Trabasso1992 PDF

This article was downloaded by: [FU Berlin]
On: 11 May 2015, At: 02:19

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK
Discourse Processes
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/hdsp20
The development of goal

plans of action in the
narration of a picture story
a b
Tom Trabasso & Margret Nickels
a
Department of Psychology , The University of
Chicago , 5848 S. University Avenue, Chicago, IL,
60637
b
Northwestern University ,
Published online: 11 Nov 2009.
To cite this article: Tom Trabasso & Margret Nickels (1992) The development of
goal plans of action in the narration of a picture story, Discourse Processes, 15:3,
249-275, DOI: 10.1080/01638539209544812
To link to this article: http://dx.doi.org/10.1080/01638539209544812
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all
the information (the “Content”) contained in the publications on our
platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of
the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and
Francis shall not be liable for any losses, actions, claims, proceedings,
demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in
relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-and-
conditions
Downloaded by [FU Berlin] at 02:19 11 May 2015
DISCOURSE PROCESSES 15, 249-275 (1992)
The Development of Goal Plans

of Action in the Narration
of a Picture Story
TOM TRABASSO
The University of Chicago
MARGRET NICKELS
Northwestern University
The construction of coherent narrations of events in a picture storybook by children was

studied. A causal network discourse analysis (Trabasso, van den Broek, & Suh, 1989)
was applied to the narrations by children, 3, 4, 5, and 9 years in age, and adults, 20 years
and older. The analysis was successfully applied at the clausal level by deriving interact-
ing causal network representations for the perspective of each character in each of the
narrations. The structure and content of the network representation of the main character
were analyzed in detail. The analysis indicated that 9-year-olds and adults narrate accord-
ing to a hierarchical goal plan of action. It also showed striking developmental differences
from age 3 to 5. The 3-year-old children described states and neutral outcomes that are
unrelated to the central theme. The 4-year-olds, in contrast, encoded actions relevant to
the central theme but omitted goals and purposes. The 5-year-olds added these purposes to
attempts.
Coherence in the narration is achieved by applying naive theories of psychological and
physical causation, reflecting knowledge about goal plans of action. The narration's
content and structure are organized according to a hierarchical set of goals and a sustained
plan of action. Unexpected obstacles and failures of goal attainment lead to the reinstate-
ment of goals in order to continue following the plan. Although comprehension of this
structure is attained by age 4, coherence begins to manifest itself most explicitly in the
narrations of the 5-year-olds.
This article presents a study on how children construct coherent narrations about
events. Coherence in narration, we argue, is achieved when children use naive
This research was supported by grants to T. Trabasso from the Smart Foundation on Early
Learning and from the National Institute of Child Health and Human Development, Grant HD 17431
to T. Trabasso and Grant HD 25742 to T. Trabasso and N. L. Stein.
We are indebted to Dan Slobin and Virginia Marchman for granting us permission to analyze the
corpus and for providing us with the 58 protocols of the American English sample, and to Judy
DeLoache for providing us with the protocols of 14 mother-child dyads in the Frog, Where Are You?
story. We thank Jeffrey Kuperman for his assistance in analyzing the corpus and Margret Park Munger
and SoYoung Suh for their assistance in data analysis.
Correspondence and requests for reprints should be sent to Tom Trabasso, Department of Psychol-
ogy, The University of Chicago, 5848 S. University Avenue, Chicago, IL 60637.
249
250 TRABASSO AND NICKELS
theories about intentional action to infer and encode information about goals and
goal plans of action into speech. This use of planning knowledge leads to nar-
rative coherence when the events are interpreted according to and structured into
a goal plan. This article uses a causal network discourse analysis (Trabasso, van
den Broek, & Suh, 1989) to identify explicitly the existence of planning struc-
tures in narrations to a picture story by children of different ages and by adults.
Comprehension of narratives is assumed to be based upon one's ability to
detect a character's goals and to infer themes and plans. These inferences allow
the interpretation of a sequence of actions according to a goal plan (Schank &
Abelson, 1977). To comprehend as well as produce goal plans, one has to be able
to form a mental representation of the events in which one anticipates possible
problems and solutions, and monitors whether the actions follow the plan and
result in successful problem resolution (Scholnick & Friedman, 1987).
The fact that children come to use knowledge of goal-directed action in both
story understanding and production is well-established (Stein & Trabasso, 1982).

Children are increasingly able to explain actions in terms of goals or purposes
(Goldman & Varnhagen, 1985; Liu, 1988; Stein & Glenn, 1979; Wimmer,
1980). In judging the goodness of stories, they come to rely upon the presence of
goals and goal-directed actions (Stein & Policastro, 1984). Finally, in producing
free stories to minimal settings, they move from describing isolated states and
actions to the temporal and then to the causal sequencing of events. This is
followed by organizing the narrations into episodes that include goals, actions,
and outcomes. The latter denote goal success or failure. Hierarchical embedding
of episodes is a characteristic of stories produced by older children (Stein, 1988).
Storytelling or narrating can also be understood as communicative acts that
follow certain narrative conventions, namely, that one should organize the telling
of events according to the rules of intentional action and causal-temporal se-
quencing. Our narrative prototype is therefore an ideal representation of this
narrative convention. Insofar as conventions imply normativeness, our pro-
totypic narrative structure is normative.
An interesting, recent approach to narrative development is one that combines
narrative comprehension with production, and hence, communication. Here chil-
dren are asked to narrate events that occur in a picture story sequence to another
person (Bamberg, 1987; Berman, 1988; Berman & Slobin, in press; Slobin,
1990). The advantage of this approach is that all the children must try to narrate
the same set of events. To do so, they must understand what is depicted in each
picture and they must encode these interpretations into a narrative form. One of
the first published studies using this approach was that by Berman (1988), who
collected narrations to a picture story by Hebrew-speaking children. The picture
story (Mayer, 1969) is about a boy's continued search in the face of failure for his
pet frog that had escaped. Berman found that most 3- and 4-year-old children and
a few 5-year-olds described the pictures in isolation. The important developmen-
tal finding was that from age 5 to age 12, the children increasingly organized the
GOAL PLANS OF ACTION 251
events sequentially in relation to an overall plot line that involved a sustained

search.
Our suggestion is that Berman's "plot line and sustained search" arose from
the children's use of knowledge about plans and goal-directed action. The picture
story may be viewed as an enactment of a hierarchical goal plan. The top-order
goal is to get the frog back, but a subordinate goal, to find the frog, dominates
the story. Searching for the frog in particular locations leads to a series of
unanticipated goal failures of not finding the frog and to a series of goal reinstate-
ments to continue trying to search for the frog. Eventually, these reinstatements
lead to success in finding the frog. The understanding of goals, attempts, goal
failures, goal reinstatements, and ultimate success lead to the child's ability to
narrate a coherent story.
Understanding and producing stories may be regarded as a bidirectional pro-
cess (Bamberg, 1987; Bamberg & Marchman, in press). Knowledge of plans
guides top-down processing. The events to be encoded by the plan constrain

