THESIS

This is to certify that the

dissertation entitled

Word Acquisition: The Effect of Perceptual
and Functional Stimulus Attributes

presented by

Maria Della Corte

has been accepted towards fulfillment
of the requirements for

PhD degree in Psychology

 

 

Major professor

 

Date 57// 3 /II3

MSU u an Affirmative Action/Equal Opportunity Institution 0— 12771

4T; ““1”
. 3.35“ 13

   

     
 

{‘ ' «4""7 ‘a
)_-qf‘0‘ 0"» ‘:¥J.':::¢

.i 1;. as" 5538‘ 145‘ l‘ir'-.Qix@

7-
, I. $3,
a;

n; “' 5". r
"i115 EL- 4

 

PJWR' .‘ffi.
. '- IJ- ,’5?~‘7?2?,",‘-7‘l‘
Ea fitswmusuu‘fir

  

 

WORD ACQUISITION: THE EFFECT OF PERCEPTUAL
AND FUNCTIONAL STIMULUS ATTRIBUTES

By

Maria Della Corte

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1983

ABSTRACT

WORD ACQUISITION: THE EFFECT OF PERCEPTUAL
AND FUNCTIONAL STIMULUS ATTRIBUTES

By

Maria Della Corte

The purpose of this study was twofold: l) to examine whether
children's early word meaning is based on perceptual features, as
stated in Clark's Semantic Feature Hypothesis, or on functional
characteristics, as stated in Nelson's Functional Core Hypothesis;

2) to clarify the equivocal results obtained in earlier investigations
of these theories. Children, whose mean age was twenty-six months,
were asked to play with one object from three categories of unfamiliar
objects. The objects within each category were similar to each other
either in shape or function. Children were asked to play with each
object twice. Based on their order of selection, the objects' salience
was assessed. Initial selections presumably based on shape provided a
measure of perceptual salience. Functional salience was assessed by
the order of selection after becoming acquainted with the object's
function. The two sets of rank orders were used to examine whether
the objects' perceptual and functional salience influenced name
learning and/or the basis upon which the names were generalized to
similar objects. After playing with the objects, the children were

asked to learn nonsense names for them. The children then were shown

the remaining objects in each category, and were asked whether the names
could be extended to any of these objects. Responses were categorized
according to object characteristics which children used as the basis of
generalization.

The results indicated that saliency was not an influencing variable
for name learning or for the basis of generalization. One category name
was learned better than the others and females learned all names better
than males. Moreover, behavioral indicators of anxiety revealed that
children who were less anxious learned the names better. Two sets of
objects were generalized more often on the basis of perceptual character-
istics for some children and on the basis of functional characteristics
for others. This investigation shows that name learning and generali-
zation are not independent of type of object. Although the results tend
to favor Clark's position, some aspects of Nelson's theory have not been

tested adequately.

To Joe

who makes it all
worthwhile

1'1

ACKNOWLEDGEMENTS

This work could not have been completed without the help of many
people.

I want to thank Ellen Strommen, my chairperson, who enabled me
to pursue my own ideas but at the same time was always ready to help.
Her support and friendship made a difficult task much more pleasant.
I also want to thank Michael Casby, Hi Fitzgerald, and Ray Frankmann
for their interest and helpful suggestions.

My observers, Jill, Joan, John, and Karen were dedicated and
competent workers.

I want to thank the Child Care Center in Spartan Village for their
participation and cooperation.

Finally, I want to thank the parents and children who participated

in this research. They actually made it fun.

TABLE OF CONTENTS

Page

LIST OF TABLES .......................... v
LIST OF FIGURES ......................... Vi
CHAPTER I: INTRODUCTION ..................... l
CHAPTER II: METHOD ....................... ll
Subjects and Experimenters ............... ll

Setting and Materials ................. l2

Anxiety Scale ................... l3

Procedure ....................... l4

Phase l ...................... 16

Phase 2 ...................... 17

Phase 3 ...................... l7

Mean Length of Utterance (MLU) .......... l8

CHAPTER III: RESULTS ...................... 20
Post-hoc Analyses ................... 28

Object Names ................... 28

Play Time ..................... 32

Anxiety ...................... 35

Sex Differences .................. 37

Birth Order .................... 38

Mean Length of Utterance (MLU) .......... 38

Summary ...................... 39

CHAPTER IV: DISCUSSION ..................... 40
APPENDIX A ............................ 48
REFERENCES ............................ 50

iv

Table

LIST OF TABLES
Page

Frequencies for the Order of Selection for Each Type
of Object During the First and Second Play Period ..... 22

Frequency of Responses in Each of the Categories of
Generalization to Similar Objects: Perceptual Similarity

(P), Functional Similarity (F), Verbal "No" Response (N),

Both Perceptual and Functional Similarity (P & F),

Neither Perceptual Nor Functional Similarity (NS),

No Response (NR) ..................... 24

Actual Frequencies and Chi Squares for Type of

Responses When Children Were Asked to Generalize to

Similar Objects in Relation to the Order in Which Each

Object Type Was Selected During the First and Second

Play Periods ....................... 27

Means and Standard Deviations for the Number of Correct
Names Learned for Each Type of Object for Males and
Females .......................... 28

Means and Standard Deviations for Length of Play Time
(in Seconds) During the First and Second Play Periods
for Males and Females for Each Type of Toy ........ 29

LIST OF FIGURES

Figure Page
1. Mean Number of Correct Name Responses for Each Object
for Males and Females .................. 3l
2. Mean Play Time (in Seconds) During First and Second
Play Periods for Each Type of Object for Males and
Females ......................... 33

vi

CHAPTER I
INTRODUCTION

This study was designed to investigate predictions from two
different theories of the development of word meaning in early language
acquisition. One theory is Clark's Semantic Feature Hypothesis which
states that perceptual characteristics are the basis of meaning. The
other theory is Nelson's Functional Core Hypothesis which states that
functional characteristics of objects are the basis of meaning. Several
studies have been conducted with the aim of obtaining evidence to
support either one of these positions. However, these earlier studies
have produced conflicting results. Methodological problems and differ-
ences in stimuli and procedures make comparisons and generalizations
across these studies difficult. The present investigation was an attempt
to determine the reasons for the conflicting results of several previous
studies as well as resolve some of the methodological problems.

In her Semantic Feature Hypothesis (1973), Clark assumes that the
meaning of words can be broken down into smaller components called
features. Clark contends that the first semantic features are likely
to be derived from perceptual input. When children first use a word,
its meaning is based on only one or two general features. Consequently,
when a child uses a word, the word will encompass a broader range of
referents than an adult's use would. For example, the meaning for

"dog" might be based on the feature, "four leggedness". Hence, cats,

1

2
horses, cows, etc., might all be called "dog". These inappropriate
uses of words have been termed overextensions. Clark obtained evidence
for her hypothesis from the nineteenth century diary studies (cited in
Clark, 1973). An examination of the overextensions revealed that they
were based on shape, movement, size, taste, texture, and sound, with
shape being used most often.

There are several problems with Clark's hypothesis, some of which
have been recognized by other investigators (Barrett, 1978; Nelson,
1974; Reich, 1976; Thomson and Chapman, 1977). First, it does not
account for underextensions which are specific, contextually bound
uses of words. For example, one child used "shoe" only for his
mother's shoes in the closet and no others (Reich, 1976). If children
always started with general features and later added more specific
ones, as Clark states, then underextensions would not occur. Second,
the basis upon which the overextension was made was inferred by Clark.
Not only is it possible that different adults might infer different
bases of similarity about a given child's overextensions, but any or
all of them might differ from that actually intended by the child.
Third, overextensions may not be the most appropriate data base for
investigating the development of word meaning. If word meaning pro-
ceeded in the manner proposed by Clark, all early words should be
overextended. However, Gruendel (1977) and Rescorla (1980) found that
less than twenty percent and thirty-four percent, respectively, of
the words used by the children were overextensions. Nelson, Rescorla,
Gruendel, and Benedict (1978) suggest that children may be commenting
on perceived similarities when they overextended words. They are not

