was 16 of the 52 trials
probes. There was a single tact trial for (30.7%).
each object during these probes. The It is interesting to note the lasting
cumulative number of correct pure effects of the reversal to the standard
tacts across sessions are presented. Fig- condition for Participant 1. When the
ure 2 shows that Participant 1 did not objects that had been in the intraverbal
completely acquire pure tacts in either condition were presented with the
probe; however, he did demonstrate a "What is that?" SD, not only did his
better overall performance on both performance decrease (see the top pan-
probes during intraverbal conditions. el of Figure 2, reversal phase, standard
During the presession probe (Figure condition), but he never recovered
INTRAVERBAL PROMPTS 97
Pre-Session Pure Tact Probe
Baseline Intervention Reversal All Intraverbal
20 -

L_ Intra-
F ~~IStandrdDOO

E I Intraverbalv
Standard
r

3 0

0 5 10 15 20 25 30 35 40 45 50

Post-Session Pure Tact Probe


Baseline Intervention Reversal All Intraverbal
30 -

d " "Standard
25 -

CD)
L 20 -

o ~~~~~~~~~Intraverbal
co

L 5 -

z 0 5 0 5I0n530 35 40 45 5

Fg2. Cmltv ubro orc at o atcpn durinbharel n osssin rbs


and
"bed."~~~~~~~~Stndr

from this condition, even when the ure 1, top panel), thus demonstrating
conditions were reversed back to in- an additional value of the probes as a
traverbal. A similar effect can be ob- separate measure of performance.
served in the postsession probes, but to Figure 3 shows that Participant 2
a lesser degree. This effect was not ob- acquired pure tacts in both conditions,
served in the within-session data (Fig- but did substantially better in the in-
98 MARK L. SUNDBERG et al.

| 1 ~ ~ ~ ~ ~ ~ ~ Intraverbal
Pre-Session Pure Tact Probe
Baseline Intervention Reversal
60
CC.0i) 50 - | |Standard
50

L(Q
40- - 1
30

D I ntr Raversal
E
Z o20 -

Standard
CD 10

E
u
0
0 5 10 15 20 25 30 35 40 45 50 55 60

Post-Session Pure Tact Probe


Baseline Intervention Reversal
80-
U)
(0-7
4-,

C.)7 Standard
I-
C. 60-
CD
L I J
C

50-
L_ 40
CD Intraverbal
E= 30 -

z
. 20

10-
E Participant1
0 ~
0 5 10 15 20 25 30 35 40 45 50 55 60
Sessions
Fig. 3. Cumulative number of correct tacts for Participant 2 during the pre- and postsession probes.
The open circles represent the objects "shoe" and "fish," and the closed circles represent "scissors"
and "bed."