processing in a bottom-up manner. In a classic top-down view, Miller, Galanter,
and Pribram (1960) viewed planning as carrying out actions that match preexist-
ing schemes. Likewise, Hayes-Roth and Hayes-Roth (1979) regarded planning
as anticipating a course of action. On the other hand, both the narrator and the
comprehender are constrained by what occurs, either as actual events or as
discourse. These experiential or linguistic data allow one to detect goals and to
infer plans in a bottom-up manner and to use these inferred plans to anticipate,
interpret, and encode later events. The narrator's and the comprehender's initial
representation of the events constrain what, when, where, and how events are to
be interpreted and encoded in a discourse. In short, the content and the structure
of a narration is determined as a result of an interaction between a person's model
of physical and psychological causation and the events to which the model is
applied.
In narrating a picture sequence, the child is assumed to use interactive top-
down and bottom-up processes. The child interprets the pictures as "data" ac-
cording to a developing knowledge about intentional action. If the child knows
about goals and goal plans and uses this knowledge to interpret the events and
actions that are depicted in the pictures, the child will first detect a goal and then
selectively encode into language actions as attempts that are relevant to achieving
that goal. The process of inferring a goal from such data is bidirectional in that
the visual information constrains possible interpretations but knowledge of goals
and plans determines the detection of the goal. Once the goal is identified, a goal
plan may be inferred. The plan constrains interpretations and encodings of subse-
quent actions and outcomes because they are evaluated against the goal. How-
ever, such interpretation is not sufficient to achieve a coherent narration because
the child must also monitor these attempts and determine whether the outcomes
that result from them lead to success or failure of the goals and goal plan. When
failures occur, the child must understand the data in the pictures as renewed
attempts to achieve the failed goal. If these causal inferences occur, the resulting
narration will become coherent. A coherent narration is thus one in which a
series of clauses is organized by the character's goal and goal plan of action. The
main character's actions are interpreted as renewed attempts with goal failures
and reinstatements that eventually lead to goal success—Berman's (1988) over-
all plot line and sustained search. The picture sequence, however, also constrains
what is narrated. If the picture sequence itself does not provide sufficient infor-
mation to detect goals and to infer causal and temporal relations between goals,
attempts, and outcomes, then a coherent narration will not result, regardless of
the person's knowledge about intentional action.
REPRESENTING A HIERARCHICAL GOAL PLAN

Given a narration to a picture sequence, how do we determine whether it reflects
the application of a goal plan? The process of analysis is similar to that which
occurs in the narration itself: One must infer goals and interpret actions according
to a plan. To decide that the narration is structured this way, one needs a dis-
course analytic model that is content sensitive, and provides criteria to identify
explicitly the existence of goals, attempts, and outcomes, and the causal-tem-
poral relations between these categories of content. The model should allow for
the possible hierarchical structuring of these categories into episodes and of the
episodes into a goal plan hierarchy.
The causal network model of Trabasso et al. (1989) is a discourse analytic
model that can represent and generate hierarchical goal plans. Figure la depicts
their model as a general causal network. Each clause's content in the narration
may be classified into one of six categories: S (setting), E (event), IR (internal
response), G (goal), A (attempt), and O (outcome). Outcomes are marked as
successful (+), unsuccessful ( - ) , or neutral (o) with respect to goal attainment.
These categories together constitute an episode.
>. GENERAL CAUSAL NETWORK
b. GOAL PLAN HIERARCHY WITH UNANTICIPATED GOAL FAILURE,

GOAL REINSTATEMENT AND SUCCESS
/
G3-»-A-W5(-)
Figure 1. (a) Causal network model of Trabasso, van den Broek, and Suh (1989). (b) Causal
network showing a goal hierarchy with unanticipated failure outcomes and successes.
The categories are assumed to be interconnected by causal relations shown as

arrows in Figure la. In the Trabasso et al. (1989) model, the connecting relations
are inferred and can be either causal, temporal, or enabling. Settings can enable
all categories. Events can physically cause other events, or they can psychologi-
cally cause internal responses or goals. Internal responses (cognitions, emotions,
beliefs) can psychologically cause other internal responses or goals. Goals can
motivate goals or attempts to achieve them. Attempts can enable other attempts
or physically cause successful or unsuccessful outcomes. Outcomes, like events,
can physically cause other outcomes or psychologically cause internal responses
or goals.
Figure lb shows a representation of a goal plan hierarchy generated from
Figure la. This particular representation has three ordered goals with attempts
that lead to unanticipated goal failures, goal reinstatements, and success. The
plan begins with a setting (S) that is followed by an event (E) that happens to a
protagonist. These, respectively, enable and psychologically cause an internal

reaction (IR). The reaction leads to a goal (Gl). This goal motivates a subgoal
(G2) to obtain it. The subgoal (G2) in turn motivates another subgoal (G3) to
obtain it. Together these three goals constitute a plan. Here, the plan is carried
out through actions motivated at the level of the third goal in the hierarchy (G3).
This goal motivates an initial attempt (A) that fails, as indicated by an outcome
(O-). This failure is monitored and psychologically causes the reinstatement of
the second order goal (G2) that is motivated by the first-order goal (Gl) that
controls the overall plan. The cycle of the top-order goal (Gl) motivating the
second-order goal (G2) motivating the third-order goal (G3) that is followed by
an attempt (A) and a failure ( O - ) is repeated until finally an attempt (A) leads to
a successful outcome (O+) for the third-order goal (G3). This outcome enables
another successful outcome (O+) at the level of the second goal (G2). This
outcome enables an attempt (A) that causes a successful outcome ( 0 + ) for the
first-order goal (Gl), completing the plan.
Filling in the content of the categories in Figure lb demonstrates why we
suggested that Berman's (1988) notion of plot line and sustained search reflects
from the application of an hierarchical goal plan. In the sequence of pictures, a
boy has a pet frog (S), loses it (E), shows concern over its loss (IR), and begins a
series of goals (G) and attempts (A) that enact a plan to get the frog back (Gl).
The plan is to find the frog (G2) by searching in particular locations (G3) for it.
The boy searches by looking or calling (A) but is met with a series of failures
( 0 - ) where animals other than the frog suddenly and unexpectedly appear. The
boy, however, resumes the search in particular locations for the frog (G3) in order
to find it (G2), and finally does find it ( 0 + ) and takes it (A) back home (O+).
Structures such as that shown in Figure lb emerge in narrations of events if
the events permit them and if the child has the requisite knowledge of goal plans
and uses this knowledge to interpret and encode the events. Because this knowl-
edge develops with age, young preschool children may not have sufficient
knowledge about intentional action to detect a goal based upon settings and
events or to encode selectively actions that follow from the goal. Rather, they
may try to describe or identify what they see in terms more familiar to them. As a
result, the descriptions of objects, actions, and states are not related to a concep-
tual plan of action, and therefore lack coherence from an older child's or adult's
perspective. The extent to which one can apply the causal network model to
capture the content and the structure of the narration as a goal plan of action
depends directly upon whether or not the child encodes the picture sequence as
such. The child who explains actions in terms of goals or purposes that are
relevant to a higher-order goal is the one who tells the more coherent tale.
Our primary purpose was to apply an explicit discourse analytic model that
entails a psychological theory of intentional action to narrations produced by
children and adults to picture stories. We wished to find out the extent to which
children reveal in their narrations a developing knowledge and theory of inten-
tional action. Our process of analysis is inductive in that we rely upon what the
narrator says as "data," but our inductions are determined conceptually by the
definitions and criteria of the causal network model. Each narration is analyzed
on a clause-by-clause basis according to these definitions and criteria. What may
emerge is a set of categories based upon clausal content encoded and structured
causally and temporally by a goal plan of action into a coherent narration.
METHOD
In order to determine explicitly whether children's and adult's narrations show

planning knowledge realized in hierarchical goal plans of action, we obtained the
American English speakers' corpus of narrations of the Frog, Where Are You?
picture story by Mercer Mayer (1969; from Dan Slobin at the University of
California, Berkeley). The corpus was collected by Tamar Renner and Virginia
Marchman at the University of California, Berkeley, as part of a cross-linguistic
study (Berman & Slobin, in press; Berman et al.,1987; Slobin, 1990). Permission
to analyze the corpus with the model was obtained form Dan Slobin and Virginia
Marchman.
In the Renner-Marchman corpus, four groups of 12 children each, and 10
adults were asked to tell a story based upon a booklet of pictures after paging
once through the booklet. The instructions were essentially the same as those
used by Berman (1988). The four groups of children were 3 to 4, 4 to 5, 5 to 6,
and 9 to 10 years in age. Hereafter they are respectively referred to as the 3-, 4-,
5-, and 9-year-old groups. The adults were 20 years and older and are referred to
variously as the 20-year-old or adult group. The children and adults were native
speakers of American English and came from middle- to upper-middle-class
backgrounds. Half of the children in each age group and half of the adults were
females.
Clauses 7 through 11
7. Then, they both go to sleep.