able to state explicitly "it looks like a " because they are

not yet able to construct sentences. Therefore, if a child calls a
slice of grapefruit "moon", he/she may be commenting on the fact that
it looks like a moon but may simply not have the productive vocabulary
to state it explicitly. Furthermore, Clark's theory does not account
for the discrepancy which exists between comprehension and production,
i.e., words that are overextended in production are not always over—
extended in comprehension (Gruendel, 1977; Thomson and Chapman, 1976).
Nelson (1974) criticizes traditional models of concept formation
because they are based on the assumption that the child learns meaning
from encounters with language rather than from encounters with the
physical and social world. In her Functional Core Hypothesis, Nelson
argues that as a result of the child's interaction with an object in
certain contexts, he/she discovers its function. It is on the basis
of this function that a concept is formed. For example, the child's
concept of ball might consist of the relations, "it bounces", "rolls
under the couch", "Mommy throws it", etc. Hence, a concept is a dynamic
set of functions and relationships. According to Nelson, a concept
begins to develop on the basis of a single instance and never starts
out as a vague, undifferentiated whole. The functional basis for meaning
is consistent with the Piagetian notion that objects are not identified
in isolation but within a system of possible relationships. Things are
considered to be similar if they can be acted on in the same way.
Vygotsky (1962) reports that if a child is asked to explain a word,
he/she will tell what the object can do or what can be done with it.
Werner and Kaplan (1963) also contend that it is through internalized
affective-sensory-motor patterns that a child comes to know objects

and reflect on them. Nelson argues that a label is used after the

4
child recognizes that a word consistently is used by others in the
context of the already formed concept. To identify new instances of
the concept, the child might utilize perceptual characteristics as
well as functional ones, especially if the function is not readily
apparent. Therefore, concept formation is based on functional
characteristics but concept identification can be based on perceptual
characteristics. Viewed in these terms, evidence differentiating the
two theories would be difficult to obtain. Nelson also suggests that
the functional core may vary from child to child since it is based on
a child's particular experience with the object. This, in addition
to the fact that the process occurs before a label is attached to the
concept, nakesthe hypothesis a very difficult one to test experimentally.
Some support for Nelson's hypothesis has been found in observational
studies (Gruendel, 1977; Rescorla, 1980) but once again, these involve
overextensions and adult inference for the basis of generalization.

In summary, Clark's hypothesis states that early word acquisition
begins with a child learning a label for an object. The meaning for
that label is based on the object's perceptual characteristics. In
contrast, Nelson argues that a child develops a concept on the basis
of an object's function while perceptual characteristics are used to
identify new instances of the concept.

Investigating the basis of early word meaning is complicated
by the fact that, in reality, perceptual and functional character-
istics are usually correlated. Anglin (1979) found that overextensions
occurred for perceptually similar objects but found it difficult to
separate form and function. Recognizing this problem, Gentner (1978)

used novel objects in which form and function could be separated. The

5

stimuli consisted of the following: a) a jiggy - a box with a face
whose eyes and nose moved up and down when a lever was pulled; b) a
zimbo - a gumball machine containing jellybeans that were dispensed
from the machine by moving a lever; c) a hybrid object which looked
like the jiggy but produced jellybeans like the zimbo. Subjects
ranging in age from 2% years to adulthood were shown the objects,
allowed to manipulate them and were told their names. They were then
shown the hybrid. After manipulating it, they were asked what it
should be called. Younger children and adults used the name based on
form (jiggy) while the middle age group (9-15 years) used the name
based on function (zimbo). Gentner presented this as evidence against
Nelson's theory, although that type of age related response pattern is
not predicted by either theory. It could be argued that the concept
might have been learned on the basis of function and that the testing
situation concerned identification of an object, in which case form
would be an important factor. However, this does not explain why the
older children's identifications still were based on function. It is
interesting to note that informal observations revealed that the young
children learned the name of the zimbo faster and displayed more interest
in it. One methodological problem with Gentner's study is that the ob-
jects were not counterbalanced. Furthermore, even though both toys had
novel names, the gumball machine was probably familiar to the children
and this may have affected the results.

In contrast to Gentner's results, Casby (1979) found that individ-
uals did not generalize on the basis of form, regardless of age. He
used wind-up toys covered with papier mache each of which was given a

nonsense name. Subjects ranging in age from two years to adulthood

6
were presented with an object which performed a certain action and were
told its name. Two additional objects were presented. One object was
similar in form to the target object but performed no action while the
other object was different in form but performed the same action. The
subjects' task was to match the target object. Casby obtained a
significantly greater number of responses based on action for all age
groups. Since objects that performed actions could be considered more
"salient,” an additional group of children (7 years) participated in an
experiment in which the shared-form object also performed an action but
one that was different from the target object. The results were the
same, i.e., they matched the object according to the action of the target,
regardless of form. It is conceivable, however, that not all age groups
would have responded in the same manner.

Using still another approach, Tomikawa and Dodd (1980) presented
preschoolers (2 to 4 years) with a sorting task involving nine wooden
toys of various geometric shapes and functions (e.g., rectangular toy
with an open-close hinge). Following the demonstration of each object's
function, the children were allowed to manipulate the toy themselves.
The test situation consisted of the presentation of two of the toys
which were different in form and function. After each was demonstrated
by the experimenter and manipulated by the children, each toy was
placed on opposite ends of a cookie sheet. Two additional objects,
which were either similar in form or function to the ones on the cookie
sheet, were presented in the same manner. The child was then told to
find the same toy and place it at the appropriate end of the cookie

sheet. The results indicated that the children made more perceptual

7
than functional groupings. Since cartoon decals had been attached to
the toys, Tomikawa and Dodd suggested that the results may have been
biased perceptually. However, a second experiment in which the decals
were removed yielded the same results. Recognizing that a sorting
task may be indirectly related to the acquisition of word meaning,
the authors performed another experiment in which the toys were given
nonsense names and made characters in a story. The training trials
consisted of the children listening to a story about the wooden toys
during which their functions were demonstrated. Two conditions were
used. In the perceptual condition, toys which looked alike had the
same name and in the functional condition, toys which functioned alike
had the same name. The children's task during the test trials was to
point to the correct toy after being given its nonsense name. Criterion
was considered to have been met when the children could correctly name
all three toys in a set in six test trials or less. More children in
the perceptual condition met criterion than in the functional condition.
There were also significantly more errors made in the functional con-
dition. In a final experiment, Tomikawa and Dodd investigated whether
familiar objects would elicit the same type of responses. Artificial
and real fruit were used in the same sorting task as the earlier ex-
periments. The children were given pieces of the real fruit to eat
and were shown that the artificial fruit could not be eaten. Grouping
according to whether the fruit was real or artificial would indicate
a functional base while grouping according to the type of fruit
(apples, oranges, etc.) would indicate a perceptual base. Once again,

the children sorted the objects on a perceptual basis.

8

The results appear to support Clark's position but there are some
methodological problems with the study. First, the authors may have
intended that the objects be comparable in functional and perceptual
saliency but that was never verified objectively from the children's
point of view. Furthermore, the term saliency is not defined.
Second, it appears that the same experimenter was used in all the ex-
periments. If she was not blind to the hypotheses, this could have
biased the results. Third, the children's ages ranged from 2 to 4
years. Since it is not clear whether all ages were equally represented
in each of the conditions, the results could have been influenced by
a particular age group. Fourth, although the toys' functions were
demonstrated during the third experiment, the children did not handle
the toys themselves as was true of the previous two experiments.
Fifth, the authors state that instructions for the experiment involving
the real and artificial fruit were the same as the first two experi-
ments, i.e., "Find the same toy. Put them together." Assuming that
"fruit" was substituted for "toy," the groupings for the fruit could
have been influenced to a greater extent by these instructions than
the nonsense objects. Children's experience with fruit may have
determined which kinds of fruit are the "same." Furthermore, the
artificial fruit did not have a function in the same sense as the
nonsense objects. Children were shown what the nonsense objects
could do whereas they had to be shown what the artificial fruit
could not do (could not be eaten).

The above studies may have obtained conflicting results as a
consequence of the variation in the stimuli which were used. The

role of the object's salience in the word acquisition process has

9

been commented on by many investigators (e.g., Anglin, 1979; Bowerman,
1978; Casby, 1979; Nelson, 1973; Tomikawa and Dodd, 1980). However,
what is meant by the term, saliency, is often left to one's imagination.
In her report of children's early vocabulary, Nelson (1973) provides
some elucidation through her observations that salient objects seemed
to be those which serve the child's basic needs, such as food, and
those which move or change. Since each study used different objects,
the role of saliency could not be assessed. It is possible that the
conflicting results were due partly to variations in the«sa1iency of
the objects across studies. For example, a hinge (Tomikawa and Dodd,
1980) might have been less salient than a mechanical toy that moved on
its own (Casby, 1979), resulting in the child's attributing names on
different bases. Greenberg and Kuczaj (1982) argue that one reason
previous investigations have failed to resolve the form-function con-
troversy is due to the lack of a measurement equating the distance
between the saliencies of particular forms and functions.

