You are on page 1of 17

The Analysis of Verbal Behavior 2007, 23, 71–87

The Induction of Naming in Children with No Prior Tact

Responses as a Function of Multiple Exemplar Histories of

Carol A. Fiorile and R. Douglas Greer, Teachers College and the Gradu-
ate School of Arts and Sciences, Columbia University

The phenomenon identified as naming is a key stage of language function that is missing in many children
with autism and other language delay diagnoses. We identified four children with autism, who, prior to the
implementation of this experiment, did not have the naming repertoire (either speaker to listener or listener
to speaker) and who had no tact responses for two- or three-dimensional stimuli. Tact training alone did not
result in a naming repertoire or echoic-to-tact responses for these students. We then provided multiple
exemplar instruction (MEI) across speaker and listener repertoires for a subset of stimuli (the teaching set)
that resulted in untaught response components of naming and the capability to acquire naming after learn-
ing tacts for subsequent sets of stimuli. We used a delayed multiple-baseline probe design with stimuli
counterbalanced across participants. The results showed that for all four students, mastery of tacts alone
(the baseline or initial training condition) was not sufficient for the naming or echoic-to-tact repertoires to
emerge. Following MEI the naming repertoire emerged for all four students for the initial set of stimuli. In
addition, we tested for naming with novel stimuli that were probed prior to the MEI and naming also
emerged following tact instruction alone for these sets. The results are discussed in terms of the role of
naming in the incidental acquisition of verbal functions as part of the speaker-as-own-listener repertoire.
Key words: naming, verbal behavior, multiple exemplar instruction, tact, learn unit, transformation of
stimulus function.

Research based on Skinner’s (1957) theory Stolfi, Chavez-Brown, & Rivera-Valdez, 2005;
has proliferated in recent years and has con- Greer, Yuan, & Gautreaux, 2005; Horne &
tributed to the identification of procedures for Lowe, 1996; Ross & Greer, 2003). One source
inducing verbal operants in children who are of verbal behavior that has been identified in
missing them. Much of this experimentation the literature is naming.
has focused on the dependence and interdepen- Naming is a phenomenon first identified by
dence of the classes of verbal repertoires Horne & Lowe (1996) that appears to be the
(Becker, 1989; Greer, Nuzzolo-Gomez, Ross, source of much of children’s verbal repertoire.
& Rivera-Valdez, 2005; Lamarre & Holland, Naming occurs when a child hears someone
1985; Lodhi & Greer, 1989; Michael, 1982; tact, or say the name, of an object that is present
Twyman, 1996a, 1996b; Yoon, 1998), the vari- in the environment, and as a result, the child
ables that functionally control verbal operants can respond to the item both as a listener and
(Chu, 1998; Karmali, Greer, Nuzzolo-Gomez, as a speaker. As a speaker, after this experi-
Ross, & Rivera-Valdes, 2005; Ross & Greer, ence, the child can emit a pure or an impure
2003; Sundberg, Michael, Partington, & tact for the object and the child can respond as
Sundberg, 1996; Tsiouri & Greer, 2003; Will- a listener to the tact for the object. As a listener
iams & Greer, 1993), and studies testing sev- the child will orient to the object when the ob-
eral theoretical explanations for, and the source ject is tacted by another, point to the object
of, productive verbal repertoires (Greer & when asked to indicate where the object is lo-
Keohane, 2005; Greer & Ross, 2004; Greer, cated, or point on hearing the tact. For example,
a parent might point to a bird and say, “Look, a
robin!” Later, on seeing a robin, the parent
might say. “Oh look, there’s another robin,”
The research reported herein was completed as and after hearing this, the child looks at or
part of a Ph.D. dissertation by the first author under
the sponsorship of the second author. Correspon- points to the robin. In addition, and at a differ-
dence concerning this article should be addressed ent time, the child will say, “Robin,” on seeing
to the first author at a robin and the parent says, “Yes, that’s a

robin.” Thus, the child acquires a speaker and presence of generative verbal behavior, such
a listener response without direct instruction. as naming, is not attributable to behavioral se-
Naming appears to be the incidental means lection because the environment does not pro-
whereby children acquire many, if not most, vide enough direct instruction (Pinker, 1999).
of their speaker and listener repertoires. If chil- Hayes, Barnes-Holmes, and Roche (2001) pro-
dren do not have naming they can acquire posed that the source of “generative” (i.e., pro-
speaker and listener responses only through ductive or emergent) verbal behavior was mul-
direct instruction. When a child has naming, tiple exemplar instructional histories (MEI) and
the child can select the stimulus as a listener that MEI offered a “purely behavior analytic
and produce the name of the stimulus as a explanation” for generative verbal behavior.
speaker when the item is present without di- Several studies have subsequently found that
rect instruction. certain types of multiple exemplar instruction
While much of the research on naming has (MEI) did result in the emergence of some gen-
concerned the question of its relation to stimu- erative or productive verbal functions in chil-
lus equivalence or categorization, other recent dren with and without disabilities. Greer, Yuan,
research has concentrated on the acquisition and Gautreux (2005) induced joint stimulus
of naming as a verbal developmental phenom- control across written and spoken spelling re-
enon in children who were found to be miss- sponses, Greer, Stolfi, Chavez-Brown, and
ing the capability (Greer, Stolfi, Chavez- Rivera-Valdez (2005) induced naming in 4- and
Brown, & Rivera-Valdez, 2005). The naming 5-year-old children with mild or no language
repertoire or capability constitutes a critical delays, Greer and Yuan (2005) induced irregu-
means for acquiring new tacts, mands, and lar and regular verb usage, Nuzzolo-Gomez and
other verbal operants as well as listener re- Greer (2004), and Greer, Nirgudkar, and Park
sponses without direct instruction and, as such, (2004) induced transformation of establishing
is a key part of inducing new verbal capabili- operations across mand and tact functions,
ties in children with verbal delays. Greer, Stolfi Speckman (2005) induced novel use of suf-
et al. (2005), found that this emergent behav- fixes, and Marianno-Lapidus (2005) induced
ior was attributable to certain exemplar expe- novel suffixes and joint control of suffixes
riences in children who were missing the nam- across saying and writing.
ing capability. Lowe, Horne, Harris, and Randle (2002)
Naming is one of three types of speaker-as- found that typically developing children dem-
own-listener repertoires that have been identi- onstrated naming after being taught only the
fied in the research literature. They are: (a) tacts suggesting that learning of tacts led to the
correspondence between saying and doing emergence of naming. Horne, Lowe, and
(Paniagua & Baer, 1982), (b) acting as speaker Randle (2004) also found that after teaching
and listener when talking to oneself aloud the listener component of naming, the tact com-
(Lodhi & Greer, 1989), and (c) naming (Horne ponent did not emerge, leading the authors to
& Lowe, 1996). Paniagua and Baer found that suggest that the listener component may be
preschool children showed correspondence present even when the speaker component is
between saying and doing when they were re- missing. Greer, Stolfi, Chavez-Brown, and
inforced for correspondence; thus, the children Rivera-Valdez (2005) found that the listener
responded as listeners to their own speaker to speaker component of naming emerged fol-
behavior. Lodhi and Greer (1989) found that lowing MEI instruction for a subset of stimuli
typically developing 5-year-olds talked aloud in children with mild disabilities. Greer, Stolfi,
to themselves in solitary play conditions in et al. (2005) did not test for whether the nam-
which anthropomorphic toys (i.e., toys that may ing response would emerge following tact in-
be endowed with human characteristics such struction alone. The present experiment iden-
as dolls or animal figures) were available. tified children with severe language delays who
When talking aloud to themselves the children had no echoic-to-tact repertoire and who, when
alternated between speaker and listener func- taught the tact repertoire, did not show nam-
tions, thus demonstrating what Skinner (1957) ing. Next, we tested whether intensive multiple
described as one example of speaker-as-own- exemplar instruction with subsets of stimuli
listener behavior. would lead to naming following tact instruc-
Some psychologists have argued that the tion alone.

