ERIC EJ859321: Emergence of Tacts following Mand Training in Young Children with Autism

2009·0.32 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2009, 42, 691–696 NUMBER3 (FALL2009)
EMERGENCE OF TACTS FOLLOWING MAND TRAINING IN YOUNG
CHILDREN WITH AUTISM
CLAIRE E. EGAN AND DERMOT BARNES-HOLMES
NATIONALUNIVERSITYOFIRELAND,MAYNOOTH
Thisstudysoughttoexaminetheeffectsoftrainingmandsontheemergenceoftactswiththe
same response forms. Results indicated that training adjective sets as mands resulted in the
emergenceofadjectivesetsastactsundermodified,butnotstandard,antecedentconditions.The
findings suggested that the apparent functional independence of mands and tacts may be
explained byalackof appropriateantecedent controlover responding.
DESCRIPTORS: antecedent control, autism, functional independence, language, mand,
tact
_______________________________________________________________________________
Skinner’s analysis of verbal behavior (1957) The view of mands and tacts developing as
classifiedhumanlanguageintodifferenttypesof functionally independent responses has in-
verbal operants, each defined by their control- formed professionals who design early language
ling antecedent and consequent variables. instruction based on Skinner’s model of verbal
According to Skinner, mands are controlled by behavior (Sundberg & Michael, 2001). Re-
relevant establishing operations (EOs) and searchers in this area have often focused on
specific reinforcers, whereas tacts are controlled establishing emergent verbal behavior, in which
by nonverbal discriminative stimuli (SD) and a response acquired under the controlling
generalized reinforcement. conditions of one verbal operant transfers to
Skinner (1957) suggested that the unique the conditions of another verbal operant
controlling variables defining mands and tacts without direct instruction (Hernandez, Hanley,
resulted in two functionally independent be- Ingvarsson, & Tiger, 2007). For example,
havioral classes. Thus, the acquisition of a Nuzzolo-Gomez and Greer (2004) tested for
particular response form under one set of emergence mands and tacts using adjective–
controlling variables (e.g., an SD for a tact) object pairs (e.g., big, medium, small bowl).
wouldnotnecessarilyresultintheemergenceof Results showed that training in one verbal
that same response form under other control- operant (e.g., mand) did not result in the
ling variables (e.g., reinforcer deprivation for a emergence of the other verbal operant (e.g.,
mand). tact) until multiple-exemplar instruction (e.g.,
training mand and tact responses separately)
was applied across one set of the adjective–
These experiments were completed as part of the object pairs. These researchers concluded that
doctoral work submitted by Claire E. Egan to the multiple-exemplar instruction provided a histo-
DepartmentofPsychologyatNUI,Maynooth.Wethank
ry of reinforcement for responding under
Olive Healy and Eva Sheehy-Perera for providing
various antecedent events for mands and tacts
interobserveragreement.WealsothankR.DouglasGreer
for allowing us to conduct the research at the Cork and thereby created contextual control over the
CABASSchool.
different functions (i.e., participants learned to
Correspondence concerning the study should be
emit mands and tacts after instruction across
addressed to Claire E. Egan, The Hong Kong Institute
for Education, Department of Educational Psychology, multiple antecedent conditions). Based on this
Counselling and Learning Needs, D1-2/F-29, Ling Po finding,thecurrentstudysoughttoexaminethe
Road,TaiPo,NewTerritories,HongKongSAR(e-mail:
role of specific antecedents in establishing a
[email protected]).
doi:10.1901/jaba.2009.42-691 context for the emergence of mands and tacts,
691
692 CLAIRE E. EGAN and DERMOT BARNES-HOLMES
using mand training procedures outlined by Preexperimental probes. Prior to training, two
Nuzzolo-Gomez and Greer (2004). probes were conducted, a 15-trial mand probe
and a 15-trial tact probe. For mand probes, the
experimenter placed a preferred item in one of
METHOD
the bowls to specify the target adjective, and no
Participants, Settings, and Materials
otherpromptsweredelivered.Correctresponses
Participantswere4boys,5years7monthsto
for mands were vocal forms that included the
7 years 7 months old, who had been diagnosed
correct adjective, with the autoclitic frame ‘‘I
with autism. During ongoing language training
want’’ (e.g., saying, ‘‘I want small’’). Incorrect
programs, the participants had learned to mand
mandsweredefinedasthosemandsthatdidnot
in the presence of objects following a period of
includetheautocliticframe‘‘Iwant’’ordidnot
deprivation and to tact when a teacher pointed
accuratelyspecifytheadjective(e.g.,didnotsay
to a target stimulus. All of the participants had
‘‘small’’).