Another difference among the stimuli is their designated function.
Some require manipulation by the child (Gentner, 1978; Tomikawa and
Dodd, 1980) while others seemed to function autonomously (Casby, 1979).
In reality, an object's function may or may not be an action (e.g.,
the function of a chair is to be sat on; the chair performs no action
on its own). Consequently, variation in conspicuousness of function
nay influence the child's response. Finally, in the studies mentioned,
the children were constrained to respond on either a perceptual or
functional basis.

In the present investigation, saliency was operationally defined

as children's demonstrated preferences for the objects as indexed by

10
their order of selection. Each stimulus type received both a per-
ceptual and functional saliency rank. Perceptual saliency ranks were
derived from children's initial selections which were based on the
object's appearance. After the children had the opportunity to play
with the objects, functional saliency ranks were determined by their
second round of object selections. These ranks were used to determine
whether perceptual or functional saliency influence both the acquisition
of object names and the way in which children generalize to new
instances of an object category. Furthermore, the present study was
conducted in a less structured fashion in order to provide the child
with the opportunity to respond in a variety of ways.

The purpose of the present investigation was to examine the role
of saliency in name learning and generalization. If children learned
and/or generalized names based on perceptually or functionally salient
characteristics, then neither Clark's nor Nelson's theories adequately
explain these processes. Moreover, if the stimulus objects differ in
saliency, then the inconsistent results obtained to date might be due
to the differences in object saliency. Therefore, the following
hypotheses were tested:

1. Objects from previous studies vary in terms of their per-
ceptual and functional saliency as indexed by the children's order of
selection.

2. Perceptual and functional saliency of objects will influence
the acquisition of names.

3. Perceptual and functional saliency of objects will influence

the basis for generalizing to new instances of an object category.

CHAPTER II

METHOD

The experiment can be divided into three phases. In the first
phase, the degree of saliency for nonsense objects was assessed by the
child's selection of the objects before and after interacting with
them. In the second phase, the child was asked to learn nonsense names
for some of the objects. The third phase was an examination of the
basis upon which the child generalizes to similar objects varying in

form and function.

Subjects and Experimenters

 

Twenty males and twenty-one females, ranging in age from 23 to 31
months with a mean of 25.66 and a standard deviation of 1.78, partici-
pated in the study. Twenty-one of the subjects were first born children
and twenty were second born. The families were middle class and lived
in the East Lansing, Okemos, Haslett, and Meridian township areas.
Potential subject names were obtained from birth records and were
solicited through letters sent to their homes. Three of the subjects
were solicited from the university day care center. Five additional
children (3 from homes, 2 from the day care center) either started
the experiment and refused to continue or refused to start at all.
They were excluded from the study.

One principal experimenter interacted with the children while

observations were made by two other individuals who were blind to the

11

12
questions being investigated. The observers were provided with a

checklist on which they recorded the children's responses.

Setting and Materials

 

The experiment took place either at the day care center (for the
three children who attended) or at the child's home (for the remaining
thirty-eight). The mother was present in most instances and both mother
and father were present in three instances. Parents were allowed to
observe but not interact with their child during the procedures.

Three types of objects, modeled after those used in the studies of
Gentner, 1978; Casby, 1979; and Tomikawa and Dodd, 1980, were used.

The objects from these studies were chosen because they provide variation
in perceptual and functional characteristics and enabled comparisons

to be made across studies. Gentner's stimuli (Type A) have noticeable
perceptual properties and require manipulation by the child before their
function is realized. Casby's stimuli (Type B) have geometric-like
shapes and move in interesting ways seemingly on their own. Tomikawa

and Dodd's stimuli (Type C) consisted of geometric shapes and have rather
obscure functions.

The toys used in the present investigation are described below:

Type A: 1) A white box with a pink and blue dog-like face on the

(marble front. The dog's ears and mouth moved up and down via

toys) a lever. 2) A white box with a stationary pink and blue

dog-like face mounted on the front. Marbles, contained
inside the box and not visible, were dispensed through
a hole via a lever. 3) A white box with a blue airplane
shape mounted on the front. The airplane moved up and

down via a lever. The box also had a transparent bowl

causes .

Type 8:

(wind-
up toys)

Type C:

(hinge/
magnet
toys)

13
on tap containing different colored, medium sized
marbles, but the marbles remained in the bowl. 4) A
white box with a stationary, blue airplane mounted on
the front. There also was a transparent bowl on top
containing different colored, medium sized marbles which
were dispensed through a hole via a lever.
The dimensions of the four boxes were each 10” x 8” x 8".
l) A wind-up toy that walked, covered with a white
papier mache cylinder. 2) A wind-up toy that hopped,
covered with a white papier mache cylinder. 3) A wind-
up toy that walked, covered with a white papier mache
cone. 4) A wind-up toy that hopped, covered with a
white papier maché cone.
The wind-up toys were 2%" tall.
1) A three inch, red, square piece of wood with an
open-close hinge. 2) A three inch, red, square piece
of wood with a magnet glued to one side. A 10'I x 4"
piece of metal was presented with this object. 3) A
red, three inch diameter circle made of wood with an
open-close hinge. 4) A red, three inch diameter
circle made of wood with a magnet glued to one side.

The same piece of metal was presented with this object.

Anxiety scale. It is conceivable that during the generalization

 

respond at all.

task (see Procedure, Phase 3 for details) some children might not

Although a lack of response could be caused by any

one of a variety of factors, it was not possible to examine all possible

Since the children were faced with three strangers in an

l4
unfamiliar situation, anxiety seemed a likely reason for response in-
hibition. Anxiety was operationally defined as those behavioral in-
dicators designated by the Preschool Observation Scale of Anxiety
(Glennon and Weisz, 1978). This particular scale was chosen since it
appears to be the only one available designed for preschoolers. Glennon
and Weisz assessed the validity of the instrument by comparing the
scores with teachers' and parents' inventory ratings of children's
anxiety. Significant positive correlations were obtained. Moreover,
children's scores on the scale were high in a session expected to pro-
voke high anxiety (mother absent) than in a session expected to provoke
minimal anxiety (mother present). Interrater reliability also produced
a significant correlation coefficient, .78, for total scores. The
scale consists of thirty items and the total score is calculated by the
number of 30 second intervals in which any of the behaviors occurred.
The method of computing anxiety scores for the purpose of the present
experiment was modified since it was not feasible to divide the session
into 30 second intervals. Furthermore, expressions of anxiety were not
critical to the central aims of the study. The anxiety score for each
subject was derived from the total number of different anxiety indicators
recorded by both observers. Each indicator was counted only once. (See

Appendix for scale items.)

Procedure

Four observers, three females and one male, were trained to record
the children's toy selection, play time, anxiety indicators, and
verbalizations. Two observers were present during twenty-seven of the
forty-one experimental sessions. One observer was present at the

remaining fourteen sessions. It was not possible to have two observers

15
present at all sessions due to scheduling difficulties. Although inter-
rater reliability for the anxiety scale was only .55, the correlations
for the measures most relevant to the study ranged from .87 to 1.00
(see Results), therefore, only one observer was deemed sufficient for
the remaining sessions. Reliability for order of selection and play
time was high initially so little formal training was necessary with
regard to these variables. The observers received practice recording
the responses with pilot subjects. Moreover, discussions with the
author prior to the present investigation clarified procedures for
possible ambiguous responses. The anxiety scale was memorized by the
observers and any instances of anxiety were recorded on the same sheet
as the other responses.

The experimenter and the two observers played with the child until
he/she seemed comfortable with them (e.g., played with the experimenters,
verbalized, expressed interest in looking at the toys, etc.). Also
during this time, the investigators, the mother, and the child played
a game in which the child pointed to each person named in response to
a song that proceeded as follows: "Where, oh where, oh where is (person's
name)?" The reason for this game was to familiarize the child with this
song since it was going to be used to assess comprehension of the object
names later in the experiment, and in addition, to confirm that the
child understood that he/she was expected to point to objects in response
to their name. However, some children disliked this game and, therefore,
it was not used with them. (Unfortunately, the exact number is not
known since this took place before the observers began recording
responses.) For children who would not respond to the song, the

experimenter simply named the people in the room and asked the child

16
to point to them. All the children demonstrated they could point to a
person or object when asked.

Phase 1. The purpose of this phase was to obtain two measures of
saliency, one perceptual and one functional, as indexed by the child's
order of selection of the objects. The length of time the children
played with each object was also recorded.

After the familiarization procedures were completed, the child
was taken out of the room and the experimenters set up the objects.
When the child reentered the room, all objects were hidden except for
three, one from each object type (see setting and materials for
description). Since it was not feasible to expose the child to all
possible combinations across types, each child was shown one set which
consisted of three objects, one from each type, randomly selected.