Table 1
Participant characteristics at the onset of this experiment.

Participant/Gender/Age Verbal Repertoires

Student B Listener: non-verbal imitation, limited one-step commands,

point Male point response topography without discrimination
2.4 yrs Speaker: emerging, with partial vocal echoics and two
independent mands (e.g., bubble and candy)
Listener/Speaker, Conversational Units, Naming,
Speaker as Own Listener, and Reader/Writer: none

Student X Listener: point response topography without discrimination,

non-Male nonverbal imitation, generalized match repertoire for
2.4 yrs identical stimuli
Speaker: emerging, with vocal echoics for single syllables,
two independent mands (e.g., Elmo and bus)
Listener/Speaker, Conversational Units, Naming, Speaker as
Own Listener, and Reader/Writer: none

Student L Listener: non-verbal imitation with teacher prompts,

Male inconsistent responding to name, point response topography
2.0 yrs without discrimination
Speaker: minimally emerging, with partial vocal
Listener/Speaker, Conversational Units, Naming, Speaker as
Own Listener, and Reader/Writer: none

Student N Listener: point response topography without discrimination,

Male non-verbal imitation of two actions, generalized
2.4 yrs matching for identical stimuli, responds to name
Speaker: minimally emerging, with partial vocal
Listener/Speaker, Conversational Units, Naming, Speaker as
Own Listener, and Reader/Writer: none

METHOD and one of a group of alternated non-target

stimuli for a minimum of 50% of the 20 probe
Participants and Settings trials with familiar stimuli; (b) demonstrated
the response topography of pointing to or
Four children with diagnoses of Autistic touching everyday objects upon request (dis-
Spectrum Disorder (ASD) who had significant crimination between stimuli was not part of the
delays in language acquisition were selected response expectation); (c) had vocal verbal rep-
to participate in this experiment out of a set of ertoires that included echoic responses or par-
ten potential candidates. The children were tial vocal approximations to teacher vocal ver-
selected for participation when they did not bal antecedents for a minimum of 50% of the
demonstrate the basic naming repertoire for trials conducted; (d) did not have echoic-to-
three-dimensional stimuli across the response tact responses or a history of acquiring tacts
components of naming. We selected three-di- via the echoic-to-tact training procedure (Ross
mensional stimuli because the children did not & Greer, 2003; Tsiouri & Greer, 2003; Will-
attend to representational stimuli in the form iams & Greer, 1993); and (e) did not have the
of pictures. We selected students who: (a) naming repertoire as identified by screening
matched three-dimensional visual stimuli when probes and the baseline training procedure. All
presented with one of a set of target stimuli experimental phases were conducted in each

child’s respective home with a second inde- Table 3

pendent observer present for a minimum of Counterbalanced presentation of stimuli sets
35% of the sessions. Characteristics of the se- for each participant.
lected participants are shown in Table 1.
Participant 1st Set 2nd Set 3rd Set
Description of Stimuli
B 1 2 3
The stimuli included nonsense bi-syllabic X 1 4 3
labels (i.e., experimenter coined tacts) paired N 3 2 none
with unusual hardware items, such as clamps L 3 1 4
and bolts. Descriptions of the sets of stimuli
(three per set) may be found in Table 2. Dependent Variables

Table 2 The dependent variables consisted of the

Description of stimuli. untaught listener response (i.e., point to target),
and untaught impure tact responses (i.e., vocal
Phoneme Item Description verbal responses under the multiple anteced-
ent control of a teacher-provided vocal ante-
Set 1 kabal green electrical cedent [What is this?] plus a visual stimulus
hook-up [the tact stimulus]) during pre-treatment and
woemup black, wavy metal 3" post-treatment probes. Responses in the probe
object trials were not consequated, that is students did
tingra gray rubber ¾" tube not receive “feedback” for their responses to
probe trials.
Set 2 nipil black clamp
keytoe large 307A silver Independent Variables
holub large, round silver The baseline or control condition consisted
washer of teaching tacts to mastery using echoic to tact
training learn units. The experimental condi-
Set 3 pakot gold caps on green tion, or independent variable, was the multiple
strip exemplar instruction (MEI) with a training set
dipoy ¾" silver object with (and in several cases, multiple training sets) of
cutout stimuli also taught to mastery using learn units.
galoe ½" drill bit This procedure will be described later in de-
Set 4 zimon 1-½" silver object with
small screw Data Collection
bikmo ½" bolt
mooga ¾" silver pipe fitting Data were collected on the student responses
with small hole during all phases using data collection forms
and a pencil. A second observer was present
The target stimuli were presented for each for a minimum of 35% of the sessions. Prior to
student in a counterbalanced manner and the implementation of the experiment, the second
positions of the target and non-target exemplar observer was trained in the response defini-
were rotated. The sequence of presentation and tions, and observers were calibrated to a crite-
combination of stimuli varied to control for or- rion of 100%.
der effects (See Table 3). For each student, four
sets of three stimuli (Sets 1, 2, 3, and 4) were Design and Procedures
selected. Set selection was counterbalanced to
control for potential difficulty biases. Within The design was a single-case, multiple-probe
each set the three stimuli were provided with a design across students. This procedure was
contrived nonsense monosyllabic or bi-syllabic selected as it provides a procedure to control
vowel-consonant combination for each item. for maturation and instructional history for non-

