VOL.9 NO.

I - SPRING 2000 Page 20

Exploring Stimulus Equivalence Formation in Infants

Martha Pelaez
Jacob L. Gewirtz
Aida Sanchez
Nadia M. Mahabir
Florida International University
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

A systmat;c replication of Ihe Devany. Hayes and Nelson The Relation Between Thought and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(1986) and the Augustson and Dougher (1991) studies was Language
conducted wilh infanls to explore the rela/ion of stimulus
equivalence to language developmenl. 9 normal inJanis. age
The debate between Piagel and VygOlSky over Ihe relalionship
21 10 15 monllu participated in 5 10 15 experimental SesS/OtIS,
each session on a different day. Infants expressIve language between language and thought raised many questions about
sleills were assessed (REEL Scale 2, 1991) before lubm/lllng their developmental sequence in humans. Piaget
each o/Ihem /0 a learning lasle consisting of malching deemphasized language by subordinating it to thought. He
different animal-liufigures in Q maJching-llNample (visual­ saw language primarily as a vehicle for expressing thoughts,
visual) conditional discriminalion formal. In a Single-subject nOI as a precursorlo Ihough!. On Ihe olher hand, VygolSky
design, "yants were taught follr conditional discrmlination.r: (1 962) argued that in the sensorimotor and early
ifA. then B; ifA. then C; ifD. then E; and ifD. then F. The preoperational stages, thought and language develop indepen­
order 0/ presentation and the lefl-right position 0/correct dently. His position was that thought is prelinguistic and
response were counterbalanced across training and tesling language is preintellectual. If educators knew the develop­
trials. Once an injant learned these mixed relations under mental point when symbolic representation and stimulus
various reinjorcers, the /ransMvity test was given Equiva­ equivalence is possible, they would be able to teach more
lence was established when a child matched Band C. in as effectively. We assume that if infants are in fact capable of
much as A had been the matching sample jor boIh. and when a thOUght and of fonning equivalence classes before language,
child matched E and F. boIh of which earlier had been poired we would not have to wait until their language has developed
to D. Every subject altained criterioll on the four independent to start teaching them concepts, numbers, etc. Moreover,
conditional discriminalions and on the mixed training j with definitive answers to how we develop thought and
subjects who altained transitivity (at 80% or abo\'e), per­ language and other human capabilities at an early age, we
jormed below chance level on at least one oftmjour symme.­ could structure more reliable tests as predictors of how infant
try tests. 8 out oj our 10 subjeclJ JX!rformed between 80 and conceptual progress would occur.
100 % correct responses in the tramitnlily tesu. We found a
significant negative correlation between Ihe 100ai number of Tbe Development of Language
trials to criterion during Ihe conditional diSCrimination
training and the combined receptive and expres.srve language Most nonnally developing infants can discriminate between
quotient. Those infants with higher language-skill scores the distinctive features of closely similar phonetic elements in
required fewer trials 10 comp/ete the condit/ofJal-ducrimina­ speech, even at I month of age (Eimas, Siqueland, Einar,
lion training. The results suggest that language skills play a Jusczyk, & Vigorito, 1 971). In natural circumstances, inrants
role ill stimulus equivalence formation can recognize some symbolic word meanings as early as 6 to 8

monlhs after birth (Bwch & League, 1991). This recognilion

There are at least three reasons why the concept of is followed shonly by demonstrated language skills for
stimulus equivalence has captured the imagination of behavior discriminating the meanings of most simple sentences by 9 to
analysts. First, the fonnation of equivalence classes is one of 1 2 months after binh. Receptive recognition vocabulary
a range of cognitive phenomena, including those that denote often rises to over I ()() words by the first year of age, and the
concept fonnalion, categorization, and rule-governed behavior first intelligibly spoken words in expressive language nonnal­
that can be addressed and organized from a behavior-analytic Iy tends to appear in the infant's repenory between 9 and 14
perspective. Second, stimulus equivalence appears anom3- months. Important for our research, however, is that expres­
lous. unexpected, and emergent. And it is not immediately sive syntactic advances, typically, do not appear in the child's
apparent that the emergence of stimulus equivalences are repertoires until after 18 months. For us, the key issue is to
direct outcomes of operant learning. And, third, equivalence­ detennine at which early developmental points stimulus
class formation appealS to be related to language development equivalence-class foonation can occur. The assumption that
(see Devany, et aI., 1986). In some way, the relations seen language skills can cause or facilitate successful equivalence­
among stimuli in an equivalence class parallel the symbolic class fonnation needs further exploration.
relations commonly said by cognitive developmental psychol­
ogists to be characteristic of language.
 Page 21 VOL.9 NO. 1 - SP!UNG 2000