The placement of objects in a set was randomly assigned for each

subject in order to eliminate position effects. If the child did not
spontaneously begin playing with one of the toys the experimenter asked
the child which toy he/she would like to play with first. The selection
and the length of time (up to a maximum of two minutes) the child played
with the toy was recorded by the observers. Based on the assumption
that the children's initial round of choices represented the objects'
perceptual salience, a child's first, second, and third selections were
assigned perceptual saliency ranks l, 2, and 3. Once children had
played with the objects, their choices should be influenced by the
saliency of what the objects do, so the rankings of functional saliency
were determined by asking the child to indicate which toy he/she would
like to play with again. The child was allowed to play with the toy

that was selected and the procedure was repeated for the two remaining

toys. The toys were then ranked 1, 2, and 3 for a second time.

17

Phase 2. In this phase, the child was asked to learn nonsense
names for the three objects with which he/she had been playing. The
experimenter named each object as follows: "This is called a
See, the looks like this and does this. This is a ."
The names of the three objects were counterbalanced resulting in six
sets of names. The order of presentation of the sets was determined
randomly for each child. Comprehension of the names was assessed by
asking the child to point to the object when given its name. The song
("Where, oh where," etc.) described earlier was used only with three
children since most children seemed to become bored with it. The
children were asked simply, "Where's the _____2 Can you point to it?”
If the child made an incorrect response, the experimenter said, "No,
this one is the ______f and then proceeded to ask for the next name.
When the child made a correct response, he/she was praised (e.g.,
"Very good, that's right.").

Phase 3. In this phase the child was asked to select similar
objects to the ones that were named but that varied perceptually or
functionally. The three remaining objects from one of the object
types was shown to the child. (The order of presentation of types and
objects within types was distributed randomly across subjects.) Each
was demonstrated by the experimenter and given to the child to play
with. The experimenter then asked the child "Are there any
here? Can you find any ? Can you point? There were several
possible responses which the child could make. He/she could select
a toy that was similar in function, in form, one that was dissimilar
in both respects, or any combination of these. For example, if the

child was presented with a square hinge, he/she could select a round

18
hinge (function) or a square magnet (form) or a round magnet (no
similarity). The child also could select both a square magnet and a
round hinge (form and function) which might imply that the child
recognized the two separate similarities. However, the child might
select no toy in which case at least two inferences might be made,
i.e., there was an inhibition of response for some reason (e.g.,
anxiety) or the child did not recognize either of the similarities.
In order to distinguish between these two responses, behavioral in-
dicators of anxiety were recorded throughout the experiment. If the
child picked all three objects or any combination of objects that
included the object which was dissimilar to the object that was named,
several interpretations might be proposed, one being that the child
simply did not understand the task. Therefore, the child's responses
were classified in one of six categories: a) perceptually similar;
b) functionally similar; c) perceptually and functionally similar;
d) no similarity; e) verbal response specifying no similar toy; and
f) no response.

Mean Length of Utterance (MLU). In order to determine whether

 

the children's stage of language development was related to performance
on any of the tasks, MLU was calculated for each child. When the
procedures described above were completed, the child was allowed to
play with any of the toys. During this time, the observers recorded
the child's verbalizations until fifty utterances were obtained.
Thirty-two of the forty-one children were used in this analysis since

a minimum of fifty utterances could not be obtained for the remaining

children due to lack of spontaneous as well as prompted utterances.

l9
MLU was defined as the mean number of morphemes per utterance. The

rules specified in Dale (1976) for designating morphemes were followed.

CHAPTER III
RESULTS

Interrater reliability was assessed for order of selection of
objects, time played with objects, comprehension of object names,
classification of responses for generalizing to new instances of a
category, anxiety indicators, and Mean Length of Utterance (MLU).
There was 100 percent agreement for the order in which the children
selected the objects during the first and second play periods.
Pearson correlation coefficients for the length of time the children
played with the toys were .99 for the first play period and .98 for
the second. The average time recorded by the two observers was used
in the data analyses. The interrater reliability for comprehension
of object names and MLU as measured by Pearsonian correlation was
.98 and .87, respectively. Once again, the average score was used
in the analyses. The reliability for responses concerned with
generalizing to new instances of a category was obtained by percent-
age agreement. The agreement was 89 percent for the marble toys (M),
96 percent for the wind-up toys (Wp), and 93 percent for the hinge/
magnet toys (H/M). When the observers disagreed, the classification
of the response was determined by a coin toss. The interrater
reliability, as measured by Pearsonian correlation, for the number of
anxiety indicators observed was .55. With reSpect to type of indi-
cator, percentage agreement between observers for each session also

was calculated. Agreement ranged from O to 100 percent with a mean

20

Jib r . 3 a a. .5 ,..v C . J l a: t . .u. .11 . k .1- ii .
LO- J o a r , .. l 1. in . J a J 1 n1. 2 . ul 4 1 o. - F- a 1 C; 9 . .r 1. $5
.3 o u a v n f .-f ‘48 F , 1. . a no A C .. I ll r IIU L y a .\u :- 6: r
.4.“ . a P at. J r. O. A 5 Ah.» «I.- II I .ul .I 151 1 \FU nu - n? «all at. Adv 1‘ I. «Lu» J
.l g 141- 11 O b .r b . ad I A 5 5.1....» (v .I a ,i t P1» P\.v .2 u I z .1. I a n l.“ ‘1» a J IHU

 

21
of 49 percent. Possible reasons for the low reliability are: a) it
was difficult for the observers to record the anxiety indicators along
with several other responses; b) since the children tended to move
around a lot, both observers did not always have an equally good
vantage point to view the children's behavior (e.g., if the child
turned his/her face, only one observer may have been able to see a
grimace that was made). Consequently, the anxiety score used in the
analyses was the total number of different anxiety indicators re-
corded by both observers.

The analyses concerned with the hypotheses being tested will
be discussed first.

H1. The objects vary in terms of their perceptual and functional
saliency as indexed by the order in which they were selected. This
hypothesis was supported by the data which indicate that the objects
differed in terms of the children's preferences for them. A Friedman
two-way analysis of variance performed on the order in which the
children selected the objects during the first play period (perceptual

E = 35.17, 2, p < .001). The same analysis

saliency) was significant (x
performed on the order of selection during the second play period
(functional saliency) also was significant (xi = 26.39, 2, p < .001).
Table 1 presents the frequencies with which each of the object types
were selected for each play period. The marble toys were selected first
most often, and these also were the toys the children most frequently
wished to play with when given the opportunity to play with any toy

they chose. However, the order of selection during the first play
period did not differ from that of the second play period as indicated

by the McNemar test for significance of changes (M: x2 = 2.58, l,

22

Table l

Frequencies for the Order of Selection for Each Type of Object
During the First and Second Play Period

 

 

 

 

 

 

First Play Period Second Play Period
Rank Order Rank Order
Object l 2 3 l 2 3
Marble 32 6 3 26 10 5
Wind-Up 8 20 l3 13 18 8
Hinge/Magnet l 15 25 2 ll 26
p > .05; Wp: x2 = 2.28, l, p > .05; H/M: x2 = .34, l, p > .05), which

would indicate that becoming acquainted with the toys' function did not
affect the children's initial selection. Perhaps they inferred the
function before they actually saw what the toy did, or perhaps,
interesting perceptual characteristics are highly correlated with
interesting functional ones, at least in these stimuli.

H2. Perceptual and functional saliency will influence the
acquisition of object names. This hypothesis was not supported. To
determine whether saliency influenced how well the children learned the
names, the rank orders were divided into two groups, high rank versus
low rank. The high rank group consisted of those subjects who selected
an object first. The low rank group consisted of those subjects who
selected the object second or third. A t-test was then performed on
these two groups in order to compare how well each group learned the
object names. Very few children selected the hinge and magnet toys
first (one subject during the first play period and two during the
second); therefore, the high rank group for these toys consisted of
those subjects who selected them second while the low rank group

consisted of those subjects who selected them third.

 

(TI

'-

23

A significance difference between children at different selection
ranks was obtained (t = 2.72, 39, p < .01) for the marble toys during
the first play period. Those who selected the toy first learned the
name for that toy better than those who selected it second or third.
This would suggest that perceptual saliency was important for learning
the name for this toy. On the other hand, comparison of the high and
low ranks for the wind-up toys indicated that those children who selected
the toy second or third during the second play period learned the name
for the toy better (t = -2.0, 39, p < .05). The reason for these results
are unclear. If functional saliency was a factor, one would expect that
children who selected the wind-up toy first during the second play
period would have learned the name better, rather than those who selected
it second or third. No significant differences were found for the
hinges and magnets.