reversible responses (Greer, Yuan, et al., 2005; experiment was the absence of the naming rep-
Horner & Baer, 1978). ertoire for three-dimensional stimuli. Using
Experimental sequence. The experimental three sets of everyday objects (other than
sequence is presented in abbreviated format stimuli used in the experiment and with which
first, followed by a detailed description of each the students had prior contact) the experimenter
step: presented targets and non-targets in an array
of two on a child-sized work-table in front of
1. A pre-experimental screening test was con- each individual prospective participant. For all
ducted to determine whether or not students probe trial sets, there were four stimuli per set,
had the naming repertoire for familiar three- and a total of 20 trials were conducted for each
dimensional stimuli. Familiar three-dimen- response repertoire (each stimulus was pre-
sional stimuli were selected because the sented five times).
students did not have, at the time of this The experimenter said the name of the ob-
study, the ability to attend to two-dimen- ject, and a correct response consisted of the
sional representational stimuli. student pointing to the target item when the
2. We then probed the students to determine child had the target and a non-target stimulus
whether or not they could match the stimuli on the table in front of them. The purpose of
used in the experiment when asked to this segment of the screening was to determine
“match.” At this point we did not tact the whether the prospective participant had the
term for the stimulus. This tested whether requisite topographical response of pointing or
or not the child could match the visual char- touching a stimulus when named. Data were
acteristic of the stimuli alone. collected on all correct and incorrect responses.
3. Next we probed for the point responses If the prospective participant responded with
when the experimenter asked the student greater than 15% accuracy across all sets of
to “Point to ____ (e.g., dipoy).” stimuli, the student was determined to be in-
4. Next we probed for the impure tact re- eligible for inclusion. An accuracy of less than
sponses (i.e., “What is this?”). 15% was arbitrarily identified since, in an ar-
5. We then taught the pure tact responses to ray of two, a 50% chance probability for a cor-
mastery for the stimuli. rect response existed. Therefore, it was con-
6. Subsequently we probed for the listener and sidered that if the student had less than 15%
the impure tact responses for the same correct responding and also satisfied the crite-
stimuli mastered in Step 5. rion for speaker responses, then the participant
7. When Step 6 showed that naming was not did not have the naming repertoire. We did not
present as result of learning the tact re- provide reinforcement for correct responses or
sponse, a teaching set of stimuli was taught corrections for incorrect responses during this
in a multiple exemplar instructional se- phase.
quence by rotating the different responses Match-to-sample trials without the tact said
to each stimulus until all of the responses by the experimenter were conducted as part of
were mastered for that particular set. That the pre-experimental screening to ensure that
is they learned to respond to the listener the students could visually match the con-
and speaker responses with match re- structed novel stimuli included in this study,
sponses while hearing the tact for the stimu- and also because the match instruction was to
lus. be integrated with the tact as a part of the ante-
8. Following mastery of stimuli in Step 7, we cedent presentation during MEI learn unit in-
again probed naming for the first set. struction to occasion the opportunity for nam-
9. If naming was not present, a second or third ing to emerge. This probe was done to ensure
set was introduced and all steps were re- that the children could match based on the vi-
peated, beginning with Step 2. sual properties of the stimuli alone. The proto-
10. Instruction was terminated when the stu- col was as follows: An array of two stimuli (one
dent achieved naming following tact-only positive exemplar and one negative exemplar
instruction (Step 6) for that training set. with the order rotated between trials) was
placed on the table in front of the student while
Pre-experimental screening. Step 1: One of the experimenter handed the sample to the stu-
the eligibility requirements for inclusion in this dent, and said, “Match.” The student matched

without hearing the name of the object during For these pre-MEI probe trials, in order to
these probe sessions. A correct response con- control for the potential of an echoic response
sisted of the student placing the sample on top following trials, a minimum of three learn unit
of the correct target within three seconds of presentations separated stimuli. For example,
the antecedent. the experimenter said, “Match” (i.e., dipoy),
All students were also assessed for speaker “What is this?” (i.e., bikmo), “Point to _____”
responses (i.e., pure tacts) for all stimuli across (i.e., mooga), “What is this?” (i.e., dipoy),
the three sets of stimuli used for the remainder “Point to _____” (i.e., bikmo), and “Match”
of the experiment. The experimenter held up (i.e., mooga). Thus, presentations for the same
the object and waited for the student to tact the stimuli were arranged such that the student
object. Data were collected on all correct and could not echo a correct response as a function
incorrect responses. If the prospective partici- of proximate stimulus presentations. When the
pant responded with greater than 15% accu- students demonstrated that naming did not
racy across all sets of stimuli, the student was emerge as a result of hearing the tact for the
eliminated as a participant. We did not provide stimulus as a function of the matching trials,
reinforcement for correct responses or correc- we began the tact training.
tions for incorrect responses during this phase. Pure tact instruction. Step 5: Following set
A total of ten students received the pre-experi- assignments for each participant (see Table 3
mental screening, and four were selected based for the counterbalanced scheme), the tact train-
on their performance. ing condition was implemented. This consisted
Pre- and post-treatment probes. Steps 2, 3, of using learn units to teach tact responses to
4 and 6: Probes for naming were conducted the stimuli. Learn units are defined by the em-
for each set of 3 stimuli for each of 4 sets (see pirical literature and include at least two inter-
Table 2). Data were collected in sets of 18 re- locking three-term contingencies for the experi-
sponses (6 exposures for each stimulus) for menter and the potential operant for the stu-
each dependent variable (i.e., 18 match trials dent (Greer, 1992; Greer & McDonough, 1999;
when the experimenter said the tact for the Greer, 2002). In the naming literature typically
stimulus, 18 point trials, and 18 impure tact developing children have been shown to dem-
trials). The probe trials were dispersed across onstrate full naming after learning the tact re-
the three responses (i.e., match, point, impure sponse alone (Lowe, Horne, Harris, & Randle,
tact) and the stimulus for each response was 2002).
rotated to avoid sequence effects. That is, a Learn units consisted of the experimenter
match response for one stimulus was followed first gaining student attention as the anteced-
by a tact response for a different stimulus, then ent for the experimenter, the experimenter then
a point response for the third stimulus in the presented the multiple antecedent to the stu-
set, and so on, in a counterbalanced rotation. dent (i.e., “Match ___” as the visual stimulus
This was done to avoid the child obtaining a was presented to the child), the student re-
correct response simply by repeating the re- sponse, as the antecedent for the experimenter
sponse for the previous trial. During all probe consequence (i.e., reinforcements for correct
trials, correct responses were not reinforced and responses or corrections for incorrect re-
no corrections were provided for incorrect re- sponses), was followed by the experimenter
sponses. In the absence of reinforcement for consequence. The correction procedure re-
correct responses and in order to control the quired the student to say the correct response
setting events for instructional responses, the after hearing the experimenter say it (i.e., the
children were reinforced for appropriate behav- correct response), and no reinforcements were
iors previously mastered. That is, we inter- provided for corrected responses. The experi-
spersed opportunities for the children to re- menter evoked the corrected response while the
spond to known items (i.e., pointing to body student viewed the stimulus as part of the cor-
parts, following a simple command) and rein- rection. If the student did not immediately echo
forced accurate responses to maintain motiva- the correct response, up to three echoic oppor-
tion. These included but were not limited to tunities were conducted, consistent with the
sitting appropriately in the chair and making learn unit protocol. No reinforcement followed
eye contact in response to experimenter re- the corrected response consistent with the learn
quests. unit protocol.

