Language comprehension strategies used by young children in script-routines

MISSING IMAGE

Material Information

Title:
Language comprehension strategies used by young children in script-routines
Physical Description:
viii, 138 leaves : ill. ; 28 cm.
Language:
English
Creator:
Harrison, Janet Snyder, 1954-
Publication Date:

Subjects

Subjects / Keywords:
Children -- Language   ( lcsh )
Comprehension in children   ( lcsh )
Attention in children   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1988.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Janet Snyder Harrison.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001093266
notis - AFH8851
oclc - 19413394
System ID:
AA00003785:00001

Full Text









LANGUAGE COMPREHENSION STRATEGIES USED BY
YOUNG CHILDREN IN SCRIPT-ROUTINES




















BY

JANET SNYDER HARRISON


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1988








































Copyright 1988

by

Janet Snyder Harrison












ACKNOWLEDGEMENTS

I would like to express my gratitude to Dr. Linda

Lombardino for her guidance in this investigation of

language comprehension. She provided invaluable informa-

tion for my studies of early language development. Her

trust in me stimulated my development as both an instruc-

tor and investigator.

Dr. Alice Dyson provided a calm perspective that

always helped me maintain the desire to continue. I am

grateful to her for her expertise in computer use and

writing style, as well as for her friendship. I am

indebted to Dr. M. Jeffrey Farrar for his knowledge of

scripts and developmental psychology. His timely arrival

at the University of Florida and his open-door policy to

me are deeply appreciated.

I am grateful to Dr. Thomas Abbott for his advice

throughout my doctoral program and for his steadfast

encouragement. Dr. Howard Rothman also served as both an

advisor and trusted friend. He gave me direction in the

study of acoustics and in recognizing that friends and

family are as important as professional commitments.


iii





Dr. Mary K. Dykes gave me valuable information and

stimulated my desire to approach my career as a transdis-

ciplinary team member. Dr. James Algina was an invaluable

committee member who willingly gave his time and expertise

to help with the research design and data analysis. His

kind guidance was extremely helpful.

I am indebted to the families who allowed their

children to participate in the study. The directors and

teachers at every day-care center were extremely coopera-

tive. I sincerely appreciate the perseverance and good

humor that Paul Thompson provided as he videotaped the

many hours of assessment.

Throughout the years, I have had many friends who

helped in many ways. Very special thanks are owed to Mary

& Richard Cummings, Cleo & Bob Hanson, and Nancy Haak. I

am grateful to my family, and most of all to my husband

Steve who has always believed in me.













TABLE OF CONTENTS


page

ACKNOWLEDGEMENTS.. .................................. iii

ABSTRACT ........... ..................... ............ vii

INTRODUCTION ..................................... .. 1

REVIEW OF THE LITERATURE ... ........................ 6

Receptive versus Expressive Language............... 6
Comprehension Strategies............................... 12
Sensorimotor Cognitive Development................. 17
Interactions of Cognitive, Social, and Language
Development.................... ........ ......... .. 19
Scripts and Language Development.................... 22
Assessing Language Comprehension................... 25
Summary.................... ...... ......... ......... 30
Statement of the Purpose ......o................... 31
Experimental Questions............................. 35

MATERIALS AND METHODS ............................... 37

Introductionr....... ............ .......... 37
Subjects....i.......................... ... ......... 37
Test Stimuli ........ ........ ......... .............. 38
Procedures.................... .............. ...... 42
Statistical Analysis..... .............................. 48

RESULTS .............................. ............ .... 49

Descriptive and Comparative Analyses............... 49
Strategy Use............... ...................... 58
Summary ................................ so ....... 79

DISCUSSION............... ...... ...... ... ...... ...... 83

Script Organization and Sentence Voice............. 85
Comprehension Strategies........................... 89
Methodological Issues.............................. 97
Conclusions.... ........... ................... 100








APPENDIX A COMBINATIONS OF ROUTINES................

APPENDIX B VERBAL STATEMENTS......................

APPENDIX C PARENT CONSENT FORM......................

APPENDIX D TEST MATERIALS............................

APPENDIX E SAMPLE ADMINISTRATION & SCORING FORM.....

APPENDIX F ERROR CLASSIFICATIONS....................

REFERENCES.... ............. .........................

BIOGRAPHICAL SKETCH..................................


105

106

124

126

127

130

131

137













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


LANGUAGE COMPREHENSION STRATEGIES USED BY
YOUNG CHILDREN IN SCRIPT-ROUTINES

By

Janet Snyder Harrison

April, 1988

Chairman: Linda J. Lombardino, Ph. D.
Major Department: Speech

The first purpose of this study was to investigate

the effects of script organization, sentence voice, and

event probability on language comprehension. The second

purpose was to describe the changes in comprehension

strategies that young children use when the level of topic

familiarity changes.

Sixty children between the ages of 2 and 5 years were

included in this study. The language comprehension tasks

were presented in four conditions: 1) probable events

with script organization, 2) probable events without

script organization, 3) improbable events with script

organization, and 4) improbable events without script

organization. Two major syntactic constructions, active

voice and passive voice, were presented in each condition.

The children were asked to enact the test stimuli using

vii





toy figures. The evaluation sessions were videotaped and

responses were judged for correct or incorrect interpreta-

tions and examined for patterns that indicated the use of

comprehension strategies.

Main effects for age and sentence type on language

comprehension were obtained. Total scores for sentence

comprehension showed significant increases with age, and

children at all age levels demonstrated better comprehen-

sion of active than passive voice sentences. Interactions

between script-organization and sentence voice were also

revealed. Higher scores were demonstrated for passive

voice sentences presented in the script condition. The

familiar context of the script organization may have

facilitated comprehension of the more difficult passive

voice.

Comprehension strategies used by the children in this

study varied with age. The child-agent strategy was used

most frequently by the youngest children, followed by the

probable-event strategy. The word-order strategy was used

most frequently by the 3- and 4-year-old children. Main

effects of script organization and event probability for

use of comprehension strategies were also revealed. The

early strategies of child-agency and probable-event were

used most frequently in the highly contextualized condi-

tions and the syntactic word-order strategy was used in

the least familiar contexts.


viii













INTRODUCTION

Paradigms for investigating early language comprehen-

sion should be founded on a theoretical framework that

includes contributions from the social and cognitive

domains of early childhood development. Such a paradigm

is needed to explore and understand the potential rela-

tionships among cognitive development, social development,

and the development of children's comprehension of

language. A description of the roles of each of these

domains could contribute to the development of effective

assessment and intervention procedures to be used with

language disordered children.

In recent years Nelson (1986) and her colleagues have

examined how scripts (Schank & Abelson, 1977), or mental

representations of routine events, contribute to cognitive

development. Observations of the language used by pre-

school children in their event descriptions led to the

more specific examination of the links between script

development and expressive language use (Carni & French,

1984; French, 1986; French & Brown, 1977). Application of

the script model to the study of receptive language

skills, although limited to date, provides a theoretical

base for the exploration of language comprehension.

1







Rumelhart (1980) has suggested that children may not

comprehend language when they lack the appropriate

schemata or organizational plan in which knowledge is

organized in memory. Comprehension of language also

involves memory and attention processes (Snyder & Downey,

1983). A distinction should be made between the categori-

zation of incoming speech sounds as members of particular

sound schemas (word recognition) and the mapping of

linguistic code onto meaning (word comprehension) (Hutten-

locher, 1974). Mental representations of communicative

interactions are described by Snyder and Downey (1983) as

the result of incoming information and an individual's

interpretation of the interaction.

Schank and Abelson (1977) have studied how people

organize knowledge as a means of understanding and knowing

the appropriate behaviors to use in particular situations.

They used the term script to refer to a structure that

describes appropriate sequences of events in a particular

context.

A script is made up of slots and requirements about
what can fill those slots. The structure is an
interconnected whole, and what is in one slot affects
what can be in another. Scripts handle stylized
everyday situations. They are not subject to much
change, nor do they provide the apparatus for
handling totally novel situations. Thus, a script is
a predetermined, stereotyped sequence of actions that
defines a well-known situation. (p. 41)

Schank and Abelson (1977) described the script concept as

a convergence of psychology, artificial intelligence, and








linguistics. Their original work involved the use of a

computer program that understood script-based situations

and primarily focused on a restaurant script.

Schank and Abelson (1977) developed the script

concept from the analysis of adult knowledge structures.

However, the development of script formats, or general

event representations, has more recently been studied in

preschoolers (Nelson, 1978; Nelson, Fivush, Hudson, &

Lucariello, 1983; Nelson & Gruendel, 1981). This work has

documented knowledge of scripts in children as young as 3

years old, and has outlined developmental changes in the

complexity of young children's scripts. The script

concept appears to be a structure that may allow inves-

tigators to more clearly discern the relationships among

social, cognitive and linguistic development. Nelson

(1986) has suggested that scripts may provide a cognitive

context for familiar situations and that the script model

of event representations is a useful analytic tool for

research in the cognitive, social, and language domains.

Procedures used in studies of script development in young

children have included enactment of routines using toys as

props, and an interview technique in which children were

asked to verbally describe what happens during familiar

situations (French, 1986; Nelson, 1978; Nelson & Gruendel,

1979, 1981; O'Connell & Gerard, 1985).







Bates, Bretherton, Shore, and McNew (1983) have

applied the script model to early language development and

described the script as the constant unit in the emergence

of naming skills. They proposed that script knowledge

serves as the foundation for the pre-referential period of

symbolic development. Scripts may also serve as a

framework for the development of early language comprehen-

sion. Although expressive language used by preschool

children in scripts has been discussed (Carni & French,

1984; French, 1986; Nelson, 1986), the role of scripts in

children's comprehension of linguistic structures has not

yet been investigated. If the primary foundation of

naming is script-knowledge, then a relationship between

language comprehension and children's knowledge of

familiar routines may also exist. Event knowledge may

facilitate comprehension of linguistic structures if it

serves as a familiar context that permits increased

processing capacity for language comprehension. Event

knowledge could also interfere with comprehension if

children allow the event knowledge to guide their actions

when they do not comprehend the language they hear. An

interference effect would only be apparent when the

meanings expressed by language conflict with the expecta-

tions of a familiar event. A descriptive and comparative

research study was designed to compare young children's

comprehension of sentences presented in script formats






5

with comprehension of sentences presented without a

familiar script context.












REVIEW OF THE LITERATURE

The vast amount of literature available on the topic

of language development considerably exceeds the scope of

this project. Therefore, this review is limited to

research in which links between early cognitive or social

development and language comprehension have been revealed.