learned vocal responses through the use of
For tact probes the experimenter pointed to
echoic modeling. Adjective sets for which the
one of the bowls to specify the adjective the
participants had no previous instruction were
participant should tact. Correct responses were
evaluated and included left, middle, right
vocal forms that included the correct adjective,
(Thomas); first, second, last (Leo); and small,
with the autoclitic frame ‘‘It is’’ (e.g., ‘‘It is
medium, big (Ben and Sean).
small’’). Incorrect responses were those that did
Eachparticipantsatatatableacrossfromthe
not specify the accurate adjective or did not use
experimenter. The materials used during all
the autoclitic frame ‘‘It is.’’ No social conse-
sessions were three clear plastic bowls of
quences were delivered during probes.
different sizes. The experimenter placed bowls
Mand training. Participants were taught to
on the table in front of the participant prior to
respond to the mand conditions described
each trial in random order.
above by vocally specifying the target adjective.
Phase 1 For correct mands, the experimenter delivered
the preferred item to the participant. For
The experimental sequence involved preex-
perimental probes (mands and tacts), mand incorrect responses or if the participant did
training, and postexperimental probes (tact and not respond, the experimenter presented the
listener probes). Probe sessions consisted of 15 vocal model (e.g., ‘‘I want small’’), and the
trials per session, with five opportunities to participantechoedthatmodel.Iftheparticipant
respond to each of the three target adjectives. did not echo the model within three opportu-
Duringmandprobeandmandtrainingsessions, nities, the experimenter put away the preferred
teacher-nominated preferred items and edible item and discontinued training for that adjec-
items (e.g., candy, toys) previously used to tiveonthattrial.Mandtrainingcontinueduntil
promote language acquisition were employed. participants achieved 14 of 15 correct respond-
To ensure that EOs were in effect during mand ing across two successive sessions or 15 of 15
trials, preferred items were restricted outside the once.
sessions on days when the experiment was Postexperimental probes. Following mand
conducted. Participants could select a preferred training, tact probes were conducted as de-
item on average every two trials by pointing, scribedforthepreexperimentalprobes.Previous
reaching,orvocallyrequestingfromamongthree research has demonstrated that listener or
or four preferred items. During mand training, receptive language commonly develops before
the opportunity to select a preferred item was speaker or ‘‘expressive’’ language in young
contingent on correct responding. children (Wynn & Smith, 2003). Therefore,
EMERGENCE OF TACTS 693
probes (e.g., ‘‘Point to the big bowl.’’) were mand or tact probes. Following mand training,
conducted to determine if listener responses Thomas, Ben, and Sean displayed no correct
were present, despite the absence of a tact responses on tact probes, whereas Leo scored 5
response (e.g., ‘‘It’s a big bowl.’’) of the same of 15 (Figure 1). All participants displayed
form. During listener probes, the experimenter some correct responding on listener probes,
presented a vocal instruction, ‘‘Point to —’’ althoughonlyThomasdisplayedcorrectlistener
(e.g., ‘‘Point to small’’) relevant to one of the responses at a relatively high level.
three targets. A correct response was defined as Phase 2 (modified tact). Thomas, Ben, and
the participant contacting the materials with Sean did not demonstrate an emergence of
any part of the fingers or hand. An incorrect untaught tact responses during standard tact
responsewasrecordediftheparticipantpointed conditions. However, when the antecedent was
toanincorrectitemormorethanoneitem.No modified, the correct tact responses emerged
differential consequences were delivered for (Figure 2). By contrast, Leo demonstrated an
correct or incorrect responses. Postexperimental emergence of untaught tact responses during
probes consisted of no mand probes (as preexperimental standard tact probes and
occurred during the preexperimental probes) postexperimental modified tact probes.
