THE EFFECTS OF VERBAL REMINDERS AND PARENTAL
CONVERSATIONAL STYLE ON LONG-TERM MEMORY IN 2-YEAR-OLDS
MICHELLE E. BOYER
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENT FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Special thanks go to the parents and children who participated in this project.
Without their support and enthusiasm, this project would not have been possible. I
would also like to acknowledge all of the undergraduate student assistants who
voluhmteered many hours of their time to work on this project as experimenters and
coders: Tara Abels, Tara Kadish, Yvonne DaVerona, Misty Kolchakian, Sandra
Kellett, Chris LaBelle, Stephanie Krantzler, Jason Wall, and Jennifer Eels. I would
especially like to acknowledge the invaluable assistance of Tara Abels, Yvonne
DaVerona, and Chris LaBelle.
I would also like to acknowledge my committee members, Dr. Patricia Miller,
Dr. Scott Miller, Dr. Ira Fischler, and Dr. Bridget Franks, for their time and
assistance. Dr. Scott Miller especially has been a wonderful source of encouragement
and guidance throughout graduate school.
Very special thanks go to my advisor, Dr. Jeffrey Farrar, for his many years
of mentoring. I gratefully acknowledge and appreciate all of his advice, time, and
encouragement throughout graduate school.
TABLE OF CONTENTS
LIST OF TABLES .........................
1 INTRODUCTION .............
Theories of Infantile Amnesia ......
Literature Review ..............
Language and Its Relation to Long-Term
Description of Current Study .......
2 METHOD ...........
3 RESULTS ...........
Parental Style ....................
Comparison of Groups for Behavioral Recall .
Comparison of Groups for Verbal Recall ....
. . . . . . . ii
. . . . . . .. v
. . . . . . . vi
4 DISCUSSION ................................... 84
Effect of Verbal Reminders on Memory ................... 85
Effect of Parental Style on Memory ...................... 87
Effect of Language Abilities on Memory ................... 88
Explanation of Results .............................. 90
Implications for Children's Memory ...................... 92
Future Directions ................................ 93
Summary ..................................... 98
REFERENCES ......................................... 100
BIOGRAPHICAL SKETCH ............................... 106
LIST OF TABLES
2-1 Making Pictures Event ........................... 39
3-1 Actions Performed by Children ...................... 59
3-2 Pairs of Actions Correctly Sequenced by Children ........... 60
3-3 Number (and Proportion) of Children by Group Performing
Each Action .................................. 66
3-4 Memory Verbalizations at Session 2 and Session 3 By Group .... 69
3-5 Summary of Correlations for All Children ................ 76
3-6 Summary of Correlations for Children in the Reminder Group ... 80
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
THE EFFECTS OF VERBAL REMINDERS AND PARENTAL
CONVERSATIONAL STYLE ON LONG-TERM MEMORY IN 2-YEAR-OLDS
Michelle E. Boyer
Chair: M. Jeffrey Farrar
Major Department: Psychology
This study examined 2-year-old children's long-term memory for an event.
Several issues were of interest regarding the relation between long-term memory and
language: (a) can 2-year-old children benefit from parental verbal reminders about a
novel event; (b) does parental conversational style (i.e., elaborativeness) affect 2-year-
old children's ability to benefit from memory talk; and (c) do children's language
abilities affect their ability to verbalize about the event? Twenty-eight 24-month-old
children experienced a novel event one time and returned 3 months later to verbally
and behaviorally recall the event. To determine parental style, at the initial session
parents engaged in memory talk with their children about novel past events. An
elaboration ratio of elaborations to repetitions was calculated for each parent. Parents
were classified as high elaborative or low elaborative based on their elaboration ratios.
To assess the effect of verbal reminders on children's memory, 12 children talked
about the event once a month with a parent during the delay period.
There was little evidence that parental memory talk facilitated 2-year-old
children's memory for the event. First, children who were reminded about the event
did not have better verbal or behavioral memory for the event than children who were
not reminded. Furthermore, the amount of information discussed by parents during
the reminder sessions did not predict children's verbal or behavioral recall.
Second, parental conversational style did not affect children's memory for the
event. Children of high elaborative parents did not evidence better memory for the
event. Also, children of high elaborative parents who were reminded about the event
were not better able to verbalize their memories.
Third, children's vocabulary at 24 months when the event was experienced
predicted their verbalizations about the event 3 months later, and there was some
evidence that children who had better language abilities at 27 months verbalized more
about the event at that time. Therefore, children's language abilities when the event
was experienced predicted their ability to verbalize about the event.
The results are discussed in relation to their implications for the development
of autobiographical memory and representational understanding.
The current study examined 2-year-old children's ability to remember a one-
time novel event over time. Of interest was the relation between memory and
language development, specifically the extent to which children's long-term memories
can be reinstated through memory talk. This issue has important implications for our
understanding of the causes of infantile amnesia and for our understanding of the
development of autobiographical memory. The autobiographical memory system
consists of memories for personally experienced and personally relevant past events,
that is those memories that tell us something about ourselves. The development of
autobiographical memory during the early preschool years signals the end of childhood
To explore the relation between memory and language in young children,
twenty-eight 24-month-old children experienced a novel event one time and returned 3
months later to verbally recall and to reenact the event. Half of the children talked
about the event with a parent during the 3 month delay period, and half did not talk
about the event. One issue of primary interest in the current study was whether
memory talk reinstates 2-year-old children's memory for events. Reinstatement occurs
if children's own memory for events is activated when those events are discussed with
others. If memory talk reinstates 2-year-olds' memory for events as it does 3- and 4-
year-olds' memories, children in the present study who are verbally reminded about
the event during the delay period should have better memories for the event than
children who are not reminded. Additionally, since parental conversational style has
been shown to affect the elaborativeness of children's event memories (Fivush, 1991),
the present study examined whether the effectiveness of verbal reminders is related to
the elaborativeness of parental conversational style during memory talk. The present
study provides additional information regarding young children's ability to benefit
from memory talk about a specific event.
Thleories of Infantile Amnesia
Although there is considerable individual variability, it is generally accepted
that most adults have difficulty reporting events experienced before the third or fourth
year of life (Kihistrom & Harackiewicz, 1982; Sheingold & Tenney, 1982). This
phenomenon has come to be known as childhood or infantile amnesia. Memories
occurring prior to age 3 are typically not recalled, whereas memories occurring after
age 3 are more easily retrieved, with retrieval improving between the ages of 3 and 5
(Waldfogel, 1948; Wetzler & Sweeney, 1986). Kihlstrom and Harackiewicz (1982)
found that, on average, adults' earliest memories were reported around the age of 3.5
years, and Sheingold and Tenney (1982) found that college age students who had
experienced the birth of a sibling when they were at least 3 years could remember that
event later, but they could not remember the birth of a sibling if it occurred prior to
The inability to recall events occurring during the first few years of life has
ramifications for how we view memory development and, more generally, cognitive
development. One important issue is whether memory can be conceptualized as
continuous or discontinuous, that is whether the memory systems of infants and young
children are qualitatively similar or dissimilar to that of older children and adults.
Research supporting the infantile amnesia phenomenon indicates that there are
qualitative differences in the memory systems of individuals at different ages. A
variety of theoretical explanations have been proposed to account for infantile amnesia
(see Pillemer & White, 1989, for a review). Pillemer and White (1989) have
classified these into several broad classes of theories. Among the earliest and most
well-known theories of infantile amnesia is the Freudian idea of repression; memories
for childhood events exist but are actively repressed because they are linked to
unacceptable childhood sexual and aggressive impulses (i.e., blockade theory)
(Pillemer & White, 1989).
A second view is related to cognitive developmental changes in the memory
system (see Nelson, 1993; Pillemer & White, 1989, for a review of these theories).
One explanation is that very young children are incapable of forming memories during
this period of time, and this is why there are few, if any, memories from this time
period. A second explanation is that because very young children lack the cognitive
abilities to construct narrative memories, memories encoded in the first few years later
become inaccessible over time (Pillemer & White, 1989; Wetzler & Sweeney, 1986).
A third view is that infantile amnesia occurs as a result of an inability to
reconstruct the traces from early memories. According to this view, autobiographical
memories undergo reconstructive processes as they are continually retrieved, such that
memories are reinterpreted in terms of the individual's present knowledge. The latter
explanation, proposed by Schachtel (1947) and Neisser (1962, 1967), holds that
amnesia for early events is the result of reorganization of children's mental structures,
necessitated by social demands on the child to become responsible, reality-oriented,
and reliant on language. This process results in mental representations or memories
that are incompatible with earlier structures, rendering early memories inaccessible.
Although these theories have tried to account for childhood amnesia, empirical
evidence provides support for none of them. First, in recent years it has become
more apparent that adults can remember events from early childhood (Sheingold &
Tenney, 1982). Second, there is evidence that very young children are capable of
forming memories during early childhood. Therefore, it is not the case that young
children are incapable of forming memories. For instance, DeLoache and her
colleagues (DeLoache & Brown, 1979; DeLoache, Cassidy, & Brown, 1985) have
demonstrated that even 2-year-olds can remember the location of stuffed toys that they
observed being hidden in a room the day before. Furthermore, there is growing
evidence that young children can, under some circumstances, remember events
occurring before the age of 3, and in some cases, before the age of 1. Some of the
earliest evidence documenting early memory was reported by Todd and Perlmutter
(1980), in which 3-year-olds recalled naturally occurring events up to 8 months
earlier, and 4-year-olds recalled events occurring 15 months earlier. Fivush, Gray,
and Fromhoff (1987) also found that 29- to 35-month-old children could recall
information from events that had occurred at least 6 months earlier, prior to the age of
2. Similarly, Fivush and Hamond (1990) found that 4-year-olds recalled as much
information about going to Disney World when they were 2.5 as they had when they
were first questioned about the event nearly 2 years earlier. Therefore, the traditional
accounts cannot adequately account for or explain this phenomenon.
The fact that events occurring in the second year of life can be remembered at
much later points in time raises a number of questions. For instance, if toddlers and
preschoolers can recall events experienced as much as a year earlier, why are these
memories apparently lost later in life? Also, how long can early memories be retained,
and what factors affect retention?
To answer some of these questions, Nelson (1988, 1989, 1992) and Pillemer
and White (1989) have proposed alternative explanations of infantile amnesia that
examine how social factors affect the development of autobiographical memory.
According to Nelson's social-functional approach (1989, 1993), the most general
function of memory is to guide present actions and to predict future outcomes based
on the past. For this reason, younger children are much more likely to remember
more routine types of events than specific events because they allow children to
understand their world and to make predictions about future occurrences (Hudson &
Nelson, 1986). Therefore, one-time events would have little predictive value, would
be of little use in guiding actions, and thus would be dropped from memory after a
period of time. They may be retained for shorter periods of time, sometimes as long
as several months (Hudson, 1990b), but they are not accessible later in life.
However, if the event is reminded or reexperienced, it is more likely to be retained.
The implication of this argument is that reminders determine whether experiences will
be retained for a longer period of time (see also Fivush & Hamond, 1989; Hudson,
Several researchers emphasize the role of social factors, such as memory talk,
in the development of the autobiographical memory system. For instance, Nelson
(1989, 1992) argues that as children become older, they begin to retain memories for
their personal significance rather than simply for their predictive value because they
understand the social value of talking about the past. Fivush (1988) argues that this
development, the development of the autobiographical memory system, is related to
the development of an enduring concept of self. She argues that although very young
children have long-term memories of the past (e.g., Fivush et al., 1987), those
memories become inaccessible over time unless they somehow are integrated into the
child's self-concept. Over time, children learn how to interpret memories in terms of
themselves (i.e., why the event was important and what it tells them about
themselves). Both researchers believe, however, that memory talk (i.e., conversations
about past events) is an important tool for the formation of durable long lasting
autobiographical memories. At some point during the preschool years, memory talk
begins to reinstate children's memory for events. That is, conversations with others
about past events begin to remind children of their memories for those events,
assuming that children have acquired the verbal skill necessary to communicate or
represent their memories (Nelson, 1993) or that they have acquired the ability to
direct their own memory search (Hudson, 1990b). For reinstatement to occur, Nelson
argues that children must be able to use the verbal representation of another person to
remind themselves of their own memories for that event. This also entails
understanding that someone else's memory of the event is different from their own.
According to Nelson, the representational abilities necessary for reinstatement are not
acquired until the late preschool years. On the other hand, Fivush and Hamond (1989)
argue that reinstatement is contingent on the ability to form coherent narrativized
versions of their experiences. Because younger children lack the language skills
necessary for this, Fivush's argument is that memory talk may not reinstate memories
in children younger than 2.5.
Although no research has directly examined when memory talk first reinstates
children's memory for events, the consequence of these more recent theories has been
an increased focus on the development of autobiographical memories and the factors
that influence the retention and duration of early memories. One way that this has
been investigated is through examination of both adults' and children's ability to recall
events and experiences from their first years of life (see Howe & Courage, 1993;
Pillemer & White, 1989, for reviews). To examine whether and how memory talk
affects 2-year-old children's memory for a novel event, the present study examined
three factors thought to affect long-term retention of early memories and the formation
of autobiographical memories: (a) verbal reminders (i.e., memory talk) about a
specific event; (b) parental conversational style; and (c) children's language
development. Each of these factors will be addressed in more detail in the following
For preschool children, memories for events have generally been examined
using naturally occurring events or laboratory-derived events. In such cases, children
simply are asked to report on what they remember about an event (Fivush, 1984).
With children younger than 3, different methods, such as naturalistic observation, in
the form of diary studies and home observation, or elicited/deferred imitation, have
typically been used. The earliest studies employing naturalistic methods, such as diary
studies (Ashmead & Perlmutter, 1980; Nelson & Ross, 1980) and home observations
(Todd & Perlmutter, 1980), found some anecdotal evidence that 3- and 4-year-olds
can remember early events. Todd and Perlmutter (1980) cite one 3-year-old who made
a reference to the chocolate- and honey-flavored milk that her mother gave her as a
baby. Similarly, Nelson and Ross (1980) noted one child who remembered the name
of a friend when driving by the friend's house after a period of months had passed.