Studies that include descriptions of methods used to

assess language comprehension are also included. This

chapter is organized under the following headings:

receptive versus expressive language; comprehension

strategies; sensorimotor cognitive development; inter-

actions of cognitive, social, and language development;

scripts and language development; and assessing language

comprehension.

Receptive versus Expressive Language

A considerable body of research has been directed

toward the description of the relationship between

expressive language development and receptive language

development. Language researchers have studied how

children understand and produce grammatical features in

speech by comparing the comprehension abilities of young

children with their production of sentences comprised of

various grammatical contrasts (Fernald, 1972; Fraser,









Bellugi, & Brown, 1963; Lovell & Dixon, 1967). Imitation,

comprehension, and production tasks have been used to

compare the language that the children understood with the

language forms that they used. Several methodological

questions have been raised in response to the results of

these studies. Fraser et al. (1963) used pictures in

receptive and expressive experimental tasks. The scoring

of novel responses as errors when the children described

the pictures may have artificially inflated the number of

errors on the expressive task. Fernald (1972) replicated

this study with a revised scoring procedure,' and the

results of his work showed no significant difference

between performance on the comprehension and production

tasks.

Cocking and McHale (1981) suggested that the investi-

gation of comprehension and production processes in

language acquisition has been confounded by the results of

studies that regarded objects and pictures as equivalent

stimuli in assessment tasks. They conducted a study to

investigate the language comprehension and production

skills of 68 4- and 5-year-old children, as demonstrated

in tasks using object or picture stimuli. The authors

hypothesized that preschool children between 4 and 5 years

of age would perform better on object tasks than on

picture tasks, and that the children would perform better

on a language comprehension task than on a language








production task comprised of the same set of linguistic

structures. Cocking and McHale (1981) also hypothesized

that an object-choice condition for either comprehension

or production would be harder than an object condition in

which there was no choice between alternatives. Their

study was designed to control the stimulus materials,

match the linguistic structures for comprehension and

production functions, and use linguistic structures

representative of English. The first task involved verbal

completion of stories that were acted out by the experi-

menter with toy props (production/object task). In the

second task, the children were required to verbally

complete a story depicted in a picture (production/picture

task). The third expressive language task was identical

to the production/object task with an added object-choice

condition. That is, the children were required to select

the appropriate objects for story completion from an array

of four to six props. The comprehension tasks were

analogous to the production tasks. The children were

required to either choose a picture corresponding to the

examiner's verbal description, or enact the verbal

directions using the toys. In a comprehension

object/choice task, the children were required to choose

from an array of four to six objects before demonstrating

their understanding by performing the action with the

objects. Results of this study were analyzed to compare








the medium of the stimuli and the types of responses. The

children demonstrated an overall superior performance on

object assessment tasks as compared to their performance

on picture assessment tasks. The object/choice condition

was the most difficult in both the production and compre-

hension tasks. Results of this study also supported the

authors' hypothesis that language comprehension tasks are

performed more accurately than language production tasks

among 4- and 5-year-old children. Cocking and McHale

(1981) suggested that the individual differences that the

children demonstrated in picture comprehension abilities

may have been the result of their varying acceptance of

social conventions. Cultural or social influences on

picture comprehension were explored in an analysis of

Peabody Picture Vocabulary Test stimuli by Chittendon

(cited in Cocking & McHale, 1981) who revealed that lower

socioeconomic status (SES) and middle SES children had

different interpretations for some of the social conven-

tions represented in the pictures. The recognition that

children's knowledge of social conventions may influence

picture comprehension tasks raises questions regarding how

children's script knowledge may affect their language

comprehension.

The hypothesis that comprehension of syntactic form

precedes production was challenged by Chapman and Miller

(1975) who concluded that strategies for comprehending







sentences lacking context are different from the strat-

egies used for producing sentences in context. They

tested 15 children ranging in age from 1 year, 8 months to

2 years, 8 months. The authors examined comprehension

without the support of a referent situation and compared

this to the ability to appropriately order subject-object

sentences in an expressive language task. The comprehen-

sion task used an enactive procedure in which the children

picked two relevant toys from an array of six, and

demonstrated the actions of sentences that were read by

the examiner. The stimulus sentences were semantically

reversible, such as 'The boy is hitting the girl' and the

reversed form of the sentence 'The girl is hitting the

boy.' In the expressive language task, the children first

watched an experimenter perform an action with two of the

six toys and then described the action in a sentence. The

children maintained correct word-order for subject and

object more consistently in the expressive task. The

authors examined patterns of performance for comprehension

of sentences without context and described two strategies

that the children may have used. They proposed an

animate-agent strategy as one possible explanation for the

misinterpretation of sentences such as 'The car is pushing

the boy.' Linguistically less advanced children tended to

interpret the animate noun (e.g., 'boy') as the subject of

the sentence and assigned object status to the inanimate









noun (e.g., 'car'). A lexical-semantic strategy was

observed to be used when the children assigned agent

status to the animate noun and object-of-action status to

the inanimate noun in a sentence such as 'The car is

chasing the dog.' Chapman and Miller described a prob-

able-event strategy that they proposed the children used

for the comprehension of sentences within referential

situations. They suggested that the children matched the

nouns and verbs of each sentence with their contextual

referents and inferred the relations among the lexical

items on the basis of the events witnessed. For example,

a contextual cue for agent status was available when the

children observed which object was moved first.

Chapman and Kohn (1978) further examined comprehen-

sion strategies in 2- and 3-year-old children using a

similar enactment procedure. They found that only 3-and-

a-half-year-old children consistently used a word-order

strategy. An examination of response preferences revealed

that the relative size of the objects used in the enact-

ment task was not a significant factor. Results of this

study also revealed that comprehension was more closely

related to age and short-term memory span than to the

subjects' mean length of utterance.

These investigations described comprehension strat-

egies as related to word-order or the probability of real-

world events. That is, authors of these studies suggested







that 2- and 3-year-old children may decode subject-object

relations lacking referential support on the basis of a

lexical-semantic strategy that assigns animate nouns to

subject status and inanimate nouns to object status.

However, replications and extensions of the early studies

provided support for a probable-event strategy in deter-

mining subject-object assignment. The proposed probable-

event strategy suggests that children's real-world

knowledge may serve as the basis for early comprehension

before syntactic decoding skills develop. When children

do not understand the message encoded by language, they

may respond to the familiar aspects of the environment or

situation. The script model will be used in this study as

a format for the investigation of the effects of real-

world knowledge on early lexical comprehension.

Comprehension Strategies

Chapman (1978) stated that comprehension is the

complete process of understanding a sentence, whereas "a

comprehension strategy is a short-cut, heuristic, or

algorithm for arriving at sentence meaning without full

marshaling of the information in the sentence and one's

linguistic knowledge" (p. 310). Thus, children may use

comprehension strategies as devices to facilitate under-

standing when they lack full linguistic competence.

Chapman (1978) proposed that children at successive

developmental levels use different comprehension








strategies. The strategies that she identified are

underlined in the following discussion. Children who are

functioning at Sensorimotor Stage VI may locate the

objects mentioned, and do what you usually do, or act on

the objects in the way they were mentioned. In the early

preoperational period, from approximately 2 to 4 years of

age, children similarly do what is usually done. Their

increasing world-knowledge allows them to select probable

locations for spatial displacement tasks and to supply

meanings that their own past experience make most probable

(a probable event strategy). Chapman (1978) described

analyses of children's incorrect answers to questions that

revealed a supply-the-missing information strategy in

which children's past experience dictated the most

probable missing element. This strategy explains the

answer "cereal" that a 2- to 3-year-old child might give

in response to "How do you eat?", "Why do you eat?", or

"When do you eat?". Chapman proposed that during the

periods of late preoperations (4 to 7 years) and concrete

operations (7 to 11 years) children may use three addi-

tional comprehension strategies. A word-order strategy

may be used to interpret agent and object of action in

simple active sentences. When this strategy is used

agent-status is assigned to the first noun and object-

status is assigned to the second. Children may use an

order-of-mention strategy to comprehend sentences








containing the conjunctions before and after. A probable-

relation-of-events strategy may be used to interpret

conjunctions such as because, if, although, and but.

Chapman (1978) concluded that most children are not able

to use syntactic cues for understanding agent and object

status in simple sentences until they are approximately 5

years of age.

Corrigan and Odya-Weis (1985) explored the comprehen-

sion of semantic relations in children between the ages of

1:11 and 2:2 (years:months) to test the hypothesis that

awareness of semantic relations does not begin until after

children produce their first word combinations. In this

training study, children were grouped according to

language proficiency in order to compare the cognitive

categories of children who were producing semantic

relations with children who were not yet expressing them.

Stimulus materials consisted of three sets of pictures

depicting sentences with actors and objects varying in

animacy. The first set depicted prototypical sentences

with animate actors acting upon inanimate patients such as

'The boy pulls the boat.' The second set depicted

sentences with inanimate actors acting upon animate

patients such as 'The bike pulls the dog.' Sentences with

inanimate actors acting upon inanimate patients such as

'The can waters the flower' were depicted in the third set

of stimulus pictures. First, the children were trained to









place a red felt triangle on 'the one who is doing the ..

(action) ', and to put a green felt square on the object

of the action (patient). The children were then asked to

do the same task on generalization pictures. The token-

placement technique was designed to investigate the

influence of actor and patient animacy in determining

which sentences the children viewed as prototypical.

Corrigan and Odya-Weis (1985) concluded that the category

of actor is usually first acquired for sentences with

animate actors and inanimate patients.

Lempert (1985) examined preschool children's sentence

comprehension with respect to animacy in a procedure that

required the children to enact their interpretation of

passive and inverted cleft sentences. Each passive

sentence (e.g., 'The cow is pushed by the horse') was

transformed into an inverted cleft sentence such as 'It's

the cow that the horse pushes.' The examiner read each

sentence aloud, and the children demonstrated the action

with toys. Lempert (1985) concluded that 3-year-old

children used a dynamic agent strategy in which they

frequently assigned agent status to the more dynamic

referent. However, in sentences that included two dynamic

inanimate objects such as 'The truck is pushed by the

car', the 3-year-old children used a word-order strategy

and assigned the role of agent to the first-named inani-

mate. The 4-year-old children also used a word-order







strategy for inverted cleft sentences but used some

structural relations in their comprehension of passive

sentences. That is, they understood animate patient

passive sentences (e.g.,'The truck is pushed by the

horse.') but interpreted inanimate patient passive

sentences (e.g.,'The horse is pushed by the car.') as

active sentences. Lempert (1985) also suggested that the

probability of the event occurring in the real world

appeared to affect the children's interpretations of

animacy.

