Content uploaded by Holly A Taylor
Author content
All content in this area was uploaded by Holly A Taylor
Content may be subject to copyright.
Interactive Dimensions in the Construction
of Mental Representations for Text
David N. Rapp
University of Minnesota
Holly A. Taylor
Tufts University
To detail the structure and format of memory for texts, researchers have examined whether readers
monitor separate text dimensions for space, time, and characters. The authors proposed that the
interactivity between these individual dimensions may be as critical to the construction of complex
mental models as the discrete dimensions themselves. In the present experiments, participants read stories
in which characters were described as traveling from a start to a final location. During movement between
locations, characters engaged in activities that could take either a long or short amount of time to
complete. Results indicate that accessibility for the spatial locations was a function of the passage of time.
The authors interpret this as evidence that the interactive nature of text dimensions affects the structure
of representations in memory.
In the biography Seabiscuit: An American Legend (Hillenbrand,
2001), the titular thoroughbred and his jockey Woolf must battle
an unfair start by rival horses during the running of the Hollywood
Gold Cup:
Leaning around the far turn, Woolf drew a bead on Specify again.
Incredibly, the horse was still rolling along. A pang of fear went
through Woolf. Ligaroti was somewhere behind him; bumped and
pinched hard on the backstretch, he had been knocked too far back,
and would finish a fast-closing fourth. Woolf knew he could easily
beat the others, but he was beginning to worry that he couldn’t catch
Specify. Dropping flat in the saddle, he gave his mount two taps with
the whip and clucked in his ear. Up in the booth, caller Joe Hernandez
saw him do it, and shouted into the microphone, “And here comes
Seabiscuit!” (p. 235–236)
The situation in this paragraph suggests movement, sounds, and
events that are only implicitly described in the narrative. For
example, Hillenbrand’s description does not mention the horses’
precise locations on the track. However readers may still construct
vivid mental representations that encode the locations of Seabis-
cuit, Specify, Ligaroti, and the rest of the field. In this example,
readers’ mental representations may be influenced by, among other
things, knowledge of horseracing history, the expected speed of
thoroughbreds, and the amount of track believed to have been
covered in the description of the race. Readers’ representations of
the situation described by this excerpt may be a function not only
of spatial information (e.g., “Ligaroti was somewhere behind
him”), but also temporal descriptions (e.g., “he was beginning to
worry that he couldn’t catch Specify”) and action-based statements
(e.g., “he gave his mount two taps with the whip and clucked in his
ear”). The combination of these spatial, temporal, and activity-
based cues may or may not influence readers’ expectations that
Seabiscuit will have enough time or ground to overcome Specify.
The goal of this article is to evaluate the interactivity of available
narrative cues and the impact of such interactivity on readers’
narrative representations.
Traditional models of text comprehension outline the types of
memory representations that readers may construct. The tri-partite
theory of text representation offered by van Dijk and Kintsch
(1983) suggests that readers encode the specific words described in
a text (a surface-level representation), the ideas conveyed in
meaning-based propositional units (a text-based representation),
and the information described by a text but not directly mentioned
in a text (a situation model representation). Situation models are
believed necessary for readers to construct inferences and ade-
quately comprehend text (Glenberg, Meyer, & Lindem, 1987; van
Dijk & Kintsch, 1983; Zwaan & Radvansky, 1998). Because
readers can potentially encode many different details (e.g., char-
acter goals, temporal sequences of events, spatial relations be-
tween locations), a considerable amount of research has been
devoted to examining which dimensions readers encode into situ-
ation models. The event-indexing model has received considerable
empirical support as a description of the dimensions readers track
while reading (Zwaan, Langston, & Graesser, 1995; Zwaan,
Magliano, & Graesser, 1995). According to this model, readers
maintain mental indices for the passage of time, the organization of
David N. Rapp, Department of Educational Psychology, University of
Minnesota; Holly A. Taylor, Department of Psychology, Tufts University.
This material is based on work supported by an Across University
Departments with Information Technology (AUDIT) grant funded by Tufts
University and by the support of the Marcia Edwards Fund and a Grant-
in-Aid of Research from the Office of the Dean of the Graduate School,
both from the University of Minnesota.
We are grateful to Richard Gerrig, Sid Horton, Jeff Long, and Paul van
den Broek for their insightful discussions as the project unfolded. We thank
Joe Magliano, D. McNamara, and Maryellen MacDonald for their com-
ments on drafts of the article. We also thank Jocelyn Gutzman for her
assistance in data collection.
Correspondence concerning this article should be addressed to David N.
Rapp, Department of Educational Psychology, 206A Burton Hall, 178
Pillsbury Drive, S.E., University of Minnesota, Minneapolis, MN 55455.