Even more striking was the fact that the child had been unable to say the friend's
name when his family relocated. Although such reports are encouraging, we cannot
conclude that young children are able to verbalize about spcif past occurrences
since such reports may simply represent more general types of memories for what
usually happens (Pillemer & White, 1989).
More recent studies have provided more convincing evidence that children can
member specific events over time. Fivush and her colleagues (Fivush, Hudson, &
Nelson, 1984; Hudson & Fivush, 1991) revealed, in a series of studies, that
kindergartners could accurately recall a trip to an archaeological museum up to 6 years
later, given specific questions and appropriate cues, such as pictures. Additionally,
Fivush and Hamond (1990) compared children's recall at 2.5 years to their recall for
the same events at age 4 and found that 4-year-olds could recall as much information
about the previously recalled events at age 4 as at age 2. Furthermore, in some
situations, age and retention interval do not seem to influence the overall amount of
information that can be recalled about a novel event by young preschoolers (Hamond
& Fivush, 1991).
Deferred imitation tasks, like those used by Bauer and her colleagues (Bauer &
Mandler, 1989; Bauer & Shore, 1987), have found that children as young as 14 to 16
months can remember the temporal and causal relations of action sequences
experienced only one time up to 6 weeks later. In elicited imitation, an experimenter
uses props to model an action or sequence of actions. Following the modeling
procedure, the props are given to the child who is encouraged to imitate the sequence
that was demonstrated by the experimenter. Children's production of target actions
and sequences immediately following modeling provides a measure of immediate
recall, whereas production of target actions and sequences after some specified amount
of time provides a measure of delayed recall or deferred imitation (Bauer, Hertsgaard,
& Dow, 1994). Much of the research which has demonstrated memory for events in
children younger than 2 has used the elicited imitation paradigm. This paradigm is
considered by many to be an acceptable alternative to cued verbal recall because the
conditions surrounding learning and testing provide no evidence that children's
behavior at testing is the result of anything but recall (Bauer & Wewerka, 1995a).
Using this paradigm, Meltzoff (1988) also has demonstrated that 14-month-old
children can recall action sequences following a delay of 1 week. Similarly Bauer et
al. (1994) have shown that 13-month-olds can retain information about temporal order
for periods of I week, and Bauer and Mandler (1989) and Bauer and Shore (1987)
have shown that 20-month-olds can reenact action sequences after periods of 2 to 6
weeks. Bauer, Hertsgaard, and Wewerka (in press) further demonstrated that 13- to
20-month-old children could "recall" specific events, that is events experienced only
one time, when they were given appropriate verbal reminders to aid retrieval (e.g.,
"You can make a windmill with that stuff. Show me how to make a windmill.").
Using another paradigm to investigate long-term memory, Myers, Clifton, and
Clarkson (1987) report that 3-year-old children who had participated in a study of
auditory space when they were 6 to 40 weeks old showed some evidence of retention
for the action sequences 2 years later, in the form of willingness to remain in the
study and number of reaches for the target object. However, with one exception,
children did not use words to relate this early experience. Perris, Myers, and Clifton
(1990) replicated the Myers et al. (1987) study using a sample of 6.5-month-old
children and a variable retention interval. Children experienced a novel auditory event
(a special rattle) one time and returned either 1 or 2 years later. Children who
returned after 1 year reached in the dark for the rattle more often than controls and
reached more than children returning after 2 years, and even children who returned 2
years later reached for and grasped the rattle significantly more often than same-age
Although these studies suggest that even very young children can remember
events substantially later in time, they do not demonstrate that children's earliest event
memories take a narrative form or that children can even verbalize about them. Such
studies also raise the distinction between implicit and explicit memory. Some
researchers (see Pillemer & White, 1989; Schacter, 1993; Squire, Knowlton, &
Musen, 1993) argue for the existence of two distinct memory systems. Implicit
memory is generally thought to be present from birth, with such memories being
unavailable for conscious reflection and not consciously encoded. Implicit memories
include knowledge for procedures and simple recognition. On the other hand, explicit
memory is thought to develop later (see Pillemer & White, 1989), with brain
maturation. The time at which the explicit memory system is considered functional is
still very controversial, with some researchers arguing for its existence as early as the
first few months of life (Mandler, 1984) and others arguing that it is not functional
until the second year of life (Pillemer & White, 1989). Explicit memories differ from
implicit memories in that they are subject to conscious reflection.
This is an important distinction to make because much of the research
examining long-term memory for infant and early childhood experiences has used
behavioral measures such as reaching, grasping, or willingness to stay in a darkened
environment, implicit measures of memory (Myers et al., 1987; Perris et al., 1990).
The clearest evidence that young children might have explicit memories for early
events comes from the work of Bauer and her colleagues (1994) and Meltzoff (1988)
which have used the elicited imitation methodology (although see Nelson, 1994).
Here it is argued that the elicited imitation paradigm is a suitable nonverbal alternative
to verbal recall (i.e., explicit memory) because (a) it measures memory for a brief,
often one-time exposure and therefore cannot be procedural memory; (b) it is unlikely
that infants' behavior is the result of priming or reinstatement of a previously acquired
response; and (c) reenactment of action sequences cannot be accounted for by
recognition because the sequences are not perceptually obvious from the available
props (Bauer et al., 1994).
Language and Its Relation to Long-Term Memory
Research examining children's verbal memories for events has provided the
best account of how memories develop and change during early childhood. One
primary research objective has been to find evidence of memory for specific events
occurring prior to the development of language. Such research would address the
structure of the memory system in younger children, and more specifically the
developing autobiographical memory system. One argument is that the reason older
children and adults have difficulty remembering events from before the age of 3 is
that memories for early events were never formed or were formed in such a way that
they are irretrievable at later periods. Such accounts purport that the memory systems
of younger and older children are not the same and that the memory process is not a
continuous one (see Pillemer & White, 1989). By examining children's verbal
memories for events occurring before the age of 3, it is possible to gather more
information concerning whether specific memories can be recalled over time. To
date, however, very little empirical research has demonstrated or tried to demonstrate
that children can verbalize about preverbally experienced events. Finding that early
nonverbal memories can be translated into more adult-like narrative forms at a later
point would provide support for the continuity of cognitive development, that is, for
the idea that all early memories are not lost with time but that some change form or
are updated to remain accessible (i.e., translated from more implicit/procedural forms
to more explicit/declarative forms, although see Mandler, 1984, regarding this
Although most research examining young children's ability to verbalize about
early occurring events is anecdotal in nature, several studies do provide limited
support that children can do so. For instance, Boyer, Barron, and Farrar (1994)
found several instances in which 3-year-olds spontaneously reported information about
the color of the laboratory room or information about the toys (color, parts) that were
present, as part of an event memory study occurring when they were 20-months-old.
However, this information was related outside of the context of the event-on the way
to the lab or getting ready to go back to the lab and not as responses to specific
questions. Furthermore, it is not the case that the information reported was stressed at
the time of the initial study nor is it likely that children encoded the names of the
colors at the time the event was experienced. Rather it seems that this information
became more verbally accessible as children became more linguistically competent and
Additionally, in the Boyer et al. (1994) study, many children anticipated some
of the actions and described what they were doing as they reenacted the event. One
child, upon entering the room, went directly to one of the props and said, "push this
down and the spaghetti comes out" (pp. 441), while manipulating the lever and
pointing to where the play-doh spaghetti should appear. At another point, he picked
up an object and said, "you need this and you put the spaghetti in this" (pp. 441). In
total, 24 of the 37 children who returned 1 to 2 years after the original event either
anticipated verbally a behavior that was to be performed or labelled the behavior as
they were performing it. It may be that even if young children are unable to construct
narrative reports of previous events on demand, they may be able to verbally recall or
anticipate some aspects of the event when adequate retrieval cues are provided (see
also Farrar & Goodman, 1992; Fivush, 1991, for similar arguments).
In the auditory space study, Myers et al. (1987) also report an instance in
which a 3-year-old child spontaneously recalled information about an event that had
occurred 2 years earlier, when instructed to "guess" what was behind the picture.
However, for the most part, they found that children were unable to verbalize about
these early experiences.
More recently, Bauer and Wewerka (1995a, 1995b) also have examined
children's ability to verbalize about a previously experienced series of activities. They
were particularly interested in whether an increased ability to verbally encode the
event (as determined by a productive vocabulary index) would facilitate 1- to 2-year-
old children's ability to access or retain events over a long period of time. Measures
of children's productive vocabulary taken when 13-, 16-, and 20-month-old children
first experienced the activities were correlated with children's memory verbalizations
about the event 1-3, 6, 9, or 12 months later, and action reenactment was correlated
with the ability to talk about the event in context. They conclude that children's
ability to verbally encode the activities when they were experienced predicted
children's ability to verbally recall them at a later point. However, memory was not
contingent on verbal encoding. For instance, although there was a relatively low
correlation between early language and later behavioral recall, children with limited
verbal ability nevertheless remembered the activities. In sum, this research suggests
that the ability to verbally encode the activities when they are experienced is not
necessary for the event to be remembered over the long term or for later verbal
expression of the memory. However, verbal elaboration does appear to facilitate
In a follow-up to the first study (Bauer & Wewerka, 1995b), children who
were 20 months when the activities were first experienced returned again when they
were 36- to 40-months old, approximately 8 months after the first delay session. The
procedure was essentially the same as in Bauer and Wewerka (1995a), with the
exception that verbal reports about the target events were elicited prior to the
opportunity for reenactment. For each activity, the experimenter held the props in
front of her, beyond the child's reach, and asked the child to verbally describe the
objects and the activity surrounding them. Children were encouraged with general
prompts such as "what else did we do?" and "and then what did we do?" Once
children either had completed their description of the actions necessary to produce the
event or appeared unable to do so, the experimenter handed the props to the child and
reminded the child of the target event: "We made an X with these. Show me how we
made an X." When compared to same-age control children, returning children
generated labels for 54 of the 108 possible activities (12 children, 9 events each);
70% of the generated labels were correct. Control children, although equally likely to
generate labels for the events, only labeled 24% correctly, suggesting that returning
children were relying on their memories for labels for the activities and were not
Returning children also were more likely to recall activities, props, and
attributes that were relevant to the activities than were control children. The amount
of information provided during the first assessment (Bauer & Wewerka, 1995a)
predicted the amount of information recalled during the second assessment 8 months
later. Overall language ability at the time the activities were experienced and encoded
(as measured by reported productive vocabulary) predicted children's ability to
verbalize about the event much later; very young children were able not only to
behaviorally indicate what they remembered about the past, but also to verbally
express what they remembered when familiar props and context were available.
However, in the Bauer and Wewerka research, it is not clear how many
children actually talked and how many and what components were verbally recalled by
the children. Nevertheless, children were able to provide verbal reports of their
memories of the activities even though they were not specifically prompted to do so.
Even the youngest children produced spontaneous verbal reports of the event.
Other research also suggests that young children can verbalize about their early
memories. Fivush et al. (1987) found that even 29- and 35-month-old children could
verbally recall specific, novel events that were experienced in the recent past (up to 3
months ago) or distant past (more than 3 months ago) before the age of 2.
Furthermore, all children were able to recount at least one event that had occurred in
the distant past, and they were able to recall as much accurate, organized information
about events in the distant past as about more recent events. Such instances of long-
term recall suggest that memories can be stored and retained for several months,
without the availability of a fully functional language system and that, in certain
instances, "language can be later superimposed on previously encoded preverbal
memories" (Nelson & Ross, 1980, p. 96).
Although there is some evidence that children can spontaneously update their
early memories to a verbal form, there is no explanation as to why some memories
are translated and others are not. It is not the case that all early memories are
expressible verbally. In support of this latter contention, Boyer et al. (1994) found
that none of the 3-year-old children who had participated in their play-doh event 1 to
2 years earlier responded to general questions about the event or to pictures or
questions regarding the objects used or actions performed in the event, prior to
reexperiencing it, although some of the children had experienced the event multiple
times at 20 months. Therefore, children could not provide an independent narrative
account of the event although it was expressible behaviorally. Also, very little
research has examined what happens to those memories over time once they are
initially recalled verbally.
The effect of memory talk on memory. One popular method of studying
children's verbal memories for events has been memory talk. By engaging young
children in conversations about past events, it has been possible to examine how
conversations about the same event change over time (e.g., content, consistency), the
duration of event memories, at what point children become active participants in
conversations about the past, the effect of parental elaborativeness during these
conversations on children's subsequent memories for these events, and how children
respond to different interviewers.
One theory that has been proposed to account for the relation between memory
talk and memory development is the social construction model. According to this
model, children learn how to remember, not what to remember, from conversations
with others about past events. This theory grew out of the Vygotskian framework
(Vygotsky, 1978) which maintains that all higher mental abilities are the result of
social interactions. These social interactions provide children with the necessary
"tools" for thought. Over time, it is thought that children internalize these tools for
personal use. Thus, much of cognitive development, in this case memory, can be
conceptualized as resulting from collaboration between children and adults, who are
more cognitively skilled.
Researchers such as Nelson, Fivush, and Hudson have attempted to account for
the development of autobiographical memory using the social construction model,
which maintains that autobiographical memories are formed through social interaction.
Through conversations about the past with their children, parents teach their children
how to remember and why it is important to remember the past. When adults first
begin to engage their children in memory talk, they provide narrative structure as well
as content information about the event. With experience, children begin to internalize
this structure and the skills for remembering and narrating events and to provide more
of the content on their own during these conversations. By the age of 3 to 4, they
generally are capable of producing independent narratives.