traverbal condition. During the pre- 11 times in 76 trials (14%). During the
session tact probe (Figure 3, top panel) reversal phase, however, his perfor-
he correctly tacted an object from the mance was better in the standard con-
intraverbal condition 32 times in 78 tri- dition. He correctly tacted an object in
als (40.1%), whereas in the standard the standard condition 26 times in 30
condition he correctly tacted an object trials (86.7%), whereas in the intra-
INTRAVERBAL PROMPTS 99
verbal condition he correctly tacted an sion 66. There was a slight decrease in
object 19 in 30 trials (63.3%). performance, but not to the degree that
During the postsession probe (Figure had occurred during the reversal with
3, bottom panel) Participant 2 correctly the first objects (see Figure 1, top pan-
tacted an object from the intraverbal el), and the participant quickly recov-
condition 43.5 times in 76 trials ered and correctly tacted with the SD
(57.2%), whereas in the standard con- "What is that?" The bottom panel of
dition he correctly tacted an object 18 Figure 4 shows that Participant 2 ac-
times in 76 trials (23.7%). During the quired the signs for the two control ob-
reversal phase he correctly tacted an jects very quickly. He met the criterion
object in the standard condition 28 of 80% or better in four sessions. This
times in 32 trials (87.5%), and in the rate of acquisition is substantially fast-
intraverbal condition he correctly tact- er than this participant demonstrated
ed an object 23 times in 32 trials with the first set of objects (see Figure
(71.8%). The failure to show a reversal 1, bottom panel).
in pre- and postsession probes was Figure 5 shows that both participants
probably due to the fact that this par- were successful under pure tact con-
ticipant (unlike Participant 1) had near- ditions. During the presession tact
ly acquired both pure tacts in the in- probe, Participant 1 correctly tacted an
traverbal condition prior to reversal. At object 14 times in 48 trials (29.1%),
this point the verbal SD "What is and during the postsession tact probe,
that?" seemed to help the discrimina- he correctly tacted an object 23 times
tion (see Figure 3, top panel, reversal in 48 trials (47.9%). This percentage of
condition) by functioning as an SD to correct pure tacting is a substantial im-
attend to the nonverbal stimulus. It provement over his performance with
should be pointed out that the improve- the first set of objects (see Figure 2),
ment in tacting during the probes by despite the slower acquisition demon-
both participants may have been par- strated with these objects. Participant 2
tially due to the use of reinforced also showed a transfer to pure tacts.
probes. However, this contingency was During the presession tact probe, Par-
in effect for both conditions, and there
still was a split in performance. ticipant 2 correctly tacted an object
Figure 4 shows the participants' per- three times in eight trials (37.5%), and
formance on the control objects. Dur- during the postsession tact probe he
ing the initial baseline conditions and correctly tacted an object six times in
the baseline conducted during inter- eight trials (75%).
vention on the other objects, neither
participant correctly tacted any of the Anecdotal Follow- Up
objects. The participants were given
training with just the intraverbal pro- Approximately 1 year after the start
cedure. The top panel shows that Par- of the study, both participants had ac-
ticipant 1 met the criterion of three quired over 50 tacts and receptive dis-
consecutive sessions of 80% or better criminations. They both were able to
after 18 sessions (M = 63.3%). Train- successfully respond to the verbal SD
ing on this condition was initially start- "What is that?" and no longer needed
ed in Session 43, but it took too long the intraverbal prompt procedure. In
to conduct the training sessions with addition, manding increased substan-
six target objects, so training was dis- tially, and both participants demon-
continued until the end of the all-in- strated improved echoic skills includ-
traverbal condition. Training on this ing vocal mands and tacts. They also
condition was extended past the crite- both began to acquire some simple in-
rion level to see if pure tacts would traverbal behavior. These advance-
emerge. The conditions were reversed ments were considered substantial for
to the standard format following Ses- these 2 participants because they had
100 MARK L. SUNDBERG et al.

Sign Language Acquisition for the Control Groups


Baseline Intervention Reversal
100
90 -

80

00

U 50 l
CD
CD
loc 40s Priipn
entraverbal Standard
L 30
-3

20 -

10- Participant 1
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75

Baseline Intervention
100
* 0-
90
80-
70 lintra-
represent
0)
the obects "og" ad "top for Prticiant 1,and "dg" an "sock for Prticiverbal
60
CD)
40--
CD
0)
a.030
20-

10
0 5 10 15 20 25 30 35 40 45 50 55 60
Sessions
Fig. 4. Percentage of correct tacts for the control objects for Participants 1 and 2. The closed circles
represent the objects "dog" and "top" for Participant 1, and "dog" and "sock" for Participant 2.

long histories of failing to acquire any sign language responses from imitative
verbal behavior. prompts. In addition, despite long his-
tories of failure with the "What is
DISCUSSION that?" procedure, the participants not
The current study demonstrated that only acquired pure tacts, but both of
the intraverbal prompt procedure can them were ultimately able to emit cor-
be an effective technique for freeing rect tacts with the "What is that?" ver-
INTRAVERBAL PROMPTS 101

Pre- and Post-Session Tact Probes for Control Objects

All Intraverbal All Standard


30
4-

2
25
Ca)
20
ol20
0
L.
a) 15
E Post-Session Probe
z0) 10 -
Pre-Sessio

4--

E
u Participant 1
0
0 5 10 15 20 25 30

All Intraverbal
C) 8

C.)
po 0Q

L Post-~~~~~SessionsPob
Fi.E Sessions-SssonPrb
uuaienme fcfettcsfrPriiat uigtepe oteso
prbsoz
an sc"
h
o
oto bet.Teojcs"o"ad"o"wr
atiiat2
n
sdfrPriiat1
n
n dg