because all participants had achieved mastery In summary, following mand training, 3 of
criteria during mand training. the 4 participants displayed emergence of
untaught tact responses during modified, but
Phase 2
not standard, tact conditions. These results
Phase 2 of the study replicated Phase 1 using
suggest that modifying the antecedent stimulus
identicalproceduresexceptfortheadditionofa
influenced the emergence of untaught tact
modified tact condition during the postexper-
responses.
imental probes. Phase 2 was conducted 4 weeks
Itispossiblethatthevocalstimuluspresented
after the conclusion of Phase 1.
in the modified tact probes (i.e., the experi-
Modified tact probes. The modified tact
menter saying, ‘‘What is it?’’) cued participants
probes were identical to the tact probes
to the change from mand to tact contingencies.
described in Phase 1 except that the experi-
Thus,thelackoftactemergenceunderstandard
menter included a vocal antecedent (saying,
tact conditions may be explained by ambiguous
‘‘What is it?’’) while pointing to one of the
antecedent control in Phase 1, rather than
bowls to specify the adjective the participant
functional independence per se (because the
should tact.
modified antecedent produced emergent re-
sponding). The expected performance for two
Interobserver Agreement
functionallyindependentverbaloperantswould
Interobserver agreement was collected for
be that training one operant (e.g., the mand)
29% of all sessions and was 100% for all
would not result in the emergence of the other
participants. It was calculated by dividing the
verbal operant (i.e., the tact) without direct
total number of agreements on the occurrence
instruction, which suggests that modifying the
of correct and incorrect responses by the total
antecedentconditionsaloneshouldnotproduce
number of agreements plus disagreements, and
emergent responding. Nevertheless, untaught
converting this ratio to a percentage.
tacts emerged when the vocal antecedent was
added during Phase 2. Future research should
RESULTS AND DISCUSSION
evaluate the functional independence of mands
Phase 1. Prior to mand training none of the and tacts under a variety of antecedent
participants displayed correct responses during conditions.
694 CLAIRE E. EGAN and DERMOT BARNES-HOLMES
Figure1. Responsestoprobesduringmandprobes,tactprobes,andlistenerprobes,andscoresduringmandtraining
trialsfor Thomas,Leo, Ben, andSean.
Figure2. Responsestoprobesduringmandprobes,standardtactprobes,modifiedtactprobes,andlistenerprobes,
andscoresduringmandtrainingtrialsforThomas,Leo,Ben,andSean.(MandtrainingdataarenotdepictedforBen.)
EMERGENCE OF TACTS 695
In general, two types of errors were made want candy’’). According to Skinner (1957), no
during tact probes. Leo and Sean consistently specified relation exists between a mand
responded without the target adjective (e.g., response and a prior stimulus; rather, the
saying, ‘‘It is a bowl’’ rather than ‘‘It is a relevant EOs and specific reinforcement define
medium bowl’’). All participants had extensive the mand. Thus, based on the occurrence of
extraexperimental training to tact objects using these responses under specific evocative condi-
the autoclitic frame ‘‘It is —.’’ Thus, this type tions, they may be characterized as mands.
of error may have resulted from a history of Anunexpectedperformancewasobservedfor
reinforcement for an alternate response form. Leo across Phases 1 and 2. In Phase 1, he had
Thomas and Ben consistently responded to been trained on mand responses but not
tact probes by saying, ‘‘I want [target autoc- standard tacts. However, during Phase 2 he
litic]’’ rather than ‘‘It is [target autoclitic].’’ emitted 15 of 15 (100%) correct responses to
Theseparticipantsappearedtorespondwiththe standardtactconditionsduringpreexperimental
response form that had been most recently probes and 0 of 15 in the mand probes.