Instruction was implemented as follows: This protocol continued until the student first
After gaining student attention, as the anteced- met the criterion for echoic responses, and then
ent for the experimenter, the experimenter then the student was required to meet criterion for
presented one of three stimuli within the vi- independent tacts. Correct responses were re-
sual field of the student. The reinforced re- inforced and incorrect responses were provided
sponse consisted of the student emitting the with a correction procedure consisting of the
correct tact for the stimulus. Each student was experimenter evoking the correct response
provided with an echoic prompt when neces- from the student. The criterion-level was 17/
sary until he/she met criterion for independent 18 or better across two consecutive sessions
responses for all three stimuli in the particular for learn unit instruction.
set. The stimuli were presented in a counter- Post-tact instruction probes. Step 6: Once
balanced order within an 18 learn unit session the student met criterion for the tacts for the
(3 stimuli presented 6 times in an 18 learn unit stimulus set, he/she was probed for listener and
session). Depending upon the learning rate of impure tact responses. Criterion for the pro-
the individual student, the echoic prompts were ductive listener responses was set at 17/18 or
systematically faded until the student indepen- 18/18 for probe trials as a measure of the pres-
dently tacted all three stimuli within the set. In ence or absence of naming. In the listener
other words, the students had echoic-to-tact probes, the student was again presented with
learn units that were reinforced. Following the the target stimuli with negative exemplars and
echoic criterion, sessions followed in which instructed to “Point to _____ (dipoy).” Each
only accurate independent tact responses to of the three stimuli was presented six times in
learn units were reinforced and incorrect re- the counterbalanced probe sessions consisting
sponses resulted in the correction procedure of 18 probe trials.
previously identified. An example of this pro- When the student demonstrated that he did
tocol follows: not meet criterion for the naming responses
(i.e., the untaught listener and speaker re-
A. The experimenter presented the first stimu- sponses for target training set did not emerge),
lus and provided the vocal tact, “Dipoy.” we introduced the MEI protocol using a train-
The student echoed the experimenter’s vo- ing set (Greer, Stolfi, et al., 2005; Greer, Yuan,
cal tact and that echoic response was rein- et al., 2005; Nuzzolo-Gomez & Greer, 2004).
forced. If the student emitted a vocal re- This phase demonstrated that the child did not
sponse other than the targeted one, or if no have naming following mastery of the tact re-
response was emitted, the experimenter sponses.
provided the student with the correct model Multiple exemplar instruction. Step 7: Mul-
using the correction procedure. tiple exemplar instruction consisted of learn
B. The experimenter presented the second units alternated between match, point, and pure
stimulus and provided the vocal tact, tact for a set of stimuli. For each set of three
“Bikmo.” The student echoed the stimuli, learn unit presentations were rotated
experimenter’s vocal tact and that echoic across these three responses for a total of 18
response was reinforced. If the student learn units per response category (i.e., 18 for
emitted a vocal response other than the tar- match, 18 for point, and 18 for pure tact). How-
geted one, or if no response was emitted, ever, while the stimuli were rotated across re-
the experimenter provided the student with sponses in the actual teaching sessions, the re-
the correct model using the correction pro- sponses were blocked for mastery measures by
cedure. response types (See Figure 2). For each stu-
C. The experimenter presented the third stimu- dent, counterbalanced sets of stimuli were se-
lus and provided the vocal tact, “Mooga.” lected to control for order and practice effects.
The student echoed the experimenter’s vo- An example of an instructional sequence was
cal tact and that echoic response was rein- as follows: 1) teach the student to match dipoy;
forced. If the student emitted a vocal re- 2) teach the student to point to mooga; and 3)
sponse other than the targeted one, or if no teach the student to tact golub. The student had
response was emitted, the experimenter rotated learn unit presentations for the match,
provided the student with the correct model the point, and the tact repertoires. Data for each
using the correction procedure. function were recorded separately according