an Participaaricpannt

bal SD. These results replicate and ex- tain types of multiple control can be
tend the findings by Partington et al. effective for establishing new types of
(1994), and support their analysis of verbal behavior.
the potential blocking effect of the ver- It is interesting to note the negative
bal stimulus "What is that?" in early effects of the standard "What is that?"
tact training. The current data also sup- tact training procedure on the 2 partic-
port Skinner's (1957) proposal that cer- ipants in this study. It appears that for
102 MARK L. SUNDBERG et al.
some children with severe language performance with the "What is that?"
delays, this verbal stimulus may ac- condition in the final phase of the con-
quire intraverbal control over a specific trol objects.
sign or word, as observed by Parting- These findings have several interest-
ton et al. (1994). In addition, this ver- ing implications and other possibilities
bal stimulus may acquire control over for further research. Perhaps most im-
a class of signs (or words) as demon- portant is that the common practice of
strated by the guessing type of errors using the "What is that?" procedure
observed in the current study. With may cause more harm than good for
both types of errors, the participants some children. However, many chil-
tended to respond quickly to the verbal dren can easily acquire tacts with the
SD without even looking at the object; "What is that?" procedure, so it is im-
an effect similar to that observed by portant to develop an assessment tool
Glat et al. (1994). At one point Partic- that would allow the early identifica-
ipant 1 was presented with each of five tion of children who do not acquire
of his reinforcers (which he could eas- tacts with the standard procedure and
ily mand for with signs) and was asked who would benefit from the intraverbal
"What is that?" He signed "bed" for procedure. This type of early identifi-
all five objects. The intraverbal proce- cation may help to avoid the establish-
dure eliminated these errors, and thus ment of error responses. In addition,
overcame this linguistic barrier, mak- procedures to fade the intraverbal
ing the further development of the tact prompt should be explored further. Fol-
repertoire possible, as demonstrated by lowing the current study, a delay pro-
current data and the anecdotal follow- cedure and a partial prompt procedure
up data. were shown to be effective for fading
The differences between the perfor- the intraverbal prompt; however, fur-
mances of the 2 participants was also ther research is necessary to examine
of interest. Although both participants these procedures empirically.
began at essentially the same verbal An important element of the current
level, their day-to-day performance research was that the intraverbal
and acquisition varied considerably. prompt was actually used in both con-
Participant l's performance was slow ditions. However, in the intraverbal
but increased steadily, whereas Partic- condition it was used as an antecedent
ipant 2 demonstrated much more vari- stimulus presented prior to the target
ability. Participant 2 engaged in a sub- response, whereas in the standard con-
stantial amount of avoidance and es- dition it was used in the correction pro-
cape behavior during sessions and was cedure following an incorrect response.
more difficult to work with, but ulti- Thus, the results of the current study
mately he demonstrated the best trans- support the view that a focus on using
fer to pure tacts. Participant 2 did not prompts to prevent errors may be more
need the all-intraverbal condition, be- beneficial than the use of prompts to
cause after a brief decrease in the re- correct errors (Terrace, 1963; Tou-
versal condition his performance under chette, 1971). Further research on the
the standard condition steadily im- location of prompts in the procedures
proved (although it was quite variable) used to teach verbal behavior to chil-
to a level higher than in the intraverbal dren with language delays would cer-
condition. This may be due to his suc- tainly be interesting.
cessful history with these objects, or to There are several aspects of the clas-
the possibility that the verbal stimulus sification of verbal behavior used in the
"What is that?" appropriately func- current study that need to be further
tioned as an SD to attend to the object. explained. First, the initial effect ob-
These possibilities are in need of fur- served in the study was the elimination
ther research, especially because Par- of imitative prompts. This step was
ticipant 1 also demonstrated successful significant for these participants be-
INTRAVERBAL PROMPTS 103