Unbenowst 10 the two steeping people in the bed,
the frog crawls out of the Jar.
The next morning, the boy and the dog wake up,
and find the empty Jar.
Figure 2. Causal network representations of four characters derived from Subject 20B's clauses 7
through 11 narrated to pictures 2 and 3.
Causal Network Analysis of Narrations

In analyzing the clauses in a narration, it is necessary to assume the point of view
of a given character. In the picture sequence oí Frog, Where Are You?, there are
10 possible points of view: boy, dog, boy and dog together, frog, gopher, bees,
owl, deer, other frogs, and narrator. For each perspective, we derived a causal
network representation for each narration in the corpus, using the Trabasso et al.
(1989) model summarized in Figure la. This resulted not only in a representation
of the perspective of each character but in a representation of a set of interacting
plans for all of the characters. From this representation, we identified the main
protagonist as the boy because his network was the longest and was the only one
that was connected from the beginning of the story until the end.
For purposes of illustrating the method, we present part of the interacting
networks derived for Subject 20B for four perspectives: the boy, boy and dog,
dog, and frog. The partial networks are shown in two figures. Figure 2 shows the
networks for clauses 7 through 11, and Figure 3 (p. 256) shows them for clauses
10 through 15. The numbered clauses are taken from the narration of Subject
20B in the Renner-Marchman corpus; the complete narration is presented in
Table 1 (p. 257).
In the two figures, a clause is coded by a letter that denotes its category in the
causal network model. Because each category belongs to an episode, the number
after the letter is the level of that episode's goal in a goal-plan hierarchy. (The
goal-plan hierarchy is explained below). The number in parentheses is the or-
dinal number of the clause in the narration. Thus, 03(7) refers to an outcome at
goal level 3 for clause 7. At the bottom of the figure is a numbered picture line;
m
m
phi
BOY G1[Get Frog Back] — K 3 2 [ Find Frog] A3(12)->O3[-] G2[ Find Frog] E3(15)-
ps> m
/psi I
psi m
BOY&DOG 01(10)—•E1(11) G3(12,13) G3(14)-»-A3(14)
phi
DOG A3(13)-M)3[-] 03(15,-;
TOG A3(9) p h ' »03(11,
PICTURE -4-
Causal Networks for Clauses 10 through 15
10. The next morning, the boy and the dog wake up 11. and find the empty jar..
12. The boy looks in his boots 13. while the dog looks in the jar for the frog.
14. They look out the window for the frog 15. The dog falls out the window.
Figure 3. Causal network representations of four characters derived from Subject 20B's clauses 10 through 15 narrated to pictures 3
through 6.
TABLE 1
Adult Subject 20B Telling Frog, Where Are You?
1. This is a story of a little kid/ before he goes to bed one night/ He looks in this jar/ and he's
checking out his frog/ and his dog is there/ and they both check out the frog.
2. Then they both go to sleep/ Unbenownst to the two sleeping people in the bed/ the frog crawls
out of the jar/
3. The next morning, the boy and the dog wake up/ and find the empty jar/ The boy looks in his
boots/ while the dog looks in the jar for the frog/
4. They look out the window for the frog/
5. The dog falls out the window/
6. and breaks the jar/ The kid goes down out the window as well/ and picks up the dog/ The dog is
happily rescued from/ having his head stuck in the jar/ But the kid's a little angry/ probably
cause he's lost his jar and his frog/
7. The little boy and the dog go outside/ looking for the frog/ The boy is calling for the frog/ and
the dog is sniffing for the frog/ But actually the dog is sniffing at a trail of bees/ coming out of a
beehive/
8. The boy is looking down a hole/ to see/ if the frog is in the hole/ and the dog is intrigued by this
beehive/
9. The boy gets his nose either bitten or sniffed at/ by some little animal living in the hole/ and the
dog is still intrigued with the beehive/
10. Now, the beehive has been knocked down out of the tree by the dog/ and the bees are intrigued
with the dog/ while the boy is sitting in a tree/ looking in the hole in the tree/ thinking/ maybe the
lost frog is there/
11. At this point/ an owl pops out of the hole in the tree/ and a bunch of bees start following the dog/
probably angry/ that the dog has knocked their hive out of the tree/
12. Suddenly, the owl—oh no—now the boy is running away from the owl/
13. and the boy is climbing up on a rock/ and now the owl is gone for some reason/ and the dog is
coming back/ Looks like he's either stunned/ or is just very frightened and ashamed/ that he was
outdone by the bees/
14. The owl is looking at the boy/ and the boy is calling for the frog/
15. Now the boy has been picked up by some antlered beast/ that looks like a deer/
16. The deer is running to a cliff/ and the dog is barking at the deer/ This dog is pretty useless/ all he
does/ is cause trouble and bark at things/
17. Now the deer has thrown the dog, no, the boy over the cliff into/ it looks like they're heading for
a pond/ and the dog goes, too/ as the dog has throughout the story/
18. And they both fall in the water/
19. And they sit up in the water/
20. And it seems like they both hear something coming from behind the log/
21. And the boy tells the dog to be very quiet/ and they crawl over the log/
22. And they find on the other side of the log two happy frogs/ one of which was the frog in the jar/
23. And then they see that there was not only two happy frogs/ but there is an entire family of little
frogs there/ and one would wonder how long—what the gestation period is for frogs/ and how
come they're so big/ when they know/ they have to go through a tadpole stage first/ However, it
seems that they've raised this happy family/ and maybe the boy was looking for these frogs for
an entire six months to a year/ Who knows?/
24. Now the boy grabs one of the frogs/ and leaves with this frog/ leaving the two big frogs and all
the tiny frogs/ So it looks like he has not taken away his big frog/ that was in the jar/ but he has
taken away one of the sibling frogs perhaps/ one really can't tell/ The end/
Note. The numbered paragraphs contain the narration for each picture in succession. The slashes
demarcate each clause.
the braces mark the clauses in the narration that were made to that picture. Causal
relations between pairs of categories are indicated by labelled arrows between
categories. The labels e, psi, m, and phi stand for enabling, psychological-,
motivational-, or physical-causal relations, respectively. Thus, for example, an
outcome, 03(7), enables (e) another outcome, 01(10). Explicit categories are
shown by parentheses around the clause number; implicit categories are shown
by brackets and a description following the letter and goal level. Successful
outcomes are denoted by a plus sign (+) in the parentheses or brackets; failed
outcomes are denoted by a minus sign (—). Because a clause can be categorized
from the perspective of more than one character, it may be represented in more
than one causal network and its category membership may change. Clauses that
are in two networks have the same number, and their equivalence is depicted by
vertical, parallel lines between them. For example, an outcome for the frog,
03(11, +), became an event, E l ( l l ) for the boy and dog.
We shall first discuss the analysis of clauses 7 through 11 depicted in Figure