In a recent study by McClellan, Yewchuck, and

Holdgrafer (1986) it was revealed that their 10 subjects,

ages 2:2 to 2:10, appeared to use a probable-event

strategy in determining subject-object assignment.

Knowledge of likely relations between objects in the real

world interfered with performance on sentence types in

which semantically unlikely relations between subject and

object were expressed. These children were asked to

demonstrate comprehension of 48 active, reversible,

present tense sentences by performing the required actions

with toys. The verbs used in the construction of the

stimulus sentences were selected to allow both animate and

inanimate subjects and objects. Four types of sentences

were included: 1) animate subject and object ('Girl bump

dog'); 2) inanimate subject and object ('Car bump boat');

3) animate subject, inanimate object ('Girl bump car');








and 4) inanimate subject, animate object ('Car bump

girl'). The four sentence types were syntactically

identical, but differed in the semantic likelihood of

occurrence. Sentences with an inanimate subject and

animate object (e.g., 'Car hit boy') were more congruent

with children's experiential knowledge than sentences with

an animate subject and inanimate object (e.g., 'Boy hit

car'). Reversed assignment of subject and object occurred

with least frequency for sentences with an inanimate

subject and animate object and with greatest frequency for

sentences with an animate subject and inanimate object.

McClellan et al. suggested that what their subjects' knew

about the relations between objects and events in the real

world interfered with their performance on the comprehen-

sion task.

Collectively these investigations offer support for

the hypothesis that real world event knowledge serves as a

frame-work for early language comprehension. The extent

to which comprehension strategies aid in the comprehension

of later acquired structures, such as embedded and

conjoined sentences, has been identified as a need for

further research (Chapman, 1978).

Sensorimotor Cognitive Development

The examination of cognitive prerequisites for

specific language development has been another principle

area of study. Evidence for a relationship between








expressive language and stages of sensorimotor cognitive

development has been identified for the onset of words

(Corrigan, 1978), the semantic aspects of word use (Dihoff

& Chapman, 1977), the use of relational words (Gopnik &

Meltzoff, 1984; Tomasello & Farrar, 1986) and the onset of

two-word combinations (Corrigan, 1978; Folger & Leonard,

1978). Correlations between sensorimotor stages of

development and levels of comprehension also have been

explored by various language researchers. Miller,

Chapman, Branston, and Reichle (1980) conducted a cross-

sectional study of language comprehension in relation to

cognitive functioning in 48 10- to 21-month-old children.

They compared performance on sensorimotor tasks with

performance on a language comprehension task in which the

examiner asked questions and gave commands with no

gestural cues. Miller et al. found that Stage VI levels

of sensorimotor functioning were not necessary for the

development of the comprehension of one-and two-word

semantic roles that they assessed. These authors sug-

gested that their findings offered evidence of the need to

consider several parameters when interpreting children's

comprehension of comments or directions such as 1) the

presence or absence of referents; 2) the number of words

used in the directions; 3) the particular semantic

relation expressed; 4) the probability of the action in

the absence of the request; 5) the availability of









nonverbal cues; and 6) the presence of supporting context.

Although Miller et al. did not discuss the use of script

routines as an assessment procedure, enactment of scripts

may provide a context that allows greater control over

some of these factors.

Interactions of Cognitive, Social, and Language Development

Bates, Bretherton, Shore, and McNew (1983) discussed the

emergence of expressive language in a framework of cognitive

and social interactions. They described the script as the

constant unit in the emergence of naming skills, proposing

that script knowledge serves as the foundation for the pre-

referential period of symbolic development. During the first

stage of symbolic development, occurring at around 10 to 11

months of age, children have some organized expectations

about events and perform appropriate actions within a

familiar script, but they do not yet use portions of their

knowledge to represent elements of the script. Bates et al.

(1983) described a "vocal procedure like 'bam' used to

accompany knocking blocks in a well-learned game" (p. 111) as

a typical Stage I symbol. They indicated that objects serve

as the central organizers for the transition to Stage II in

which referential symbols emerge. During this second period,

occurring between 13 and 20 months, children use portions of

their script knowledge to represent a referential segment of

the script. Bates et al. suggested that during Stage II,

children use manual or vocal symbols to indicate some







20

specified segment within the whole script. The pre-existing

Stage I link to the script is used in a new way. For

example, the vocal procedure that previously represented the

entire activity of knocking down blocks may now serve as an

indicator of just the banging action. Stage III typically

occurs between the ages of 16 and 22 months and is charac-

terized by relational speech. During this period verbal

naming is used to identify relations among segments of the

scripts. According to Bates et al. (1983) "a case relation-

ship, already implicit in the child's knowledge, is recon-

structed at a symbolic level. acts of indirect reference

reflect an increasing 'distancing' of the symbolic routine

from its script base" (p. 112). Bates et al. concluded that

action names and relational terms that were the most diffi-

cult symbols at the earlier stages are more easily manipu-

lated at the naming level.

The role of social development in language acquisition

has been described by various investigators. Bruner (1975)

discussed the development of shared attention in joint

activities as a foundation for the development of communica-

tion. Bates (1979) observed that children are social

participants in various interactive contexts, and the

functions of their early communicative signals are seen in

early gestures and first words. Nelson (1986) suggested that

the context for learning is set as children directly exper-

ience the ongoing events of the real world and gain social









and cultural knowledge. The structure of event

representations may serve as a tool for understanding basic

characteristics of the child's cognitive system (Nelson,

1986). The interaction of social and cognitive development

was described by Lucariello, Kyratzis, and Engel (1986), who

have proposed that children form cognitive representations of

social-interactive events. These authors stated that:


Insofar as the child must represent the roles of others
and reciprocate with appropriate actions in order to
participate competently in these interactive formats, it
can be said that representations have been built up to
guide interaction in these exchanges. Event representa-
tions, then, can be thought of as the cognitive counter-
parts to formats or routines. (p. 139)



French (1985) has described real-world knowledge as the

basis for social and cognitive development with the premise

that mental representations of personally experienced events

constitute young children's primary knowledge base. Child-

ren's event representations (scripts) guide "their under-

standing of others' social roles, and their expectations

regarding what objects can be expected to be associated with

the event, and how they are to be used" (French, 1985, p.

182). Research to date has been directed toward revealing

the effects of scripts on recall and memory (Haberlandt &

Bingham, 1984; Mistry & Lange, 1985; Nelson, Fivush, Hudson,

& Lucariello, 1983; Walker & Yekovich, 1984) and as a basis

for categorical structures in semantic memory (Lucariello &

Nelson, 1985).







22

As noted earlier, 2- and 3-year-old children typically

use a probable-event strategy to interpret sentences that

exceed their level of syntactic comprehension (Chapman, 1978;

Lempert, 1985; McClellan, Yewchuck & Holdgraffer, 1986).

Bauer and Shore (1987) found that event-familiarity appeared

to affect immediate and delayed recall abilities of 21-month-

old children. The relationship between items in a sequence

also influenced the children's recall of events. The

children demonstrated superior recall for items within

sequences that were linked by causal or enabling relations,

and significantly poorer recall of novel sequences charac-

terized by arbitrary relations between items. For example, a

causal or enabling relation exists between the action of

drying and the action of dressing in a script for taking a

bath. An arbitrary relation exists between actions such as

putting on a shirt and putting on pants because the order of

these events may be reversed. Although Bauer and Shore did

not assess receptive language, it is feasible that compre-

hension of linguistic structures that describe causing or

enabling relations may be superior to comprehension of

linguistic structures that describe arbitrary relations.

Scripts and Language Development

The interaction of language and script development has

been preliminarily described by investigators of early

childhood development. Evidence for script development in

children has been described by Nelson (1978, 1986), Nelson








and Gruendel (1979, 1981), and Slackman and Nelson (1984).

Nelson and Gruendel (1981) used a verbal interview format to

explore evidence for children's script-like knowledge. Eight

children within the age range of 48 to 61 months were

interviewed to reveal knowledge of three situations: eating

lunch at a day care center; eating dinner at home; and eating

at McDonald's fast-food restaurant. The children were asked

a sequence of questions to reveal their knowledge about these

routine events and were then interviewed a second time with

toys available to encourage acting out of scripts. According

to Nelson and Gruendel (1981) the children typically reported

the acts of each event in the correct sequence and generally

focused on the event that was specified in the interview.

Nelson (1986) summarized studies that have been con-

ducted to determine how scripts develop relative to the

amount of experience a child has with an event. In general,

children with prior experience give longer accounts of

events. Further, the development of a script appears to

proceed from a skeletal structure to a gradually more

elaborated script. Nelson (1986) has also described the

language terms that children used in describing activities.

She stated,


We had observed in the first studies that children
tended to use the timeless present form of the verb
rather than the past tense and also used the general
'you' form, as in 'you eat.' Although these forms are
appropriate to general timeless accounts, their use by
preschoolers had not been previously reported. (p.32)






24

Lucariello and Nelson (cited in French, 1985) examined 24- to

29-month-old children's responses to their mother's question-

ing in three different contexts: 1) free play; 2) novel

play; and 3) a routine event such as eating or bathing. They

found that the children's highest proportion of appropriate

responses occurred during participation in a routine event.

These authors also proposed that children's own scripts for

routine situations guide their understanding and use of

language within such situations. They suggested that when

children participate in unfamiliar situations, most of their

processing capacity is required by the ongoing activity and

little is available for language use and comprehension.

Lucariello, Kyratzis, and Engel (1986) further examined

how event representations support the use of language and

proposed that the actual process of meaning acquisition

develops through mother-child interactions as they occur

within event contexts. In a summary review of script

research, French (1986) described the verbal protocols

provided by preschoolers as evidence of both cognitive and

linguistic abilities. Relational terms such as before,

after, because, so, if, or, and but have been used by

children in their verbal descriptions of scripts at an

earlier age than previous research had suggested. Carni and

French (1984) revealed evidence that comprehension of the

relational terms before and after proceeds in a context-

sensitive manner. Comprehension initially appears to be








highly dependent on semantic context, and the ability to

display comprehension is influenced by the particular

paradigm used.

Carni and French (1984) assessed the degree to which

comprehension of relational terms depends upon the child's

familiarity with the specific relationship expressed. The 32

children included in this study were between the ages of 3:2

and 4:11. These subjects were required to answer questions

that referred to pictured event sequences having invariant

and arbitrary real-world temporal orders, such as shopping in

a grocery store. The 4-year-old children demonstrated

comprehension of both types of sequences; however, the 3-

year-old children only performed well with the invariant or

script-like sequences. Carni and French (1984) concluded

that it is not lexical knowledge that develops within this

age range, rather it is the ability to coordinate lexical

knowledge with the cognitive demands of responding to

questions referring to arbitrary sequences.