H3. Perceptual and functional saliency will influence the basis
for generalizing to new instances of an object category. Before dis-
cussing the data pertaining to this hypothesis, some preliminary infor-
mation is required. Table 2 shows the frequency with which the children
responded in each of the categories. As indicated, the responses
differed depending on the type of object. With regard to perceptual
and functional similarity, the marble toys and the hinges and magnets
were generalized most often on a perceptual basis while the wind-up
toys were generalized on a perceptual basis by some children and a
functional basis by others. To examine these differences, the
Cochran 0 test for matched samples was used to compare the frequency
of responses in the perceptual similarity category and the functional

similarity category across objects. Responses based on perceptual

24
Table 2

Frequency of Responses in Each of the Categories of Generalization
to Similar Objects: Perceptual Similarity (P), Functional
Similarity (F), Verbal "No" Response (N), Both Perceptual

and Functional Similarity (P & F), Neither
Perceptual Nor Functional Similarity (NS),
No Response (NR)

 

 

 

 

Object Response Category

P F N P&F NS NR
Marble l7 3 5 O 7 9
Wind-Up 14 11 2 1 6 7
Hinge/Magnet 13 4 10 O 7 7

 

similarity did not differ significantly across objects (Q = 1.13, 2,

p > .05). However, responses based on functional similarity did differ
significantly (Q = 6.71, 2, p < .05), with children responding on the
basis of functional similarity more frequently for the wind-up toy than
the other two toys. Although the index of saliency used in this in-
vestigation would indicate that the wind-up toy was not the most
functionally salient toy, this analysis suggests that, in some way, the
function of the wind-up toy influenced children's responses to a greater
extent than the function of the other two toys.

To test whether the frequencies of each response for each object
type were random or differed in some systematic way, a chi square good-
ness of fit test was performed. Response frequencies in each of the
categories (perceptual similarity, functional similarity, perceptual
and functional similarity, no similarity, verbal response specifying
no similar toy, and no response) were compared with an expected frequency.
Three separate tests were performed, one for each object type. A signi-

ficant difference was found for the marble toys (x2 = 25.3, 5, p < .001),

‘IDI
P"v
1.vv

9
0‘?“

(ll

r4

,..
B“
.

'in-

‘4‘
;.
‘j

I,"

25

2 = 17.51, 5, p < .01), and for the hinges and

for the wind-up toys (x
magnets (x2 = 15.04, 5, p < .02), indicating that the distribution of
responses was not random.

Considering just the responses based on perceptual similarity and
functional similarity, since these were the responses of primary in-
terest, a significantly greater number of responses based on perceptual

2

similarity were made for the marble toys (x = 9.8, 1, p < .01), and for

the hinges and magnets (x2 = 4.76, l, p < .05). No significant difference
was found for the wind-up toys (x2 = .36, l, p > .05). These analyses
indicate that although the basis of generalization depended on the type

of toy, generalization was more often based on perceptual than functional
similarity for two of the object types.

The main reason for including the anxiety scale was for the purpose
of differentiating between children who did not respond at all and those
who gave a verbal ”no" response. However, when behavioral anxiety in-
dicators for these two groups were compared, there was no difference
(M: t = -2.03, 12, p > .05; Wp: t = -.l4, 7, p > .05; H/M: t = -.69, 15,
p > .05).

In order to determine whether the saliency rankings (order of
selection) influenced the basis for generalizing to new instances of an
object category, the chi square goodness of fit test was performed on
the frequency in each response category in relation to the order in
which the toy was selected. For example, eighteen children selected the
marble toy based on perceptual similarity during the first play period.
Of these eighteen, fifteen children had selected the toy first, three
had selected it second, and none had selected it third. These three

frequencies were compared using the chi square test in which the actual

26

frequencies are compared with an expected frequency. In this example,
the expected frequency was six. Table 3 presents actual frequencies
and chi-squares for each response category. (The response category
which was defined as the selection of two toys, one based on perceptual
similarity and one based on functional similarity, was eliminated from
the analysis since it occurred only once.) The marble toys and the
hinges and magnets produced significant differences for perceptual

similarity (M: x2 = 18.19, 2, p < .001; H/M: x2

= 6.62, 2, p < .05).
This indicates that the marble toys were generalized most often on the
basis of perception when they were selected first. However, the hinges
and magnets also were generalized on the basis of perception but unlike
the marble toys, this occurred when they were selected second and third.
Since it was assumed that selecting a toy second or third during the
first play period indicated that the toy was less perceptually salient
than one selected first, it would appear that perceptual saliency did
not affect the basis of generalization. Significant differences also
were found for other response categories during the first play period
and for several response categories during the second play period
(Table 3), but most could not be related to perceptual or functional
saliency (order of selection). Therefore, the third hypothesis was

not supported.

Summarizing thus far, it seems that the objects differed in terms
of the children's preferences for them. However, the order of selection
during the first play period assumed to be an indication of perceptual
saliency, and the order of selection during the second play period

assumed to be an indication of functional saliency were the same.

Whatever attracted the children to the toys did so right from the

27

Table 3

Actual Frequencies and Chi Squares for Type of Responses1 When Children Were

Asked to Generalize to Similar Objects in Relation to the Order in
Which Each Object Type Was Selected During the First and
Second Play Periods

 

 

 

 

 

 

 

Object Response First Play Period Second Play Period
Frequencies Frequencies
Order of Selection Order of Selection
1 2 3 x2 1 2 3 x2
P 15 3 0 18.19**** 11 5 1 8.51**
F 2 l 0 1.50 3 0 O 3.33
M N 5 0 0 9.98*** 4 O l 5.19
NS 5 0 1 6.15‘ 5 0 1 6.15*
NR 5 2 2 1.99 3 4 2 0.66
P 4 5 5 0.14 5 8 l 5.27
F 1 5 5 2.86 5 4 2 1.27
Wp N O 1 1 0.99 1 1 0 0.99
NS 1 3 2 1.00 0 5 1 7.00*
NR 1 6 0 8.87** 1 2 4 2.01
P 0 6 7 6.62* 0 4 9 9.40***
F O 1 3 3.51 1 1 2 0.50
H/M N 0 4 6 5.60 0 3 7 7.40*
NS 0 2 5 5.44 O 3 4 3.72
NR 1 2 4 2.01 l 2 4 2.01
* < .05
** < .02
it! < .01
**** < 001

P - _
t P: perceptual similarity; F: functional similarity; N: verbal “no" response; NS: neither
perceptual nor functional similarity; NR: no response.

28
start and for the most part, this did not change after playing with the
toy. The orders in which the toys were selected were associated with
learning the names for the toys, but this association differed depending
on the type of toy. Furthermore, the basis of generalization differed
as a result of type of toy, but not as a result of the order in which

the toys were selected.

Post-hoc Analyses
Several other questions of interest can be addressed from these
data.

Object names. Did the children learn the names for certain

 

objects better than others? Children were asked to point to the correct
toy when given its name. This was done six times for each toy.
Therefore, a perfect score for all three toys would be eighteen. Table
4 shows the means and standard deviations for the number of correct
responses for each type of object. A two-way ANOVA involving repeated
measures was performed on the accuracy of learning the names for the
three types of objects. Sex was a between groups factor and type of

object was a within subject factor. Significant main effects were

Table 4

Means and Standard Deviations for the Number of Correct Names
Learned for Each Type of Object for Males and Females

 

 

 

 

Marble Wind-Up Hinge/Magnet
Sex X s X s X ' s
M 2.13 1.67 4.03 1.41 1.93 1.05
F 3.45 1.16 4.57 1.54 3.07 1.96
Total 2.80 1.75 4.30 1.48 2.51 1.67

 

M"

0"

:A‘

ibi

«‘1-

.A.

.-

I‘p

.-

 

29
obtained for sex of subject and type of object, F (1, 39) = 8.39,
p < .006; F (2, 78) = 20.84, p < .000. There was no significant inter-
action. As shown in Figure l, the name for the wind-up toy was learned
the best and females performed better overall.

The names for each of the types of toys were not randomly assigned
for each subject, and therefore, the name is confounded with the type of
toy. One possible explanation for better performance for the wind-up
toy is that the nonsense name for this toy had a high association value
with a real name, and consequently, was easier to learn. All the
nonsense names used were considered to have low association value
according to Hilgard (1951). However, the association explanation
cannot be dismissed completely since the associative norms were obtained
for adults and may not be applicable for children. The possibility of
phonological biases can be discounted since Leonard, Schwartz, Folger,
Newhoff, and Wilcox (1979) found that subjects showed the same tendency
to imitate a nonsense word that contained phonemes that were not
produced spontaneously as those that contained phonemes that were pro-
duced spontaneously. Moreover, the nonsense words in the present study
that produced the most correct responses, wug (hues/), contained
phonemes that are often produced later in development than the nonsense
words that produced fewer correct responses, mep (finap/), and zib (/e:b/).
In a follow-up study intended to determine whether learning the name for
the wind-up toy was due to the name itself or some aspect of the toy,
the names were randomly assigned (Della Corte, in progress).