There is evidence that supports the social construction model. First, Eisenberg
(1985) proposed that children develop linguistic forms for memory talk by engaging in
conversations with others. In one study, she monitored the language development of
two children from the time they were 18 months until they were 36 months. She
identified three phases that children pass through in learning how to talk about the
past. In the first phase, adults provide the content and structure of these
conversations, requiring that children simply respond with a "yes" or "no" to
participate. With time, however, children begin to contribute memory information and
to become more active participants in these conversations, eventually initiating
conversations on their own. In another study, McCabe and Peterson (1991) found that
children's ability to provide a coherent narrative at 42 months was related to how their
parents structured earlier conversations at 27 and 31 months.
Furthermore, past research with parent-child memory talk has found evidence
for the existence of two different parental conversational styles of talking about the
past (Fivush & Fromhoff, 1988) and has found that style differences affect the
development of autobiographical memory (Fivush, 1991; Fivush & Fromhoff, 1988;
Hudson, 1990b; McCabe & Peterson, 1991; Reese & Fivush, 1993; Reese, Haden, &
Fivush, 1993). These styles have typically been distinguished by the types of
questions asked and the amount of contextual support and elaboration provided within
the conversation. Elaborative (or high elaborative) parents talk about the past more
often with their children and ask fewer questions (most of them open-ended), while at
the same time providing their children with many details about the event being
discussed. They often continue to question their children about an event by providing
additional details with each question until the entire event has been discussed. This
style is illustrated in an example from Reese and Fivush (1993), in which a parent and
child are discussing a trip to the aquarium (p. 606). Although this child contributes
very little information, the parent continues to provide additional information:
P: Did we see any big fishes? What kind of big fishes?
C: Big, big, big.
P: And what's their names?
C: I don't know.
P: You remember the names of the fishes. What we called them. Michael's
favorite kind of fish. Big, ugly fish.
P: What kind is it?
C: urn, ba.
P: A ssshark?
P: Remember the sharks?
P: Do you? What else did we see in the big tank at the aquarium?
C: I don't know.
P: Remember when we first came in, remember when we first came in the
aquarium? And we looked down and there were a whole bunch of birdies in the
water? Remember the names of the birdies?
P: Nooo! They weren't ducks. They had on little suits. Penguins. Remember
what did the penguins do?
C: I don't know.
P: You don't remember?
P: Remember them jumping off the rocks and swimming in the water?
P: Real fast. You were watching them jump in the water, hm?
On the other hand, characteristics of repetitive (or low elaborative) parents
include (a) asking repetitive questions without providing many details; (b) topic
switching; (c) asking simple yes-no questions; and (d) pursuing one specific answer.
P: How did we get to Florida, do you remember?
P: How did we get there? What did we do? You remember?
P: You want to sit up here on my lap?
P: Oh, Okay. Remember when we went to Florida, how did we get there? We
went in the____?
C: The ocean.
P: Well, be-, when we got to Florida we went to the ocean, that's right, but
how did we get down to Florida? Did we drive our car?
P: No, think again, I don't think we drove to Florida. How did we get down
there, remember, we took a great big ? Do you remember? (Reese & Fivush,
1993, p. 606)
Stated differently, parents who use a more elaborative style tend to emphasize
the wher, when, why, and how of an event, whereas parents who are more repetitive
and less elaborative emphasize the who and what (Bauer & Wewerka, 1995a).
Importantly, research has consistently demonstrated that parents who are more
elaborative and provide more information have children who later recall more
information and construct more elaborate narratives themselves (Hudson, 1990b;
Fivush, 1991; McCabe & Peterson, 1991; Reese & Fivush, 1993; Reese et al., 1993).
For instance, Fivush (1991) examined the effect of mother-child conversations about
the past on the development of narrative abilities for those same events. Specifically,
she examined whether the amount of information that mothers provide about past
events in their conversations influences the amount of information that 2.5-year-old
children later provide and whether mothers' narrative organization during early
conversations affects children's later verbal organization of an event. She found no
evidence that the amount of information mothers provided influenced the amount of
information that children could later recall about an event. However, she did find a
relation between the amount of information that mothers provided per conversational
turn (i.e., narrative density) and the amount of information per turn that children were
later able to provide.
Also, it is not the case that children of more elaborative parents are simply
recalling the information provided by their parents in the earlier conversations (rote
learning). Rather it seems that more elaborative parents are teaching their children
what information is important to include in memory narratives and how to structure
memory narratives. They may also be providing their children with subtle messages
regarding the social value of reminiscing about the past (Fivush, 1991; Nelson, 1989,
1993). For instance, parents who provide considerable information about the past
during memory talk may convey to their children that memory of one's past is
Interestingly, all parents demonstrate characteristics of both styles when
engaged in talk about the past with their children. However, parental differences are
most evident when children indicate a willingness to continue the conversation, such
as responding to a yes-no question or repeating after the parent, without providing
memory information. In this case, elaborative parents are more likely to provide
additional information, and repetitive parents either repeat the same question or switch
to a new topic. When children provide information or do not respond, no parental
differences are evident.
Although research on parental style differences in memory talk emphasizes the
role of social factors in the development of autobiographical memory, little research
has examined the role that children may have in eliciting a particular interactional
style from the parent. That is, this effect may also be bidirectional (Barron, 1994).
Not only may parents be influencing how children learn to think about events by the
international style they use, but children may also be influencing the style that their
parents use when they talk to them. For instance, McCabe and Peterson (1991) argue
that the relation between children's discourse skills at 27 months and parental language
style at 42 months was the result of parents fine-tuning their talk to best suit their
children's abilities, or what they believed to be their children's abilities.
On the other hand, although Reese and Fivush (1993) found that maternal
elaborativeness during memory talk was related to children's responses at each of four
time points, there was very little evidence that children's responses during earlier
interactions affected mothers' level of elaboration for any of the conversations.
However, it is possible that children who are more interested in social interaction than
others are more likely to elicit an elaborative style (Barron, 1994; Fivush & Reese,
1992) and that children who are highly impulsive, active, or easily distracted may
elicit more directive speech from parents, resulting in a more repetitive parental style.
It is also possible that the language abilities of children may affect parental style
although attempts to establish a relation between language skill and maternal
conversational style have not been informative. For instance, Hudson and Sidoti
(1988, as cited in Hudson, 1990b) found no relation between maternal style and the
mean length of utterance (MLU) of 24- and 30-month-old children. Also, Reese et al.
(1993) concluded that children's language ability was not related to maternal style
because they found no relation between 40-month-old children's MLU and maternal
style. However, it is still possible that children's ability to initiate a conversation and
to maintain a topic may affect parents' perceptions of their children's language and
cognitive skills, and these perceptions may be reflected in how they talk about the past
with their children (Barron, 1994; see also McCabe & Peterson, 1991). Therefore,
one possibility is that parents of children who are more linguistically competent try
harder to elicit responses from their children by providing more content and additional
details to help their children.
Not only do social construction theorists advocate the importance of memory
talk for the development of narrative skills, but they propose that memory talk about
events also reinstates memory for events over time (Nelson, 1992). This idea,
borrowed from research with physical reactivation (Rovee-Collier & Hayne, 1987),
suggests that memory for an event is strengthened each time the event is discussed. In
general, the reactivation model considers the effects of experience more broadly and
has typically used physical reactivation to demonstrate the effect of experience.
According to this view, it is not necessary for an event to be reexperienced in its
entirety; rather, the information that is reexperienced activates additional memory, and
the entire memory is strengthened (see also Sheffield & Hudson, 1994). From an
information-processing perspective, recall of an event memory may activate nodes and
paths in a semantic network and strengthen them in memory. Those nodes and paths
that develop through reactivation make the memory more accessible to additional
reactivation over time. Consequently, events that have been recalled or reactivated
more frequently are more accessible to later recall because there are more and stronger
available paths to facilitate retrieval of information. Furthermore, nodes of
information that are activated just by thinking about the event are also strengthened.
Therefore, repeated experience with an event strengthens memory and extends the
length of time that a memory will be accessible (see Hudson 1990b, for a complete
Research examining physical reactivation supports the contention that the more
often an event is experienced or part of it is reexperienced, the more memorable that
event becomes and the more resistant to forgetting (e.g., Bauer et al., 1994; Boyer &
Farrar, 1995; Farrar & Boyer, 1995; Hudson, 1990a, 1991a). Most of the support
for this particular model comes from research by Rovee-Collier and her colleagues
(Rovee-Collier & Hayne, 1987), which examined infant memory using the mobile
conjugate reinforcement paradigm. Using this paradigm, 3-month-old infants can
remember a novel event, making a mobile move, up to 4 weeks later if they are
reminded using reactivation.
Simply reexperiencing an event once can also facilitate memory. Rovee-
Collier and Hayne (1987) demonstrated that a learned behavior can be strengthened
and retained for longer periods of time if infants are reexposed to the original context
prior to when forgetting occurs. Similar effects also have been found for older
children (Fivush & Hamond, 1989; Sheffield & Hudson, 1994). For instance, Fivush
and Hamond (1989) found that physical reinstatement facilitates 24- and 28-month-old
children's memory for a novel event. In this particular study, children participated in
a novel event one time. Then, half of the children returned 2 weeks later to
reexperience the event. All children returned 3 months later to reenact the event.
Children having repeated experience with the event performed better 3 months later
than children having only a single experience, indicating that reinstatement in the form
of additional experience facilitates long-term memory.
Crucial for the success of physical reactivation is the amount of time that has
passed since the event was last experienced. Rovee-Collier and Hayne (1987) and
Hudson and Sheffield (1993) argue that reactivation is time-window dependent and
that reminders are most effective when they are presented close to the point of
forgetting; this length of time is age-dependent because children remember for longer
intervals as they grow older. Hudson and Sheffield (1993) found that 18-month-old
children who reenacted an event either 2 or 8 weeks later recalled more event
components 2 months later than children who reenacted the event immediately after it
was originally experienced. Furthermore, 8 to 12 months later, children who had
reenacted the event 2 weeks after the event recalled more activities than children who
had reenacted the event immediately after it occurred.
Verbal reinstatement and its relation to long-term memory. More recently, it
has been proposed that memory talk about events can reinstate children's memory for
events (Nelson, 1992). Research examining physical reactivation supports the
contention that the more often an event is experienced or part of it is reexperienced,
the more memorable that event becomes and the more resistant to forgetting (e.g.,
Bauer et al., 1994; Boyer & Farrar, 1995; Farrar & Boyer, 1995; Hudson, 1990a,
1991a). One way that early memories may remain accessible is through verbal
reinstatement using memory talk. Whereas Rovee-Collier and Hayne (1987) argue
that it is theoretically possible to keep a memory alive through periodic reexperience
with the event, it may be that memory talk or verbal reinstatement can also act to
make the translation of memories from child form to adult form more likely.
Whereas Nelson (1993) has argued that children cannot use language as a
means to reinstate memory until the age of 4, both Fivush and Hamond (1989) and
Hudson (1993) believe that this may occur earlier. According to Hudson (1990b,
1993), memory talk is likely to reinstate young children's memories when they can
participate in conversations about past events, some time between the ages of 2 and 3
years. She found that 24- and 30-month-old children's repeated conversations with
their mothers affected their recall with an experimenter, such that events that had been
discussed with the mother were more readily discussed with the experimenter than
new events. However, prior discussions with the mother did not increase the amount
of information that children provided when talking with the experimenter; that is,
children did not incorporate information the mothers provided into their own accounts.
She concludes that memory talk may affect 2-year-olds' memory differently than
preschoolers' because 2-year-olds do not remember more about events that have
previously been discussed. Rather, she argues that memory talk does facilitate
memory for events but that the effect is more general, for instance, affecting
children's ability to contribute information during memory conversations. In her
study, children more readily discussed events that had been discussed than events that
had not been discussed, but repeated conversations did not affect the amount of
information that they verbally recalled.
According to Nelson, an event that is discussed during memory talk is more
likely to be remembered at a later point or to be incorporated into the autobiographical
memory system than an event that is not discussed. However, whereas researchers
examining the effects of physical reactivation can document the effectiveness of
additional experience in infants as young as 3 months, there is currently some debate
regarding when memory talk or verbal reinstatement becomes an effective means of
extending the duration of early childhood memories. Additionally, less is known
about verbal reinstatement in general, for instance whether verbal reinstatement also is
time-window dependent and if verbal reinstatement and physical reactivation operate
within the same time frames.
In sum, to date the dominant theory explaining young children's memory
development is the social construction model because of its developmental emphasis
on how children learn to think and talk about the past as a function of
environmental/social factors. Although there is some discussion of how memory talk
reinstates memory (cognitive factors), there is little empirical research to support this.
Because so little research has examined verbal reinstatement, it is unknown whether
verbal reminders function in the same manner as physical reminders and whether
young children benefit from them.
To date despite the fact that most adults are unable to remember events
occurring prior to age 3, a number of studies have demonstrated that, at least during
childhood, very young children do retain memories of novel events over relatively
extended periods of time. Amount of experience with the event and the amount of
time that has passed since the event was experienced or reinstated have been identified
as important variables which affect the retention and ultimate recall of early event
memories. The ability to use language competently may also be a prerequisite skill
for permanent long-term memories (Nelson, 1989), but the relation between language
and memory, including the contribution of parental conversational style, is not well
understood. Although anecdotal evidence indicates that children can remember events
experienced while they are preverbal or transitional, few laboratory studies have
examined children's ability to verbally recall events occurring prior to the onset of
language or mature narrative abilities, and none have investigated the role of verbal
reminders in young children's verbal and behavioral memory. The age at which
children can benefit from verbal reminders about events has implications for the
development of autobiographical memory.