Objectives of the Present Study conditional discrimination training. Devany et al. concluded
that the failure of the language-deficient children to form
The main aim of this study is to explore how early in equivalence classes could not. be explained on the basis of an
human development equivalence relations may appear and to inability of those children to learn conditional discriminations
examine the relation between stimulus equivalence and per se: That is. because all the retarded/no language subjects
language competency. Previous studies have shown stimulus In thell' study did learn the 4 conditional discriminations, while
equivalence in nonnal children (Augustson et aI., 1992), only those with language skiJI were able to show transitivity.
language-able mentally-retartled children (Devany et aI., 1986, Devany et al. attributed the inability of the retarded non­
Saund.,. & Spradlin, 1993), and adults (Hayes, Thomas, & language children to fann equivalence class to their language
Hayes, 1989 ). An equivalence class is shown if the stimuli in deficit. However, theDevany et al. (1986) study was based
the class show the three defining relations of rencxivity, on the constraint sample of mentally-retarded infants that
posits some potential confounds explain below. Based on the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

symmetry, and transitivity (Sidman & Tailby 1982). Match­

Devany et al. study, one can not arrive at a definitive conclu­
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ing-t�sample procedures have been used to establish and test

for these relations. In this test, the subject is presented with a sion about equivalence-class fonnation in 1 - to 2-year-old
children and its relationship to differential language (or
sample visual stimulus, and given an array of three compari­
cognitive) skills.
son visual stimuli underneath the sample stimulus. They are
then asked to point to the comparison stimulus that goes with Oill'study was designed as a systematic replication of the
Devany et 81. (1986) and Auguston and Dougher (1992)
the onc above, the sample. That is the child will learn that in
studies. The same materials were used, mode of presentation,
the presence of the sample visual stirnuL� if they ptck a certain
comparison visual stimuli they will be reinforced. order of training the conditional relations and testing, as well
as same reinforcement schedule. One divergence was that we
Reflexivity is generalized identity matching, by matching a
used a more stringent criterion for training the conditional
novel stimulus to itself under conditions of no reinforcement.
discriminations than did Devany et al. Instead of9 correct
That is to say, if a child is shown a picture of a car, the
out of ten responses, we used9 consecutive COlT'eCt responses
sample, and then asked which one of these comparisons go
to index the learning of each conditional discrimination. The
with the above, and they have to choose between a bat, b'ee or
reader is cautioned to look carefully when viewing the
car. If the child picks the car, (comparison), 10 match the car,
information given in the figures of theDevany et a!. results.
(sample), then this shows an identity relationship ( "ifA, then
Their charts represent individual training and testing data for
A"). Symmetry refers to the functional reversibility of the
the 12 children (4 of whom were classified as "normal" 4 as
conditional relation; "ifA, then B, and if B, thenA". This
means that if the sample visual stimuli is a car, and the
"retarded-with-Ianguage," and 4 as "retarded-withou Ian· t'
guage"). Their classificatton, based on the Bayley Scale and
comparisons are a house, truck and swimming pool, the child
the Slant0rd-Binet standardized intelligence test score, led to
would be asked "Which one goes with the car?" If the child
conclUSions that: (I) the retarded/language skilled and the
picks the truck then they will receive reinforcement for that
normal children all required fewer trials to complete the
behavior. This is called visual-visual conditional discrimina­
disaimination training than did the retarded children with no
tion.
language skills; and (2) that all their language-able children
Some previous matching-to-sample tests have used
performed better than the language-deficient children on the
auditory stimuli and asked the subject to match visual stimuli
stimulus-equivalencc test However, when one examines the
with it; this is called auditory-visual conditional discrimina­
Devany et al. results showing the transitivity test scores of
tion. In a matching to sample lest, in order to show transitivi­
their 12 Ss, at least a question remains: Was transitivity indeed
ty at least 3 stimuli are required. If after the relations "ifA
then B" and "if B, then C" have been taught with conditiona i shown by the 8 language-skilled children? If we look at the
results in theIvsl block: of every language-able subject on this
discrimination trials, and then when tested for the relation "if
test, we see that none of these subjects attained greater than 70
A, then C" and the relation emerges without ever being
explicitly trained with reinforcement or paired together,
% c�)r'rect responses in that first block, with many subjects
sconng between the 50 and the 60 % level. One problem in
transitivity has been demonstrated.
interpreting theDevany et al. results is the way in which the
da� were calculated and plotted. In their graphs, each data
Problems with Previous Studies with Young
pomt represented the number of COITCct responses out of the
Children number of responses anempted, in a block of ten trials. This
way of plotting presents an inconsistency in the representation
From their findings, Devany, et al. concluded that language of those data that can be misleading. For example, in a
skills are related to the demonstration of stimulus equivalence. sample block of ten trials, there may have been only 4 trials
In their study, three groups of children were ttained and tested with responses anempted and6 trials with no attempts at all.
on equivalence relations: (I) normally developing preschool­ Out of the 4 responses attempted, 3 may have been correct,
ers, (2) children tenned "retarded" showing expressive speech resulting in a correct score: of 750/•. The eRQrnlOUS variability
or signs, and (3) children termed "retarded" showing lan­ of the denominators yielding these percentages may provide a
guage-deficiency. Devany et al. reported the acquisition of questionably-valid representation of performance during the
visual-visual conditional discrimination and transitivity in transitivity tests. To correct for this problem, in the present
every one of their language-able children (retarded and study denominator values were held constant, always with 10
nonnal), whereas none of the language-deficient "retarded" responses attempled. In general, subjects improved their
children exhibited transitivity after having succeeded in the perfo:mance from the first to the last transitivity test block of a
 VOL. 9 NO. I - SPRING 2000 Page 22