2. These clauses describe the boy and the dog going to sleep, and while they are
asleep, the frog crawls out of the jar. The next morning, the boy and dog wake up
and find the jar empty. Going to sleep is an outcome from the perspective of the
boy and the dog because they achieve the goal of sleep. This outcome, identified
as 03(7), in turn, psychologically causes a state of cognition in which they lack
awareness. This cognition is regarded as an internal response and is noted as
IRC3(8). Shifting the perspective from the boy and dog to that of the frog, we
reclassify the boy and dog's outcome and internal cognitive response as a pair of
events that are observed by the frog. These two events psychologically activate
an inferred desire in the frog to try to escape from the jar, labeled G3[Escape].
This goal is inferred from the two prior events and from the frog's explicit
attempt, A3(9), to crawl out of the jar. The arrow from G3[Escape] to A3(9)
indicates that the implied goal motivated the attempt. In order to infer a goal, we
required that it be anchored by an explicit causal antecedent or antecedents (e.g.,
E3(7) and E3(8)), and an explicit causal consequent (e.g., A3(9)). In inferring
other categories, particularly outcomes, and, rarely, attempts, we required the
presence in the narration of explicit, direct causal antecedents and consequents.
Returning to the analysis, the attempt, A3(9), results in a successful outcome
of the frog escaping and leaving the jar empty, O3(ll, +). At this point, the
analysis shifts back to the perspective of the boy and the dog in which they wake
up, an outcome, 03(10), that is enabled by the initial state of sleeping, 03(7).
This change in state enables them to experience an event, E3(l 1), namely, that of
finding the jar empty.
We now introduce the analysis of relations between categories. On the boy
and dog line, the outcome, 03(7), psychologically causes an internal response
that is cognitive, IRC3(8). That is, going to sleep results in a psychological state
of unawareness. We believe that there is a causal relation between the outcome
and the reaction because if the boy and dog had not gone to sleep, then they
would have been aware of their surroundings. This counterfactual reasoning tests
the assumption that causes are necessary in the circumstances of a story (Mackie,
1980; Trabasso, Secco, & van den Broek, 1984). Counterfactual tests were
applied to all identified direct causal relations throughout the analysis. To label
the type of causal relation, one must know what the categories are because the
categories constrain the kind of relation. For example, goals motivate actions or
other goals, whereas actions enable other actions or physically enable or cause
outcomes (see Trabasso et al., 1989, for rules and criteria for relating categories,
and the previous discussion on the different kinds of causal relations that may be
obtained between categories).
Identifying a Goal-Plan Hierarchy

Consider now clauses 10 through 15 in Figure 3. In this series, the boy and dog
wake up and see the empty jar. The boy looks in his boots, while the dog looks in
the jar for the frog. Then they both look out the window for the frog. The dog
falls out the window. The narration contains an outcome in clause 10, an event in
clause 11, two purposeful attempts in clauses 12 and 13, another purposeful
attempt in clause 14, and a failed outcome or an event in clause 15.
In our analysis, we separated purposeful attempts into two categories, a goal
and an attempt, which together make up a local goal plan. The goal is based upon
the purpose of the action and is expressed frequently as a prepositional phrase
(e.g., for the frog) or as an infinitive phrase (e.g., to see if the frog is in there).
The presence of the frog in the phrase indicates that it is the desired goal object.
We assume that the goal exists prior to the attempt and motivates it. Thus, the
action of looking in the boots, A3(12), is motivated by the goal of looking for the
frog in certain locations, G3(12). The purposeful attempt in clause 14 is another
local goal plan. Here the goal is to look for the frog in another location, namely
at the window, in G3(14), and is carried out as an attempt in A3(14).
Why do the boy and dog look for the frog out the window after looking for the
frog either in the boots or in the jar? Apparently, they did not find the frog in the
first locations. We therefore infer a failed outcome from each of their perspec-
tives, O3[—]. (The brackets indicate that the goal or outcome is inferred.) This
outcome was enabled by the initial attempts and it contributes psychologically to
another, local goal plan. We now ask: What motivates these two local goal plans
of searching for the frog in specific locations? These local plans are best under-
stood as being motivated by an inferred, higher order goal, labeled G2[Find the
Frog]. This higher order goal of finding the frog motivates local goal plans in
which the boy and dog search in particular locations for the frog and it is
reinstated each time the local goal plan fails. We now ask: Why is the goal of
finding the frog reinstated? The answer is that the boy has developed yet another
inferred, higher order goal, namely, to get back his pet, labelled GlfGet Frog
Back]. Together, these inferred and explicit goals constitute a goal-plan hier-
archy, where the top-order goal, GlfGet Frog Back] motivates a second-order
goal, G2[Find Frog] which, in turn, motivates a local goal plan, G3fSearch in
Particular Locations] that motivates attempts at this level. We now understand
why levels 1, 2, and 3 were assigned to the categories. We finally ask: What
caused the goal-plan hierarchy to arise? The answer lies, in part, in the Event,
E l ( l l ) in which the boy discovers the empty jar, indicating that the frog is lost
and psychologically causes, in the circumstances of the story, the boy's goal of
wanting to get the frog back, Gl. Note that E l ( l l ) is not sufficient to cause this
goal. Rather, it is necessary to know that the boy values the frog as a pet,
established as a setting in the first picture. It is the loss of a valued goal object
that results in the desire to reinstate the relationship with the frog. The relation
between goal states and changes in goal states as determinants of emotions and
goals is central to Stein and Levine's (1989) theory of emotional understanding
and is consistent with the present analysis.
Extracting Episodic Structures

In the analysis, episodic structures emerge when the narration permits identifica-
tion of a set of categories that depict goal-directed action. A complete episode
would contain one or more of each of the categories in the general recursive
transition network in Figure la, namely, an event, a goal, an attempt, an out-
come, and a reaction. Figure 3 shows a well-formed episode as part of the frog's
network. The episode consists of two events, a goal, an attempt, and a successful
outcome, and these categories are causally related.
In Figure 3, one can also discern basic goal-attempt-outcome episodes in the
boy's or dog's networks. For example, clauses 12 and 13 contain a local goal and
an attempt that are followed by a failed outcome for the boy. Similarly, for the
dog, clauses 14 and 15 contain another local goal and an attempt followed by a
failed outcome. The main action sequence in the narrative occurs at level 3 in
which the boy is trying to find the frog in particular places.
Identifying Goal Plans