Description of the Current Study
The current study addressed several issues regarding the relation between long-
term memory and language: (a) can 2-year-old children benefit from verbal reminders
about a novel event; (b) does parental conversational style (i.e., elaborativeness) affect
the extent to which 2-year-old children benefit from memory talk; and (c) do
children's language abilities affect their ability to remember, and more specifically to
verbalize about the event?
The goal of the present study was to examine more specifically how language
affects young children's memory for events. As an extension to the social
construction model, it was proposed that talking about an event serves several
functions for children: (a) it teaches children how to think about events and remember
them; (b) it reactivates memory for the event so that the memory remains accessible
for longer periods of time, making it easier for children either to independently recall
it in the future or to recall it with someone else; and (c) it models narratives about
events so that children can better express their memories using the correct labels,
temporal order, and descriptions for the more novel components. One partial
explanation for why children may have difficulty talking about events experienced
preverbally, except in rare instances, is that they do not have experience mapping
language onto those memories. Many memories that children do spontaneously update
linguistically have some significance for them or may have been reactivated at some
point after they became more linguistically competent, making it easier for those
memories to be translated to more mature and enduring forms.
Although hypotheses have been proposed regarding the potential relation
between language and memory, little research has been done to support them, with the
exception of research examining memory talk and the development of narratives. In
the current study, verbal reminders, in the form of parent-child memory talk about a
novel event, were used to examine this relation and the claim that children younger
than 3 or 4 do not benefit from verbal reminders.
In the present study, twenty-eight 24-month-old children engaged in a
moderately complex, novel event. Half of the children (12) were randomly assigned
to a verbal reminder condition in which they talked about the event once a month with
the same parent across a 3-month delay interval. Parents of the remaining children
(16) were instructed not to talk to their children about the event (no reminder
condition). Three months later, children returned to the laboratory to verbally and
behaviorally reenact the event.
The present study is the first to examine the effect of memory talk and
parental conversational style on young 2-year-olds' long-term memory for a novel
event and to examine the claims about reinstatement set forth by other researchers.
Nelson (1993) argues that if memory talk functions as rehearsal only, repeated
conversations should strengthen memory only for those components that are discussed,
and children should begin to incorporate information discussed previously into their
own accounts. Research with older preschoolers indicates that repeated conversations
do not rehearse specific memories but that children recall new information at
consecutive sessions and do not incorporate information provided by their mothers into
their own accounts (Fivush & Hamond, 1990; Fivush, Hamond, Harsch, Singer, &
Wolf, 1991; Hudson, 1990b).
Although Hudson (1993) examined memory talk in 24- to 30-month-olds and
also found that young children do not incorporate information provided by their
parents into their own memories of an event, the present study differs from that
research in several important ways. First, in that research the mean age of children at
the time the events were first discussed was 27 months. Therefore, the age of the
children in the present study is somewhat younger. Second, parents discussed
naturally occurring events with their children for the measure of memory talk. Thus,
there was very little control over which events were discussed because all children
experienced different events. Also, there was little control over when the events were
experienced as long as half of them had been experienced in the recent past (1 to 6
months previously) or in the remote past (6 to 10 months previously). Therefore,
some of the events were discussed as soon as 1 month after they were experienced,
and others were not discussed until 6 or 10 months later. There was also no way to
determine whether and how often parents had already discussed these events with their
children prior to the study. In the present study, all children experienced the same
event in the same way at the same age. There was also more control over
circumstances surrounding the event and memory talk about the event. For instance,
parents of children in the reminder condition were instructed not to discuss the event
with their children except during designated times, and other parents were instructed
not to discuss the event with their children during the delay period at all. Therefore,
there was more control over whether the event was discussed and how often it was
The present study further differs from previous research because it examines
long-term effects of memory talk on young children's memory for the event. Memory
for the event was assessed 3 months after the event was experienced allowing for the
examination of long-term effects of memory talk on children's memory. Also, the
present study allowed for an assessment of the effect of memory talk independent of
children's verbalizations about the event. Past research with memory talk has used
children's verbalizations during memory talk as the primary measure of children's
memory. However, in the present study, it was possible to examine behavioral
memory for the event, in addition to verbal recall, to assess the effect of verbal
reminders on children's memory. Finally, the present study allowed for an
examination of the relation between children's language skills and their ability to
verbalize about the event and to benefit from verbal reminders about the event.
The current study also controls for problems of novelty, complexity, and
maturation that have plagued other studies which have used laboratory events to
examine young children's long-term event memory. Specifically, maturation problems
have existed because an event that is age-appropriate for younger children becomes too
easy for them over time, thus making it difficult to determine if performance is
attributable to memory. One way to control for this potential problem is to create an
event that is novel for both younger and older children by using objects in a way that
would not be obvious to older children (Meltzoff, 1988). The current study utilized
activity names and both common and uncommon objects that 2-year-olds would find
interesting and not too familiar. Additionally, some more common objects were used
in ways that were not typical of how children would ordinarily use them or see them
used (e.g., painting with combs).
Hypotheses. Several patterns of results regarding children's ability to benefit
from verbal reminders and to verbally and behaviorally recall the event were possible.
First, if 2-year-old children benefit from verbal reminders about a novel event, 2-year-
old children who talked about the event with the parent should verbally report and
reenact more of the event than children who did not. However, if verbal reminders
do not affect 2-year-olds' memory, children in both the reminder and no reminder
conditions should perform comparably 3 months later. Such a finding would provide
support for Nelson's theory that 2-year-olds cannot use language to represent or to
reinstate an event because they lack the representational abilities necessary to do so.
However, because memory talk research has not examined the relation between
memory talk and behavioral recall, even if verbal reminders affect 2-year-olds'
memory, it was not known whether verbal reminders also would affect their
behavioral memory for the event. For instance, if verbal reminders only affect
children's verbal expression of the event, behavioral recall should not differ between
the two groups, since presumably all children would have experienced the event the
same way. On the other hand, if memory talk reinstates memory, children in the
verbal reminder condition should also have better behavioral recall of the event than
other children, since presumably their memory for the event would have been
activated and strengthened each time the event was discussed.
Second, studies with preschoolers have shown that parental conversational style
affects children's ability to verbalize about the event; children of more elaborative
parents have more elaborate verbal memories of events than children of low
elaborative parents. If parental conversational style affects 2-year-olds' memory as it
does preschoolers', children of high elaborative parents should have more elaborate
verbal memories for the event than children of low elaborative parents.
Third, it has been proposed that children's ability to verbalize about event
memories is a function of their current language abilities and the amount of experience
that children have verbalizing about the event, and the inability to verbalize about
memories is not the result entirely of retrieval deficits (Mandler, 1990). If children
simply have difficulty expressing themselves because they have not talked about a
specific experience, talking about the event with the parent after it occurs should help
children formulate ways to verbally express their memories and increase the likelihood
that they will report more novel aspects of the event or that their verbal recall will
correspond to their behavioral recall. Also, it would be expected that children having
more sophisticated language and narrative skills would be better able to verbally
describe their memories.
The examination of whether children's language abilities are related to their
ability to verbalize their memories has important implications for infantile amnesia. It
is possible that verbal memory is contingent on verbal skills and that children who
have more sophisticated language skills are the ones who provide verbal accounts of
their memories. Such a finding would suggest that the verbal translation of early
memories is dependent on exposure to the verbal form of that event. One implication
then is that many early memories become irretrievable because they were not
translated to more verbal forms.
Twenty-eight 24-month-old children (14 females and 14 males) participated in
the current study (M = 24 months, 7 days on the first visit; range = 23 months, 3
days 25 months, 11 days). There were 8 males and 4 females in the verbal reminder
condition and 6 males and 10 females in the no reminder condition. Children were
recruited from an existing pool of volunteer parents who had responded to
advertisements or solicitations. The majority of the children and parents who
participated were white, middle-class residents of a small, university-affiliated
Eight additional 27-month-old children (5 girls, 3 boys; M= 27 months, 0
days; range = 26 months, 26 days 27 months, 4 days) were recruited as controls in
order to assess the validity of performance during reenactment at the memory session.
An additional 12 children were recruited but were not included in the final
sample. Five children (1 from the no verbal reminder condition and 4 from the verbal
reminder condition) participated in the first session or the first two sessions but did
not return to complete the final session--four were too busy, and one had moved. A
sixth child, who participated in the no verbal reminder condition, was not included in
the final sample because he did not speak English. A seventh child, who participated
in the verbal reminder condition, was not included in the final sample due to
uncooperativeness. The remaining five children were in the verbal reminder condition
and participated in all sessions but were not included in the final sample because of
problems with the verbal reminder sessions. One parent lost her child's tape after
making the recordings, one parent taped over the memory talk conversations, for two
parents, the verbal reminder sessions were not recorded, and one child destroyed the
Half of the children were randomly assigned to the verbal reminder condition
(12), and half (16) were assigned to the no verbal reminder condition prior to the first
session. All parents were asked to return with their children 3 months later for verbal
and behavioral recall. Parental conversational style of talking about the past was
determined at the time of the first session.
The event was a moderately complex set of four activities organized around a
central theme of "making pictures." Table 2-1 provides a description of the event.
The length of the event was similar to the overall number of activities that researchers
have typically used with young children. Each individual activity within the event
consisted of three actions, making each activity approximately the same length as
Making Pictures Event
Preliminary action: put on art smock
Activity 1: "make a glue picture"
a) squeeze colored glue on paper plate
b) fold plate in half and press
c) glue on colored construction paper shapes
Activity 2: "make a collage"
a) put paper clips on edge of paper
b) roll stamp on paper
c) glue straws on paper
Activity 3: "do sand art"
a) remove "sticky paper" from clear shoe box container
b) sprinkle colored sand with salt shakers onto sticky paper
c) prop picture on easel
Activity 4: "paint a picture"
a) add glitter to paint dishes
b) dip comb in paint and paint on paper
c) dip sponge shapes in paint and paint on paper
Closing actions: take off art smock and get a sticker
individual activities used in elicited and deferred imitation studies with young
children. All children experienced the event the same way with a trained research
assistant to prevent differences in the way that the event was experienced.
All children participated in three laboratory sessions over a 3 to 4 month
period of time: (a) the initial visit when the event was experienced; (b) a verbal recall
session 3 months later (M = 3 months, 7 days delay); and (c) a behavioral recall
session 3.5 months later (M = 3 months, 21 days delay). The behavioral recall and
verbal recall sessions were scheduled 2 weeks apart to avoid interference of verbal
recall with behavioral recall or participant fatigue. The verbal recall session always
occurred first so that the objects and actions from behavioral recall would not affect
Session 1. Children's first visit occurred when they were 24 months old.
Following a warm-up play session to establish rapport, all children participated in the
event with a research assistant in an adjoining laboratory room. The event was 15 to
20 minutes in length and required children to be active participants. One parent was
present in the room to increase the child's comfort level. A chair was provided in the
room so that parents could watch unobtrusively. Parents were instructed not to
prompt specific actions; however, they were allowed to encourage their children to
follow the experimenter's instructions.
Following the event, parents and children returned to the first room. To assess
parental conversational style, parents were asked to talk to their children about two or
three novel one-time events that had occurred within the past several months. Parents
and children either sat on the floor or sat on a couch to talk. Topics that parents
could discuss with children were limited to one-time events that lasted no more than
one day. They were asked to avoid discussing routine events or those that were
extended in time, such as an extended vacation. Also, they were asked to discuss
events that they had personally experienced with their children and to avoid talk of
unshared events. The experimenter instructed parents to discuss the events with their
children as they normally would discuss past events at home. No restrictions were
placed on the amount of time that parents were to talk. Parents were instructed to
indicate when they were finished talking to their children. After providing the parent
with instructions, the experimenter left the parent and child alone in the room with
some books. Conversations were audiotaped for purposes of later analysis.
Parents of children in the verbal reminder condition were further instructed to
talk about the target event with their children at this time. The initial conversation
between the child and the parent provided an indication of the child's initial ability to
engage in talk about the event and provided an account of all event-related information
discussed by parents (i.e., reminded). This was considered to be the first reminder
Because it was thought that children's own language abilities would affect the
way that the event was encoded and children's subsequent ability to verbalize about
the event, parents also were asked to complete the MacArthur Communicative
Development Inventory for toddlers (CDI) (Fenson et al., 1993). This instrument is
appropriate for assessing productive vocabulary, words children say, in children aged
16 to 30 months. The inventory consisted of a checklist of 680 vocabulary terms
from 22 different categories (e.g., animals, body parts, descriptors). Parents were
asked to indicate the terms their children produced by checking off the appropriate
words. This instrument has been externally validated through comparison with
observational measures (e.g., Bates, Bretherton, & Snyder, 1988; Bretherton,
McNew, Snyder, & Bates, 1983; Fenson et al., 1994) and is considered to be an
accurate assessment of the linguistic abilities of young children (Bretherton et al.,
1983). An estimate of each child's productive vocabulary was obtained by counting
the number of words that parents indicated their children currently produced.
To assess the impact of verbal reminders on 2-year-old children's long-term
memory for the target event, at the close of the session parents of children in the
verbal reminder condition were provided with an audiotape and asked to discuss the
event with their children once a month in their home as if it were any other event that
they would discuss during periods of reminiscing. To ensure that all parents in this
condition engaged their children in monthly conversations about the event, parents
received a reminder phone call each month (after one month and two months) prior to
the time that the event was to be discussed. At this time, parents also were reminded
about the activities and props from the event. Conversations between children and
parents were audiotaped. Parents of children in the verbal reminder condition were
asked not to talk to their children about the event except during the designated times.
Parents of children in the no verbal reminder condition were asked not to talk to their
children about the event by initiating or encouraging such conversations.
Finally, to control for out-of-lab intervening experience, all parents were asked
not to engage their children in any activity that was similar in nature to the activities
of the event. They were also provided with forms to record any event-related verbal
utterances or behaviors that their children made or initiated during the 3-month delay
interval. Because only one parent indicated that her child evidenced memory for the
event outside of the lab, this measure will not be further addressed.