series. Even so, the answer as to whether or not infants relation. We expected that this addition in the design would
subjetted to this control procedure were able to fonn a new penn it us to assess for symmetry routinely after each subject
relation without implicit or explicit training remains unan­ has attained the response aiterion in a conditional-discrimina­
swered. tion training task. Thus, one purpose in our study was to test
Figure I shows results from our fllSl subject (N1) when for symmetry, not to train it (see Figure I).
we tried to conduct a direct replication of the Devany el al. Devany et aL used mentally-retarded subjects as a means
procedure. Notice that using their procedure, we were only of dctennining if language skills are prerequisite for stimulus
able successfully to train the 2 conditional relations (A:B, equivalence. We believe it is difficult to determine if behavior
D:E) independently and then mixed, in this 2S-mos.-old patterns denoting "mental retardation" are associated with the
language-skilled girl, with receptive and expressive language failure of the subjects to derive equivalence relations (even
scores nonnal for her age. However, our subject was unable when the groups are matched using mental age), or if the
to demonstrate the emergence of the new relation during the deficit in language skills per se could be responsible. Very
transitivity test We were about to run additional subjects 10 imponantly is that the data of Devany et a!. revealed a
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

verify this contradictory finding. when Augustson's and significant difference in the number of unattempted trials
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Dougher's systematic replication of me Devany e1 aJ. study between the retarded language and the retarded n�language
came to our attention (EAHB, 1992). By failing to replicate groups. The retarded language group averaged 5 unattempted
the Devany et 01. results, Augustsoo and Doogher corroborat­ responses per child, whereas the retarded no-language group
ed our concerns about the complexity of the Devany et a!. averaged 10. This could be another possibility for the failure
procedure. But Augustsoo and Doogher left us with additiooal of the retarded no language group to show the transitive
concerns about their own design. Their subjects were nn'eT relation. in addition, Devany et al. made only an infonnal
tested for transitivity because they could not attain criterion assessment of the language and speech skills of their subjects.
during the mixed-training phase. We believe that this oc­ A fannal methcx1 for assessing the language skills of infants
curred because Auguston and Dooghcr used tIu<e comparison was employed in our study.
stimuli instead of two. Several reasons might explain why the ALI of these aforementioned issues prompted us to modifY
subjects in these two studies failed to derive the new stimulus the proccdW"e from that used in the earlier studies and to look
relation: for alternative tactics in the study of equivalence relations in
First, there is the possibility that children 24 mos. and very young children. Apon from coocems with methodologi­
YOWlger may have difficulty in leaming (and/or remembering) cal problems, our main long-Ienn interest is to detect at which
more than two conditional discriminations at a time before developmental points infants are able to demonstrate the
being tested for stimulus equivalence. In the Devany et al emergent relations denoting symmetry and transitivity. We
design, al14 conditional discrimination relations (A:B, D:E, are interested in detennining if any panicular sequence in the
A:C, D:F) were trained in a complex sequence, before the conditional-discrimination training can facilitate the acquisi­
equivalence lest was presented. Augustson and Dougher also tion of equivalence relations. We are open to the possibility
trained the 2 conditional discrimination relations independent­ that stimulus-equivalcnce class fannation would be manifest­
ly, before they introduced the mixed task.. However, as in the ed in young children at early developmental points, even
aforementioned, their subjects were not able to reach criterion before they show expressive language skill.