to repertoire, and three columns were set up sponses (i.e., probes and learn units) for each
for data collection purposes prior to the ses- student. The percentage of agreement was cal-
sion. Eighteen learn units for each repertoire culated in 55% of sessions for Student B, with
constituted a teaching session, and the re- a mean agreement of 99%; 35% of sessions
sponses were blocked according to response for Student X, with a mean agreement of 99%;
types (i.e., match, point, and tact). Mastery cri- 46% of sessions for Student N, with a mean
terion for each match, point and tact repertoire agreement of 98%; and 46% of sessions for
was set at 17/18 or better across two consecu- Student L, with a mean agreement of 100%.
tive sessions, or 100% for one session. It is
important to note that when a student met mas- RESULTS
tery criterion for a particular repertoire, pre-
sentations for the mastered repertoire contin- During pre-experimental probes, three out
ued to be rotated with non-mastered repertoires of four students who participated in this ex-
as part of the antecedent condition, consistent periment emitted match to sample responses
with the MEI protocol cited in Greer, Stolfi, et (requiring visual to visual discrimination only,
al. (2005). This was done in order to occasion without the experimenter saying the tact for the
the joint stimulus feature of the multiple ex- stimulus) across generalized stimuli (an array
emplar instruction across the three repertoires. of common environmental stimuli other than
Post MEI probes. Step 8: Once the student those in Table 2) with 100% accuracy, which
met criterion for the tacts for the stimulus set, satisfied the first component of the naming rep-
listener and impure tact responses were probed. ertoire. Therefore, these three students could
Criterion for the listener and speaker responses match common stimuli based on visual prop-
was set at 17/18 or 18/18 for probe trials as a erties alone even with the lack of an instruc-
measure of the presence or absence of nam- tional history with these specific target stimuli.
ing. One student (Student L) did not have the match
Step 9: When the student met the criterion response for common identical three-dimen-
for naming in Step 8, the student was taught to sional objects and was taught that repertoire
independently tact the three stimuli in a differ- prior to initiation of experimental conditions
ent set (repeat all steps from Step 2). When the using stimuli other than those used for this
student met mastery criterion for the pure tact study. It is essential to note that, since the
for the alternate set, the student was then probed stimuli were not tacted by the experimenter
for the listener and impure tact responses. during pre-experimental match probe trials,
Step 10: When the student met criterion for there did not exist an incidental condition that
the untaught listener and impure tact responses could occasion the children to learn new words
following speaker (tact) only instruction, we if, in fact, they did have the naming higher or-
decided that the student had demonstrated joint der operant.
stimulus control across response categories and None of this group of participants could point
instruction was terminated. to the target stimuli when requested during the
first set of probes with more than 11% accu-
Interobserver Agreement racy level (below the level of chance respond-
ing), nor could any of them tact the stimuli.
Independent observers were present for be- When the student was requested to point to the
tween 35% and 55% of the probe and learn stimuli in an array of two, the probability of
unit sessions. Indices of interobserver agree- chance responding at 50% correct responses
ment (IOA) were taken on the numbers of cor- existed. Results of post match probes are in-
rect responses emitted during probe and learn cluded in the graphs for purposes of establish-
unit sessions. A second observer independently ing that none of the children had responses to
recorded the students’ responses following a the independent variables prior to implemen-
calibration training period. IOA was calculated tation of experimental conditions.
for each session by dividing the total number Following the instruction that led to the mas-
of point-to-point agreements by the total num- tery of the tact for one set of stimuli (counter-
ber of agreements plus disagreements and balanced across students to control for diffi-
multiplying by 100%. Interobserver agreement culty effects), none of the four participants ac-
data were collected across all categories of re- quired the naming repertoire at the preset cri-

Figure 1. Delayed multiple baseline for naming probes for all students.

terion as a function of learning the tact response results did not satisfy the preset criterion for
alone. Figure 1 shows that following tact learn the naming repertoire. Thus, while the child
unit instruction for two sets of three stimuli per had the speaker response he did not have the
set, neither Student B nor Student X subse- listener response. We then introduced the MEI
quently acquired naming for either set. In ref- protocol, and the student met criterion for nam-
erence to Figure 1, for Student B (see first row ing (i.e., he acquired the listener half of nam-
of data) post-tact instruction probes for the first ing). Again, for Student B, post-tact instruc-
set (Set 1) were 6/18 correct responses for the tion probes for the second set of stimuli (Set 2)
point response, and 16 out of 18 correct re- were 12/18 correct responses for point, and 11/
sponses for the impure tact response. These 18 correct responses for the impure tact. These

Table 4 sponses for the point response, and 14/18 cor-

A comparison of the number of learn units rect responses for the impure tact response.
required for tact and MEI. These results did not satisfy the preset crite-
rion for the naming repertoire. We then intro-
Participant/Sets Tact MEI duced another MEI set, and the student met
criterion for naming. Student X was taught a
Student B third and novel set of stimuli (Set 3) for the
1st Set 216 none tact response only. The post-tact instruction
2nd Set 180 126 probes for this set were 16/18 correct responses
3rd Set 216 none for the point response, and 18/18 correct re-
sponses for the impure tact. These results were
Student X very close to criterion-level performance but
1st Set 216 54 did not absolutely satisfy the requirements. We
2nd Set 144 72 went on to present MEI instruction for a third
3rd Set 108 36 teaching set for this student and the student met
Student N Figure 1 also shows that Student L (see third
1st Set 288 54 row of data) did not learn the naming reper-
2nd Set 72 none toire for two sets of stimuli following the tact
only instruction. Following tact instruction for
Student L the first set of stimuli (Set 3), Student L was
1st Set 162 108 probed and had 7/18 correct point responses
2nd Set 162 108 and 13/18 correct impure tact responses. These
3rd Set 126 none results did not satisfy the preset criterion for
the naming repertoire. We then taught another
Total Learn Units 1926 558 MEI set, and the student met criterion for nam-
Mean 175.0 79.7 ing for the first set. We introduced a second set
Range 72–288 36–126 (Set 1) and taught the tact to criterion for inde-
pendent responses. We probed for the untaught
point and impure tact responses, and the re-
results did not satisfy the preset criterion for sults showed 7/18 and 18/18, respectively. We
the naming repertoire. We then introduced MEI then introduced MEI again for another teach-
instruction, and the student met criteria for ing set, and the student met criterion for nam-
naming for both repertoires tested (i.e., point ing for this set. We went on to teach indepen-
and impure tact). After learning to tact a third dent tact responses for a third set of stimuli
set of stimuli (Set 3), Student B did show mas- (Set 4) and probed for the untaught point and
tery criterion for the untaught point and im- impure tact responses. Results for Student L
pure tact responses following tact only instruc- following post-tact probes showed that this stu-
tion. We decided that this satisfied the compo- dent had 17/18 correct point and 18/18 correct
nents for naming. impure tact responses. We decided that this stu-
Again in reference to Figure 1, for Student dent had the naming repertoire based on our
X (see second row of data), post-tact instruc- preset criteria.
tion probes for the first set of stimuli (Set 1) Following pre-experimental probes, it was
were 8/18 correct responses for the point re- determined that for Student N a monosyllabic
sponse, and 12/18 correct responses for the phoneme would be chosen rather than a bi-syl-
impure tact response. These results did not sat- labic phoneme because he was emitting only
isfy the preset criterion for the naming reper- partial vocal approximations to teacher ante-
toire. We then introduced MEI, but the student cedent echoic stimuli. An alternate set of re-
did not meet criterion for naming for both rep- sponses was identified for Set 3 (pakot, dipoy,
ertoires tested (i.e., point and impure tact). We and galoe), and the corresponding monosyl-
introduced a second set of stimuli (Set 4) and labic phonemes were ee, ay, and mm. In refer-
taught the tact response for all three stimuli in ence to Figure 1, Student N (see the fourth row
that set. Probes following tact instruction of data) acquired the naming repertoire follow-
showed that Student X had 13/18 correct re- ing one exposure to MEI. Following tact in-