Five female experimenters conducted the initial sessions with the children at 24
months and were aware of children's assignment to a reminder condition. However,
because parental style was determined using information calculated during the first
sessions, all experimenters were unaware of parents' conversational style.
Session 2. The second session occurred at the lab 3 to 3.5 months after the
first session, when children were 27 months old. The purpose of this session was to
determine if children could verbalize about the target event after an extended delay
and to get an updated account of children's verbal abilities. After rapport had been
established using picture books, children were interviewed by an experimenter about
the "making pictures" event, using a core set of general and specific questions (e.g.,
"Can you tell me how we made pictures before?" "Tell me how we made the glue
picture." "Did you use colored glue?"). All children were asked about all four
activities. This interview also served as a verbal reminder about the event for all
Following the verbal interview, the experimenter left the room, and children
and parents participated in a 5-minute play session that was used to assess children's
MLU at 27 months. Parents were instructed to play with their children as they
normally would at home. Parents and children were provided with a box of Fisher
Price airport toys to facilitate play and conversation. These sessions were videotaped
and later transcribed for purposes of analysis.
Finally, parents once again completed the vocabulary inventory by indicating
the words that their children currently produced. Parents of children in the verbal
reminder condition returned their verbal reminder audiotapes at this time.
Session. All children and parents returned to the lab approximately 2 weeks
following the verbal interview to reenact the event. At this time, all parents also
returned any records that were kept regarding their children's memory for the event.
The purpose of this session was to determine if children would evidence behavioral
memory for the target event. After the experimenter had established rapport with the
children, they were taken back into the laboratory playroom and asked to demonstrate
how to make the pictures they had made on their first visit. Children were
encouraged to look around the room at all of the activity stations and to make pictures
as they had before. The arrangement of the room was the same. Children's
behavioral recall and verbal recall were assessed at this time. Parents were present
but were instructed not to cue or prompt their children in any way. All cuing or
prompting was done by the experimenter using general memory probes (e.g., "What
did we do to make pictures before?") and prompts (e.g., "How did we make this
picture?" or "Do you need to do anything else?") (see Boyer et al., 1994). In
general, specific behaviors were not prompted. Exceptions were made if children
became frustrated because they did not know how to use a particular object. Actions
performed as the result of a specific prompt were excluded from analysis (e.g., if
child used the glue bottle correctly after being shown), but actions performed
following the performance of a prompted action were not excluded, for instance if the
child folded the plate after the experimenter showed her how to use the glue bottle.
Control children participated in an equivalent session. After rapport had been
established, control children were told that they were going to go to a different room
where they would be able to make pictures. Children were then taken into the
laboratory room and given the opportunity to explore and to engage in any of the
activities. The arrangement of the room was the same. Children were encouraged to
examine objects at each of the four tables and to play with the toys as they liked. All
target behaviors occurring within a 15 to 20 minute period of time or until children
indicated boredom or unwillingness to cooperate were scored.
Four of the same experimenters and three additional experimenters conducted
the second and third sessions. When possible, the same experimenter conducted both
session 2 and session 3 for the same child to increase the child's comfort level. To
prevent experimenter bias effects, only the primary investigator knew children's
assignment to a reminder condition. Experimenters were not told of a child's
assignment to a reminder condition prior to that child's follow-up sessions, and in
many cases, a different experimenter conducted the first session and the follow-up
sessions. Also, the primary investigator made reminder phone calls to parents of
children in the reminder condition and scheduled all of the follow-up sessions so that
the remaining experimenters would be unaware of children's assignment to a reminder
Children's linguistic ability. To determine the complexity of children's
language abilities, specifically the extent to which they would be able to narrate about
an event or to elaborate on an event (Bauer & Wewerka, 1995a), children's productive
vocabulary as determined by parental report on the vocabulary inventory from the first
and second sessions was scored by counting the total number of words that each
parent indicated his or her child produced (possible range = 0 to 680). Additionally,
children's utterances during the parent-child play session from session 2 were scored
for mean length of utterance (MLU) by dividing the total number of morphemes
produced by the total number of utterances children made in that period of time. A
morpheme was defined as the smallest unit of analysis, such as an individual word or
word endings (e.g., cat = 1 morpheme, cats = cat + s = 2 morphemes). Mean
length of utterance provides an indication of the complexity of sentences being used
by a child at a particular time and was used in the present study to determine the
complexity of children's narrative abilities at 27 months.
Parental syle. All tapes were transcribed verbatim, according to the guidelines
of the CHILDES system (MacWhinney & Snow, 1990). Transcripts were then scored
according to the following coding schemes.
It was hypothesized that parental conversational style, specifically the
elaborativeness with which parents talked to their children about past events, would
affect children's long-term memory for events. To classify parental style, all parent-
child conversations occurring during the first session were transcribed. As with Reese
and Fivush (1993), the beginning and ending of each conversation about each event
was determined. Because some parents talked about fewer than three events or more
than three events, only the first two conversations about events provided by each
parent that fit the criteria were scored. Past research examining parent-child
conversations about the past has required that children provide at least two codable
unique pieces of information during each event conversation to be included.
However, because children in the present study were younger than children in other
studies and thus less verbal, and because the primary interest was in parental
conversational style rather than children's ability to contribute to conversations, this
was not a requirement in the current study. Excluded from analysis were routine
events (e.g., going to the store), those that had a storyline, such as movies (e.g., the
Lion King movie), or events that lasted more than one day (e.g., an extended family
vacation). Exceptions to this were made if parents talked about a portion of a
vacation (e.g., getting on the airplane to go see Grandma) or if they talked about
going to a movie when it was a unique experience and the storyline was not discussed.
For purposes of the current study, talk of the child's 2-year birthday, Christmas
(e.g., Christmas day or going to see Santa Claus), and Halloween was not excluded
because it was not likely that such events would be confusable in time for 2-year-olds
although such events have been excluded in studies of older preschoolers.
Parental utterances were scored according to the following categories (Reese &
Fivush, 1993; Reese et al., 1993). The coding units for most types of utterance
codes were independent clauses, with each unique or implied verb constituting a new
proposition (Barron, 1994; Reese & Fivush, 1993). Although only a few utterance
types, specifically repetitions and elaborations, were important for the current study,
coding was exhaustive, with all codes being mutually exclusive.
1. elaborations: utterances which structured conversation by focusing on a new
aspect of the event or adding additional information. Elaborations took the form of
questions or statements.
2. repetitions: questions or statements which repeated the gist or the exact
content of previously mentioned information. An utterance was scored as a repetition
only if it provided no new information about the event under discussion.
3. information requests: any who, what, where, when, and why questions that
requested specific information about the event (e.g., "Who came with us to Disney
4. clarifications: utterances which required the child to provide acoustical
clarification and generally took the form of a question (e.g., "you did what?" or
5. memory prompts: questions or statements that requested additional
information from the child without providing any new information (e.g., "do you
remember? and "tell me about it.").
6. yes-no questions: questions which required the child to provide affirmative
or negative response (e.g., "Was Sally there?"). For the purpose of the present study,
questions fit this categorization if the information presented in the questions was not
new information. If new information was presented, the question was scored as an
7. evaluations: utterances which commented on the accuracy of the child's
previous statement or confirmed the child's statement (e.g, "that's right." or "very
good."). Evaluations were coded by instance of occurrence and limited to one per
conversational turn since parents tended to provide evaluations in succession (e.g.,
"that's right, very good.").
8. associative talk: utterances that were related to the event being discussed but
not specifically about it. Such utterances took the form of future talk, fantasy talk, or
general information about the event under discussion. Associative talk was coded by
instance of occurrence and limited to one per conversational turn.
9. metamemory comments: comments that referred to the process of
remembering (e.g., "you probably don't remember that because it was so long ago.").
10. off-topic talk: comments that were not related to the event under discussion
(e.g., "we'll go to the store when we are done talking.").
Each proposition was considered to be a separate codable unit, even if it
occurred in the same conversational turn with another proposition. Utterances
containing tag questions were not considered to be questions and were scored as
statements (e.g., "We went to Disney World, didn't we?"). Additionally,
"statements," that is utterances that do not take the standard form of a question, were
scored as questions if they functioned as such, that is if the intonation of the utterance
was question-like and the child either was expected to respond, as determined by at
least a 2-second pause, or if the child did respond as if it were a question. A 2-
second pause is the established amount of time that a child has after a question has
been asked to answer a question without being credited with a "no response" (Reese &
Fivush, 1993). Therefore, if the parent provided an utterance and paused for at least
this period of time, this was assumed to imply a question and was scored as such.
Additionally, parental repetitions of children's previous utterances were scored as
questions if they sounded like questions, that is followed a question-like intonation
pattern. Based on these scores, an elaboration ratio score (the measure of parental
style) was calculated for each parent-child dyad by dividing the total number of
elaborations provided across the scorable conversations by the total number of
repetitions used in those same conversations (Barron, 1994).
Parent target event codes. Parents' conversations with their children about the
target event (session 1 and reminder conversations) were scored for content
informatinoin. Each unique unit of information discussed about the event was scored:
(a) activity-any of the activities in the event (4 possible); (b) prop--any of the props
directly related to the event (17 possible); (c) action--any of the actions related to the
activities (12 possible); and (d) setting and related information--any information
pertaining to the building or room where the event took place or to the circumstances
surrounding participation in the event (14 possible). It was possible for an utterance
to contain multiple units of information. Content was scored independently for each
of the three conversations and also summed across the three conversations to obtain an
overall account of information discussed at least once overalll content).
Consistency measures were calculated using content information. In general,
consistency was scored by summing the number of overlapping units of information
discussed in consecutive conversations, divided by the total number of unique units of
information produced in the two conversations (Hudson, 1991b). A score of 1.00
indicated that the information provided across conversations was identical, and a score
of 0 indicated that there was no overlap (Hudson, 1991b).
Two different consistency scores were calculated using the content information
obtained from parents. First, an overall consistency score for each parent was
derived. This measure provided an account of how often the same units of
information were discussed by parents on each of the three conversations during the 3-
month delay interval. Also, consistency measures were used to examine the relation
between the information that was recalled by children at different sessions and the
information previously discussed by parents. In this case, the parent measure was
The total number of unique event-related units of information discussed by
parents across the three conversations comprised the compilation score (i.e., this was
the same as the sum of overall content). This score was used to determine how many
pieces of information about the event were discussed by the parent at least one time in
the 3-month delay period.
Child target event codes. To determine the extent to which children in the
reminder group contributed to conversations about the target event, children's
responses were scored according to the following categories:
1. no response-children did not respond. A no response was scored if the
child did not respond to a parent's question within 2 seconds of its being asked.
2. response-children responded to the question but did not provide new
memory information. Some examples of responses included memory placeholders ("I
don't know."), evaluations (answering "yes" or "no" to a question), and unintelligible
utterances as part of an appropriate conversational turn.
3. memory response-children provided new information about the event in
response to a question (usually a memory prompt or request for information).
4. offers of information-children provided new information spontaneously or
in response to a general prompt (i.e., not in response to a specific question) (Fivush &
All children's memory responses and offers of information about the targti
event at session 2 and session 3 also were scored for content, using the same protocol
described for parents. For the reminder children, a compilation score was derived to
determine how much of the event (total number of unique pieces of event-related
information) was recalled by each child across all verbal reminder sessions. To
examine the extent to which children incorporated information provided by the parent
during the reminder sessions into their own recall, consistency scores were calculated,
as described earlier, for the percentage of parental contributions that were repeated by
children in later sessions (Hudson, 1991b).
The purpose of session 2 was to determine the extent to which children could
verbalize about the target event after 3 months had passed. Children were interviewed
by an experimenter regarding the specifics of the target event. The content of
children's recall about the target event was scored for event-related units of
information. Children's "yes" responses to yes-no questions asked by the experimenter
were not scored as memory responses unless children were able to provide additional
information about the event.
Behavioral memory codes. The purpose of session 3 was to determine whether
children could behaviorally recall the target event after an extended period of time.
There was also the opportunity to assess event-related verbalizations made by children
while engaging in the event. If 2-year-olds are able to verbalize about a previously
experienced event, it was expected that they would be more likely to verbalize about
the target event in the presence of the objects (in context) than in a different context
(session 2) when contextual support was minimal.
Children's reenactment of the event was scored for (a) total number of actions
performed and content, and (b) correct sequencing of target actions within each of the
four activities (see also Boyer et al., 1994; Boyer & Farrar, 1995). For sequencing
scores, only the first occurrence of each target action was scored in order to prevent
artificial inflating of temporal sequencing resulting from chance or trial and error. As
with similar studies, the number of actions recalled is known to affect the production
of correctly sequenced pairs of actions, rendering the two dependent measures not
completely independent of one another.
For sequencing, children were given credit for correct performance of a pair of
actions if sequencing approximated the event's order (see Boyer et al., 1994; Boyer &
Farrar, 1995). There were eight possible pairs that could be sequenced, with each of
the four activities having two possible pairs of actions that could be sequenced. This
measure allowed children to receive credit for correct sequencing without being
penalized for omitting an action. If children repeated an action or performed two
actions out of order, credit was only given for the first occurrence of the action or not
given at all.
Children's verbalizations during behavioral recall were of two types: (a)
verbalizations indicative of memory for the event (mnemonic); and (b) verbalizations
not indicative of memory for the event (nonmnemonic) (Bauer & Wewerka, 1995a).