reinforced (i.e., the autoclitic frame ‘‘I want’’ Although the reasons behind this unexpected
that was reinforced during mand training prior outcome remain unclear, it may be that the
tothetactprobesessions).Asecondexplanation experimental procedures in Phase 1 provided
for this type of error might be that the ambiguous antecedent control for the standard
participants continued to request preferred tact probes. Likewise, Thomas and Leo both
items due to reinforcer deprivation that had scored0of15onpreexperimentalmandprobes
been established during earlier mand training in Phase 2, despite reaching criterion during
(i.e., the relevant EO was still in place). mand training in Phase 1. It may be that the
Whereas Skinner (1957) defined pure tact training criterion of 15 of 15 once was
responses as having nonverbal antecedent insufficient to ensure that the mand response
controls, he also characterized tact responses would be maintained over the 4 weeks between
evoked by both nonverbal and verbal stimuli as Phases 1 and 2.
having impure or multiple antecedent controls. It is notable that scores for the listener
During the modified tact probes, the emergent probes across Phases 1 and 2 increased for 3 of
tacts may be more accurately classified as the 4 participants, despite the absence of direct
impure tacts, due to the presence of an training.The4thparticipant,Sean,maintained
additional (vocal) stimulus. To date, all studies high scores across both experiments. These
that have examined tact emergence following data suggest that relations existed among
mand training have employed vocal (e.g., mand, tact, and listener responses despite
Lamarre & Holland, 1985) or nonvocal (e.g., initial evidence that they were functionally
Nuzzolo-Gomez & Greer, 2004) verbal ante- independent. However, the nature of these
cedents during tact testing and may thus have relations remains unexplained. It may be the
employed impure antecedent controls. Future case that the introduction of the modified tact
research might differentiate emergent perfor- probes clarified the stimulus conditions in the
mances for tacts under pure (nonverbal) versus listener probes. It is also possible that the
impure (verbal and nonverbal) antecedent additional mand training provided in Phase 2
conditions. contributed to the emergence of the listener
Inthepresentstudy,participantsweretaught behavior.
to mand by specifying the bowl containing the One limitation to the present study was that
preferred item (e.g., ‘‘I want big bowl’’) rather the two phases employed the same participants
than directly specifying the reinforcer (e.g., ‘‘I and training stimuli. It could be argued that
696 CLAIRE E. EGAN and DERMOT BARNES-HOLMES
the effectiveness of the modified antecedent in REFERENCES
evoking tact responses in Phase 2 was, in part,
Hernandez,E.,Hanley,G.P.,Ingvarsson,E.T.,&Tiger,
a result of the testing conditions in Phase 1. A J. H. (2007). A preliminary evaluation of the
second limitation to the current study was the emergence of novel mand forms. Journal of Applied
BehaviorAnalysis,40,137–156.
single pre- and posttest probe design em-
Lamarre, J., & Holland, J. (1985). The functional
ployed, which limits the demonstration of independence of mands and tacts. Journal of the
functional control. Further, a preexperimental Experimental AnalysisofBehavior,43,5–19.
Nuzzolo-Gomez, R., & Greer, R. D. (2004). Emergence
assessment of modified tacts was not conduct-
of untaught mands or tacts of novel adjective-object
ed. Consequently, it is unclear what effect pairs as a function of instructional history. The
mand training had on the performance of the AnalysisofVerbal Behavior,20,63–76.
Skinner, B. F. (1957). Verbal behavior. New York:
participants in the modified tact condition.
Appleton-Century-Crofts.
Thus, the current results should be viewed as
Sundberg, M. L., & Michael, J. (2001). The benefits of
preliminary, and future studies should address Skinner’sanalysisofverbalbehaviorforchildrenwith
these procedural issues by including preexperi- autism.BehaviorModification,25(5),698–724.
Wynn,J.W.,&Smith,T.(2003).Generalizationbetween
mental measures of modified tacts and by
receptive and expressive language in young children
employing an experimental design that allows withautism.Behavioral Interventions,18, 245–266.
evaluation of the variables responsible for
Received January 15,2008
performance under modified and standard tact
Final acceptance January13, 2009
conditions. Action Editor,Henry Roane