Figure 2. Delayed multiple baseline for learn unit instruction for Students B and X.

struction for the first set of stimuli (Set 3), Stu- who acquired the naming repertoire following
dent N was probed and had 9/18 correct point only one exposure to MEI. It is important to
responses and 15/18 correct impure tact re- note that he was required to emit only partial
sponses. These results did not satisfy the lis- vocal approximations for all sets of stimuli.
tener criterion for the naming repertoire. We Two students required mastery of two sets with
then introduced MEI, and the student met cri- MEI, and one student required mastery of three
terion for naming for this first set. We intro- sets with MEI in order to establish the joint
duced a second set (Set 2) and taught the tact stimulus control across speaker to listener and
to criterion for independent responses. We listener to speaker repertoires.
probed for the untaught point and impure tact Figures 2 and 3 show the progression of in-
responses, and the results showed 17/18 and struction using learn units for the tact and MEI
18/18, respectively. These results satisfied our responses for all four students across elapsed
preset criterion for the naming repertoire. time. A comparison is shown (see Table 4) be-
Student N was the only participant of the four tween the number of learn units required for

Figure 3. Delayed multiple baseline for learn unit instruction for Students L and N.
criterion-level performance for tacts and mul- The mean for the tact function was 175.0 and
tiple exemplar instruction, and the total, mean, for the MEI function was 79.7. The range for
and range for both conditions. The total num- the tact function was 72–288 (4 to 16 training
ber of learn units required for criterion-level sessions), and the range for MEI was 36–126
performance for all students for learning the (2 to 7 training sessions).
tact function was 1,926; in contrast, the total Prior to MEI none of the four students had
number of learn units for the MEI was 558. the naming capability nor did they have an

echoic-to-tact repertoire. All four students re- for the presence of transformation of stimulus
quired more than one set of stimuli presented function across both listener to speaker and
using the MEI protocol in order to learn the speaker to listener repertoires for typically de-
naming repertoire. For Student N, the joint veloping 2-year-old children using three-di-
stimulus control across repertoires occurred mensional stimuli. Maturation and instructional
with less instruction than for the other three history were controlled for by using a combined
students. Thus, following mastery of one to multiple probe design within groups and an ex-
three sets of training stimuli in MEI sessions, perimental control group design. The experi-
the students acquired naming from learning the mental group, following pre-treatment probes,
tact function for the probe set, which they could was taught using MEI, while the control group
not do prior to the MEI experiences. These stu- was not initially exposed to MEI. The results
dents acquired the capability for naming from showed that MEI functioned to establish the
tact instruction as a direct result of exposure, transformation of stimulus function across re-
in some cases multiple exposures, to MEI. sponses required for naming to emerge for the
initial group, but this transformation did not
DISCUSSION emerge for the control group who did not re-
ceive MEI during this same time frame. Sub-
This experiment was conducted to test the sequently, the control group received MEI, and
conditions leading to transformation of stimu- naming emerged for them also. These children
lus function for the naming repertoire. In a pre- were not, however, tested for whether or not
vious study that tested the effects of multiple naming would emerge after tact instruction
exemplar instruction on the transformation of alone.
stimulus function (Greer, Stolfi, et al., 2005), Nuzzolo-Gomez and Greer (2004) showed
three students who did not have the listener to transformation of establishing operations as-
speaker component of the naming repertoire sociated with the verbal operant functions of
were taught to identify sets of pictures by teach- the mand and tact as a function of relevant MEI
ing the match response first. Following expo- conditions. In the first condition, four students
sure to MEI, all three students showed joint were taught either the mand or tact function
stimulus control from listener to speaker for for a set of forms and were then probed for the
untaught sets. untaught function, which they did not show.
Four significant distinctions between the They were then taught a training set of forms
Greer, Stolfi, et al. (2005) study and this cur- alternating between mand and tact establish-
rent study are that the children included in the ing operation functions using MEI until the
first study had more extensive verbal reper- students mastered both functions. Subsequently
toires than the children in the current study, they could emit untaught functions for the
none in the current study had any vocal tact mands or tacts in the original set and then dem-
repertoires or the echoic-to-tact capability. The onstrated transformation of establishing opera-
Greer, Stolfi, et al. study used pictures of real tion function for a novel set. The establishing
objects that were unfamiliar to their participants operations came to control an untaught func-
whereas in the present study all the stimuli were tion for forms learned in the alternate function.
contrived and the stimuli were three-dimen- It is important to distinguish between (a)
sional. The students in the Greer, Stolfi, et al. stimulus generalization across stimulus classes
study tested listener to speaker transformation and (b) the transfer of stimulus function to dif-
of function (i.e., learning the listener compo- ferent response repertoires. The former term,
nent led to the speaker component) for 2-di- stimulus generalization, identifies the spread
mensional stimuli, whereas in this study of the effect of reinforcement for a single re-
speaker to listener was tested (i.e., learning the sponse emitted in the presence of a stimulus
tact led to the listener component) for three- that maintains some property of the original
dimensional stimuli. Another point of distinc- stimulus presented under extinction conditions
tion between these two studies is that in this (Catania, 1998; Cuvo, 2003). Therefore, a tar-
study the participants were at least a year get stimulus may be considered to have gener-
younger than the participants in the Greer, alized stimulus control if the stimulus has prop-
Stolfi, et al. (2005) study. erties of the original condition but differs from
In a recent dissertation, Gilic (2005) tested the original stimulus (Cuvo, 2003). The dis-