Assessing Language Comprehension

Factors that must be controlled when testing language

comprehension have been highlighted by the previously

described studies. A basic distinction must be made among

the definitions of comprehension that are used by various

investigators. The meaning of the term comprehension has

varied from highly specific descriptions to characterizations

that are too broad to be scientifically quantified (Lord,








1985). Chapman (1978) has defined comprehension as the

complete process of understanding a sentence. This defini-

tion was used in most of the investigations of comprehension

of word-order and syntactic form. Bransford and Nitsch

(1978) described comprehension as a "function of the recipro-

cal relationships between inputs and one's current cognitive-

perceptual situation" (p. 298) and suggested that the minimal

unit of analysis must be a situation plus an output. Yet the

few evaluation tools that assess language comprehension of

children younger than 3 years are primarily based on picture-

identification as a mode for testing comprehension. For

children under the age of 2 years videotaped observations and

maternal interviews have been used to supplement formal test

data obtained on comprehension abilities (Bretherton, McNew,

Snyder, & Bates, 1983; Snyder, Bates, & Bretherton, 1981).

Golinkoff, Hirsh-Pasek, Cauley, and Gordon (1987) tested

comprehension of nouns, verbs, and word-order in a paradigm

that relied on differential visual fixation of simultaneously

presented video events. Young children between the ages of

1:4 and 2:4 were seated on their mothers' laps as videotapes

were presented simultaneously on two side-by-side video

monitors. A message that matched only one of the video

displays was played through a speaker. For example, the

auditory message, 'Which one is drinking?', was played as one

video screen showed a woman blowing on a sheet of paper and

the other video screen displayed a woman drinking from a








coffee cup. Hidden observers recorded the infant's visual

fixation, or total time that the infant watched each video

display. A correct response was determined by whether the

matching video screen was watched longer and more quickly

than the non-matching screen. In the first of a series of

three experiments, Golinkoff et al. investigated whether

infants would search for the video screen that matched an

auditory message to search for a noun such as 'Where's the

cookie?.' Verb comprehension was assessed in the second

experiment. The infants were required to watch dynamic events

depicted on the video screens that matched the linguistic

stimuli. The same paradigm was used in the third experiment

to examine comprehension of word-order. These authors

suggested that this series of controlled tasks more accurate-

ly tested language comprehension because the children were

not required to perform overt actions and the video represen-

tations of dynamic events were naturalistic. Enactment tasks

using toy props have been used in studies of symbolic play,

language development, and script development. Bower (1978),

in a response to Nelson's early studies of children's script

development, criticized the use of toy props while examining

children's descriptions of routine events. He questioned

whether the influence of the pretend world created by

enacting scripts with toys might affect expectations of the

scripts. Bates, Bretherton, Shore, and McNew (1983) have

provided a partial answer to this question. To examine vocal








and gestural development in 20- to 28-month-old children,

they used elicited symbolic play tasks in which the level of

abstractness of the objects used was varied. Scripts for

having breakfast, putting a teddy bear to bed, and giving a

doll a bath were modeled for the children. The scripts were

presented at three levels of abstractness. In the first

level, realistic toy objects were used. A wooden cylinder

was substituted for toys in the scripts at the second level.

An inappropriate object was used in place of the toys for the

most abstract level of each script. Although these investi-

gators were interested primarily in comparing the lexical and

grammatical measures of play obtained in each of the three

conditions, the degree to which each child's imitation

preserved the order of the elements in the scripts was also

measured. No differences were reported for the ordering of

elements of scripts across the varying levels of abstraction.

Methodological considerations were described by O'Con-

nell and Gerard (1985) in an investigation of temporal

sequencing abilities that used enactment tasks with young

children. They used an elicited imitation enactment task in

which children between the ages of 20 and 36 months were

required to enact six scripts or routine events. They

concluded that sequencing abilities develop in an orderly

progression that suggests an interaction between familiarity

of the sequence and general constraints on information-

processing capacities. The youngest children in their study








were unable to reproduce the order of a three-act sequence,

regardless of the meaningfulness of the modeled event.

Although ability to use ordered information increased between

the ages of 24 and 28 months, the 28-month-old children

adhered to the conventional order of familiar events,

indicating that their organizing principle was dependent on

the familiarity of the sequences. Only the 36-month-old

children demonstrated an ability to reproduce the scrambled

or reversed sequences that were modeled.

O'Connell and Gerard (1985) conducted a pilot study to

identify the routine events familiar to children within the

20- to 36-month age range. They identified six events: 1)

taking a bath; 2) going to the grocery store; 3) eating a

snack; 4) cooking; 5) going to the toilet; and 6) going to

bed. Recognizing that there are family variations in

specific routines, O'Connell and Gerard included only the

most prototypical acts from each routine event. These

authors also noted that the use of three-dimensional objects

in an enactment task to assess children's sequencing skills

appeared to be a more sensitive method than picture sequenc-

ing. The 3-year-old children in the O'Connell and Gerard

study reproduced reversed sequences using objects while 4-

and 5-year-old children were unable to reproduce the order of

reversed picture sequences in an earlier study by Fivush and

Mandler (cited in O'Connell & Gerard, 1985).








Summary

The preceding literature review included descriptions of

research in cognitive, social, and language development.

Descriptions of event representations, in which social

cognition is recognized as a possible context for language

development, support the use of the script-model as a

theoretical framework for the assessment of early language

comprehension. Enactment procedures using toy props have

been successfully used in studies of script development,

symbolic play, and language comprehension. Use of the

enactment procedure in the assessment of language comprehen-

sion within scripted and nonscripted routines could provide a

paradigm that controls the social and cognitive contexts

within which comprehension is assessed. If scripts serve as

contexts that minimize cognitive demands and facilitate

language comprehension, then differential performance may be

observed on language comprehension tasks in which event

probability and script organization are controlled. Addi-

tionally, evidence has been presented that supports child-

ren's use of comprehension strategies to interpret sentences

that exceed their level of syntactic comprehension. Children

may also use real-world knowledge to guide their actions when

they lack comprehension of syntactic structures. The degree

to which event familiarity or script organization facilitates

language comprehension may be a developmental issue.








Statement of the Purpose

The relationship between world knowledge and language

comprehension has not yet been fully explored. The

purpose of this research was to examine the role of world

knowledge in children's comprehension of linguistic

structures, particularly as it relates to script-based

knowledge. A methodology for examining script-based

language comprehension in young children has been devel-

oped. Although scripts have been proposed to serve as

cognitive contexts for early language development (Bates,

Bretherton, Shore, & McNew, 1983; Nelson, 1986), research

is needed to describe the development of script-based

language comprehension. The procedures used in this study

were designed to facilitate a description of young

children's language comprehension abilities in contexts

that vary in their degree of familiarity. If scripts

serve as contexts that minimize cognitive demands, then

performance on language comprehension tasks presented in

familiar scripts should result in better performance than

comprehension of the same linguistic structures without

script presentation. This study is designed to explore

the relationship between the development of language

comprehension and real world knowledge.

In defining real world knowledge, a distinction must

be made between script-knowledge and probable event-

knowledge. A script has an ordered sequence of actions









that are appropriate to a particular spatial-temporal

context (Schank & Abelson, 1977). The actions are

organized around a goal and the actors, actions, and props

used to carry out the goal are specified. For example, a

young child's script for having lunch at school may

include the teacher who calls the children and gives them

food, as well as the actions of eating and cleaning up,

all of which fulfill the goal of eating lunch. Scripts

often have embedded subscripts that may occur, but do not

necessarily occur in all circumstances. For example, in a

lunch script, the children may sometimes go to another

room to get their lunches. Therefore, script-knowledge is

defined as awareness of the actors, actions, and props

that are typically involved in a particular routine, an

expectation that the primary events will occur, and an

understanding that secondary scenes may occur. The

hierarchical nature of script-knowledge provides inferen-

tial power (Nelson, 1986); that is, the necessary compo-

nents of a script are predictable, and the secondary

scenes are optional components that fit the script user's

expectations. Event-knowledge is defined as awareness of

the probability that some events are more likely to occur

in the real world than are other events. It is similar to

script-knowledge in that it is also the result of exper-

ience with the real world. However, event-knowledge

differs from script-knowledge because it includes only the








expectations associated with specific actors, actions or

objects. For example, when a ball is a prop in a scene,

children know from real world experience that the ball is

likely to be hit, thrown, bounced, or rolled. The term

event-knowledge is used in reference to the expectations

that apply to the actors, actions, and props within a

secondary scene. Event-knowledge does not invoke an

entire script as a cognitive context; therefore, event-

knowledge lacks inferential power and does not include

expectations that any other primary event will occur.

In accordance with the proposal that scripts serve as

cognitive contexts that facilitate language development

(Bates, Bretherton, Shore, & McNew, 1983; Nelson, 1986),

young children's comprehension of linguistic structures

should be affected by their script-knowledge as well as by

their probable event-knowledge. Script knowledge should

facilitate comprehension of linguistic structures pre-

sented in sentences that are organized within a script

format. Event-knowledge should facilitate comprehension

of linguistic structures presented in sentences describing

probable events that are not organized in a script format.

Performance on language comprehension tasks that require

young children to interpret sentences that describe

improbable events embedded in non-script formats should

most accurately reveal the children's levels of syntactic









comprehension because here linguistic comprehension is

decontextualized.

Previous investigations of the development of

language comprehension have revealed that when children

lack understanding of the syntactic form of a sentence,

they frequently used a probable-event strategy (Chapman,

1978; Lempert, 1985; McClellan, Yewchuck, & Holdgraffer,

1986). Use of this strategy results in inferior perform-

ance on language tasks in which the event specified

conflicts with the subject's knowledge of the probable

event. For example, the sentence 'The table hits the boy'

describes an improbable event. The use of a probable-

event comprehension strategy results in an incorrect

interpretation of the sentence (i.e., 'The boy hits the

table). Children have also been reported to use a word-

order strategy in which the word order, or order of

mention serves as a cue for interpretation of agent and

object status. Use of this strategy results in inferior

performance on language tasks in which the order of the

sentence structure is reversed, as in a passive voice

sentence such as 'The table is bumped by the pig.'