To be of the first type, the child needed to (a) ask a question or make a statement
regarding a nonobvious target action, (b) verbally anticipate a target behavior with or
without performing the accompanying action, or (c) request an unseen target object or
action (e.g., Where is the ...?). The second category included verbal behaviors
wherein children (a) labeled a perceptually available obvious object; (b) labeled or
commented on a nontarget action that they could perform or were performing with an
available object; (c) made a general request or comment; (d) immediately repeated
something the experimenter had said, even if it was event-related; or (e) made an
For each coding scheme, at least two coders established reliability. For
parental style, one coder scored all transcripts for parental style, and a second coder
independently scored approximately 20% of the transcripts (6). Reliability between
the two coders was 84%.
Reliability was also established for information discussed during the verbal
reminder sessions by children and their parents and for children's performance during
the behavioral recall session. For information discussed by parents and children
during the reminder sessions, one coder scored all of the verbal reminder sessions,
and a second coder scored all sessions for three of the children (25%). Reliability for
information discussed by parents was 88%, and reliability for information recalled by
children was 88%. Because contributions by children were so low, this number
represented one disagreement.
For behavioral recall, one coder scored all of the behavioral recall tapes, and a
second coder scored approximately 20% (6). The percentage of agreement for
number of actions recalled was 100% and for actions sequenced was 95%.
Analyses were conducted to address the following issues regarding the relation
between long-term memory and language: (a) can 2-year-old children benefit from
verbal reminders about a novel event; (b) does parental conversational style (i.e.,
elaborativeness) affect 2-year-old children's ability to benefit from memory talk; and
(c) do children's language abilities affect their ability to remember, and more
specifically to verbalize about the event? Parental style will be discussed first,
followed by the results for behavioral recall and verbal recall.
In general, parental style was calculated from the first two events discussed at
session 1 that fit the designated criteria. For six parents, style was classified using
only one novel event; five parents talked about only one event that fit the criteria, and
one other parent in the verbal reminder condition only talked about the target event.
The elaboration ratio was determined by dividing the number of elaborations
provided during these conversations by the number of repetitions provided. An
elaboration ratio of 1.50 was used to classify parents as high elaborative or low
elaborative; this score fell at approximately the median point when all scores were
ranked from lowest to highest. Past research has used either a median split (Barron,
1994) or a set value to assign parental style classification (Hudson, 1993; Reese et al.,
1993). Reese et al. (1993) used a set value of 1.00 to classify parental conversational
style. In the present study, an elaboration ratio of 1.50 indicates that high elaborative
parents used at least 50% more elaborations than repetitions in the present study.
The average elaboration ratio was 1.81 (range: .07 7.25); fifteen parents
were classified as low elaborative (M = 0.80), and 13 were classified as high
elaborative (M = 2.98). The average elaboration ratio for the reminder condition was
2.35 and for the no reminder condition was 1.42. Although children were randomly
assigned to a reminder children prior to determining style, parents of children in the
reminder condition were more elaborative than parents of children in the no verbal
reminder condition. However, this difference was not significant, E (1, 26) = 2.24,
12 > .05, and resulted from a few extremely high elaborative scores in the reminder
group and because there were more low elaborative than high elaborative parents in
the no reminder condition. Five parents of children in the reminder condition were
classified as low elaborative (M = 0.58), and seven parents were classified as high
elaborative (M = 3.61). In the no reminder condition, 10 parents were classified as
low elaborative (M = 0.92), and 6 parents were classified as high elaborative (M =
2.25). There was no significant difference between the reminder-low elaborative and
no reminder-low elaborative groups or between the reminder-high elaborative and no
reminder-high elaborative groups, ps > .05. These differences between the groups
were not considered extreme enough to affect the outcome of any of the analyses,
particularly since many of the analyses also examined general effects of parental style
as a continuous variable.
Comwnparison of Groups for Behavioral Recall
This section consists of three different sets of analyses: (a) analyses of
variance to examine group differences for behavioral recall; (b) correlational analyses
to examine the relation of language and parental style to behavioral recall for all
children and for children in the reminder condition; and (c) consistency measures to
examine the relation between information discussed during the reminder sessions and
children's behavioral recall.
To assess the effect of verbal reminders and parental conversational style on
behavioral recall (action recall and sequencing), children's behavioral recall was
analyzed in two 2 (Parental style: high elaborative vs. low elaborative) x 2 (Reminder
condition: verbal reminder vs. no verbal reminder) between-subjects analyses of
variance. There were no significant effects for reminder condition (reminder
condition: M = 6.00 actions, 2.25 pairs; no reminder condition: M = 6.06 actions,
2.13 pairs) or for parental style (high elaborative: M = 5.42 actions, 1.87 pairs; low
elaborative: M = 6.57 actions, 2.47 pairs) for number of actions performed (12
possible) or for number of actions sequenced (8 possible), s > .05. See Tables 3-1
and 3-2 for the number of actions recalled and sequenced.
An additional comparison of same-age control children (8) to experimental
children (30) was analyzed in a one-way analysis of variance to ensure that
experimental children's performance was attributable to memory. These results show
that experimental children remembered the event after a 3 month delay and were not
simply "figuring out" the event. Experimental children performed significantly more
Actions Performed byv Children
Reminder condition M SD N M SD N
Reminder 5.57 2.47 7 6.60 2.97 5
No reminder 5.25 1.08 6 6.55 1.26 10
Note. The maximum number of actions that could be recalled was 12.
actions than control children, E (1, 34) = 26.84, p1 < .01 (control group: M = 2.25
actions; range: 0.0 4.0 actions; experimental group: M = 6.03; range: 2.0 9.5
actions) and sequenced the event significantly better than the control group, E (1, 34)
= 28.80, p < .01 (control group: M = 0.13 pairs; range = 0- 1 pairs; experimental
group: M = 2.18; range =1 4 pairs). In sum, although parental style and verbal
reminders did not facilitate behavioral recall of the event, there was evidence that
experimental children were remembering the event 3 months later, when compared to
Pairs of Actions Correctly Sequenced by Children
Reminder condition M SD N M SD N
Reminder 2.14 .90 7 2.40 1.34 5
No reminder 1.50 .55 6 2.50 1.18 10
Note. The maximum number of pairs that could be sequenced was 8-- 2 pairs for each
of the 4 activities.
Overall relation of language and style to behavioral recall. Although there
were no significant group effects, correlational analyses were carried out to detect any
patterns of responses or behaviors that were not apparent in the analyses of variance.
Because parental style is thought to have a general effect on memory, it is possible
that children of high elaborative parents may have more elaborate memories for events
than children of low elaborative parents, even for events that have not been discussed
with others. For children in both the reminder and no reminder groups, correlations
were used to examine the relation between behavioral recall at session 3 and parental
style at session 1. Caution should be used in interpreting the following correlations
given the number of correlations that are reported and the small sample size of some
of the groups.
There was no relation between behavioral recall and parental style, r (26) = -
.20, V > .05, indicating that parental elaborativeness did not facilitate or affect
behavioral memory for the event. Furthermore, vocabulary at session 1 and session 2
was not related to behavioral recall, r (26) = .05 and .07, ls > .05, respectively.
Therefore, neither parental style nor children's language abilities predicted children's
behavioral memory for the event at 27 months.
Relation of language and style to behavioral recall for the reminder group.
Although there were no significant differences between the reminder and no reminder
groups, indicating that verbal reminders did not facilitate behavioral recall, it was still
possible that repeated conversations about the event were affecting what information
children in the verbal reminder condition recalled. The following correlations were
conducted to examine further the relation between verbal reminders and children's
behavioral recall: (a) children's behavioral recall with parental style; (b) children's
behavioral recall with vocabulary at 24 months and 27 months; (c) amount of
information discussed by parents during the reminder sessions (using the parent
compilation score) with information behaviorally recalled by children; and (d) amount
of information recalled by children during the reminder sessions (using the child
compilation score) with information behaviorally recalled by them.
First, regarding the relation between parental style and behavioral recall, there
was not a significant relation for the reminder children, r (10) =-.21, V > .05.
Second, vocabulary at 24 months did not predict behavioral recall, r (10) = -.09, p
> .05, and vocabulary at 27 months when children recalled the event did not predict
behavioral recall, I (10) = -.14, 1 > .05. Therefore, there was no relation between
parental style or children's language skills and children's ability to behaviorally recall
To examine whether the reminder conversations affected children's behavioral
recall, the total amount of information discussed by each parent was correlated with
children's behavioral recall. A compilation score was derived for each parent to
reflect the total number of event-related units that were discussed during the three
reminder conversations (e.g., setting information--'we went to the University a long
time ago," activities-"you made a glue picture," props--'there was colored sand in
salt shakers," and actions--'you put the glue on the plate'). The mean number of
units discussed across all conversations was 20 (range: 7-29); parents with lower
scores may also have been talking about the same information each time rather than
incorporating new information on subsequent conversations. Although the correlation
between the amount of information discussed by parents during the reminder sessions
and behavioral recall was not significant, r (10) = .31, p1 > .05, this correlation was
in the expected direction, indicating that increased exposure to information during the
reminder sessions may facilitate behavioral recall.
The relation between the amount of information recalled by children during the
verbal reminder sessions was also correlated with children's behavioral recall at 27
months. A compilation score was derived for each child to reflect the total number of
event-related units that were discussed during the three reminder conversations, as
described earlier. The mean number of units discussed across all conversations was
1.5 (range: 0 5). Eight of the 12 children in the reminder group recalled
information during the verbal reminder sessions. There was a tendency for children
who contributed more information during the reminder sessions to behaviorally recall
more information, r (10) = .43, p > .05.
Consistency of behavioral recall with reminded information. In order to
determine whether children in the reminder group were behaviorally recalling
information discussed earlier either by parents or themselves, consistency scores were
computed for (a) information discussed by parents during the verbal reminder sessions
and behavioral recall at session 3; and (b) information recalled by children during the
verbal reminder sessions and behavioral recall at session 3. For two children,
consistency scores did not include information discussed from session 1; for one
parent, this conversation was not recorded, and for the second parent, the conversation
about the target event was omitted from session 1. For all other parents, consistency
was calculated between session 1 and the 1-month conversation and between the 1-
month conversation and the 2-month conversation. Overall consistency (i.e.,
information provided at all three sessions) was .30 (range: 0.00 to 1.00). Therefore,
in general, parents did not provide their children with the same information each time
they talked about the event with their children.
Also, it is important to note that children's consistency across the verbal
reminder sessions was very low. Only five children talked on more than one session,
and four children did not contribute any information during these discussions.
Eighteen units of information (e.g., props, action, activities) were contributed by
children during these conversations. Of the eight children who contributed
information, 58% of the information that was provided had been previously mentioned
by parents at an earlier session. Therefore, children were recalling some units of
information that had been reminded at least 1 month earlier.
Regarding the relation between what was discussed by parents and behaviorally
recalled by children, consistency ranged from 0.00 to .78 (M = .45), indicating that
children were performing many actions that had not been specifically discussed by
parents. However, there was considerably higher consistency when the relation
between what children remembered and what parents discussed was examined more
generally. For instance, nearly all of the actions performed by children (M = .91)
had been discussed more globally by the parents. That is, if children performed an
action from the paint picture activity, it was very likely that parents had discussed
making the paint picture with the child, even if the specific action had not been
discussed. For 10 of the 12 children, there was 100% agreement between behavioral
recall of specific actions and general activity information discussed by the parent.
However, these findings are considerably less impressive when children's performance
is examined in relation to the performance of children who were not reminded about
the event. Children in the no reminder condition were as likely to perform the actions
as children who had been reminded about them. Also, 10 of the 12 parents in the
reminder condition discussed all four of the activities during the reminder sessions,
increasing the likelihood that children would recall information from an activity that
had been discussed earlier. Table 3-3 provides an account of the number of children
in each group who performed each action at session 3.
A second consistency analysis compared information discussed by children
during the verbal reminder sessions and behaviorally recalled at session 3, using the
eight children who provided information during the verbal reminder sessions. There
was 13% agreement between the actions recalled and the information that children
provided during the verbal reminder sessions. Therefore, many of the actions that
were behaviorally recalled were not recalled by children during the reminder sessions.
Given the earlier finding of the relation between parents' recall and children's
performance, it is likely that the information verbally recalled here by children was
the same information discussed by parents during those same conversations.
Summary The results of the analyses with behavioral recall indicate that 24-
month-olds can remember a complex novel event for periods of up to 4 months.
However, in general, verbal reminders and parental conversational style did not
facilitate their recall of this event. Children who were verbally reminded about the
event did not evidence better behavioral memory for the event than children who did
not receive verbal reminders. Additionally, elaborativeness of parental style during
memory talk when children were 24 months did not affect children's behavioral
memory for the event, and children having higher vocabularies when the event was
experienced did not have better behavioral memory for the event.
Furthermore, amount of information discussed by parents during the verbal
reminder sessions did not facilitate behavioral memory. However, information
discussed during these conversations was reflected in behavioral recall. Many of the
children performed the same actions or activities that were discussed during these
conversations, and only two children performed activities and actions that had not
been discussed. However, these results are less impressive when compared to the
performance of the no reminder group. Children who were not reminded were as
likely to perform as many actions and the same actions as children who were
Comparison of Groups for Verbal Recall
This section consists of four different sets of analyses: (a) analyses of variance
to examine group differences for verbal recall; (b) correlational analyses to examine
the relation of language and parental style to verbal recall for all children and for
children in the reminder condition; (c) consistency of verbal recall for all children;
and (d) consistency of verbal recall with reminded information.
There is evidence from other studies that how parents talk about the past with
their children influences how children come to be able to talk about the past
themselves (e.g., what types of information they provide and how much information
they provide). For instance, more elaborative parents may provide better language
models for their children, making it more likely that children will be able to verbalize
their memories. To assess the effect of verbal reminders and parental style on
children's verbal memory for the event at sessions 2 and 3, children's verbal memory
responses were analyzed in two 2 (Parental style: high elaborative vs. low elaborative)
x 2 (Reminder condition: reminder vs. no reminder) between-subjects analyses of
variance. For session 2, there were no significant effects of reminder condition
(reminder condition: M = .58; no reminder condition: M = .38) or parental style
(high elaborative: M = .23; low elaborative: M = .67) on children's verbal recall, all
ps > .05. However, in general, verbal recall at session 2 was low because very few
children were able to provide information. Only six children recalled information at
this session, and four of these children were in the reminder group.