Recall that the representation of the narration consists of interacting causal net-
works from the perspective of each of the characters. It is usually the case that
one perspective dominates a narrative. The causal network analysis enables us to
determine from whose perspective the narration is told. Essentially, the main
protagonist is the character with the largest network. In the present case, this was
clearly the boy, although the boy and the dog actually appeared together in all but
one picture.
In order to compare the narrations to the extent they indicate a goal-plan
hierarchy, we extracted the boy's network from each analyzed narration in the
corpus. It is not convenient to show the complete network of the boy for any one
narration. However, to motivate the data analyses following, we show part of the
episodic structure of the boy's network for Subject 20B in Figure 4 (p. 262). The
corresponding clauses for the explicit categories in Figure 4 are presented in
Table 2 (p. 263) to assist in the reading of Figure 4.
Figure 4 and Table 2 show evidence that Subject 20B's narration is structured
according to a hierarchical goal plan strikingly similar to the idealized plan in
Figure lb. The network begins in the upper left with an event in which the boy
finds the jar empty (E), causing his desire to get his frog back (Gl) in episode 2.
This goal generates a goal plan to find the frog (G2) and entails a subordinate
goal to search in particular locales (G3). The first realization of the local goal is
in episode 9 in which the boy wants to search for the frog outside [G3] and so he
calls for the frog (A) but fails to find it [O—]. The boy's goal plan is reinstated in
episode 10 in which a hole in the ground (S) is discovered and the boy wants to
see if the frog is in it (G3). He looks (A), but a small animal that lives there (S)
bites him on the nose, a failed outcome ( O - ) .
The goal plan (G2) is reinstated once again in episode 11 in which the boy is
sitting in a tree (S) and there is a hole in the tree (S). The goal plan to find the
frog [G2] psychologically causes the boy to think about these two conditions as a
possible location in which the frog might be searched for and found (IRC). This
cognition psychologically causes the activation of the goal to look in the hole
(G3). This goal motivates an attempt to look (A) which is disrupted when an owl
pops out and the outcome is consequently a failure to find the frog (O—). Later,
in episode 14, the boy finds himself sitting in water, an outcome from episode 13
in which a deer has thrown him over a cliff into the pond. An event occurs in
which he hears (IRC) some sound coming from the other side of the log (E). This
cognition, in conjunction with the reinstatement of the goal plan (G2), causes the
local goal of wanting to see if the frog is on the other side of the log (G3). This
motivates him to crawl over the log (A) where he finds some frogs (O+), one of
which was the frog that had been in the jar (O+). This outcome completes
episode 3 in that it finally satisfies the goal of finding the frog. The finding of his
frog completes our criteria for inferring a second-order goal (G2), and enables
the boy to attempt (A) to achieve his top-order goal of getting the frog back home
(Gl). The attempt of taking the frog back home (A) constitutes part of the second
episode. This explicit attempt supports our inference that there is a top-order goal
(Gl) of wanting to have the frog back. Once again, we require an antecedent, the
initiating event of the frog leaving the jar empty (E), and a consequence, the
attempt to bring the frog back (A), for the inferred goal. The inferred goal-plan
hierarchy is anchored by explicitly narrated clauses, either an event that causes a
goal and an attempt to achieve it, or by failed outcomes and instances of goal
plans to search in particular locales.
Analyses of the Content and Structure in the Representations of the

Boy's Perspective in the Narrations
If a narrator interprets the pictures as a sequence of actions that follow a goal-
plan hierarchy, then we expect the causal network representation to approximate
in form and content the ideal structure shown in Figure lb. If this is the case, the
network representation would contain first- and second-order goals (Gl and G2)
and be organized into episodes that each consist minimally of a level three goal
Level 1 E-*-[G1]-
psy
Level 2 [G2] [G2] [G2l [G2]
Level 3
\
3
m e
7
-G3-+-A—*O-«—S S
m e psy pSy
S-»-|RC —*{G3]-*-A-*- O- E —MRC -»-[G31 *• A
m e e kPsy
0(0)
Episode 10 11 14 14
Figure 4. Selected episodes showing a goal hierarchy with unanticipated failure outcomes and success taken from the boy's perspective in the causal network representation
of Subject 20B's narration.
TABLE 2
Selected Episodes (Boy's Perspective) of Subject 20B's Narration
Episode
2 The boy finds the frog's jar empty. (E)
He wants his pet frog back home. [Gl]
3 The boy wants to find his frog. [G2]
8 He wants to search for his frog outside. [G3]
So he called for the frog. (A)
But the frog did not come. [ O - ]
10 There is a hole in the ground. (S)
The boy wants to see if the frog is in there. (G3)
The boy is looking in the hole. (A)
Some little animal is living in the hole. (S)
The boy gets his nose bitten by the animal. ( O - )

11 The boy is sitting in a tree. (S)
There is a hole in the tree. (S)
The boy thinks that maybe the lost frog is in there. (IRC)
He wants to see if the frog is in the hole. [G3]
So he looks into the hole. (A)
But at this point an owl pops out of the hole in the tree ( O - )
14 The boy sits up in the water. (O)
Some sound is coming from behind the log. (E)
The boy can hear the sound. (IRC)
14 He wants to check if there are frogs behind the log. [G3]
So he crawls over the log. (A)
And he finds on the outer side of the log two happy frogs. (O+)
One of which was the frog in the jar. (O+)
2 Now the boy leaves with his frog. (A)
Note. The letters correspond to the content categories of the general causal
network. The numbers on the goals refer to their level in the hierarchy. The
positive and negative signs for outcomes indicate goal success or failure. The
bracketed goals and outcomes are inferred.
(G3), an attempt (A), and an outcome (O), hereafter called GAO episodes. For
these properties, we report analyses with age as a factor in order to assess the
development of the use of knowledge of goal plans of action to organize a
narration.
RESULTS
Evidence for a Goal-Plan Hierarchy
The existence of first- and second-order goals in the representation provides
evidence that narrators structure their narrations in terms of a goal-plan hier-
archy. We analyzed each person's network representation as to whether it con-
e
20
Figure 5. Proportion of subjects with a level 1 goal (Gl) and at least one level 2 goal (G2) for each
age group.
tained a first-order goal (Gl) or one or more second-order goals (G2). The
respective proportions of subjects having these goals in the representations of
their narrations are shown in Figure 5.
Figure 5 shows that second-order goals first appeared at age 4 and first-order
goals at age 5. The 9-year-old narrators showed nearly the same number of these
goals as did the adult narrators. The ordering of the age groups 3 = 4 < 5 < 9 =
20 was statistically significant, \ 2 (2) = 37.45, p < .01.
Episodic Structures
In order to assess the degree to which the narrations were structured episodically,
we identified two kinds of units. The first unit is termed a goal-attempt-out-
come (GAO) episode, and has been previously defined and illustrated. The
second, termed a non-GAO unit, consists of a clause or a set of clauses that
maintain a topic but does not allow the inference of a goal. The clauses within
each unit may or may not be connected. For each representation, we counted the
number of GAO episodes and non-GAO units. Comparisons between the age
groups on these two measures were done by analyzing them in a 5 x 2 analysis
of variance with age as a between-subject factor and the type of unit as a within-
subject factor.
In the analysis, there were no statistically significant age differences as to the
mean number of combined episodes and units, F(4, 53) = 1.28, p = .29. There
was, however, a significant interaction between age and type of structural unit,
F(4, 53) = 298.96, p < .01. The means for age, and for the two kinds of units in
the interaction with age, are shown in Figure 6.
In Figure 6, the GAO episodes followed the same pattern as the higher order
goal data. The GAO episodes were rare for both the 3- and the 4-year-olds. They
12
10
GOAL-ATTEMPT-OUTCOME
8-
m
o
z
III
o
111
5
NON-GAO UNITS
1 3 5 7 9 11 13 15 17 19 21
AGE
Figure 6. Average number goal-attempt-(+, — )outcome episodes, mean number of units, and
non-GAO units for each age group.
increased substantially in frequency for 5-year-olds and approached the level of

adult frequency by 9 years of age. For the non-GAO units, the pattern was
completely reversed, and quite symmetrical compared to the GAO episode data.
That is, the majority of the units for the two younger groups were non-GAO
units. These units declined, were still a slight majority at age 5, and became very
rare for the 9-year-olds and the adults. Individual comparisons between the types
of units at each age indicated that all differences were statistically significant (p
< .05).
Categories in GAO Episodes