Since age was found to be correlated with the accuracy of learning

the names (r = .38, 39, p < .02), an ANCOVA was performed using age as

3O

 

 

 

5 '1-
4.m.
NUMBER
CORRECT 3 'T’
2‘!-
l'-~
n I 1
I ”r ”T
M Wp H/M
OBJECTS

_ ma] es

Figure 1. Mean Number of Correct Name Responses for Each Object for
Males and Females

31
the covariate. Males and females still differed on how well they learned
the names even when age was taken into account, F (1, 39) = 6.06,
p < .02.

Play time. Since the order in which the children selected the
objects did not influence name learning or generalization, the length
of time the children played with the objects was examined with respect
to these variables. Using a 2 (sex) x 3 (type of object) x 2 (play
period) ANOVA, significant main effects were found for sex of subject,

F (l, 39) = 6.15, p < .03, type of object, F (2, 78) = 36.93, p < .000,
and play period, F (l, 39) = 5.17, p < .03. No significant interactions
were found. Females played with the objects longer, and children played
with the marble toys the longest, the wind-up toys next, and the hinges
and magnets least (Figure 2). Means and standard deviations are
presented in Table 5. Children played with the objects longer during
the first play period. This reduction in play time may have been due

to a novelty effect.

When length of play time was examined depending on the order in
which the object was selected, a simple one-way ANOVA revealed signi-
ficant differences for all three types of objects (M: F (2, 78) = 7.05,
p < .01; Wp: F (2, 78) = 5.46, p < .01; H/M: F (2, 78) = 5.33, p < .01)
during the second play period. For the marble toy, paired comparisons
using the Duncan test showed significant differences in play time
between those children who selected this toy first and those who selected
it second, with the children who selected the toy first playing with it
for a longer period of time. Similarly, those children who selected
the marble toy first played with it significantly longer than those who

selected it third. Unexpectedly, the children who selected the marble

 

 

 

90 .1- k‘ M
0“ ‘0
0C T‘JL
80 c1- ‘§‘ ~~~
\N~%fi~ 0
. \“
-o
70 j.
60 --
TIME ”p
50 -_
(in seconds)
40 .1. 0“
‘“\
x.
"’07
30 di- \\
“~
‘No
20 -- ”M
10 ..
l l
j 1
First Second
Play Play
Period Period

0.--- ..0 females

Figure 2. Mean Play Time (In Seconds) During First and Second Play
Periods for Each Type of Object for Males and Females.

33

 

 

 

 

 

 

 

 

¢¢.¢~ NP.NF om.¢e mo.~m cw.mq m~._w mw.om om.mm em.oe ow.mc om.o¢ em.mw »
we.Pm o~.o~ mo.~e en.mo em.me FN.- om.pm mo.Pe mo.mm w¢.mw oe.mm mm.mm u
o~.mp Pe.mp n¢.m¢ ms.me am._e Fm.ow -.~N m~.¢~ m~.oe m¢.mm em.me mm.mw z
m .x m w m w m .x m M m .x.
E: a: E: a: z
vowcma zaps ncoomm newton amps umewu xmm

ace to maxh seem so» mmposmu new mm—mz com mcowgma xmpm vacuum new amcwu
mg» mcwcao Amccoumm cwv damp xmpa mo cameo; commcowamw>mo ucmucmum new menu:

m mpnmh

34
toy third played with it longer than those who selected it second. The
reason for this may be due to the fact that sometimes the lever on this
toy was difficult for the children to press, and consequently, they
would discontinue trying to make the toy work. Eventually, the children
were able to press the lever and when this was accomplished they en-
joyed playing with the toy. Perhaps these children were the ones to
select the toy last during the first play period. However, when they
finally were able to Operate the toy, they played with it for a greater
period of time. Unfortunately, the observation concerning the children's
difficulty in operating the toy was not systematically recorded.
Therefore, this explanation must be viewed as speculation since it
cannot be confirmed at this time.

Paired comparisons for the wind-up toy revealed that children who
selected this toy first played with it significantly longer than those
who selected it third, and those who selected it second played with it
significantly longer than those who selected it third. Unlike the
marble toy, no significant differences were found between children who
selected the wind-up toy first and those who selected it second. For
the hinges and magnets, a significant difference was found between
children who selected these first and those who selected them third with
the former playing significantly longer. There also was a significant
difference between children who selected the hinges and magnets first and
those who selected them second with those who selected them first
playing with them longer. No significant differences were found for
children who selected the hinges and magnets second and those who selected

them third.

35

To examine whether the length of time that children played with a
particular toy had any effect on how well they learned the name for
that toy, a 2 (sex) x 3 (type of object) ANCOVA was performed with
play time as a covariate. Play time had no effect on how well the
children learned the names for the objects.

To investigate whether the length of play time affected the
basis of generalization, the play time of children who generalized
on the basis of perceptual similarity was compared with those who
generalized on the basis of functional similarity. Only the marble
toy produced significant differences for both the first play period
(t = -2.88, 18, p < .01), and the second play period (t = -4.56, 18,
p < .001). The children who played with this toy longer tended to
generalize on the basis of function. These results must be viewed with
caution since the number of children comprising the group who generalized
on the basis of function was very small while the group who generalized
on the basis of perception was large. Play time may have an effect on
generalization whereas the order in which the toy was selected did not.

Anxiety. Anxiety scores as measured by the total number of anxiety
indicators recorded by both observers ranged from O to 9 with a mean of
3.83 and a standard deviation of 2.52. To examine whether anxiety was
related to children's performance in learning the object names, the
anxiety scores were correlated with how well they learned the object
names. A Pearson correlation coefficient was significant (r = -.36,
39, p < .05). Children who were less anxious learned the names better.

To explore whether anxiety had any effect on the responses
children made during the third phase (generalizing to new instances of

an object category) of the experiment, their responses were grouped

36
according to how "responsive" they were, i.e., to what extent did they
comply with the investigator's request to select an object. Since
anxiety can have a disorganizing effect on behavior, children who did
not make a selection may have been more anxious than children who did.
To investigate this possibility, children who made some kind of
selection response regardless of whether the object chosen was similar
to the one that was named were considered most responsive and given a
score of 1. Those who verbally replied there was no object that was
similar were given a score of 2. Those who gave no response at all
were considered least responsive and were given a score of 3. When
these scores were correlated with the anxiety scores for each type of
object, the marble and wind-up toys produced significant correlation
coefficients of .38 (p < .02) and .41 (p < .01), respectively, indicating
that the children who were more anxious were the least responsive.
Similarly, a single factor ANOVA comparing the three responsiveness
groups on anxiety also revealed significant differences for the same two
object types (M: F (2, 78) = 3.49, p < .05; Wp: F (2, 78) = 3.98,
p < .05). Paired comparisons using the Duncan test showed that for the
marble toy, children who gave no response displayed significantly more
anxiety than children who gave some kind of response. Children who gave
no response also were significantly more anxious than the children who
verbally replied that no object was similar. No significant difference
in anxiety was found for children who verbally responded that no object
was similar and the children who made some kind of selection. For the
wind-up toy, children who verbally responded that no object was similar
were significantly more anxious than children who made some kind of

selection. Children who made no response were significantly more

37
anxious than children who made a selection. No significant difference
was found between children who made no response and children who verbally
responded that no object was similar. The reasons for the differences
between the marble and wind-up toys are not clear, but once again, the
children's responses differed depending on the type of object.