A second analysis assessed children's memory verbalizations made during
behavioral recall at session 3. These verbalizations consisted of anticipatory labelling
of actions, activities, and props. Once again, there were no significant effects for
reminder condition (reminder condition: M = .92; no reminder condition: M = 1.50)
or parental conversational style (high elaborative: M = 1.15; low elaborative: M =
1.33) on children's ability to talk about the event, ps > .05. These results suggest
that children who were verbally reminded about the event were no more likely to
verbally recall the event 3 months later than children who were not verbally reminded
and that parental elaborativeness did not facilitate verbal memory for the event. See
Table 3-4 for the number of memory verbalizations provided by children in session 2
and session 3. Fourteen children verbally recalled information at session 3; six of
these children were in the reminder group.
Memory Verbalizations at Session 2 and Session 3 by Group
M SD N M SD N
Reminder 0.43 0.54 7 0.80 1.79 5
No reminder 0.00 0.00 6 0.60 1.08 10
Reminder 1.00 1.29 7 0.80 0.84 5
No reminder 1.33 0.82 6 1.60 3.06 10
Note. The means represent the total number of actions (12 possible), props (17
possible), activities (4 possible), and related information (14 possible) verbally
recalled by children about the event.
Overall relation of language and style to verbal recall, Correlations also were
used to examine further children's verbal recall for the event. First, there was no
relation between parental style and verbal recall at either session 2, r (26) = -.06, or
session 3, r (26) = -.13, ps > .05, indicating that once again, parental
elaborativeness did not affect memory for the event.
One possibility that was proposed is that children's ability to talk about an
event may be less a function of children's memory for the event than their language
abilities. Therefore, it would be expected that children having higher MLUs and
vocabularies at 27 months would be better able to express their memories. It was
important to address whether children's language abilities when the event was
experienced and when they were required to recall the event were related to their
ability to verbally recall or talk about the event. As with behavioral recall, the
relation between vocabulary at 24 months and vocabulary and MLU at 27 months and
verbal recall was examined. The mean vocabulary for children at 24 months was 249
words (range: 9 542) and at 27 months was 407 words (range: 31 619). By 27
months, the average MLU was 2.47 (range: 1.0 4.6), as determined by free play at
session 2. There was a significant relation between vocabulary at 24 months and
verbal recall at session 2, r (26) = .57, p <.01. There also were marginally
significant relations between vocabulary at 24 months and verbal recall at session 3, r
(26) = .35, and between vocabulary at 27 months and verbal recall at session 2, r
(26) = .36, ps < .10. However, there was not a significant relation between
vocabulary at 27 months and verbal recall at session 3, r (26) -= .19, p > .05, or
between children's MLUs at 27 months and their verbal recall at session 2 and at
session 3, ps > .05. Therefore, vocabulary at the time the event was experienced
predicted children's ability to verbalize about the event at session 2 more than
children's vocabulary at the time of recall.
The results indicate that children with higher vocabularies at the time the
event was experienced were better able to express their memories verbally to the
experimenter (see Bauer & Wewerka, 1995a, for a similar argument). However, it
should be noted that, in general, verbal recall at session 2 was very low and did not
occur often; only four children in this condition expressed any verbal memory for the
event at session 2. In sum, children's language abilities at the time the event was
experienced predicted their ability to verbally recall the event 3 months later, but
language abilities in the form of MLU and vocabulary at 27 months did not strongly
predict verbal recall at that time although all correlations were in the expected
There was also no relation between the amount of information verbally recalled
at session 2 and session 3, r (26) = -.01, p > .05, indicating that children were not
responding similarly at both sessions. Correlations also were used to examine the
relation between behavioral recall and verbal recall at sessions 2 and 3. First, there
was not a strong relation between behavioral recall at session 3 and verbal recall at
session 2, r (26) = .31, p > .05. Also, the relationship between behavioral and
verbal recall was not higher for verbal recall at session 3, r (26) = .19, p > .05.
Thus, there was not a strong relation between what children verbally recalled and what
they behaviorally recalled, indicating that children were not necessarily recalling the
same information about the event over time.
Given that no effects were found for parental style, and other studies have
found a positive relation between elaborativeness of the parent and elaborativeness of
older children's memory at later points, additional analyses were conducted to account
for the discrepancies in findings. One possible explanation concerned the elaboration
ratio, which consisted of both an elaboration score and a repetition score. For
instance, a parent who was classified as high elaborative could actually have provided
many elaborations or only a few, as long as the parent provided at least 50% more
elaborations than repetitions. Therefore, it was possible that some high elaborative
and low elaborative parents actually provided their children with the same number of
elaborations. However, there was no relation between number of elaborations and
verbal recall at either session 2, r (26) = -.07, or session 3, r (26) = .08, ps > .05.
Similarly, there was no relation between the number of repetitions and verbal recall at
either session 2, [ (26) = .04, or session 3, r (26) = .14, ps > .05. Therefore,
there was very little about the parents' actual style at 24 months (elaborations and
repetitions) that predicted children's subsequent verbal recall.
It was also possible that parental style represented nothing more than how the
parent talked to the child at that particular point in time and was not representative of
a global conversational style characteristic. Given that few studies have examined the
relation between parental style and children's event recall in young 2-year-olds, little
is known about the relation between the two for children this age. It is possible that
although parental style has positive effects for older children, parental style may have
little or no effect on children this young. It is also possible that parents of children
this age have not engaged in memory talk often enough to have developed a stable
style. To address the possibility that parental elaborativeness was related to a child
characteristic, such as language, rather than to the parent, the relation between
parental style and child language (vocabulary at 24 and 27 months and MLU at 27
months) was examined.
There were significant negative correlations between parents' style (as
determined at session 1) and children's vocabulary at session 1, r (26) = -.39,
children's vocabulary at session 2, r (26) = -.50, and children's MLU at session 2, r
(26) = -.47, all ps < .05. These results suggest that parents who were more
elaborative at session 1 also had children who had lower vocabularies at session 1
(i.e., at the time parents and children talked about events) and that these same children
continued to have lower language skills than other children 3 to 4 months later. One
explanation is that some of the parents of low language children may have
compensated for what they perceived to be their children's inability to contribute to
the conversation by providing more elaborations and asking fewer questions that
requested specific information (i.e., more repetitive parents are typically after specific
information and ask the same questions repeatedly in an attempt to get their children
to provide that information.). Although the traditional interpretation of parental
elaborativeness with older children has not demonstrated this pattern, it is possible that
some parents who perceived either that their children were competent language users
(i.e., higher vocabularies) or had good memories provided fewer elaborations and
repeated themselves more often, indicating an expectation that their children should be
able to answer.
To address the possibility that low elaborative/repetitive parents were asking
more questions in general than high elaborative parents, parental style was correlated
with the number of questions asked at session 1. Given that elaborations could be
questions or statements, high elaborative parents may not only be providing their
children with more information, but they may be asking fewer questions. This
hypothesis was not supported. First, there was no relation between parental style and
the number of questions asked by parents, r (26) = -.19, p > .05, indicating that
high elaborative and low elaborative parents were not asking different numbers of
questions. Second, additional analyses also indicated a positive relationship between
the number of questions asked and the number of elaborations provided, r (26) = .57,
and between the number of questions asked and the number of repetitions provided, r
(26) = .66, ps < .01, respectively. Thus, it was not the case that elaborations were
less likely to be questions, and high elaborative parents were as likely to ask their
children questions as low elaborative parents.
Another possibility was that elaborative parents were asking questions but not
waiting for (i.e., expecting) answers or that they answered their own questions more
often. To examine whether there were differences in children's ability to respond to
parental questions, the number of children's responses to parental questions was
examined. Responses were any verbalizations that encouraged parents to continue
with the conversation and included memory responses, repetitions, and unintelligible
comments that were not clearly off topic. Analyses revealed no relation between
parental style and children's responses, r (26) = .04, p > .05; low elaborative
parents were not more likely to have children who responded to their questions.
Additionally, there was no relation between parental style and children's failure to
respond, r (26) = .02, p > .05; high elaborative parents were not more likely to
have children who failed to respond. There was, however, a positive relation
between the number of questions asked and the number of children's responses, r (26)
= .84, p < .05; that is, children were likely to respond in some way to their parents'
questions that were asked. Therefore, the second hypothesis that high elaborative
parents either did not ask their children as many questions or provided more answers
themselves in response to what they perceived to be their children's language
inadequacies was not supported.
Although the present research did not analyze the types of questions that
parents asked their children, one possibility is that these parents asked their children
questions that they could answer without needing many language skills (e.g., yes/no
questions). Research also has demonstrated that parents ask their children fewer
yes/no questions and more open-ended questions as children acquire more verbal skills
(Reese & Fivush, 1993). In the present study, parents may have been more
elaborative and may have asked easier questions to compensate for their children's low
vocabulary and language skills. See Table 3-5 for a summarized account of
significant and nonsignificant correlations.
O\ 00 *vt
0 x-, 0
r d 0 e (S
%0 %O 0 1"-- o -,
e- 0 0
ir- mt (o 0 rM -
I I I I i
^ s s * ,
S .> .c 'u u
Relation of language and style to verbal recall for the reminder group.
Although there were no significant differences between the reminder and no reminder
groups, indicating that verbal reminders did not facilitate verbal memory for the
event, it was still possible that repeated conversations about the event affected what
information children in the verbal reminder condition recalled (Hudson, 1991b, 1993).
Therefore, correlations also were used to examine further whether verbal reminders
affected children's verbal recall. These correlations included: (a) correlating
children's verbal recall at sessions 2 and 3 with the amount of information discussed
by parents during the verbal reminder conversations; (b) correlating children's verbal
recall at sessions 2 and 3 with the amount of information discussed by them during the
verbal reminder conversations; (c) correlating parental style with verbal recall at
sessions 2 and 3; (d) correlating information verbally recalled by children at session 2
and session 3; and (e) correlating amount of information recalled verbally at sessions 2
and 3 with information recalled behaviorally.
To determine whether information discussed by parents during the verbal
reminder sessions affected children's verbalizations at sessions 2 and 3, children's
verbal recall at each session was correlated with the amount of information discussed
by parents during the verbal reminder conversations (parent compilation score).
Amount of information discussed did not predict verbal recall at session 2, r (10) =
.24, p > .05, and was not related to verbal recall at session 3, i (10) = -.04, p >
.05. These findings suggest that parental elaborativeness did not affect children's
verbal recall of the event, and that parents who talked about more information did not
have children who recalled more information at recall.
To determine whether information discussed by children during the verbal
reminder sessions affected their verbalizations at sessions 2 and 3, children's verbal
recall at each session was correlated with the amount of information that they recalled
during the verbal reminder conversations (child compilation score). First, there was
a significant relation between the amount of information that children discussed during
the reminder sessions and the amount of information recalled at verbal session 2, i
(10) = .79, p < .05. There also was a marginally significant relation between
children's recall during the reminder sessions and information recalled at verbal recall
session 3, r (10) = .54, p < .10. Therefore, children's ability to contribute to
memory talk about the event during the reminder sessions predicted their ability to
recall the event with an experimenter 3 months after experiencing the event.
Because it was originally predicted that children having high elaborative
parents would be more likely to have elaborate memories of the event, particularly if
they talked about the event over the delay period, parental style also was correlated
with verbal recall at the second and third sessions for children in the reminder group.
If parental conversational style affects children's recall, children having parents with
higher elaborative styles would be expected to verbalize more about the event. Once
again, there was no relation between style and verbal recall at session 2, r (10) =-.08,
or between style and verbal recall at session 3, r (10) = -.22, all ps > .05, indicating
that children of high elaborative parents who were reminded were not more likely to
have better verbal memories of the event than children of low elaborative parents.
Although there also was not a significant relation between verbal recall at
session 2 and behavioral recall, r (10) = .27, or between behavioral recall and verbal
recall at session 3, r (10) = .37, ps > .05, all correlations were in the expected
direction. Children who verbally recalled the event did not demonstrate superior
behavioral memory for the event.
Once again, there was no correlation between the amount of information
verbally recalled at session 2 and session 3, r (10) = .26; p > .05; amount of
information recalled at one point did not predict the amount of information recalled at
a later point. Therefore, children who talked at session 2 were not more likely to talk
at session 3.
In sum, although there was no relation between what was discussed during the
sessions and later recall, and elaborativeness of parental style had no effect on
children's subsequent memory for the event, children's vocabulary at the time the
event was experienced and children's ability to participate in the reminder sessions
predicted their memory for the event. See Table 3-6 for an account of the significant
and nonsignificant correlations for the reminder group.
Consistency of verbal recall. Consistency of verbal recall was examined for all
children to determine relations between information recalled at different time points.
First, only three children provided information on both sessions, and only one of them
recalled the same information on both sessions. Therefore, verbal recall across
IS IS S-
00 -a <- c -<
I-. V- 0o (s C- i- w
I I i I I I
> e >a M (
sessions was very inconsistent, with most of the children who recalled information
only providing information at one of the two sessions. For all children, consistency
of verbal recall at both sessions to behavioral recall also was assessed. Given that the
number of children who verbally recalled information was so low, consistency was
only calculated for those 6 children who verbally recalled information at session 2 and
those 14 children who verbally recalled information at session 3. Consistency between
verbal recall at session 2 and behavioral recall was .11, and consistency between
information that was both verbally anticipated at session 3 and also behaviorally
recalled was .23. Therefore, only about 25% of the actions performed by those
children were verbally anticipated or labeled, indicating that, in general, consistency
between verbal and behavioral recall was quite low.