on the mixed task. Thus, Augustson and Dougher neither
tramed the next two relations nor tested for the transitive METHOD
relation. Although Augustson and Dougher attempted to
replicate the Devany et al. study, their results are not compa­ Participants. 9 normal subjects, 7 males and 2 females
rable to those of Devany et al. because those researchers made aged 21 to 2S months, panicipated in this study. The mostly
the learning more difficult for the children by adding a third daily experimental sessions lasted between 20 and 30 minutes.
comparison stimulus. They used an array of3 comparison The number of trials presented was not fixed, but usually
stimuli (one correct and two incorrect) on each trial. It is involved at least 20 trials. The number of sessions ranged
possible that this methcx1ological change in the Augustan and between S and IS.
Dougher design increased the complexity of the mixed
training task beyond the skills of their 2 year-old subjects. Asstssmtnt Materials and Experimelltal Setting.
Augustson and Dougher, in the second phase of their study, lAnguage Skills and Equ"'a/I!nce Re/atiolls. Before the first
extended the mixed task and found that their subjects contin­ training session, each child's receptive and expressive lan­
ued to perform at chance level even after two hundred trials, guage skills were assessed through the administration of The
so they terminated the mixed training. But, even had the Bzoch-League Receptive-Expressive Emergent Language
training been extended beyond 200 trials, one wonders if any Scale (1991), for the Measurement of Language Skills in
of their child subjects would have succeeded in attaining Infancy (REEL Scale 2). For reliability purposes, this
criterion in the mixed training task, much less derived a new language-skill assessment was made twice, one time by the
equivalence relation? experimenter alone, and a second time using the mother as an
Is imponant to note that neither of those two studies informant (with the experimenter rcading each item to the
tested for symmetrical responding. The possibility of such mother). The average scores of the two informants were
symmetrical responding was only inferred. In the present used for analysis of the data. This language test includes a
study we break down the complexity of the training sequence I 32-item checklist and uses observational information to
by training only one conditional-discrimination at a time and identify the level of language skills in infants.
testing immediately for symmetry before training a new Two functional language systems, receptive and expres-
 Page 23 VOL.9 NO.1 - SPRING 2000

sive, are assessed directly_ To the test developers, receptive ringing of a bell, or the delivery of food (cereal, M&Ms).
language refers to the unified activity of all the sensory-neural Incorrect responses were not reinforced. Physical prompting
associations and auditory-perceptual processes that are (guiding the child's hand to the correct choice) and visual
involved in the decoding and understanding af the intended prompting (placing the experimenter's finger on the correct
meaning of oral languages (i.e., auditory comprehension). In choice) were used with some children at the beginning of
contrast, express;ve language refers to all of the underlying training. Initially, a continuous reinforcement schedule was
sensory-neural processes and also to the motor neural skills of used and was gradually thinned until a programmed conse­
the breathing, phonation, resonance, and articulation mecha­ quence was delivered only after every three or four correct
nisms of the body that are involved in communicating with responses. Reinforcement was not delivered for the target
others through the mediation of spoken symbolic languages. response during testing. Instead, the child was praised for
cooperation, good sitting, and the like two or three times
Setting. In the experimental room, two video cameras during a block of ten trials. The mixed training and the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

recorded all activities. The subject sat at a table facing the equivalence test were administered within the same session.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ex.perimenter. Six stimulus figures were used in the experi­