Sex differences. There was no sex difference for the order in which

 

the objects were selected for either the first play period (M: t = 1.81,
38, p > .05; Wp: t = -1.87, 38, p > .05; H/M: t = .58, 38, p > .05), or
for the second play period (M: t = -.22, 38, p > .05; Wp: t = -.43, 38,

p > .05; H/M: t = .81, 38, p > .05). However, there was a difference in
the length of time males and females played with the toys as measured

by a 2 (sex) x 3 (type of object) ANOVA, F (l, 39) = 6.15, p < .02)

with females playing with the wind-up and hinge/magnet toys longer than
the males. Females also learned the names for the toys better (t = -2.82,
38, p < .01). Since females were more accurate in learning the names

and played with the toys longer, Pearson correlation coefficients were
computed between number of correct responses and play time for each

type of object for the first play period (M: r = .39, 19, p > .05; Wp:

r = -.35, 19, p > .05; H/M: r = .09, 19, p > .05), and the second play
period (M: r = .22, 19, p > .05; Wp: r = -.30, 19, p > .05; H/M: r = -.24,
19, p > .05). Although none were significant, it is interesting to note
that for females during the second play period, there was a positive
correlation coefficient between time and correct naming responses for

the marble toy (r = .39), but there was a negative correlation for the
wind-up toy (r = -.35). For males, just the opposite was the case

(M: r = -.32; Wp: r = .36). Fisher's Z transformation revealed the

correlation coefficients for males versus females for marble toys did

38

not differ significantly from one another (t = 1.63, 39, p > .05), but
for the wind-up toys, they did differ significantly from one another
(t = 2.04, 39, p < .05). It would appear that children react differently
to the wind-up toy depending on sex, although the cause is undetermined
at present.

With regard to other variables, no sex differences were found for
generalizing to new instances of an object category, anxiety scores, or
MLU.

Birth order. Since there were almost equal numbers of first and

 

second born children in the sample, they were compared on the various
dependent variables to determine whether birth order was an influencing
factor. No differences were found for the order in which the objects
were selected, the length of time the children played with the objects,
accuracy of learning the object names, basis of generalization, or

MLU.

Mean Length of Utterance (MLU). The mean and standard deviation

 

for MLU was 2.37 and .63, respectively. Children ranged from 1.34 to
4.19. MLU was computed for each subject in order to examine whether
this variable was related to the accuracy with which children learned
the object names. A Pearson correlation coefficient was not
significant (r = .19, 30, p > .05). It should be noted that MLU was
difficult to obtain since many of the children were silent most of
the time, and often their speech was difficult to understand. The
obtained MLU for each subject probably does not reflect accurately
the children's language development since much of their speech was

in response to questions which tend to elicit one word responses.

39

Summary. The objects differed in terms of the children's pre-
ferences for them, but the index used to measure saliency (order of
selection) had little effect on the accuracy of learning the names or
the basis of generalization. On the other hand, play time was related
to generalization and differed between males and females. Sex differences
also were found for name learning. Females learned the names better than
males. Furthermore, all children learned the name for the wind-up toy
better than the other two toys. Finally, the children's anxiety may

have interfered with their performance on some of the tasks.

CHAPTER IV

DISCUSSION

The idea that differences in object saliency might explain the
conflicting results of previous studies which have investigated the
basis of early word meaning was not clearly supported. When measured
by the order in which children selected the toys, saliency was not an
influencing variable. Either the index used to measure salience was
inadequate or the conflicting results of earlier studies are due to
other factors. Although play time did not have an effect on how
well the children learned the different object names, there was some
indication that it influenced the basis of generalization (functional)
for one of the object types (marble toy).

The children generalized the marble toys and the hinges and magnets
on the basis of perceptual characteristics which is consistent with
studies using similar objects (Gentner, 1978; Tomikawa and Dodd, 1980).
The results for the wind-up toys were not totally consistent with
Casby's (1979) findings. Some children in the present investigation
generalized on the basis of perceptual characteristics and others
generalized on the basis of functional characteristics, whereas
children in Casby's study generalized on the basis of functional
characteristics. This discrepancy may be due to the fact-that some
toys in Casby's study had no action while others did. As stated
earlier, this was corrected in a follow-up study in which all the

toys were given actions, and children persisted in generalizing on

40

41
the basis of function. However, the children were seven years of age,
and hence, older than the children in the present investigation. At
present, it would seem that the type of object involved makes a
difference in how children will generalize to similar objects. There
may not be one, overriding basis for generalization.

Unlike previous, similar investigations, the present investigation
showed that females learned the object names better than males. However,
* no sex differences were found for the basis of generalization.

The children in this study preferred playing with some of the
objects more than others, but their preferences had only a slight
influence on how well they learned the names for the objects, and no
influence on how they generalized to new instances, However, the
objects did elicit different responses. The children learned the
name for the wind-up toy better and generalized the name for it on
the basis of function more often than the other two toys. The fact
that the name for the wind-up toy was learned the best is consistent
with Nelson's (1973) findings that children acquire words for things
that move before they acquire words for static objects. The explana-
tion typically given for this is that dynamic objects are more
"salient," but when salience is operationally defined as the order in
which the objects are selected, the wind-up toy is not the most
salient. Ross, Nelson, Wetstone, and Tanouye (Note 1) also found
that object attractiveness was not related to receptive name learning
for novel objects. In their study, object attractiveness was measured
by "the number of five second intervals during which children contacted

the object type." Children in the Ross et a1. study learned the names

42
better for the objects with which they performed more specific inter-
actions; i.e., behaviors that were related to the particular properties
of the object (stretching out a slinky) as opposed to undifferentiated
behaviors (mouthing, banging, etc. the object). Their results differ
from the present study in that the object whose name the children
learned the best (wind-up toy) was a toy with which they tended to
interact the least. Most children just watched the toy walk or hop
because they were unable to wind it up themselves.

Although one of the purposes of the experiment was to obtain an
estimate of both perceptual and functional saliency, some uncertainty
exists as to whether the index used (order of selection) adequately
reflected these two aspects of saliency since the two sets of rank
orders were not significantly different from one another. One possible
explanation is that the two sets of orders did indeed reflect the two
aspects of saliency, but the perceptual and functional characteristics
were strongly correlated, i.e., the same toys were high or low in both
perceptual and functional salience. Therefore, the order of selection
did not change. Another is that the children inferred the objects'
function even before they interacted with them and thus, the first set
of rank orders was not based solely on perceptual characteristics.

This possibility is illustrated by the children's spontaneous comments.
When presented with the marble toy in which marbles were visible but no
marbles were released, many of the children said it was broken. This

comment was made even before they had seen the toy that did dispense

marbles. The same comments were made in reference to the magnets that
looked like the hinges but did not open. It appears that the children
had expectations of what the objects should do just by looking at them.

43

This is consistent with Nelson's (1974) view that perceptual character-
istics are used to predict the object's function, and with Rosch and
Mervis' (1975) argument that perceptual and functional characteristics
are difficult to separate since, in reality, they are strongly
correlated.

The reason for females' better performance on name learning is
not clear. Although Nelson (1973) found that girls develop vocabulary
more rapidly than boys, experimental tests of comprehension between
the ages of two and four years have not found strong sex differences
(Dale, 1976). Since MLU did not differ between boys and girls in the
present investigation, girls' better performance on name learning
cannot be attributed to a more advanced stage of language production.
However, it is possible that the female subjects in this particular
investigation were more advanced in comprehension than the male sub-
jects. An alternative interpretation is that the female subjects
payed closer attention to the task resulting in more correct responses.
Another intriguing difference between males and females was the opposite
sign correlation coefficients between play time and accuracy of learning
the name for the wind-up toy. For females, a greater number of correct
responses was negatively correlated with play time whereas for males,
there was a positive correlation. The reason for the shorter play time
for females is not clear. Seven of the children did express fear of
the wind-up toy and six of these were females, although the binomial
test was not significant. Furthermore, if more of the female subjects
were indeed afraid of the toy, but did not express this explicitly, one
would expect the play time to be shorter during the first play period

when the toy was first encountered. However, the significant difference

44
between males and females was obtained during the second play period.
Moreover, the shorter play time did not hamper the acquisition of names
by the female subjects. It would seem that the amount of time spent
playing with an object is not crucial for acquiring its name.

The names for the marble toy and hinge/magnet toys were generalized
most frequently on the basis of perceptual characteristics. In contrast,
the wind-up toy differs from the other two toys in that its shape is
less well defined by the papier maché covering, perhaps resulting in
some children noticing the action more readily than the shape. Another
possibility for the greater number of responses based on function is
that when the objects were being demonstrated during the phase of the
experiment in which children were asked to select new instances of the
object category, it was possible to have all the wind-up toys moving
at the same time, whereas the other toys could only be demonstrated
singly even though they were all present for the children to see. Perhaps
it was difficult for some of the children to remember which toy performed
which function. This may have led them to generalize on the basis of
shape since this was the most obvious. On the other hand, they did not
have to remember the function of the wind-up toy since the toys were
moving right in front of them. Despite this, some children still
generalized on the basis of shape with respect to the wind-up toy. This
may be due to individual differences in cognitive ability but that can-
not be ascertained from the available data.

Other investigators have obtained similar findings with regard to
classification of objects. Bowerman (1978) pointed out that children
do not necessarily classify objects on the basis of one single principle.