Consistency of verbal recall with reminded information. Consistency measures
also were used to examine the relation between information discussed during the
verbal reminders and information verbally recalled. To determine whether children
were recalling the same information from the event across recall or reminder sessions,
consistency of verbal recall was examined: (a) between information provided by the
parents over the verbal reminder sessions and children's verbal recall at session 2 and
at session 3; (b) between information provided by the child over the verbal reminder
sessions and their verbal recall at session 2 and at session 3; and (c) between verbal
recall at sessions 2 and 3.
The important question of course was whether information discussed during the
reminder sessions was later recalled during verbal recall. First, regarding the
consistency between information provided by the parents and children's verbal recall
at session 2, of the four children who verbally recalled information at session 2, 63%
of the information verbally recalled at session 2 had been discussed by parents during
the verbal reminder sessions. For the six children who verbally recalled information
at session 3 during behavioral reenactment, 47% of this information had been
discussed by parents during the verbal reminder sessions.
The relation between information provided by children during the verbal
reminder sessions and their verbal recall at both recall sessions was also examined.
For session 2, 38% of the information verbally recalled by the four children had been
recalled by them previously during the recall sessions. For session 3, 17% of the
information recalled had been recalled by them previously during the verbal reminder
sessions. It is interesting to note that nearly all of the information recalled by children
at session 2 had been previously discussed by either themselves, the parent, or both.
However, a greater proportion of information recalled by children at session 3 was
new information and had not been previously discussed during the verbal reminder
Finally, the relation between information recalled at session 2 and at session 3
was examined. Only two children provided information at both sessions. One of
them provided the same information at both sessions; 20% of this information was the
The results for verbal recall and behavioral recall suggest that verbal reminders
and elaborativeness of parental style at the time the event is initially experienced do
not facilitate long-term memory for a novel event. Children who received verbal
reminders were not more likely to verbalize about the event at either session or to
behaviorally reenact the event than children who did not receive verbal reminders. It
also was not the case that parents having high elaborative styles had children who
were better able to verbalize about the event 3 months later. Vocabulary at 24 months
predicted how well children verbalized about the event at session 2 and marginally
predicted verbal recall at session 3. However, even when children did verbally recall
information about the event, verbal recall at different sessions reflected different
For the reminder group, the amount of information discussed by parents during
the reminder sessions did not predict how much children verbally or behaviorally
recalled. Although what was discussed and recalled during these sessions was strongly
related to what children actually recalled later, children who were not reminded
performed as many actions and the same actions as children who were reminded.
Therefore, the results from the present study provide little evidence that memory talk
improves memory for an event in young 2-year-old children.
The purpose of this study was to examine the relation between long-term
memory and language in 2-year-olds. As an extension to the social construction
model, it was proposed that talking about an event could potentially serve several
functions for children: (a) it could teach children how to think about memories and
why it is important to remember; (b) it could reactivate memory for the event so that
the memory remains accessible for longer periods of time, making it easier for
children either to independently recall it in the future or to recall it with someone else;
and (c) it could model memory talk so that children can better express their memories
using the correct labels, temporal order, and descriptions for the more novel
components of the event.
Several issues were addressed in the current study. The first issue was whether
2-year-old children benefit from verbal reminders about a novel event. A second
issue was whether parental conversational style (i.e., elaborativeness) affects the extent
to which 2-year-old children benefit from memory talk. Memory talk about events
has been shown to facilitate older preschoolers' memory for events (Nelson, 1992);
more specifically the elaborativeness of parental style during memory talk has been
shown to affect children's subsequent memories for events. A third issue concerned
the relation between children's language skills and their ability to recall and verbalize
about an event. Finding that memory talk benefits 2-year-old children's memory for
events that are experienced prior to the development of a functional language system
has implications for the development of autobiographical memory and infantile
amnesia. For instance, one reason that young children may not be able to retrieve
early experienced events may be that those events were never reminded or translated
to a more verbal form (Nelson & Ross, 1980; Pillemer & White, 1989).
In general, the results of the present study provide evidence that 24-month-olds
can remember a complex novel event for periods of up to 4 months, and that they can
verbalize about the event as well as demonstrate behavioral recall. In the present
study, half of the children verbalized about the event during reenactment, and all of
the children demonstrated some degree of behavioral recall.
Effect of Verbal Reminders on Memory
Two patterns of results were possible regarding verbal reminders. First, if 2-
year-old children benefit from memory talk, or verbal reminders, about the target
event, the verbal and behavioral memory of children who engaged in memory talk
about the event should be superior to that of children who were not reminded about
the event. Furthermore, if memory talk actually reinstates the 2-year-olds' memories
as it has been proposed to do with older preschoolers (Nelson, 1992), children should
remember not just the information that parents recalled but additional information
reflecting independent memories of the event. That is, children would be expected to
remember information that their parents did not talk about.
However, if verbal reminders do not affect 2-year-old's memory, children in
both the reminder and no reminder conditions should perform comparably 3 months
later. Such a finding would provide support for Nelson's theory that 2-year-olds
cannot use memory talk to reinstate a memory because young children lack the
representational abilities necessary to reinstate their memories using the verbal
accounts of others. It is worth noting that little work has examined this proposal
regarding the relation between children's memories and their representational
understanding. Finding that children younger than 3 benefit from verbal reminders in
the present study would suggest that children do not need complex representational
abilities to benefit from memory talk.
Regarding children's behavioral recall, there was no indication that children
who were reminded about the event performed more actions or sequenced the event
better than children who were not reminded. Children who were not reminded
performed the same types of actions and the same number of actions at behavioral
recall as children who had been reminded. The same results were obtained for
children's verbal recall. Therefore, 2-year-old children did not appear to benefit from
memory talk about the event although certain patterns of responses made by children
in the reminder condition indicated that they may have benefited from memory talk
about the event. For instance, children who were reminded were highly likely to
remember or verbally recall information that had been discussed during the three
memory talk conversations. Also, children's behavioral recall matched parents'
discussions of activities nearly perfectly, indicating that children were recalling many
of the same activities that parents had discussed. However, because many parents
were discussing all of the activities, it was very likely that if children remembered the
event, there would be overlap between what parents discussed and what children
There was also no indication that verbal reminders increased children's verbal
recall of the event. Children who were reminded about the event did not provide
significantly more information about the event at either recall session than children
who were not reminded. Also, children who verbally recalled information at session
2 were not more likely to verbally recall information at session 3. Only three children
recalled information at both sessions, and two of these children were in the verbal
reminder condition. Additionally, it was not the case that children behaviorally
recalled information that they had previously recalled verbally.
Effect of Parental Style on Memory
A second objective of the present study was to determine whether parental
conversational style affected children's memory for the event or affected the extent to
which children benefited from verbal reminders. Therefore, it was important to
examine whether children's ability to benefit from memory talk was contingent on the
elaborativeness of memory talk. There is evidence that how parents talk about the past
with their children influences how children come to be able to think and talk about the
past themselves (e.g., what types of information they provide and how much
information they provide). One idea was that more elaborative parents provide better
language models for their children. Children of high elaborative parents were
expected to provide more information verbally than children of low elaborative
parents, given that they would have experienced a better model of how to think and
talk about event memories, in general.
Regarding children's behavioral recall, there was no indication that children of
more elaborative parents performed more actions or sequenced the event better than
children of low elaborative parents. There were no differences between the groups
whether parental style was examined as a dichotomous variable or as a continuous
variable. Therefore, it was not the case that parents who have high elaboration ratios
as determined at session 1 were more likely to have children who had better memories
for the event. It was also not the case that children in the reminder condition who had
high elaborative parents had better memories for the event. The same patterns held
for verbal recall; parental style did not predict verbal recall at session 2 or at session
3. There was no relation between parental elaborativeness and later memory for the
Effect of Language Abilities on Memory
A third objective was to determine whether children's ability to recall and
verbalize about an event is related to the amount of experience that they have had
verbalizing about it and to their current language and narrative abilities. If children
simply have difficulty expressing themselves because they have not talked about a
specific experience, talking about the event with the parent should help them formulate
ways to verbally express their memories and increase the likelihood that the more
novel aspects of the event would be reported or that verbal recall would correspond to
Although verbal reminders and parental style as a whole did not predict
children's memory for the event, there was some evidence that children's language
abilities was an important determinant of children's memory for the event.
Specifically, children's vocabulary at 24 months predicted their ability to verbalize
about the event 3 months later. Children with higher vocabularies when the event was
initially experienced were more likely to verbalize about the event at session 2 and
session 3. In the present study, the child who had the highest vocabulary at 24
months verbalized the most about the event at both sessions and provided the same
information. These results partially corroborate those of Bauer and Wewerka (1995a,
1995b) in that they also found that vocabulary at the time the event was experienced
predicted children's ability to verbalize about the event at a later time. However, in
the present study, vocabulary did not significantly predict children's memory
verbalizations during reenactment, possibly because many of the children talked during
reenactment regardless of their initial vocabulary. Talking about an event when there
is contextual support is less difficult for children than talking about an event when
there is no contextual support. Research with young preschoolers also indicates that
verbal recall is poorer when there is no contextual support or little contextual support
(Farrar & Goodman, 1992). In the present study, one possibility is that children who
could "verbally elaborate" on the event at the time it was experienced were better able
to express their memories verbally to the experimenter. This is an important finding
because it suggests that the ability to verbalize about an event when it is first
experienced has an effect on subsequent memory for the event. Such a finding could
potentially explain why memories from certain time periods are more difficult for
children to verbalize about than others.
However, it should be noted that, in general, verbal recall at session 2 was
very low and did not occur often; only four children in this condition expressed any
verbal memory for the event at session 2. Interestingly, language abilities at 27
months (vocabulary and MLU) predict children's ability to verbalize about the event
less well. Therefore, there was little support for the idea that the ability to verbalize
about an event is related to the children's current narrative abilities. Children who
were more verbal at the time of recall were not always more likely to verbalize about
Explanation of Results
There are several possible explanations for why children did not benefit from
verbal reminders and for why parental elaborativeness did not affect memory or
verbalizations about the event. First, one explanation is that children need to be able
to contribute to these conversations before they can use them to facilitate recall. This
explanation necessitates a bidirectional effect. It may not be sufficient for children to
simply listen to what others are telling them about an experience; rather, they may
need to play a more active role in these conversations. Few children made memory
responses during the verbal reminder sessions, and those who did provided only about
one piece of information on average.
Given that very little is known about the time-dependent nature of verbal
reminders, it may also be the case that verbal reminders were not effective for these
children because the event either was not discussed enough or was discussed too often
or too soon after the event was experienced. Research examining physical reminders
suggests that a reminder is most effective when it is presented close to the point that
the event would most likely have been forgotten (Rovee-Collier & Hayne, 1987).
Because little is known about how verbal reminders work in children, a 1-month
interval may not have been effective. Only future research will be able to address this
Other explanations for the lack of effects for parental style and memory talk
include individual differences in memory, language abilities, and motivation. For
instance, although some children evidenced good memory for the event and found the
event very entertaining, other children were either uninterested in participating or
failed to perform the actions correctly. Within the reminder group, one-fourth of the
children performed fewer than 4 of the 12 actions; however, it was not the case that
they were not using the props, only that they were not using them in the correct
manner. Some of the children, then, may have been unmotivated to use the props the
way the experimenter had demonstrated previously. For others, the event may not
have been distinctive enough to distinguish it from other types of arts and crafts
activities that they engaged in at school, although parents were interviewed about their
children's activities and the typical activities of 2-year-olds in day care.
Implications for Children's Memory
The results from the current study suggest that young 2-year-old children do
not benefit from verbal reminders about an event, and more specifically, that verbal
reminders do not act to reinstate memory for events in children this young. Although
the current study did not provide direct evidence that the inability to benefit from
verbal reminders was the result of representational deficits, as proposed in Nelson
(1992), this is a possibility. Changes in children's understanding of mental
representation are compatible with Nelson's (1992) views on the development of the
ability to use verbal reminders to reinstate memory. She argues that children cannot
use language to reinstate memory when they first begin to talk about past events with
others because they lack the facility with language needed to use another person's
verbal account of the event to set up a representation of their own. The ability to use
language as a representational tool makes it possible to exchange representations with
others and to compare one's own representations with those of other people. Prior to
age 4, she argues that talking about the past with others may only be useful to
organize information rather than to reinstate memories.
Other researchers, such as Hudson (1993), argue that what is necessary to
benefit from memory talk about an event is the ability to participate more fully in
these conversations and to direct their own memory searches. If this is the case,
memory talk should begin to reinstate children's memories for events between the ages
of 2 and 3 years.
The age at which children can use memory talk to reinstate their memories for
events has great relevance for the development of the autobiographical memory system
and for the subsequent decline of infantile amnesia. Prior to the formation of
autobiographical memory, it is claimed that memories for specific events persist only
for periods of days, weeks, or months. However, with the advent of the
autobiographical memory system, memories are organized and interpreted in relation
to their importance to the self (Fivush & Reese, 1992; Nelson, 1992). Memory talk
is useful for the interpretation of events in relation to the self and for the reinstatement
of memories, which serves to safeguard important memories against forgetting.
Without memory talk and reinstatement, many memories are subject to forgetting over
time and are not retained across the lifespan. Finding that memory talk reinstates
memories for events in children younger than 3 or 4 would suggest that durable, long
lasting memories can be formed prior to the development of the autobiographical
memory system or that the autobiographical memory system is functional at this time.
Future research will need to address the age at which verbal reminders, or
memory talk, become effective at reminding children about events and increasing the
stability and duration of memories. Although verbal reminders do not appear to have
facilitated memory for the picture making event, it is possible that more naturalistic
types of events may be more conducive for reinstatement.
Also, future research needs to examine whether children's ability to benefit
from verbal reminders is related to the development of certain representational abilities