E-mail: rappx009@umn.edu
Journal of Experimental Psychology: Copyright 2004 by the American Psychological Association
Learning, Memory, and Cognition
2004, Vol. 30, No. 5, 988–1001
0278-7393/04/$12.00 DOI: 10.1037/0278-7393.30.5.988
988
space, the relations and intentions of characters and objects, and
the causal structure of events in narratives.
The event-indexing model, however, makes no claim about the
processes by which these indices are encoded into complex mem-
ory representations. To address this issue, researchers have ap-
pealed to Gernsbacher’s structure-building framework (1990).
This framework outlines the processes by which readers construct
representations of events described in texts. When new events are
read, new mental substructures are constructed, resulting in a more
elaborated representation. As a result of this process, old structures
may become less accessible from memory. The process theory
suggested by the structure-building framework has been mapped
directly onto the structural theory outlined by the event indexing
model (e.g., Zwaan, Langston, et al., 1995; Zwaan, Magliano, et
al., 1995). The combination of the two theories provides one
account of the underlying cognitive processes at work during
situation model construction. As a text is read, information rele-
vant to a specific dimension is encoded. For each dimension or
index, a new substructure along that dimension may be developed,
making the old information less accessible from memory. Discon-
tinuities or expectancy violations for specific dimensions further
decrease the accessibility of prior information.
Researchers have examined the linguistic cues in texts that help
readers build mental structures for particular dimensions. To illus-
trate, one area that has received extended interest has been the use
of temporal markers in linguistic discourse. It has been suggested
that these cues can operate as signals for the encoding, application,
reactivation, and deactivation of information from mental repre-
sentations (Bestgen & Vonk, 2000; Carreiras, Carriedo, Alonso, &
Fernandez, 1997; Madden & Zwaan, 2003; Rapp & Gerrig, 2002;
Zwaan, 1996; Zwaan, Madden, & Whitten, 2000). For example,
Magliano and Schleich (2000) presented stories that included
sentences indicating either an ongoing imperfective aspect or a
completed perfective aspect. After reading one version of the
story, participants determined whether a verb phrase had appeared
in the text. Participants took longer to identify verb phrases from
stories containing completed activities than from stories containing
ongoing activities. These results suggest that the degree to which
readers encode events as completed can influence the accessibility
of those events from memory.
Temporal cues are but one dimension for which readers may
construct representations in situation models. Our contention is
that text representations are constructed through the interactivity of
multiple text dimensions that include temporal cues as only one
possibility. According to this view, other dimensions including
space, characters, and plot-based causality can impact the structure
of temporal representations. To take a specific example, character
actions in a narrative can provide cues for the temporal qualities of
an event. Activity is often associated with expectations about
behavior that are grounded in personal experience, including the
likelihood of an activity being completed, the difficulty of an
activity, and the amount of time encompassed by an activity (e.g.,
Barsalou, 1999a; Barsalou, Huttenlocher, & Lamberts, 1998; Kas-
chak & Glenberg, 2000; Rapp & Gerrig, 1999; Rapp & Gerrig,
2002; Zwaan, 1996; Zwaan, 1999). These temporal durations can
be perceived as relative to a specific, explicit period of time (e.g.,
“He cooked the popcorn in the microwave,” which may take a
predetermined amount of time as mandated by cooking directions)
or as implicit and associated with more general expectations (e.g.,
“He cut down the tree” should take an extended amount of time to
complete, whereas “He cut down the string” should take a shorter
amount of time, while neither is associated with a specific tempo-
ral duration). The implicit durations associated with character
activity may be as critical a cue for structuring temporal events in
memory as explicit descriptions of time shifts and verb aspect.
In our view, then, it should be the case that the interactions
between text dimensions matter. We suggest, therefore, that these
dimensions function interactively to influence readers’ construc-
tion and updating of situation models. According to this hypoth-
esis, interactivity will impact the construction and application of
situation models by defining the structure and organization of
events as, for example, whether those events are ongoing or
completed or whether they encompass one activity or multiple
activities. The earlier excerpt from Seabiscuit provided an example
for which real-world knowledge about the passage of time, char-
acter movement, and qualities of the environment collectively
impact readers’ expectations about the spatial situation. That is,
readers’ knowledge about a particular text dimension (e.g., the
temporal duration implied by the speed of racehorses’ movement)
may inform expectations about other dimensions (e.g., the spatial
layout or distance between the horses in the race). As we have
outlined, information in one dimension may cue readers to seg-
ment or structure memory representations along a second dimen-
sion. In Seabiscuit, information about speed may help readers to
generate expectations about where the racehorses are located on
the track.