What kinds of categories were included in the GAO episodes and how many,
relatively speaking, were there of each kind? The number of categories in a
narration is approximately equal to the number of clauses, because every clause
could, in fact, be categorized using the definition and criteria of Trabasso et al.
(1989). Recall from the examples given in the previous analysis, that, on occa-
sion, a clause can have more than one classification, for example, a purposeful
attempt. The proportion of each kind of category to the total number of catego-
ries was found for each narrative representation. Separate analyses of variance
with age as a factor were carried out. The first analysis was on goals, attempts, or
outcomes; the second was on settings, events, neutral outcomes, or internal
responses.
Figure 7 (p. 266) shows the data on goals, attempts, and outcomes that
0.4
ATTEMPTS
0.1
5 7 9 1 1 1 3 1 5 1 7 19 21
Figure 7. Average proportion of clauses classified as a goal, attempt, and success or failure
outcome for each age group.
encoded success or failure. Consider the data on attempts first. Note that the
proportion of clauses that were attempts was relatively constant from age 4 to
adulthood. Only the 3-year-olds produced a significantly lower proportion of
attempts relative to all the other groups, F(4, 53) = 3.14, p < .05, for age as a
variable, and by individual comparisons (p < .05). Note also that the attempts
were relatively more frequent than either goals or outcomes at all age levels.
Consistent with previous findings on higher order goals, level three goals and
outcomes increased with age, F(4, 53) = 23.88 and 27.80, p < .01; the ordering
from lowest to highest across the ages was 3 = 4 < 5 < 9 = 20, by individual
comparisons (p < .05).
In order to identify the categorical content that is more representative of the
younger children's narrations, similar analyses were carried out for settings,
events, neutral outcomes (those not involving success or failure with respect to
the goal of finding the frog), and internal responses (emotions or cognitions).
Figure 8 shows these proportions of the total number of clauses produced in the
narration.
In Figure 8, the most striking finding was that the 3-year-old children exhib-
ited a very high proportion of neutral outcomes (e.g., "he fall down"). These
declined up to age 5 F(4, 53) = 6.05, p < .01, and were ordered 3 > 4 > 5 = 9
= 20, by individual comparisons (p < .05). Settings were the next most frequent
for the youngest children and this category declined, somewhat, for 5- and 9-year
olds, but increased for adults, F(4, 53) = 6.28, p < .01. Events increased and
then decreased slightly with age, F(4, 53) = 3.47, p < .02. Internal responses
increased slightly with age, but their effect was not significant (F < 1) and was
overall relatively infrequent.
0.6
Settings
0.5- Events
Internal Responses
Neut Outcomes
O
O
Q.
O 0.3
ce
CL
0.2
1 3 5 7 9 11 13 15 17 19 21
AGE
Figure 8. Average proportion of clauses classified as a setting, event, neutral outcome, or internal
response for each age group.
Local Cohesion Within Episodes

The previous analyses suggest that the narrations became increasingly coherent
at a global level with development: They include more higher level goals and
greater structuring of the narration into GAO episodes. Local cohesion at the
third goal level in the hierarchy can also be obtained and allows the narrators to
increase the length of the narration by adding categories at this level of organiza-
tion. These added categories are causally related to the other categories of the
GAO episode.
For each narration, the total number of additional categories and the total
number of different categories added to the minimal GAO episodes were found
and entered into separate analyses of variance with age as a factor. The analysis
showed a significant effect for both the total number of categories, F(4, 53) =
29.04, p < .01, and different categories, F(4, 53) = 14.22, p < .01. Figure 9 (p.
268) shows the respective means as a function of age.
The proportion of different categories declined with age: .94, .99, .85, .87
and .77. This means that as the narrators grew older they added more of the same
kinds of categories at a faster rate than they added different kinds of categories to
their episodes.
5- TOTAL CÄTB30RES „
IT ^^^^
HI
O ^^^^
E
Z
4-
HI
a
211 1 ****"
5 3-
L ^ ^ DIFFERENT CATEGORIES
3 5 7 9 11 13 15 17 19 21
AGE
Figure 9. Average total or different number of categories per episode per subject for each age
group.
Global Coherence: Causal Connections Within

and Between Episodes
To build a coherent narration, the narrator needs to provide the listener with the
information necessary to infer causal and temporal relations among the clauses.
Given that the narrations are increasingly organized into GAO episodes and that
the number and diversity of the episodic categories increase with age, the next
question addressed was to what extent the clauses within and between the GAO
episodes are connected by these kinds of relations. To understand this analysis,
refer to Figure 3. For the GAO episode representing the narration of clauses 12
and 13 from the perspective of the boy, one can see that there is one connection
between the goal and the attempt and one between the attempt and the outcome.
That is, the goal, G(12, 13), to look for the frog in a particular location motivates
the attempt to look in the boots. However, this attempt fails, O[—]. Thus, there
are two connections within this GAO episode. Between-episode connections
arise when outcomes of previous episodes either cause or enable states or actions
in subsequent episodes. In Figure 3, the failed outcome, O [ - ] , for the GAO
episode involving clauses 12 and 13 psychologically causes the reinstatement of
G2[Find Frog] which, in turn, motivates the local goal, G3(14), to look for the
frog at the window. These two connections are considered to be between-episode
relations because they connect two different GAO episodes. For each narrator,
the number of within- or between-episode connections per episode was found
and analyzed in different analyses of variance with age as the between-subject
factor. Figure 10 shows the average number of each kind of connection for the
age groups.
The analysis of variance indicated that there were significant effects due to
age for within, F(4, 53) = 48.99;p < .01, and between, F(4, 53) = 31.71, p <
.01 connections. Individual comparisons on the means (p < .01) ordered the age
2
CE
U.
QJ
2-
1 3 5 7 9 11 13 15 17 19 21
AGE
Figure 10. Average number of connections within or between episodes per subject for each age
group.
groups in the same sequence for each kind of connection: 3 = 4 < 5 < 9 < 2 0 .
The ratio of between to within connections increased with age up to 9 years: .00,
.23, .30, .51, and .47.
The analytical model identifies four kinds of connections: enabling, psycho-
logical, motivational, and physical. The respective proportions of each type of
relations for the within or between connections were found for each person's
narrative representation. The proportions for the within-episode connections are
given in Figure 11.
In Figure 11, with the exception of the 4-year-olds, the relations were ordered
from the most frequent to the least frequent as follows: enabling, motivation,
psychological, and physical. Aside from the 4-year-olds, there are no changes in
0.8
WITHIN CONNECTIONS
0.6-
ENABLING
Z
O
O 0.4 i MOTIVATIONAL
a.
o
ce
a.
PSYCHOLOGICAL
PHYSICAL
1 3 5 7 9 11 13 15 17 19 21
AGE
Figure 11. Proportion of connections within episodes per subject for each age group.
these relative frequencies across age. The 4-year-old narrations show slightly
more enabling and slightly fewer motivational relations compared to the other
groups. For the between-episode relations, from age 5 to adulthood, the propor-
tion of enabling, psychological, and motivational relations were equal (.33) and
physical relations were nonexistent; for 4-year-olds, enabling relations were
more frequent and motivational ones were less frequent, as before.
DISCUSSION
Development of Plans of Action
The causal network representations reveal striking developmental differences in
the content and structure of the narratives. The changes were both qualitative and
quantitative in nature. From 3 to 5 years in age, the children's narrations changed
first from predominantly describing states and identifying objects to encoding
isolated actions. The encoding of the actions was, however, related to a goal plan
to find the frog. The next developmental change was the explicit encoding of the
goals and purposes behind these actions. Older children and adults more ex-
plicitly organized actions according to a goal plan by encoding them along with
goals and outcomes into episodes of increasing diversity and causal interrelated-
ness. This developmental sequence was first reported on children ranging in age
from 5 to 11 years by Stein (1988) and Stein and Policastro (1984) for both free
story generation to setting statements and for judgments of what constitutes a
story.
The 3-year-olds in the present study were clearly descriptive in their narra-
tions, a result consistent with Berman's (1988) claims about this age group. Their
descriptions consisted mainly of state changes classified as neutral outcomes
unrelated to the goal of finding the frog. Settings constituted the second most
frequent category. These descriptive statements were primarily deictic in nature.
The 4-year-old children showed a similar but lesser trend. These descriptive
categories accounted for a large percentage of the nonepisodic units in these two
younger age groups.
The 4-year-olds did show evidence of an awareness of the goal plan in that
they encoded a relatively large proportion of their clauses in terms of actions that
may be interpreted as attempts relevant to the goal of finding the frog. They did
so without the inclusion of phrases with goal objects or purposes so that we were
not provided with a basis by which we could infer these motivational categories.
The result is that the 4-year-olds appear to be describing rather than explaining
the boy's actions. This description appears to be temporal rather than causal,
corresponding to the a second stage of development reported by Stein (1988) in
story production by children.
One explanation for the 4-year-olds' omission of goals is that the goal object
(the frog) is absent from the pictures in which the goal-directed search occurs.
Four-year-olds may encode only what is shown with respect to goal-directed