Children at an earlier level of symbolic development have

also been reported to use a child-agency strategy in which

they perform actions without involving demonstration with

another object. For example, children who use the child-

agent strategy might pretend to eat the cookie when asked








to demonstrate 'The bear eats the cookie.' The design of

this study allows investigation of the effects of script-

knowledge and event-knowledge on the use of comprehension

strategies.

Experimental Questions

The study was designed to answer the following

experimental questions:

1. How does language comprehension vary for sentence

structures presented in the context of script

routines, non-script routines, improbable events, and

probable events in 2-, 3-, and 4-year-old children?

2. How does comprehension of passive sentences compare

to comprehension of active sentences in children who

are 2-, 3-, and 4-years of age?

3. What comprehension strategies are used to interpret

sentence structures presented in the context of

script routines, non-script routines, improbable

events, and probable events by 2-, 3-, and 4-year-old

children?

4. How are error strategies affected by the contexts of

script-organization, event probability, and sentence

structures?

The language comprehension tasks were presented in

four conditions: 1) probable events with script organiza-

tion, 2) probable events without script organization, 3)

improbable events with script organization, and









4) improbable events without script organization. Two

major syntactic constructions, active voice and passive

voice, were presented in each condition. Comparison of

children's performance on the two sentence types in four

conditions of world-knowledge was made to obtain informa-

tion regarding the following hypotheses:

1. Children at all age levels will demonstrate highest

performance levels of language comprehension for

sentence structures presented in script formats.

2. Children at all age levels will demonstrate higher

performance levels of language comprehension for

sentences that describe probable events than for

sentences that describe improbable events.

3. Children at all age levels will demonstrate higher

performance levels of language comprehension for

active voice sentences than for passive voice

sentences.

4. Children at all age levels will demonstrate different

use of comprehension strategies in the four exper-

imental conditions.













MATERIALS AND METHODS

Introduction

The purpose of this study was to investigate the

possible effects of script organization and event proba-

bility on language comprehension. An experimental design

was used that allowed control of the presentation of

active voice and passive voice sentences in two levels of

script organization and in two levels of event proba-

bility.

Subjects

Sixty children served as subjects in this study, 18

between the ages of 26 months and 36 months, 21 between

the ages of 38 and 49 months, and 21 between the ages of

50 and 61 months. Six males and 12 females were included

in the 2-year-old range, 13 females and 8 males were

included in the 3-year-old range, and 9 females and 12

males were included in the 4-year-old range. Participants

were located through play groups, nurseries, or day-care

centers in the Gainesville, Florida, area. Subjects

passed a hearing screening of pure tones presented at 30

dB for the frequencies of 500 Hz, 1000 Hz, 2000 Hz, and

4000 Hz. All subjects were monolingual and spoke English

as their native language. Children who had previously









been referred for speech, language, or mental aptitude

testing were excluded from participating in this study.

The Peabody Picture Vocabulary Test-Revised (Dunn & Dunn,

1981), Form L, was administered to subjects who had met

the eligibility criteria. Children who did not pass the

hearing screening, or who scored below the fourth stanine

on the Peabody Picture Vocabulary Test-Revised, were not

included in this study.

Test Stimuli

A total of 16 possible routines, or sets of test

stimuli, were developed. Eight of the sets were organized

within script formats. The script formats were based on

four themes: cooking, eating, taking a bath, and going to

bed. Each theme was assessed in two script formats:

probable event format and improbable event format. The

eight nonscript routines were also developed with four

probable event formats, and four improbable event formats.

Sentences used in the script formats were scrambled for

both order and theme to develop the nonscript formats.

For example, if children were tested with the script-

probable formats for the themes of cooking and taking a

bath, and script-improbable formats for the themes of

eating and going to bed, then they received two nonscript

formats developed from the improbable versions of the

themes of cooking and taking a bath, and two nonscript

formats developed from the probable versions of the themes








eating and going to bed. The nonscript formats were com-

prised of the same sentences included in each script, but

were presented in random order without script organiza-

tion. The sentences were presented in the same alter-

nating pattern of active, passive, and nonscored sen-

tences. Six possible combinations of the script and

nonscript routines were developed. Each child received

one combination of the script and nonscript formats.

Administration of the combinations was randomly ordered

(see Appendix A).

Each set of test routines was composed of 10 verbal

statements. Six of the statements served as actual test

stimuli. Three of the test stimuli within each set were

active voice sentences, and three were passive voice

sentences. Figure 1 depicts the organization of the test

stimuli within the experimental conditions. The remaining

four statements did not serve as stimuli but were included

to contribute to the script organization. These nonscored

statements were also included in the nonscript formats to

maintain a constant length across routines. The nonscored

statements were not presented in an order that developed a

script theme when included in the nonscript formats.

Verbal statements accompanying the four script-

probable routines were constructed to describe events that

are typically expected within each script and events that

may logically occur within a particular script. All of






40






to =- 4


E
c
o







<0 e *





E -Y u
4 -- 4





= -




hT ..4
S- u


(. 0





0 0 it
E a t o -I r4








0 2/ -4 -
E 4 1 8 2






S'-





S0 r.
t rua ------O)- a
SE
o














p a S,-...
(B \/ 4
i-\ > "
\ ~ /g\< 3
\ / ^\^ S >








the actions described were performed with the appropriate

objects. For example, 'The light is turned off by the big

rabbit' was a passive sentence included in the script-

probable version of the theme 'going to bed.' The verbal

statements accompanying the four script-improbable

routines reflected theme organization, but the test

stimuli described events and actions that were incongruous

with the script or required actions to be performed with

unlikely objects. For example, 'The light is hugged by

the big rabbit' was a passive sentence included in the

script-improbable 'going to bed' routine.

The nonscript routines were comprised of a series of

statements that described characters performing actions

that were appropriate for particular objects. These

routines did not have script organization; therefore, the

verbal statements describing these events were not

temporally or thematically organized. The presence of a

particular action did not suggest the occurrence of any

other event. The four nonscript-probable routines

included statements that described actions that could

occur with the specified actors or objects. For example,

the verbal statements accompanying the nonscript-improb-

able routines described actions that were unlikely to

occur because of the characteristics of the objects

involved. Inanimate objects were sometimes described as

performing actions, or toy characters were described as








performing actions that were improbable. For example, an

active voice sentence from a script-improbable format

required a toy animal to 'eat a bib' instead of the

logical action of wearing a bib.

To test the validity of the probable and improbable

categorizations of the statements, the test stimuli were

rated by 12 graduate students in speech-language pathol-

ogy. Only items that were rated with at least 80%

agreement were retained. Items that did not meet this

criterion were revised. The verbal statements for all 16

routines are included in Appendix B.

Procedures

Parent permission was obtained for each subject

included in this study. The purpose and procedures of the

study were given to the parents of each subject (see

Appendix C).

Out of the 16 possible routines, each child enacted a

total of eight routines: two script-probable routines,

two script-improbable routines, two nonscript-probable

routines, and two nonscript-improbable routines. The

administration order was counterbalanced across subjects.

The test materials included toy objects, stuffed animals

and toy characters that varied from 4" to 12" in size (see

Appendix D).

Testing was conducted in a quiet room in the center

that each child attended or at the University of Florida









Speech and Hearing Clinic. The test environment consisted

of a table of the appropriate height for the child's

comfort, two chairs, and the toys that were needed to

enact each script. The examiner was seated opposite to

the child at the table, and kept toys that were not being

used out of the child's line of vision. A student

assistant operated a video camera at a distance from the

table that allowed a clear view of the child's manipu-

lation of the toys.

Training Procedure

The examiner introduced the routines by saying, "I'm

going to tell you about some different toys. I'll show

you the toys and then tell you about them. You show me

what the toys do. Let's try a little story now." Because

a variety of toy animals were used in the routines, the

examiner used four representative toys in the training

items. The instructions for the training items are

described in the following section.

Training item #1

Examiner: "Show me, 'The lion is walking'."

Acceptable response: Child must pick up the lion, and move

it in a horizontal motion across the table. The child may

move the legs individually, but this discrete movement is

not necessary for an acceptable response. If the child

picks up the lion, but does not enact the movement, the

examiner will repeat the statement with a revision, such









as "Show me how the lion walks." If the child still does

not enact the direction, the examiner will say "Let me

show you 'the lion is walking'." The examiner will then

demonstrate a walking motion with the lion.

Training item #2

Examiner: "Now show me, 'The bear is sitting'."

Acceptable response: Child will place the bear in a

sitting position on the table. If any of the children

make the bear sit on their lap, the examiner will prompt

them to "put the toy on the table". Failure to enact the

direction to place the bear in a sitting position will

result in reinstruction and then a demonstration, if

necessary.

Training item #3

Examiner: "Good job. Sometimes these animals do funny

things. Now show me, 'The pig jumps on the bed'."

Acceptable response: Child must place the toy on the bed

and repeat a vertical movement. Repeated instruction and

demonstration will be provided as described for Training

Items #1 and #2, if a child did not enact the direction

correctly.

Training item #4

Examiner: "The bed bumps the pig."

Acceptable response: Child must pick up the bed and touch

the pig with the bed. Reinstruction was provided if a

child failed to attempt any action with the two toys.








After training was completed the examiner presented

the eight routines. The selection of routines presented

was systematically varied so that the script/nonscript,

and probable/improbable conditions were balanced across

the 60 subjects. At the completion of the first routine,

the examiner verbally praised each of the children for

his/her cooperation by saying, "You did really well. You

showed me how the bear (for example) 'takes a bath.' That

was fun! Now I'll get some more toys. If you get tired,

tell me and we will stop." The examiner removed the toys

used in the first routine and replaced them with toys to

be used in the next routine. If the child appeared to be

distracted or did not remain seated at the table, a short

break was taken. When the child was seated at the table

and attending to the task, the examiner introduced the

next routine. Script formats were introduced by saying

"I'm going to tell you another story about (for example)

some rabbits. I want you to show me how they do things,

just like you did before. This time, the rabbits are

going to (for example) 'go to bed.'" The same procedure

was used to introduce each of the subsequent scripts.

Nonscript formats were introduced by saying "Now we're

going to play with some different toys. I'll tell you

about them and you show me what they do, just like we did

before." No theme or story topic was mentioned in the

introduction to nonscript routines. The specific








instructions and guidelines for acceptable responses are

described for a sample script in Appendix E.

Scoring Procedures

Each item was judged as correct or incorrect and a

percentage of accuracy score was determined for each

routine. Incorrect responses were further described by

recording the exact action that was performed (see

Appendix E).

A training tape for verification of accuracy in

scoring was made before scoring of the experimental data

was initiated. Three children, one from each of the

designated age levels, were videotaped as they completed

eight different routines. The examiner and student

assistant practiced scoring until a reliability level of

90% agreement was reached for classification of errors.