In her study of overextensions, Rescorla (1980) also found that children

45
overextended words on several bases. If children do not always generalize
on the basis of one feature, then it would be useful to specify the
circumstances under which certain features are used. Saliency, as de-
fined in the present investigation, did not seem to be an influencing
factor.

On the basis of children's responses when asked to generalize to
new instances of an object category, the results are consistent with
Clark's theory which states that perceptual characteristics are the
basis of word acquisition. Both the marble and hinge/magnet toys were
generalized mostly on the basis of a perceptual characteristic (shape).
On the other hand, the wind-up toy was generalized on the basis of
a perceptual characteristic (shape) and on the basis of function.
However, the function of the wind-up toy was the action it performed.
Action or movement can also be considered a perceptual characteristic.
When viewed this way, even the name for the wind-up toy was generalized
on the basis of perceptual characteristics, albeit two different ones,
shape and action. Moreover, defining an object's function is not as
simple as it appears at first, especially since a child's notion of an
object's function may not always coincide with that of the experimenter.
For example, one child opened the hinge and used it like a gun. Another
child did not want to watch the wind-up toy move but rather wanted to
put it into the marble toy.

Nelson argues that children will identify new instances of a con-
cept by perceptual characteristics especially when the function is not
apparent. In the present experiment, the objects' functions were
demonstrated for the children but they still generalized the names based

on perceptual characteristics. However, one might argue that children

46

have already formed a concept when they are being asked to generalize
to new instances. When children are identifying concepts, predictions
on the basis of Nelson's theory are the same as that of Clark's theory.
The present study also showed that comprehension of the names varied
depending on the object, however, the reason for these results are
still unclear.

The finding that behavioral expressions of anxiety were related
to comprehension of object names indicates that the children's responses
were not a result solely of the task at hand, at least in the context
of an unfamiliar, experimental situation. Consequently, it may be un-
wise to assume that personality variables do not influence performance
on cognitive tasks.

Since the children responded differently to the various types of
objects, the characteristics of objects used in future investigations
of the basis of word meaning should not be ignored. Although the pre-
sent investigation was an attempt to obtain separate measures of per-
ceptual and functional salience, it is not certain this was accomplished
since perceptual and functional salience may have been correlated in
the same objects. In the future, it would be necessary to construct
objects which are alternatively high in one measure of salience and
low in the other. Furthermore, it will be necessary to specify what
is meant by function. Another aspect of this study which needs to be
explored further is the reason for children's better performance in
learning the name for the wind-up toy. A current investigation
(Della Corte, in progress) in which the names for the toys were randomly

assigned may help to clarify the issue. The role of anxiety also

47
should be considered in future investigations. Perhaps children who
are the least anxious will more accurately reflect the processes in-
volved in early word acquisition. As Ross and her colleagues indicate,
the process of word acquisition seems to be much more complex than

previously had been recognized.

APPENDIX

10.
11.
12.

13.
14.

15.

APPENDIX

Preschool Observation Scale of Anxiety

Physical complaint: Child says he or she has a headache, stomach-
ache, or has to go to the bathroom.

Desire to leave: Child says he or she wants to leave the testing
room or makes excuses about why he or she must leave; desire or
"need" to leave must be explicit.

Expression of fear or worry: Child complains about being afraid of
or worried about something; must use the word "afraid," "scared,"
"worried," or a synonym.

Cry: Tears should be visible.

Scream.

Whine or whimper.

Trembling voice.

Stutter.

Whisper: Child speaks softly, without vocal cords; should not be
a playful whisper.

Silence to one question in the interval.

Silence to more than one question in the interval.

Nail-biting: Child actually bites his or her nails in the testing
room.

Lip-licking: Tongue should be visible.

Fingers touching mouth area: Not counted if bites nails while
touching mouth.

Sucking or chewing object: Not fingernails.

48

l6.
17.
18.
19.
20.

21.

22.

23.

24.

25.
26.

27.

28.

29.

30.

49
Lip contortions.
Trembling lip.
Gratuitous hand movement at ear area.
Gratuitous hand movement at top of head.
Gratuitous hand movement at an object separate from body or at
a part of clothing separate from body.
Gratuitous hand movement at some part of body (not ear, hair,
mouth, or genitals).
Gratuitous hand movement.
Gratuitous leg movement.
Gratuitous foot movement: Below ankles, distinguish from foot
merely moving along with leg.
Trunk contortions (e.g., arching back).
Rigid posture: Part of body is held unusually stiff or motionless
for the entire 30-sec interval.
Masturbation: Touches genital area.
Fearful facial expression.
Distraction: Must be indicated by a verbal reminder by the
examiner to the child to pay attention.
Avoidance of eye contact: Examiner should be having clear trouble

making eye contact with child.

Glennon and Weisz, 1978

=___-_

REFERENCES

REFERENCE NOTE

Ross, 6., Nelson, K., Wetstone, H., and Tanouye, E. Acquisition and
generalization of novel object concepts by young language

learners, submitted to Child Development.

 

50

REFERENCES

Anglin, J.M. The child's first terms of reference. In N.R. Smith

and M.B. Franklin (Eds.), Symbolic functioning in childhood.

 

NJ.: Lawrence Erlbaum Associates, 1979.
Barrett, M.D. Lexical development and overextension in child

language. Journal of Child Language, 1978, 5, 205-219.

 

Bowerman, M. The acquisition of word meaning: An investigation into
some current conflicts. In N. Waterson and C. Snow (Eds.), The

development of communication. NY.: John Wiley and Sons, 1978.

 

Casby, M.W. Comparison of static form and dynamic action as the

basis of children's early word extensions. Kansas Working Papers

 

in Linguistics, 1979, 4, 71-77.

 

Clark, E.V. What's in a word? On the child's acquisition of
semantics in his first language. In T.E. Moore (Ed.), Cognitive

development and the acguisition of language. NY.: Academic Press,

 

1973.

Dale, P.S. Language development. NY.: Holt, Rinehart, and Winston,

 

1976.
Gentner, D. What looks like a jiggy but acts like a zimbo? Papers
and Reports in Child Language Development, 1978, 15, 1-6.

 

Glennon, B. and Weisz, J.R. An observational approach to the

assessment of anxiety in young children. Journal of Consulting

 

and Clinical Psychology, 1978, 46) 1246-1257.

 

51

52

Greenberg, J. and Kuczaj, S.A., II. Towards a theory of substantive
word meaning acquisition. In S.A. Kuczaj, II (Ed.), Language

development, Vol. 1, Syntax and Semantics. NJ.: Lawrence Erlbaum

 

Associates, 1982.
Gruendel, J. Referential extension in early language development.

Child Development, 1977, 48, 1567-1576.

 

Hilgard, E.R. Methods and procedures in the study of learning. In 5.5.
Stevens (Ed.), Handbook of experimental psychology. NY.: John

 

Wiley and Sons, 1951.
Leonard, L.B., Schwartz, R.G., Folger, M.K., Newhoff, M., and Wilcox,

M.J. Children's imitations of lexical items. Child Development,

 

1979, $9 19-27.

Nelson, K. Structure and strategy in learning to talk. Monograph of

 

the Society for Research in Child Development, 1973, 38, (1-2,

 

Serial No. 149).
Nelson, K. Concept, word, sentence: Interaction in acquisition and

development. Psychological Review, 1974, 81, 267-285.

 

Nelson, K., Rescorla, L., Gruendel, J., and Benedict, H. Early lexicons:
What do they mean? Child Development, 1978, 42, 960-968.

Reich, P.A. The early acquisition of word meaning. Journal of Child
Language, 1976, 3, 117-123.

Rescorla, L. Overextension in early language development. Journal of

Child Language, 1980, Z, 321-336.

 

Rosch, E. and Mervis, C.B. Family resemblances: Studies in the internal

structure of categories. Cognitive Psychology, 1975, Z, 573-605.

53

Thomson, J.R. and Chapman, R.S. Who is 'Daddy' revisited: The status
of two year olds' overextended words in use and comprehension.

Journal of Child Language, 1977, 4, 359-375.

 

Tomikawa, S.A. and Dodd, D.H. Early word meanings: Perceptually or

functionally based? Child Development, 1980, 51, 1103-1109.

 

Vygotsky, L.S. Thought and language. Cambridge, MA.: The M.I.T.

 

Press, 1962.

Werner, H. and Kaplan, 8. Symbol formation. NY.: John Wiley and

 

Sons, 1963.

   

llllllllllllllllllll'es
46 8270

0

mlllllllllllllllllllllllll
3 1293 03