activity. This interpretation is supported in a recent undergraduate honors thesis
by Munger (1989) directed by the senior author. When the 4-year-olds were
asked to narrate the Frog, Where Are You? story in which the frog is absent,
children encoded goals as purposes in only 17% of the pictures that contains an
attempt to find the missing frog. In contrast, when they were asked to narrate
another picture story by Mayer (1967), entitled A Boy, A Dog and A Frog, in
which the frog is present, they encoded goals in 51% of the pictures that con-
tained an attempt to catch the frog.
Evidence for the narration being organized according to a hierarchical goal
plan of action appears most clearly in the data of the 5-year-olds. The 3-year-olds
did not provide sufficient explicit information in their narrations to permit in-
ferences for either first- or second-order goals. Second-order goals, however,
were inferred for two (17%) of the 4-year-olds and for four (33%) of the 5-year-
olds. By age 9, the vast majority of the narrators included both kinds of goals.
These American English data are consistent with those of the Hebrew speakers in
Berman's (1988) report. We believe that the development and use of planning
knowledge accounts for the present data and for Berman's (1988) particular
findings of an increase in overall plot line constituents and in sustained search.
The Development of Narrative Skills

The data suggest that the children are not only learning to apply their knowledge
about intentionality to their narrations of events but also that they are learning
about the requirements of narration as a conventional, communicative form.
Listeners need information in order to establish a foundation for constructing a
coherent interpretation of what is narrated. Developmental differences in orient-
ing the listener occurred in how the children narrated the beginning of the story
and in how the children narrated actions that occurred in different scenes
In their setting statements, the 3-year-olds merely point to and identify either
the boy or the dog or the frog and seldom mentioned all three together or the fact
that the animals were the boy's pets. The 4-year-olds, however, introduce all
three characters in relation to one another, frequently including new information
that one or both of the animals are the boy's pets. This information is very
important to the listener or reader because it establishes who are the main charac-
ters in the story, and points to the relationship between the boy and the frog.
Later, when the frog escapes, one can infer that the boy is upset and wants to find
the frog so as to have him back as a pet. Without knowing that the frog is of
value, inferences about the goals of wanting it back and finding it are unwar-
ranted. The 5-year-olds also orient the listener when scenes shift in the pictures.
These shifts happen to co-occur with renewed attempts. By providing reasons in
the form of goals or purposes for these attempts in a new context, the 5-year-old
narrator is both orienting the listener and providing explanatory coherence.
Narrative Skill and Story Complexity

The Frog, Where Are You? story (Mayer, 1969) presents the child with a very
complex task. There are several characters whose goal plans interact. The pro-
tagonist's plan is interrupted by several unexpected obstacles. These charac-
teristics contrast with more typical stories for young children in which one
character seeks to attain a single goal with one or two blockages. Stein (1988),
for example, reports that the typical story generated by a 5-year-old is a goal-
based story with two episodes with one or no obstacles. The complexity of
Mayer's (1969) story may have prevented the younger children from demonstrat-
ing what they know about simple goal plans. Our data indicate that the children
were able to deal with this complexity. By age 5, several children encoded
sustained goal plans of action. By age 4, however, they encoded only the relevant
actions to a goal plan. The 4-year-olds may understand goal plans of action even
if they do not explicitly encode goals and purposes. They do encode goal plans
when the goal object is explicit and present in the situation being encoded. In the
study by Munger (1989) cited previously, two picture stories were contrasted on
explicitness of goal objects but were similar in complexity. Both stories portrayed
a sustained plan of action with multiple obstacles. When the 4-year-olds saw an
action directed towards a explicit goal object, they encoded purposes with at-
tempts. If the goal object was not present, they rarely encoded purposes explain-
ing the action. However, in both cases, when Munger asked them w/iy-questions
on the actions, they stated the relevant purposes. Four-year-olds thus appear to
know about goals but do not encode them if the goal object is not present. Thus,
the underestimation of the 4-year-olds' knowledge does not appear to depend
upon the complexity of the story but does depend upon the physical availability
of the goal object and other knowledge about narrative conventions. Our conjec-
ture is that the 4-year-olds have not as yet acquired narrative conventions of
explaining behavior in terms of goal plans.
The Episode as a Unit in Working Memory During Production

The fact that the number of structural units was relatively constant across the age
groups is striking. Of equal interest is the fact that the number of clauses per unit
increased with age, leading to an increase in the length of the narration. The
constant number of units suggests some upper limit in what can be narrated. The
picture story, as conceived by Mayer (1969) probably constrains the number of
episodes one could generate in order to encode it. In addition, the demands of the
task to try to narrate every picture also constrained how many units were pro-
duced. The episode, however, may also be thought of as a kind of working
memory unit, a "chunk," according to Chi (1978), that organizes a number of
categories into a single psychological structure for purposes of production and
storage in memory. The sizes of these chunks in memory are known to increase
with one's knowledge in a domain (Chi, 1978). The more knowledge children
have about goal plans, the more information they can include and organize into
the same number of narrative units. Further, we know that both in memory and in
on-line processing of stories, the episode is an efficient psychological unit (Black
& Bower, 1979; Haberlandt, 1980). Episodic structuring allows the narrator to
work with an efficient unit of memory in production because it allows the
narrator to expand and diversify the information within the unit without increas-
ing the demands on working memory. Our data on the constancy in the number of
units encoded into the representation of a narration, along with the developmen-
tal increase in the total number of categories and different categories added to
episodes, and the increase in the number of relations within and between epi-
sodes are consistent with this interpretation. These developmental changes first
become evidence by age 5.
CONCLUSION: THE USE OF A CAUSAL NETWORK