The examiner observed and recorded responses during

the evaluation sessions of 54 children. Additionally, the

examiner scored separately all of the evaluations from the

videotapes. Response scores obtained during the sessions

were later compared with the scores assigned from the

video tapes of the evaluations and the videotapes were

reviewed again to resolve differences in scoring. The

student assistant also scored the responses of 15 children

from the videotapes. An interjudge reliability level of

99% agreement was achieved for the 15 evaluations that

were scored by the student assistant and the examiner.








Error responses were analyzed to identify use of

comprehension strategies. Specific responses that

suggested the use of strategies were listed on the score

forms (see Appendix E). Classification of the probable

event strategy and word-order strategy was based on the

descriptions proposed by Chapman (1978). The probable

event strategy was indicated when it appeared that the

child's past experiences or world knowledge made the error

response more probable than the correct response. For

example, in the script-improbable cooking routine, an

active voice stimulus sentence directed the child to

demonstrate 'The monkey stirs the soup with a hammer.' If

the child depicted the monkey stirring the soup with a

spoon, the response was classified as the probable event

strategy. A word-order strategy was identified when it

appeared that word order served as a cue for interpreta-

tion of status for agent or object-of-action. Responses

to passive voice sentences were examined to determine when

the word-order strategy was used. For example, if the

child made the toy Cookie Monster paint the monkey in

response to 'The Cookie Monster is painted by the monkey',

the response was classified as a word-order strategy. The

third classification of error responses was used to

indicate child-agency. When a child performed the

required actions instead of demonstrating the action with

a toy, the response was classified as a child-agent









strategy. For example, if the child pretended to drink

from a cup in response to 'The bear drinks from a cup',

the response was coded as a child-agent strategy. When

descriptions of responses could not be clearly identified

as belonging to any of the three strategies described

above, error responses were placed in a category labeled

"other." The error responses of each subject were

analyzed for use of comprehension strategies, and a total

number of strategies used in each condition was tallied.

The definitions of error types are listed in Appendix F.

Statistical Analysis

The study was based upon a 3 (age) x 2 (script

organization) x 2 (event probability) x 2 (sentence type)

factorial design with repeated measures on the last three

factors. Analyses of variance appropriate for repeated

measures were performed. An error analysis was performed

on each subject's data to identify the use of compre-

hension strategies. Follow-up comparison tests were

performed for main effects revealed with the analyses of

variance.












RESULTS

The purpose of this study was to examine how script

organization, sentence voice, and event probability affect

language comprehension. Analyses of variance were

performed to identify effects of age, script organization,

event probability and sentence voice on language compre-

hension. The results are organized under the headings of

descriptive and comparative analyses, strategy use, and a

summary section.

Descriptive and Comparative Analyses

Comprehension scores were submitted to a 3 (age) x 2

(script organization) x 2 (event probability) x 2 (sen-

tence voice) mixed analysis of variance in which script

organization, event probability, and sentence voice were

the repeated measures. An alpha level of .05 was used for

interpretation of the statistical analyses. The mean

comprehension scores for each of the experimental condi-

tions are displayed by age level in Table 1. There was a

significant main effect for age (F(2,57)=70.66, P=0.0001).

For each child the total comprehension score was the total

number of sentences correct in all conditions, with a

total possible score of 48. The mean total comprehension








Table 1

Mean Comprehension Scores for each Experimental Condition



Age


Condition 2 3 4


Script Probable M (SD) M (SD) M (SD)

Active Voice 2.83 (1.72) 5.18 (.88) 5.92 (.28)

Passive Voice .28 (.57) 3.12 (2.34) 4.96 (1.24)

Script Improbable

Active Voice 1.83 (1.38) 4.71 (1.69) 5.72 (.54)

Passive Voice .44 (.62) 3.18 (2.48) 4.52 (1.33)

Nonscript Probable

Active Voice 2.94 (1.95) 5.06 (1.20) 5.88 (.33)

Passive Voice .44 (.78) 2.29 (1.79) 4.52 (1.73)

Nonscript Improbable

Active Voice 2.78 (1.73) 5.00 (1.32) 5.64 (.49)

Passive Voice .61 (.98) 2.76 (2.14) 4.52 (1.55)

Total Scores 12.15 31.29 41.68







scores for each age group were 12.15 for the 2-year-old

children, 31.29 for the 3-year-old children, and 41.68 for

children at the 4-year level. The total comprehension

scores at each age level are depicted in Figure 2.

Differences between age groups were tested using the

Bonferonni t-test. The difference between total scores of

the 2-year-old children and total scores of the 3-year-old

children was statistically significant (t= -6.99, p <

.05). The difference between total scores of the 3-year-

old children and total scores of the 4-year-old children

was also statistically significant (t= -4.07, p < .05), as

was the difference between 2- and 4-year-old children's

scores (t= -11.77, p < .05).

The main effect for sentence voice was significant

(F(1,57)=132.59, p=.0001), as was the interaction between

sentence voice and age (F(2,57)= 4.91, p=.0108). This

relationship is depicted in Figure 3. The 2-year-old

children had a mean comprehension score of 10.38 for

active voice sentences and 1.77 for passive voice sent-

ences. Three-year-old children demonstrated mean compre-

hension scores of 19.95 for active voice and 11.35 for

passive voice stimuli. The 4-year-old children had mean

comprehension scores of 23.16 for active voice and 18.52

for passive voice sentences. Significant differences were

found between scores on the active voice sentences and
















(41.68)


(31.39)


(12.15)


2 Years 3 Years 4 Years


Figure 2
Total Comprehension Scores at Each Age Level






















Active voice


24-
23-
22-
21-
20-
19-
18-
17-
16-
15-
14-
13-
12-
11-
10-
9-
8-
7-
6-
5-
4-
3-
2-
1-


2 years


3 years


4 years


Figure 3
Mean Comprehension Scores for Sentences of
Active Voice and Passive Voice


Passive voice











passive voice sentences for the 2-year-old children,

(t=7.23, p=.0001), the 3-year-old children (t=6.68,

p=.0001) and the 4-year-old children (t=5.40, p=.0001).

Mean comprehension scores for each sentence voice are

displayed in Figure 4. Children at all ages had more

correct interpretations of active voice sentences than of

passive voice sentences.

A significant interaction was revealed between age

and script organization (F(2,57)=5.46, p=.0068). This

relationship is depicted in Figure 5. The script compre-

hension score was determined by the number of correct

responses with a total possible score of 24. The mean

script comprehension scores were 5.38 for 2-year-old

children, 16.19 for 3-year-old children, and 21.12 for 4-

year-old children. The nonscript comprehension score was

determined by the number of correct responses in the non-

script routines. Mean nonscript comprehension scores were

6.77 at the 2-year level, 15.11 at the 3-year level, and

20.56 at the 4-year level. Post hoc comparison testing

did not result in statistically significant differences

between the script and nonscript condition for scores of

the 2-, 3-, or 4-year-old children.

The interaction between script organization and

sentence voice was significant (F(1,57)=6.84, p=.0114).

The mean comprehension scores for active and passive voice

sentences are depicted by script condition in Figure 6.










24 (23.16)

22-

[n(19.95)
20 Active Voice (1
1
18 -
8 Passive Voice
16-

14 -

12 (11.35)
(10.38)
10


8

6

4

2 (1.77)



2 Years 3 Years 4 YE


Figure 4
Mean Comprehension Scores for Sentence Voice


18.52)


3ars




























script


I--i


nonscript


3 years
3 years


4 years
4 years


Figure 5
Mean Comprehension Scores for the Script
and Nonscript Conditions


2 years
2 years


















Active voice






Passive voice


onscit
Nonscript


I
Script


Figure 6
Mean Comprehension Scores for Sentence Voice
by Script Condition


12-

11-

10-











The mean comprehension scores for sentence voice by script

condition reflect the numbers of correctly interpreted

active and passive voice sentences in scripts and non-

scripts. There were 12 sentences of each voice in each

script condition; therefore, a total correct score of 12

was possible.

A significant interaction was also revealed between

probability and sentence voice (F(1,57)=9.77, p=.0028).

The mean comprehension scores for active and passive voice

sentences in the probable and improbable conditions are

depicted in Figure 7. The mean comprehension scores for

sentence voice by probability condition are the numbers of

correctly interpreted active and passive voice sentences

in the probable and improbable conditions. There were 12

sentences of each voice in each probability condition.

Higher comprehension scores were demonstrated for active

voice sentences presented in the probable condition,

whereas comprehension scores for the passive voice were

higher in the improbable condition.

Strategy use

Comprehension strategy scores were submitted to mixed

analyses of variance. A 3 (age) x 2 (script organization)

x 2 (event probability) x 2 (sentence voice) analysis of

variance was used to examine use of the child-agent

strategy. Because the probable-event strategy cannot be

inferred in the probable condition, only responses to




















Active Voice







Passive Voice
0------


Improbable
Improbable


Probable
Probable


Figure 7
Mean Comprehension Scores for Sentence Voice
by Probability Condition









sentences in the improbable condition were inspected to

detect this strategy. A 3 (age) x 2 (script organization)

x 2 (sentence voice) analysis of variance was used to

examine use of the probable-event strategy. Use of a

word-order strategy could only be identified in passive

voice sentences; therefore, the responses for active voice

sentences were not included in the analysis for the word-

order strategy. Responses that did not fit the descrip-

tion of any strategy were classified as "other." The

results for active voice sentences were not examined for

use of "other" responses, because very few errors were

demonstrated by 3- and 4-year-old children for active

voice sentences. A 3 (age) x 2 (script organization) x 2

(event probability) analysis of variance was used to

examine the word-order strategy and "other" responses.

Differences between age groups were tested using the

Bonferonni t-test.

Mean comprehension scores for use of the child-agent

strategy are displayed in Table 2. A significant main

effect for age (F(2,57)=29.69, p=.0001) on use of the

child-agent strategy was revealed. Child-agent comprehen-

sion scores reflect the number of times the child-agent

strategy was used in all of the experimental conditions.