tinction is made in the case of the experiment response came under the control of an untaught
presented herein in that the transformation of establishing operation condition.
stimulus function does not refer to a particular In the present study, the tact training condi-
stimulus or set of stimuli but rather to a reper- tion preceded the MEI condition and when the
toire of responses or response topographies that participants demonstrated they could not learn
were present in the post-treatment probe con- naming from tact instruction alone, they were
dition but not present in the training condition. exposed to the MEI intervention. After mas-
That is, the stimulus control was transformed tering two to three MEI training sets, the stu-
from control over a single topography to mul- dents demonstrated untaught listener response
tiple topographies. When the children mastered after tact instruction alone. In the Lowe, et al.
the tact repertoires, they still did not have the (2002) study, the typically developing children
listener repertoire until they had experiences emitted naming after learning the tact responses
with a subset of stimuli that occasioned the joint only. In the Horne, et al. (2004) study their par-
control across speaker and listener responding ticipants acquired the listener component but
with novel stimuli. The acquisition of this did not demonstrate the speaker component.
higher order operant, or possibly a relational In the present study, the children could not dem-
frame (Hayes, Barnes-Holmes, & Roche, onstrate naming as a result of learning tacts
2001), provides the students with the capabil- alone until after they had received two or more
ity to learn incidental responses that they could MEI training sets for three of the four students.
not learn incidentally prior to the multiple ex- This suggests that the MEI experiences pro-
emplar experiences. vided the means for the children to learn lis-
In order for the new repertoire to be identi- tener responses after learning speaker re-
fied as a relational frame, the new higher order sponses. In the Greer, Stolfi, et al. (2005) study,
operant would have to demonstrate the sub- the children acquired the listener to speaker
components of a frame or mutual entailment, component after MEI training. These studies
combinatorial entailment, and derived relations and the Gilic (2005) study suggest that either
(Hayes, et al., 2001). However, since no tests listener to speaker or speaker to listener com-
were made of the presence or absence of these ponents may be missing and that MEI can lead
components of a frame, it cannot be determined to both. We suspect that one may not speak but
that the new higher order operant was a frame. attain a listener component of naming as Horne,
Whether or not it is a relational frame, the new et al. (2004) have demonstrated. We are cur-
capability may still be categorized as a higher rently working on this with children who have
order operant (Catania, 1998) and the partici- no speaker responses.
pants did acquire joint stimulus control across While this and prior studies show that inci-
speaker and listener response when a single dental learning of speaker and listener compo-
stimulus response relation was taught. Thus, nents of naming emerged for particular types
the stimulus control was transformed from con- of stimuli following intensive MEI, it does not
trol of a single response (the taught speaker necessarily follow that the students have nam-
response) to the listener response (the untaught ing for other types of stimuli. Future research
listener response). needs to test for this. It is possible that MEI
Within the category of higher-order verbal across different types of stimuli (i.e., two- and
operants it appears that there are several dis- three-dimensional stimuli, abstractions, or print
tinct types. Some of these include the transfor- control) may be necessary to evoke broad based
mation of stimulus function in which a single naming such that children can acquire a wide
stimulus gains control over more than one re- range of tacts or other verbal operants inciden-
sponse class as in the case of the present study tally. This remains to be investigated. But these
and in the cases of the other studies (Greer, results to date are promising and it is apparent
Stolfi, et al., 2005; Greer, Yuan, et al., 2005; that without a fluent naming repertoire, chil-
Lamarre & Holland, 1985; Tsiouri & Greer, dren cannot acquire verbal operants inciden-
2003; Twyman, 1996a & b). Nuzzolo-Gomez tally. This repertoire is critical if children with
(2004) showed transformation of establishing language delays are to be incidental verbal
operations associated with the verbal operant learners (Greer & Ross, 2004).
functions of the mand and tact following rel- Anecdotally, it was observed by an indepen-
evant MEI conditions. In the latter case, a single dent data collector (a parent) that Student B

had generalized within the class of stimuli sub- al. (2002) study, acquiring the tact did not lead
sequent to experimental conditions. While to naming until the MEI experiences. In this
watching his father work on the family auto- study, results showed that the independent
mobile in the driveway, this student picked up speaker and listener repertoires came under
a hardware item similar in color and shape but joint stimulus control following MEI experi-
dissimilar in size to the stimulus used in this ences with subsets of stimuli, and these stu-
experiment, and the student shouted the cor- dents emitted naming responses to novel
rect contrived tact topography used in the ex- stimuli. MEI was an effective treatment for
periment. This child had learned transforma- establishing the naming capability for students
tion of stimulus function across repertoires (i.e., who, previous to experimental conditions, did
speaker to listener and vice versa) and across not have this capability. Future related research
classes of stimuli (generalization), neither of should also be extended to test the effects of
which was present before his participation in MEI on students who have no speaker reper-
this study. toires to determine if the listener component
Prior to onset of experimental conditions, of naming alone can be induced with relevant
Student L required a mean of 400 instructional MEI experiences.
opportunities to meet one instructional objec-
tive across instruction in all of his curricula REFERENCES
programs taught in home instruction. However,
subsequent to experimental conditions, fewer Becker, B. J. (1989). The effect of mands and
numbers of instructional opportunities (a mean tacts on conversational units and other ver-
of 138) were required to meet one objective, bal operants. (Doctoral dissertation, 1989,
with the greatest improvement in the area of Columbia University). Abstract from UMI
listener repertoires. It is also important to note Proquest Digital Dissertations [on-line].
that, prior to his participation, this student did Dissertations Abstracts Item: AAT 8913097.
not reliably point to target stimuli on the table. Catania, C. A. (1998). Learning. Upper Saddle
Subsequent to termination of this study, he re- River, NJ: Prentice-Hall.
liably pointed across all of his instructional Chu, H. C. (1998). A comparison of verbal be-
programs. Student L showed an educationally havior and social skills approaches for de-
significant improvement in acquisition rates velopment of social interaction skills and
across all programs within the listener reper- concurrent reduction of aberrant behaviors
toire. In addition, this student responded to of children with developmental disabilities
teacher presented vocal echoic prompts dur- in the context of the matching theory. (Doc-
ing the first phase of tact instruction for the toral dissertation, 1998, Columbia Univer-
third set of stimuli with clear and consistent sity). Abstract from UMI Proquest Digital
bi-syllabic phonemes, which he had been un- Dissertations [on-line]. Dissertations Ab-
able to do for the first two stimuli sets. stracts Item: AAT 9838900.
Two significant contributions of this study Cuvo, A. J. (2003). On stimulus generalization
were the identification of a protocol for induc- and stimulus classes. Journal of Behavioral
ing the bidirectional listener to speaker and Education, 12, 77–83.
speaker to listener repertoires that constitute Gilic, L. (2005). The development of naming
the higher-order operant of naming and the in two-year old children. Unpublished doc-
induction of naming as a function of instruc- toral dissertation, Columbia University, NY.
tional history. These findings replicate and ex- Greer, R. D. (1992). Teaching Operations for
tend those of the prior study on the induction Verbal Behavior. Unpublished manuscript.
of naming as a function of MEI history. In the Greer, R. D. (2002). Designing Teaching Strat-
present study, the students were younger and egies: An Applied Behavior Analysis Sys-
had significantly more deficits, suggesting that tems Approach. San Diego, CA: Academic
naming may be induced even with children sig- Press.
nificantly more impaired than those in the ini- Greer, R. D., & Keohane, D. D. (2005). The
tial study. Also, in the present study, even evolution of verbal development in young
though the students mastered the tact, mastery children. Behavior Development Bulletin, 1,
of the tact did not result in naming. Unlike the 31–48.
typically developing children in the Lowe, et Greer, R. D., & McDonough, S. H. (1999). Is