DISCOURSE ANALYSIS
The causal network model of Trabasso et al. (1989) was successfully applied to
narrations of a picture story by children ranging in age from 3 to 9 years and by
adults. This analysis, applied from the perspective of every character in the
picture story, allowed us to depict the narration as a set of interacting goal plans
from which we extracted the goal plan of the main protagonist. The boy's goal
plan corresponded to a hierarchical set of three goals and a sustained plan of
action at the level of the third goal in the hierarchy to search for the frog in
particular locations. Unexpected obstacles and failures of attaining the third goal
led to the reinstatement of the second-order goal of continuing the search. Suc-
cess in locating the frog enabled the attainment of the goals of finding it and
taking it (or a substitute pet) back home.
The explicit representation of narratives as a causal network showed how
children from 3 years onward progressively move from identification and de-
scription of states to actions and then to explanation of actions carried out
according to a goal plan of action. This developmental sequence corresponds to
Berman's (1988) taxonomic findings on plot line and sustained search develop-
ment, and to Stein's (1988) and Stein and Policastro's (1984) classification of
stories produced by children as descriptive sequences, temporal action se-
quences,^ or episodic structures organized by goals and causal relations. The use
of knowledge of goal plans of action to interpret and encode actions as attempts
demonstrates the importance of the use of knowledge of psychological causation
in narration and narrative knowing (Polkinhorne, 1988). The choice of which
lexical terms should be used to encode the actions appears to be dependent upon
this knowledge of intentionality. Further, the development of a well-formed
overall plot line and a theme that is sustained over the course of the narration
depends directly on the development and use of this and other knowledge of
narrative communicative conventions.
REFERENCES
Bamberg, M. (1987). The acquisition of narrative: Learning to use narrative. Berlin: Mouton de
Gruyter.
Bamberg, M., & Marchman, V. (1991). Binding and unfolding: Towards the linguistic construction
of narrative discourse. Discourse Processes, 14, 277-305.
Berman, R. A. (1988). On the ability to relate events in narrative. Discourse Processes, 11, 469-
499.
Berman, R. A., & Slobin, D. I. (in press). Narrative structure. In D. I. Slobin & R. A. Berman
(Eds.), Different ways of relating events in a narrative: A cross-linguistic study. Hillsdale,
NJ: Erlbaum.
Berman, R. A., Slobin, D. I., Bamberg, M., Dromi, E., Marchman, V., Neeman, Y., Renner, T., &
Sebastian, E. (1986). Coding manual: Temporality in discourse (rev. ed.). Berkeley: Univer-
sity of California, Cognitive Science Program.
Black, J. B., & Bower, G. H. (1979). Episodes as chunks in narrative memory. Journal of Verbal
Learning and Verbal Behavior, 18, 187-198.
Chi, M. (1978). Knowledge structures and memory development. In R. S. Siegler (Ed.), Children's
thinking: What develops? Hillsdale, NJ: Erlbaum.

Goldman, S., & Varnhagen, C. K. (1985). Memory for embedded and sequential story structures.
Journal of Memory and Language, 25, 401-418.
Haberlandt, K. (1980). Story grammar and reading time of story constituents. Poetics, 9, 99-118.
Hayes-Roth, B., & Hayes-Roth, F. (1979). A cognitive model of planning. Cognitive Science, 3,
275-310.
Liu, L. (1988). Using causal questions to assess and promote children's understanding and memory
for narratives that vary in causal structure. Unpublished doctoral dissertation, University of
Chicago.
Mackie, J. L. (1980). The cement of the universe: A study in causation. Oxford: Clarendon Press.
Mayer, M. (1967). A boy, a dog and a frog. New York: Dial Press.
Mayer, M. (1969). Frog, where are you? New York: Dial Press.
Miller, G. A., Galanter, E., & Pribram, K. (1960). Plans and the structure of behavior. New York:
Holt, Rinehart & Winston.
Munger, M. P. (1989). Children's understanding of intentional action: Evidence from picture story-
telling. Unpublished bachelor's thesis, Department of Psychology, University of Chicago.
Polkinhorne, D. E. (1988). Narrative knowing and the human sciences. New York: State University
of New York Press.
Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ:
Erlbaum.
Scholnick, E. K., & Friedman, S. L. (1987). The planning construct in the psychological literature.
In S. L. Friedman, E. K. Scholnick, & R. R. Cocking (Eds.), Blueprints for thinking. New
York: Cambridge University Press.
Slobin, D. I. (1990). The development from child speaker to native speaker. In J. W. Stigler, R. A.
Shweder, & G. Herdt (Eds.), Cultural psychology: Essays on comparative human develop-
ment. New York: Cambridge University Press.
Stein, N. L. (1988). The development of children's storytelling skill. In M. B. Franklin & S. Barten
(Eds.), Child language: A reader. New York: Oxford University Press.
Stein, N. L., & Glenn, C. G. (1979). An analysis of story comprehension in elementary school
children. In R. O. Freedle (Ed.), Advances in discourse processes: Vol. 2. New directions in
discourse processing. Norwood, NJ: Ablex.
Stein, N. L., & Levine, L. (1989). The causal organization of emotion knowledge: A developmental
study. Cognition & Emotion, 3, 343-378.
Stein, N. L., & Policastro, M. (1984). The concept of a story: A comparison between children's and
teachers' perspectives. In H. Mandl, N. L. Stein, & T. Trabasso (Eds.), Learning and

comprehension of text. Hillsdale, NJ: Erlbaum.
Stein, N. L., & Trabasso, T. (1982). Children's understanding of stories: A basis for moral judgment
and dilemma resolution. In C. J. Brainerd & M. Pressley (Eds.), Verbal processes in children
(Vol. 2). New York: Springer-Verlag.
Trabasso, T., Secco, T., & van den Broek, P. (1984). Causal cohesion and story coherence. In H.
Mandl, N. L. Stein, & T. Trabasso (Eds.), Learning and comprehension of text. Hillsdale,
NJ: Erlbaum.
Trabasso, T., van'den Broek, P., & Suh, S. (1989). Logical necessity and transitivity of causal
relations in stories. Discourse Processes, 12, 1-25.
Wimmer, H. (1980). Children's understanding of stories: Assimilation by a general schema for
actions or coordination of temporal relations? In F. Wilkening, J. Becker, & T. Trabasso
(Eds.), Information integration by children. Hillsdale, NJ: Erlbaum.

Trabasso1992 PDF

Uploaded by

Copyright:

Available Formats

Trabasso1992 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Trabasso1992 PDF

Uploaded by

Copyright:

Available Formats

This article was downloaded by: [FU Berlin]

On: 11 May 2015, At: 02:19

The development of goal

To link to this article: http://dx.doi.org/10.1080/01638539209544812

PLEASE SCROLL DOWN FOR ARTICLE

The Development of Goal Plans

The construction of coherent narrations of events in a picture storybook by children was

story understanding and production is well-established (Stein & Trabasso, 1982).

events sequentially in relation to an overall plot line that involved a sustained

guides top-down processing. The events to be encoded by the plan constrain

REPRESENTING A HIERARCHICAL GOAL PLAN

>. GENERAL CAUSAL NETWORK

b. GOAL PLAN HIERARCHY WITH UNANTICIPATED GOAL FAILURE,

The categories are assumed to be interconnected by causal relations shown as

protagonist. These, respectively, enable and psychologically cause an internal

In order to determine explicitly whether children's and adult's narrations show

7. Then, they both go to sleep.

and find the empty Jar.

Causal Network Analysis of Narrations

TOG A3(9) p h ' »03(11,

Causal Networks for Clauses 10 through 15

We shall first discuss the analysis of clauses 7 through 11 depicted in Figure

Identifying a Goal-Plan Hierarchy

Extracting Episodic Structures

Identifying Goal Plans

Analyses of the Content and Structure in the Representations of the

3 The boy wants to find his frog. [G2]

increased substantially in frequency for 5-year-olds and approached the level of

Categories in GAO Episodes

Local Cohesion Within Episodes

Global Coherence: Causal Connections Within

Four-year-olds may encode only what is shown with respect to goal-directed

The Development of Narrative Skills

Narrative Skill and Story Complexity

The Episode as a Unit in Working Memory During Production

CONCLUSION: THE USE OF A CAUSAL NETWORK

thinking: What develops? Hillsdale, NJ: Erlbaum.

teachers' perspectives. In H. Mandl, N. L. Stein, & T. Trabasso (Eds.), Learning and

You might also like