Two-year-old children had the highest mean score of 9.16,

whereas 3-year-old children demonstrated a mean score of

4.05. and 4-year-old children had a mean score of 1.20. A








Table 2

Child-Agent Strategy Scores


Age

Condition 2 3 4


Script Probable M

Active Voice 1.67

Passive Voice 2.38

Script Improbable

Active Voice 1.17

Passive Voice .33

Nonscript Probable

Active Voice .94

Passive Voice 1.61

Nonscript Improbable

Active Voice .78

Passive Voice .28


(SD)

(1.15)

(1.50)



(1.10)

(.49)



(1.35)

(.98)



(.94)

(.46)


M

.41

1.35



.35

.12



.24

.88



.41

.29


(SD)

(.62)

(1.54)



(.70)

(.33)



(.75)

(1.11)



(.79)

(.59)


M

.00

.56


(SD)

(.00)

(1.12)


.08 (.28)

.08 (.28)



.04 (.20)

.40 (1.08)



.00 (.00)

.04 (.20)









significant difference was found between frequency of use

of the child-agent strategy for the 2- and 3-year-old

children (t=6.03, p<.05), and the frequency of use between

the 2- and 4-year-old children (t=20.93, p<.05). Children

at 2 years of age used the child-agent strategy most

frequently. A significant difference was also revealed

between 3- and 4-year-old children in their use of the

child-agent strategy (t=7.37, p<.05). The 3-year-old

children used the child-agent strategy more frequently

than the 4-year-old children.

A main effect for script organization on use of the

child-agent strategy was obtained (F(1,57) = 5.98,

p=.0175), and is depicted in Figure 8. The mean scores

for use of the child-agent strategy were 2.45 in the

script condition and 1.80 in the nonscript condition.

A main effect for probability on use of the child-

agent strategy (F(1,57)=65.23, p=.0001) was also observed

and is depicted in Figure 9. The mean scores for use of

the child-agent strategy were 1.18 in the improbable

condition and 3.07 in the probable condition. A signifi-

cant interaction was also revealed between probability and

age for use of the child-agent strategy (F(2,57)=10.99,

p=.0001). This relationship is depicted in Figure 10.

The 2-year-old children (t=7.11, p=.0001) and the 3-year-

old children (t=3.59, p=.0025) used the child-agent

strategy more frequently in the probable condition.

















3.0

2.5 -(2.45)

2.0 (1.80)

1.5 -

1.0

.5

0
Script Nonscript


Figure 8
Child-Agent Strategy Use in the
Script Conditions















3.5
(3.07)
3.0

2.5

2.0

1.5
(1.18)
1.0

.5

0
Probable Improbable

Figure 9
Child-Agent Strategy Use in the
Probable and Improbable Conditions
























Probable


Improbable


2 years


3 years


4 years


Figure 10
Mean Scores for Child-Agent Strategy in Probable
and Improbable Conditions









A significant interaction was revealed between

probability and sentence voice for use of the child-agent

strategy (F(1,57)=44.55, p=.0001). This relationship is

depicted in Figure 11. A significant interaction was also

observed between probability, sentence voice, and age

(F(2, 57)=5.41, p=.007) for use of the child-agent

strategy. The relationship between probability and

sentence voice at each of the three age levels is depicted

in Figures 12, 13, and 14.

A main effect for sentence voice on use of the child-

agent strategy (F(1,57)=9.75, p=.0028) was also obtained

and is depicted in Figure 15. The mean total score for

use of the child-agent strategy with active voice senten-

ces was 1.67, as compared to a mean total score of 2.58

for passive voice sentences.

Mean scores for use of the probable-event strategy in

the experimental conditions are displayed in Table 3. The

probable-event strategy score is the total number of

instances that the probable-event strategy was used to

interpret sentences in the probable script and probable

nonscript conditions. A main effect for age was obtained

on the probable-event strategy (F(2,57)=18.85, p=.0001)

and is depicted in Figure 16. Mean probable-event

strategy scores were 4.06 for the 2-year-old children,

1.64 for the 3-year-old children, and 1.04 for the 4-year-

old children. This strategy was used significantly more




















Passive voice


2.4-

2.2-

2.0-

1.8-

1.6-

1.4-

1.2-

1.0-

.8-

.6-

.4-

.2-


I
Improbable


Probable
Probable


Figure 11
Mean Child-Agent Strategy Scores for Sentence
Voice by Probability Condition


Active voice















Passive Voice


4.0-

3.8-

3.6-

3.4-

3.2-

3.0-

2.8-

2.6-

2.4-

2.2-

2.0-

1.8-

1.6-

1.4-

1.2-

1.0-

.8-

.6-

.4-

.2-


Improbable


Probable


Figure 12
Mean Child-Agent Strategy Scores at the 2 Year
Level for Sentence Voice by Probability Condition


Active voice




























Passive voice


Active voice


I
Improbable


I
Probable


Figure 13
Mean Child-Agent Strategy Scores at the 3 Year
Level for Sentence Voice by Probability Condition


2.6-

2.4-

2.2-

2.0-

1.8-

1.6-

1.4-

1.2-

1.0-

.8-

.6-

.4-




























Passive voice


2.0-

1.8-

1.6-

1.4-

1.2-

1.0-

.8-

.6-

.4-

.2-


Improbable


Probable


Figure 14
Mean Child-Agent Strategy Scores at the 4 Year
Level for Sentence Voice by Probability Condition


Active voice





















3.0
(2.58)
2.5 -

2.0-
(1.67)
1.5

1.0-

.5

0
Passive Voice Active Voice

Figure 15
Child-Agent Strategy Use in the Active
and Passive Sentence Voices








Table 3

Probable-Event Strategy Scores


Age

Condition 2 3 4


Script M (SD) M (SD) M (SD)

Active Voice 1.28 (1.02) .41 (.87) .12 (.33)

Passive Voice 1.06 (.80) .70 (1.26) .48 (.59)

Nonscript

Active Voice .78 (.88) .24 (.44) .20 (.41)

Passive Voice .94 (1.16) .29 (.59) .24 (.44)


Note:


Only responses to the improbable sentences were
examined for use of the probable event strategy.















4.5
(4.06)
4.0

3.5

3.0

2.5

2.0
(1.64)
1.5
(1.04)
1.0

.5



2 years 3 years 4 years


Figure 16
Probable-Event Strategy Use at
Each Age Level








frequently by 2-year-old children than by the 3- (t=

10.71, p<.01) and 4-year-old children (t=14.65, p<.01).

Additionally, the probable-event strategy was used signif-

icantly more frequently by the 3-year-old children than by

the 4-year-old children (t=2.9, p<.01).

A main effect for script organization was also

revealed on use of the probable-event strategy (F(1,57)=

5.34, p=.0245) and is depicted in Figure 17. Mean

probable-event strategy scores were 1.27 in the script

condition and .85 in the nonscript condition.

Mean scores for use of the word-order strategy in

each experimental condition are displayed in Table 4. The

word-order strategy score is the total number of instances

that the word-order strategy was used to interpret passive

voice sentences in each of the experimental conditions. A

main effect for probability was obtained for use of the

word-order strategy (F(1,57)=9.85, p=.0027) and is

depicted in Figure 18. The mean word-order strategy score

was .82 in the probable condition and 1.4 in the improb-

able condition.

Mean scores for use of "other" responses are dis-

played in Table 5. The mean score for use of "other"

responses is the total number of "other" responses that

were used to interpret passive voice sentences in all of

the experimental conditions. A significant main effect

for age was observed (F=(2,57)48.24, p=.0001) and is













2.0

1.8

1.6

1.4
(1.27)
1.2

1.0 (.85)

.8

.6

.4

.2

O --------- --
0
Script Nonscript

Figure 17
Probable-Event Strategy Use in
the Script Conditions








Table 4

Word-Order Strategy Scores


Age


Condition 2 3 4




Script M (SD) M (SD) M (SD)

Probable .39 (.78) .24 (.44) .32 (.56)

Improbable .61 (.98) .65 (.79) .64 (.81)

Nonscript

Probable .33 (.59) .65 (.86) .52 (.65)

Improbable .61 (1.04) .88 (.93) .80 (1.19)


Note:


Only responses to the passive sentences were
examined for use of the word-order strategy.















2.0

1.8

1.6

1.4

1.2

1.0

.8

.6

.4

.2

0


Probable


(1.4)


& &


Improbable


Figure 18
Word-Order Strategy Use in the
Probable and Improbable Conditions


(.82)









Table 5

Mean Scores for Use of "Other" Responses


Age

Condition 2 3 4



Script M (SD) M (SD) M (SD)

Probable 2.94 (1.83) 1.24 (1.44) .16 (.37)

Improbable 3.56 (1.42) 1.35 (1.66) .28 (.68)

Nonscript

Probable 3.61 (1.50) 2.18 (1.78) .56 (1.12)

Improbable 3.56 (1.46) 1.76 (1.99) .40 (.71)


Note:


Only responses to the passive sentences were
examined for use of "other" responses.








depicted in Figure 19. The mean total score for the use

of "other" responses was 13.67 for the 2-year-old child-

ren, 6.53 for the 3-year-old children, and 1.40 for the 4-

year-old children. The 2-year-old children used sig-

nificantly more "other" responses than the 3- (t=5.18,

p<.05) and 4-year-old children (t=9.72, p<.05). Further,

the 3-year-old children used significantly more "other"

responses than the 4-year-old children (t=3.99, p<.05). A

main effect for script organization on "other" responses

(F(1,57)=11.67, p=.0012) was also obtained and is depicted

in Figure 20. The mean total score for use of "other"

responses was 9.53 in the script condition and 12.07 in

the nonscript condition.

Summary

Main effects for age and sentence type on language

comprehension were revealed. Total scores for sentence

comprehension showed significant increases with age, and

children at all age levels demonstrated better comprehen-

sion for active than for passive voice sentences. The

predicted effect of script organization was not obtained

across all age levels. However, there was an interaction

of script organization and sentence voice and effects of

script organization on the use of comprehension strateg-

ies were obtained. Event probability had neither main nor

interactive effects on comprehension. The interaction

between script organization and sentence voice suggested









(13.67)


2 Years


Use of "Other"


(6.53)


3 Years


(1.4)


4 Years


Figure 19
Responses at Each Age Level






81
14

13
(12.07)
12

11

10 (9.53)

9

8

7

6

5

4

3

2

1



Script Nonscript

Figure 20
Use of "Other" Responses in
the Script Conditions








voice had a larger effect in the nonscript condition than

in the script condition. However, the interaction between

probability and sentence voice suggested sentence voice

had a larger effect in the probable condition than in the

improbable condition.

Two-year-old children used the child-agent, probable-

event, and "other" response strategies more frequently

than the 3- and 4-year-old children. A main effect for

probability on use of the child-agent and word-order

strategies was also obtained. The child-agent strategy

was used significantly more in the probable condition.

The word-order strategy was used more frequently in the

improbable condition. A main effect for script organiza-

tion was obtained for use of the probable-event strategy

and "other" responses. The probable-event strategy was

used most frequently in the script condition. Children at

all ages used "other" responses more frequently in the

nonscript condition.