the learn unit a functional measure of peda- Lowe, C. F., Horne, P. J., Harris, F. D. A., &
gogy? The Behavior Analyst, 22, 5–16. Randle, V. R. L. (2002). Naming and cat-
Greer, R. D., Nirgudkar, A., & Park, H. (2003). egorization in young children: Vocal tact
The effect of multiple exemplar instruction training. Journal of the Experimental Analy-
on the transformation of mand and tact sis of Behavior, 78, 527–549.
functions. Paper presented at the Interna- Marianno-Lapidus, S. (2005). (Doctoral dis-
tional Conference of the Association for Be- sertation, 2005, Columbia University). Ab-
havior Analysis, San Francisco, CA. stract from: UMI Proquest Digital Disser-
Greer, R. D., & Ross, D. E. (2004). Research tations [on-line]. Dissertations Abstracts
in the induction and expansion of complex Item.
verbal behavior. Journal of Early Intensive Michael, J. (1982). Skinner’s elementary ver-
Behavioral Intervention, 1 (2), 141–165. bal relations: Some new categories. The
http://www.jeib Analysis of Verbal Behavior, 1, 1–3.
Greer, R. D., Stolfi, L., Chavez-Brown, M., & Nuzzolo-Gomez, R., & Greer, R. D. (2004).
Rivera-Valdez, C. (2005). The emergence Emergence of untaught mand or tacts with
of the listener to speaker component of nam- novel adjective-object pairs as a function
ing in children as a function of multiple ex- of instructional history. The Analysis of Ver-
emplar instruction. The Analysis of Verbal bal Behavior, 24, 30–47.
Behavior, 21, 123–134. Paniagua, F. A., & Baer, D. M. (1982). The
Greer, R. D., Yuan, L., & Gautreaux, G. (2005). analysis of correspondence training as a
Novel dictation and intraverbal responses chain reinforceable at any point. Child De-
as a function of a multiple exemplar instruc- velopment, 53, 786–798.
tional histories. The Analysis of Verbal Be- Pinker, S. (1999). Words and rules. New York:
havior, 21, 99–116. Perennial.
Hayes, S. C., Barnes-Holmes, D., & Roche, B. Ross, D. E. & Greer, R. D. (2003). General-
(2001). Relational frame theory: A post- ized imitation and the mand: Inducing first
Skinnerian account of human language and instances of speech in young children with
cognition. New York: Kluwer/Academic autism. Research in Developmental Dis-
Plenum. abilities, 24, 58–74.
Horne, P. J. & Lowe, C. F. (1996). On the ori- Skinner, B. F. (1957). Verbal Behavior. Acton,
gins of naming and other symbolic behav- MA: Prentice-Hall.
ior. Journal of the Experimental Analysis of Speckman, J. (2005). (Doctoral dissertation,
Behavior, 65, 185–241. 2005, Columbia University). Abstract from:
Horne, P. J., Lowe, C. F., & Randle, V. R. L. UMI Proquest Digital Dissertations [on-
(2004). Naming and categorization in young line]. Dissertations Abstracts Item: AAT
children: II. Listener behavior training. 3159757.
Journal of the Experimental Analysis of Sundberg, M. L., Michael, J., Partington, J. W.,
Behavior, 81, 267–288. & Sundberg, C. A. (1996). The role of au-
Horner, R. D., & Baer, D. M. (1978). Multiple- tomatic reinforcement in early language
probe technique: A variation on the multiple acquisition. The Analysis of Verbal Behav-
baseline. Journal of Applied Behavior ior, 13, 21–37.
Analysis, 11, 189–196. Tsiouri, I., & Greer, R. D. (2003). Inducing vo-
Karmali, I., Greer, R. D., Nuzzolo-Gomez, R., cal verbal behavior through rapid motor
Ross, D.E., & Rivera-Valdes, C. (2005). imitation responding in children with severe
Reducing palilalia by presenting tact cor- language delays. Journal of Behavioral
rections to young children with autism. The Education, 12 (3), 185–206.
Analysis of Verbal Behavior, 21, 145–154. Twyman, J. S. (1996a). An analysis of func-
Lamarre, J. & Holland, J. G. (1985). The func- tional independence within and between
tional independence of mands and tacts. secondary verbal operants. (Doctoral dis-
Journal of the Experimental Analysis of sertation, 1996, Columbia University). Ab-
Behavior, 43, 5–19. stract from: UMI Proquest Digital Disser-
Lodhi, S., & Greer, R. D. (1989). The speaker tations [on-line]. Dissertations Abstracts
as listener. Journal of the Experimental Item: AAT9631793.
Analysis of Behavior, 51, 353–359. Twyman, J. S. (1996b). The functional inde-

pendence of impure mands and tacts of ab-

stract stimulus properties. The Analysis of
Verbal Behavior, 13, 1–19.
Williams, G., & Greer, R. D. (1993). A com-
parison of verbal-behavior and linguistic-
communication curricula for training devel-
opmentally delayed adolescents to acquire
and maintain vocal speech. Behaviorology,
1, 31–46.
Yoon, S. Y. (1998). Effects of an adult’s vocal
sound paired with a reinforcing event on the
subsequent acquisition of mand functions.
(Doctoral dissertation, 1998, Columbia
University). Abstract from: UMI Proquest
Digital Dissertations [on-line]. Dissertations
Abstracts Item: AAT9839031.