cause of their history of dependence on aversive stimulus (the experimenter


imitative prompts, but the responses at and an object presented in front of the
this point can only be classified as mul- participant) that is terminated upon a
tiply controlled responses because it response. Although no measures were
was unclear whether they were con- taken on any of these variables, it is
trolled by the English word (intraver- probably the case that one or all of
bal) or the object (tact). It was clear them played a role in evoking the re-
that neither stimulus alone evoked the sponse during the probe conditions.
response initially, but when presented However, although they are not totally
together they did control a correct re- pure tacts, for purposes of the current
sponse. Eventually the object alone study they were a significant approxi-
evoked a tact, but no data were taken mation to pure tacts for these partici-
on the intraverbal prompt alone during pants.
training to determine at what point this In conclusion, the data show that
stimulus could correctly evoke the tar- participants who had long histories of
get response. These data would be of failing to acquire tacts could be suc-
interest in future research on this pro- cessful with a modification of the
cedure. teaching procedure. The analysis pre-
Second, both procedures involved a sented in this study and the interven-
type of multiple control in that they tion procedure that proved to be suc-
both used a verbal and nonverbal stim- cessful were directly derived from Ver-
ulus. However, the "What is that?" bal Behavior (Skinner, 1957). The re-
stimulus, although clearly a verbal sults of this study lend further support
stimulus, typically cannot control a to Skinner's point that his analysis of
specific verbal response form without verbal behavior "is inherently practical
the accompaniment of a nonverbal and suggests immediate technological
stimulus. That is, in the absence of any applications at almost every step" (p.
object to attend to (e.g., if someone is 12).
blindfolded) one cannot successfully
answer the question "What is this?" REFERENCES
Thus this type of prompt is perhaps Bijou, S. W., & Ghezzi, P. M. (1999). The
more correctly identified as simply a behavioral interference theory of autistic
verbal prompt. On the other hand, the behavior in young children. In P. M. Ghez-
verbal prompt "sign shoe" can suc- zi, W. L. Williams, & J. E. Carr (Eds.), Au-
cessfully evoke a specific intraverbal tism: Behavior analytic perspectives (pp.
response in the absence of the nonver- 33-43). Reno, NV: Context Press.
bal stimulus. Thus, this type of prompt Carroll, R. J., & Hesse, B. E. (1987). The
is classified as an intraverbal prompt in effects of alternating mand and tact training
on the acquisition of tacts. The Analysis of
the current study. Verbal Behavior, 5, 55-65.
Finally, the classification of the pre- Drash, P W., High, R. L., & Tutor, R. M.
and postsession tact probes as pure (1999). Using mand training to establish an
tacts suggests that there were no other echoic repertoire in young children with
sources of control for the response oth- autism. The Analysis of Verbal Behavior,
er than the relevant nonverbal stimulus. 16, 29-44.
Glat, R., Gould, K., Stoddard, L. T, & Sid-
Although there were no verbal stimuli man, M. (1994). A note on the transfer of
such as "What is that?" or "sign dog," stimulus control in the delayed-cue proce-
there were possibly other variables, dure: Facilitation by an overt differential
such as the presentation of the object response. Journal of Applied Behavior
in front of the participant (an SD to re- Analysis, 27, 699-704.
spond), the presence of an audience Guess, D., Sailor, W. S., & Baer, D. (1976).
A functional speech and language training
with a history of reinforcing verbal re- program for the severely handicapped.
sponses, and the possibility of a reflex- Lawrence, KS: H & H Enterprises.
ive conditioned EO (Michael, 1993b) Kent, L. (1974). The language acquisition
that would involve the presence of an program. Champaign, IL: Research Press.
104 MARK L. SUNDBERG et al.
Lovaas, 0. I. (1977). The autistic child: Lan- Skinner, B. F (1957). Verbal behavior. New
guage development through behavior mod- York: Appleton-Century-Crofts.
ification. New York: Irvington. Smith, T. (1993). Autism. In T R. Giles
Lovaas, 0. I. (1987). Behavioral treatment (Ed.), Handbook of effective psychotherapy
and normal educational and intellectual (pp. 107-133). New York: Plenum.
functioning in children with autism. Jour- Sundberg, M. L., & Partington, J. W. (1998).
nal of Consulting and Clinical Psychology, Teaching language to children with autism
55, 6-9. or other developmental disabilities. Pleas-
Maurice, C., Green, G., & Luce, S. C. ant Hill, CA: Behavior Analysts, Inc.
(1996). Behavior interventions for young Terrace, H. S. (1963). Discriminative learn-
children with autism. Austin, TX: Pro-Ed. ing with and without errors. Journal of the
Michael, J. (1 993a). Concepts and principles Experimental Analysis of Behavior, 6, 1-
of behavior analysis. Kalamazoo, MI: So- 27.
ciety for the Advancement of Behavior Touchette, P. E. (1971). Transfer of stimulus
Analysis. control: Measuring the moment of transfer.
Michael, J. (1993b). Establishing operations. Journal of the Experimental Analysis of
The Behavior Analyst, 16, 191-206. Behavior, 15, 347-354.
Ulman, J. D., & Sulzer-Azaroff, B. (1975).
Partington, J. W., Sundberg, M. L., New- Multielement baseline design in education-
house, L., & Spengler, S. M. (1994). Over- al research. In E. Ramp & G. Semb (Eds.),
coming an autistic child's failure to acquire Behavior analysis: Areas of research and
a tact repertoire. Journal ofApplied Behav- application (pp. 377-391). Englewood
ior Analysis, 27, 733-734. Cliffs, NJ: Prentice Hall.
Singh, N. N., & Solman, R. T (1990). A Urcuioli, P. J. (1984). Overshadowing in
stimulus control analysis of the picture- matching-to-sample: Reduction in sample-
word problem in children who are mentally stimulus control by differential sample be-
retarded: The blocking effect. Journal of haviors. Animal Learning & Behavior, 12,
Applied Behavior Analysis, 23, 525-532. 256-264.

Você também pode gostar