DISCUSSION

The purpose of this study was to investigate the

effects of varying script organization, event probability,

and sentence voice on language comprehension. Syntactic

difficulty was controlled by presenting stimulus sentences

of the active and passive voices in varying levels of

contextual support provided in the script and nonscript

formats. In addition to the script organization, event

probability was systematically varied by presenting

probable and improbable sentences. The comprehension

scores that were obtained from administration of the

experimental procedures were analyzed to identify effects

of sentence voice, script organization, and event proba-

bility on language comprehension. Error responses were

further examined for possible effects of script organiza-

tion, event probability, and sentence voice on children's

use of comprehension strategies.

Script organization and event probability resulted in

significant interactions with sentence voice. Comprehen-

sion of the active voice appeared to decrease in the

script condition and comprehension of the passive voice

increased in the script condition. The effect of proba-

bility interacted with sentence voice in a different

manner. Comprehension of the active voice increased in

83








the probable condition and comprehension of the passive

voice decreased in the probable condition. The improved

comprehension of passive voice in the script condition and

improved comprehension of active voice in the probable

condition were in agreement with the prediction that

script organization and event probability would facilitate

language comprehension. The effects of the nonscript

condition on comprehension of the active voice and of the

improbable condition on the passive voice were not

expected. These effects will be further discussed as

possibly resulting from the experimental definition of

"improbable," and from the children's use of comprehension

strategies.

The most obvious finding of this study was the

confirmation that comprehension of sentence structure

follows a developmental pattern increasing with age.

Overall scores compared among age groups indicated that

each successive age level from 2 years to 4 years demon-

strated a higher level of comprehension. This study also

presented additional evidence for developmentally earlier

comprehension for the active voice than for the passive

voice. Mastery of the passive voice has long been

recognized as a later development than comprehension of

the active voice (Baldie, 1976; Horgan, 1978). The

results supported the hypothesis that children at age

levels included within this study would demonstrate higher







performance levels of language comprehension for active

voice sentences than for passive voice sentences.

Evidence was obtained that indicates effects of

script organization and event probability may differ with

age. That is, the youngest children may have responded to

the contextual information that was provided by script

formats to the point of excluding the linguistic informa-

tion. The amount of contextual information may have

influenced the older children by facilitating their

comprehension of passive voice sentences and increasing

their use of probable-event and word-order strategies.

The script organization may have increased information

processing capacities and allowed the older children to

more accurately interpret linguistic messages. Whereas

the youngest children demonstrated fewer off-task

("other") responses in the nonscript condition, script

organization may have also increased their information

processing capacities, and allowed them to more consist-

ently attend to the experimental tasks.

Script Organization and Sentence Voice

Sudhalter and Braine (1985) described comprehension

of passive sentences with verbs that denote physical

actions as systematically easier than passive sentences

that include non-actional or experiential verbs such as

'hear', 'like', or 'believe.' All of the stimulus senten-

ces included in this study were comprised of action verbs







that could be clearly demonstrated in the enactment tasks.

Therefore, the passive voice sentences used in this study

were of the type that are comprehended earlier in the

developmental progression for passive voice.

In previous studies (Baldie, 1976; Lempert, 1985) the

types of passive sentence construction were systematically

controlled to examine effects of verb types or animacy.

All of the passive voice sentences in this study followed

a pattern of subject (e.g., 'The monkey') + past tense

verb (e.g. 'was painted'), + past participle (e.g. 'by the

Cookie Monster'). Noun animacy was not systematically

varied. The construction of the sentences used in this

study included reversible passive sentences such as 'The

little rabbit is washed by the big rabbit.' In this type

of sentence both nouns are animate and action could be

performed by either animate agent. This type of sentence

construction was reversed when it occurred in the improb-

able condition. The sentence was modified to violate an

expected role or action of a routine, or to conflict with

the probability of an event. For example, when this

sentence was reversed to become 'The big rabbit is washed

by the little rabbit,' the role of the parent or caregiver

bathing the smaller animate agent was reversed. This

situation was noted by several children who verbally

indicated that the bigger animal should perform the

washing action. The children's comments indicated that







they were aware of the script-role violation. However,

most of the children who commented that this was incorrect

then performed the task described in the sentence and

received a correct score.

In active voice improbable sentences such as 'The

monkey stirs the soup with a hammer,' or passive voice

improbable sentences such as 'The pillow is kissed by the

big rabbit,' the described actions were unlikely because

the instrument or object of the action was not expected.

Some children commented on the illogical aspects of these

sentences. For example, several children said "No, you

stir soup with a spoon" in response to the first sentence.

One child commented, "My pillow doesn't kiss," indicating

awareness of the improbable condition for the second

sentence. However, after children commented on the

unexpected aspect of a direction, they often correctly

performed the action. Most of the older children appeared

to have the ability to suppress the effect of script

organization or event probability. The interaction of age

and script organization suggests that children at the

youngest age level were guided by their script knowledge.

The 2-year-old children demonstrated more errors in the

script condition, resulting in highest total scores in the

nonscript condition. The youngest children typically

performed script actions in response to nonscript state-

ments. For example, when a sentence such as 'The little








rabbit puts on pajamas' preceded the sentence 'The milk is

poured by the bear,' the 2-year-old children often began a

routine of putting the rabbit to bed and either refused to

perform the next action by saying "No" or ignored the

examiner's prompts to follow the directions.

Comprehension of active voice sentences was consist-

ently better than comprehension of passive voice senten-

ces, supporting the premise that active voice sentences

are easier to comprehend at all age levels. The 2-year-

old children demonstrated their best total scores in the

nonscript condition. Because their mean score for passive

voice sentences was only 1.77, this total score more

accurately reflects performance on active voice sentences.

As previously discussed, the youngest children may have

comprehended more of the easier active voice sentences in

the nonscript condition because conflicting contextual

cues (script) were not present.

Script organization appeared to facilitate comprehen-

sion of the passive voice across all ages. This result

supports the prediction that script organization would

facilitate language comprehension. When children had

emerging syntactic competence, as indicated by their

developing comprehension of the passive voice, the

presence of script organization appeared to facilitate

their interpretations of the stimulus sentences. The

facilitating effect of script organization may be the








result of increased information processing capacity that

can be used for linguistic comprehension.

Comprehension Strategies

The data obtained in this study expands the descrip-

tion in previous literature of comprehension strategies

that children use to interpret sentences. The develop-

mental progression of the strategies outlined by Chapman

(1978) was supported by the results of this investigation.

Children at the 2 year level used the child-agent strategy

with the highest frequency. A review of the mean scores

(2 years: 9.16, 3 years: 4.05, 4 years, 1.2) supports the

observation that child-agency may be considered an early

strategy and that it is rarely used by the age of 4 years.

However, children at all age levels used the child-agent

strategy more frequently in the probable and script

conditions. Additionally, the interaction of age and

probability with use of the child-agent strategy suggests

that the 2-year-old children may have been more influenced

by event probability than were the older children. The 2-

year-old children demonstrated a mean child-agent strategy

score of 2.56 in the improbable condition as compared to

6.11 in the probable condition. Four-year-old children's

mean child-agent strategy scores were .2 and 1.0 respect-

ively in the improbable and probable conditions. The

youngest children used this strategy most frequently in

the probable condition.







The child-as-agent response pattern was described by

Bridges (1980) as a primitive response pattern used by

children between 2 and 2 1/2 years of age. The child-

agent strategy is indicated whenever the child performs an

action without involving the designated agent-of-action.

For example, for the stimulus sentence 'The bear drinks

from a cup,' child-agency was indicated if the child

pretended to drink from the cup. The child-agent strategy

was applied to an improbable stimulus sentence such as

'The little bear eats the toy boat' if the child pretended

to eat the boat. Children at all age levels used the

child-agent strategy significantly less frequently in the

improbable condition. The children appeared to be less

willing to demonstrate the unlikely actions or use the

incorrect objects that were described in the improbable

sentences.

The significant effect of sentence voice on use of

the child-agent strategy indicated that this strategy was

used more frequently to interpret passive voice sentences.

Because these sentences were consistently more difficult

to comprehend, it appeared that the children used this

strategy more frequently for the challenging task of

passive voice interpretation. The interaction of sentence

voice, probability, and age also indicates that children

at all age levels used the child-agent strategy more with








passive voice sentences presented in the probable condi-

tion.

Use of the probable-event strategy followed a similar

developmental progression with frequency of use decreasing

with age. This pattern was expected since the probable-

event strategy has been documented (Chapman & Kohn, 1978)

as a commonly used response that precedes use of a word-

order strategy. A main effect for script organization was

obtained for use of the probable-event strategy, indicat-

ing that this strategy was used more in the script

condition. The mean total score for use of the probable-

event strategy in the script-condition was 2.04 as

compared to a mean score of 1.35 in the nonscript condi-

tion. This suggests that the script context might have

affected language comprehension by facilitating the

children's use of their real-world experiential knowledge

as they attempted to interpret sentences that exceeded

their level of syntactic comprehension.

Although a main effect of age was not obtained for

use of the word-order strategy, it was used with less

frequency than any of the other strategies used by

children of all ages included in this study. These

results also support the findings of Chapman and Kohn

(1977) and Chapman (1978) who described it as a later

emerging strategy. Lempert (1985) found that 3-year-old

children used both a word-order strategy and a dynamic








agent strategy in which agent status was assigned to the

dynamic inanimate object (e.g., truck) in reversible full

passive sentences such as 'The truck is pushed by the

horse.' The 4-year-old children in the Lempert study also

used a word-order strategy, but replaced the dynamic agent

approach with an animate agent approach. That is,

children assigned agent status to the animate noun in

sentences such as 'The car chases the bunny.' The purpose

of this study was not to examine sentence comprehension

with respect to animacy; however the construction of

passive sentences included two main combinations of

animate nouns as agent or object. For example, 'The

little rabbit is kissed by the big rabbit' was comprised

of an animate object and subject. The improbable version

of this sentence ('The pillow is kissed by the big

rabbit') was comprised of an inanimate object and an

animate agent. This pattern of changing animacy was not

consistently altered with event probability. For example,

the script-probable sentence 'The washcloth is picked up

by the big bear' was constructed with an inanimate object

and animate agent. The script-improbable version of this

sentence maintained an inanimate object and animate agent,

but the verb was changed to alter probability: 'The

washcloth is eaten by the big bear.' The passive sent-

ences in this investigation were constructed to include

various combinations of animate and inanimate agents and