ment. The tasks consisted of matching animal-like figures Recordin g. Behavior in each trial was scored as "cor­
using a matching-ta-sample fonnat. Each stimulus figure was rect", "incorrect" or "no response". A "correct" response is
colored with one of six watercolor paints (red, brown, green, defined as touching the correct comparison stimulus while
purple, yellow, and orange). Color assignment was random, refraining from touching the incorrect comparison or the
except that all six colors had to be used in each stimuli set. sample stimulus. An "incorrect" response is defined as
touching the incorrect. comparison or touching both the correct
Procedure. A single-subject design was implemented. and incol'1'CCt comparisons. Any other behavior was scored as
The children were taught four conditional discriminations: if "no response."
A, then B; if A, then C; ifD, then E; and ifD, then F. This
order was counterbalanced for half of the subjects. During RESULTS
training, either the A orD stimulus was presented as a sample
with either B and E or C and F as comparisons. The left-right As seen in Figure I, the number of correct responses divided
order of presentation of the comparison stimuli was counter­ by ten (ten-trial block) was calculated to give a percentage of
balanced across trials. Each child was trained and tested using unprompted correct responses for each ten-trial block. The
a different stimulus set, made by randomly selecting from a first block of each training period may renect fewer than I 0
pool of items. (The stimuli used in the equivalence test were trials with responses. In symmetry and transitivity testing
identical to those used in the previous training phase, except periods, 01/ blocks are comprised of 10 trials, each involving a
that the sample stimuli were stimuli that previously had been response.
comparisons during the conditional discrimination training.) Figure I shows that all 9 subjects attained criterion (9
Equivalence was indicated by matching B and C, inasmuch as consecutive correct responses) on the four independent
A had been the matching sample for both, and by matching E conditional discriminations and on the mixed training.
and F, both of which earlier had been matched toD. Subjects required from 34 to 242 trials (mean=103) to learn
First, a child was taught to select B in the presence of A, the two relations (A:B, A:C,) and the mixed training. In
the sample stimulus (A:B). The criterion for tenninating contrast, theDevany e1 al. subjects required from 50 to 70
training is nine consecutive (unprompted) correct responses. trials, with a mean of 68, to reach their response criterion.
Once the child reached criterion on this relation, testing for the This difference in number of training trials to criterion may
symmetrical relation (B:A) was conducted. Symmetry tests have to do with the fact that our subjects were less develop­
consisted of a block of ten trials in which responses were mentally advanced, than the subjects of Devany et al. Our
emitted. (Trials in which the child did not respond were not children ranged from 21 to 25 months and theirs from 25 to 52
included in the ten trial blocks.) months. The average number of correct responses for all
After the symmetry test, the child was taught the second subjects in the symmetry tests was 6.5. Surprisingly, 5
conditional discrimination, which was to select C in the subjects who attained transitivity (at 80% or above), failed at
presence of A (A:C). When the child reached mastery least one of the four symmetry tests, that is, perfonned below
criterion on this task, testing for the symmetrical relation chance level. Eight out of our 9 young subjects perfonned
(C:A) was conducted. (Notice that at this point in the between 80 and 100 % correct responses in the transitivity
sequence the relationD:E was not trained asDevany et a!. tests.
and Augustson andDougher had; instead, A:C was trained). A significant negative correlation of -.84 (at better than p
Then, a mixed task followed. In the mixed training the < .01 alpha level) was found whcn the total number of trials to

stimulus cards from both sets of the two previously acquired criterion during the conditional discrimination training was
relations were mixed together and presented in a random related 10 the combined receptive and expressive language
order. Once the child reached criterion on this mixed task, the quotient. That is, those children with higher language-skills
transitivity equivalence test was presented. The entire scores required fewer trials to complete the conditional­
procedure was then repeated for theD:E,D:F relations. discrimination training.
At the stan of each trial, the experimenter pointed to the
sample stimulus and said, "See that one? Which one at the
bottom goes with it?" Correct responses during training were
reinforced with either praise, the blowing of bubbles, the
 VOL.9 NO. I - SPRING 2000 Page 24

j;;:'
,.�-- 101,11 __

- �

;;:.t::J I
ru.

�
I:
M
M
III I
I I I-I I I ,-
I-I e
•

, , t l " ,
•

, , , , I
• '.'
,
•
, , "
, , , ,
•
, ,
, ,
•
, ,
•
,
� �, .
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

� .. .. � ,

tv,� I
This document is copyrighted by the American Psychological Association or one of its allied publishers.

j IleiI
",,,_-
101,11.__

� Iff) � �
'I.I I I l I II -,I I
:;,

J
e l
'. '. ' " '
I I· . I
, , I
' ' I I I I I I I I I
I. I I I I I ,
•

I I I ,
. I
, I I I
, , , , I I I I
•
,
. ,
, ., ,
• , •
,
,
. ,
,

, ,�
,� J,u' 'N ..
'0.,'
_. N ,
.. -

-
.
. ,
..._
,.,- -
� ..
I!fJ ..

� �E1
ru, N' � �

fl
= = -
,
.....
' e'e' ' .' ,. ,

r
•

, , . , "
,
, ,
•
• , , .
, ,

"
' , , . : ,
.. ,� N ... '.-./ .. � ,-,

......- -

0I:J j ·:V: :v':

" ,"----

:;;. � I� i "
� ,­
: ·: � ' .. ' '. .
'el I
.: :I":.: : , ,
II II , , I I I I I I I , ,
II I I
I I II II I
I lei I , ,

I I I " ,
��,. L---�"--T-' T'----T'-T
It

• '-'�-------T-T--��-- ,
I I II , , ,

" , , , . ,
•
I ..I I 1 I
•

, �
.. ••
,- .. -,

�
.. "

�
..

/'
ru, = �

'/'
J:
.,
Figure 1.
, , I
'I"' I
I I ,
, , ,. I -, 1-
' ,
.
, ,
: '
.
,
,
. ,

,
" ,

, ,

.
.

BLOCKS OF TEN TRJALS

 Page 25 VOL.9 NO.I - SPRING 2000

naming and other symbolic behavior." Journal o/Experimen­

DISCUSSION tal Analysis 0/ Behavior, 299-301.
Saunders, K.J., Spradlin, J.E. (1993). Conditional
From this study. we concluded that children as young as 21 discrimination in mentally retarded subjects: Programming
months appear to be able to form equivalence relations. The acquisition and learning set. Journal ofExperimental Analysis
young children who participated in this study had some olBehavior. 60. 571-585.
receptive and limited expressive language skills. At this Sidman, M., Tailby,W. ( 1982). Conditional discrimina­
point,we do not know about the exact nature of the relation tion vs. rnatching-to-sample:An expansion of the testing
between stimulus equivalence and language skills. However, paradigm, Journal oj/heExperimental Analysis o/Behavior.
many have taken the Devany et al. results to show that the 37,5-22.
subjects used acquired transitivity and that this is related to Vygotsky, L. (1962). Thought and Language. Cam­
their language abilities. Our results suggest primarily that bridge, MA: MIT Press.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

stimulus equivalence can be acquired by infants via condi­

This document is copyrighted by the American Psychological Association or one of its allied publishers.

tional-discrimination training and suggest that the phenome­

non or transitivity may be related to language skills.
We are only beginning to understand the meanings of the
relationship between language and the ability to derive new
equivalence relations. We should not discard the possibility
that stimulus equivalence formation and language skills could
be concurrent systems as readily as that one system is the
"cause" of the other. Some researchers have assumed implicit­
ly or explicitly a direction of causality, such that verbal skills
are a precondition to the formation of equivalence relations
(Home & Lowe, 1996). Specifically, they have seemed to
suggest that certain language-skill level cause, facilitate,or
serve as a precondition for successful equivalence-class
formation, This assumption is not obvious to us yet. We
must be open to the possibility that the formation of equiva­
lence relations may precede the development of language,or
that equivalence relations and language may be alternative or
parallel outcomes of the same process (Pelaez, 1996),
Clarification of this issue is important given the clear rele­
vance of stimulus equivalence as a model for such other
cognitive phenomena as concept formation and categorization
that also occur during early childhood.

REFERENCES

Augustson, K. G., & Dougher,M. 1. (1992). Teaching

conditional discrimination to young children: Some method­
ological successes and failures. Experimenlal Analysis oj
Human Behavior Bulletin, 9,21-24,
Bzoch, K. R., & League, R. (1991). Receptive-Expres­
sive emergent language test: A method Jor assessing the
language skills ol in/ants. (2nd. edition). Pro-ed: Austin,
Texas.
Devany, J.M., Hayes, S.C., & Nelson, R.O. (1986).
Equivalence class formation in language-able and language­
disabled children. Journal o/ the Experimenlal Analysis 0/
Behavior. 46,243-257.
Eimas, P. D., Siqueland, E. R., Juszyck, P. W., &
Vigorito, J. (1971). Speech discrimination in the infant.
Science, /7/, 303-306.
Hayes, L., Thompson, S., Hayes, S. (1989). Stimulus
Equivalence and Rule Following, Journal 0/ Experimental
Ana/ysis olBehaviar, 52, 275-291.
Home, P. J., & Lowe,C. F. (1996). On the origins of
naming and other symbolic behavior, Journal 0/Experimental
Analysis ojBehavior, 65, /85-24/.
Pelaez-Nogueras,M. (1996). Thought without naming:
A commentary on Home and Lowe's article "The origins of