The present research begins to examine this notion of interac-
tivity among text dimensions and whether this interactivity mat-
ters. Although there is some suggestion that readers can encode
multiple discrete text dimensions, researchers have not explicitly
investigated how one dimension (e.g., time, space, or character-
based information) may be useful for encoding a second dimen-
sion. For example, consider the following story:
1. Joe was working diligently on his term paper.
2. Joe had been laboring for quite some time in the library.
3. He began to feel kind of hungry.
4. He decided to get something to eat from the diner.
5. Joe gathered his things and left.
6. Outside, he noticed a stick on the ground.
7. He pulled out a small pocket knife from his bookbag.
8. Joe began to whittle away at the stick while he walked.
9. He carved the stick into a small flute.
10. He put the finishing touches on it just as he arrived.
11. The smells of cooking burgers and French fries made
his mouth water.
12. He stepped up to the counter and ordered a grilled
cheese sandwich.
989
INTERACTIVE DIMENSIONS IN TEXT
After reading this story, what expectations might readers have
about the relative locations of the library and diner? Evidence
suggests that readers conduct mental simulations (cf. Barsalou,
1999b; Kahneman & Tversky, 1982) to evaluate aspects of the
situations described in texts. Readers are likely to conduct a similar
form of mental simulation when considering Joe’s activity during
movement between the two locations. If Joe engages in an activity
associated with a long temporal duration (e.g., carving a stick into
a flute) while moving between the library and the diner, the
implication is that the two locations are spatially distant from one
another. Suppose, however, that Sentence 9 described a different
activity:
He carved his initials right on the stick.
In this case, readers may believe that the spatial distance between
the two locations is not very large because the process of carving
initials into a stick should take a relatively short time to complete.
For both versions of this story, the temporal durations of activities
can serve as a cue for the spatial qualities of text locations. In other
words, these dimensions can interact in the construction of readers’
situation models. Specifically, expectations and beliefs about the
durations of these events structure whether they should take a long
or short amount of time to complete, and inferences about traveled
distance are a function of those temporal estimates.
To examine the validity of our interactive claim, we evaluated
whether expectations about character activity can provide readers
with an indication about the relations between locations. We did
this by examining the accessibility of text information following
the introduction of an event shift or change in the spatial or
temporal situation described in the ongoing text. One potential
outcome of this process, which we will call the rigid-boundary
hypothesis, suggests quite broadly that event shifts result in uni-
form decreases in the accessibility of prior text information. Ac-
cording to this view, for the story about Joe described above,
accessibility for the start location should decrease in a consistent
fashion regardless of the passage of time or amount of spatial
movement by the character. There is some evidence for this view.
Certain types of temporal shifts (e.g., explicit time shifts such as
“an hour later” vs. “a day later”) result in similar decreases in
accessibility for previously mentioned information despite appar-
ent differences in temporal magnitude (Zwaan, 1996).
A competing hypothesis suggests a more interactive influence of
text dimensions on situation models following event shifts. This
view argues that readers’ real-world expectations and beliefs about
the temporal and spatial qualities of events will influence the
structure of situation models. According to this view, changes in
accessibility should be a function of the relevant magnitude, du-
ration, or size associated with an event shift. This flexible-
boundary hypothesis suggests a graded or continuous effect of
narrative shifts compared with the categorical shift suggested by
the rigid-boundary hypothesis. Locations interpreted as spatially
distant from the protagonist’s current location should be less
accessible from memory than narrative locations in close proxim-
ity to the protagonist (Levine & Klin, 2001). The flexible-
boundary hypothesis suggests that representational boundaries are
not strictly a function of the introduction of new text events and
event shifts but are also a function of readers’ background knowl-
edge and expectations about the concomitants of time, space,
characters, and causality. Again, there is some evidence for this
view: Zwaan (1996) demonstrated that particular discrepancies in
chronological distance (e.g., explicit time shifts such as “a minute
later” vs. “a day later”) can differentially influence the accessibil-
ity of text information prior to the time shift. The following studies
compared these two hypotheses as well as the degree to which shift
statements need to explicitly demarcate the distances traveled in
order to influence memory for text.
The rigid- and flexible-boundary hypotheses differ in their pre-
dictions concerning the accessibility of narrative information pre-
ceding the introduction of an event shift. However, it is important
to note that both hypotheses similarly rely on the notion that new
text events initiate the construction of mental substructures in
memory. In addition, they both share the deictic view that infor-
mation in the narrative here-and-now should be more accessible
than information from the past or future (Bower & Morrow, 1990;
Morrow, 1994; Morrow, Bower, & Greenspan, 1989; Morrow,
Greenspan, & Bower, 1987).
Thus, our experiments were designed to test the notion that
nonspatial text dimensions influence representations of spatial
locations. We examined whether interactivity between dimensions
occurs, and if so, whether it influences construction processes in a
specific way (on the basis of expectations about events and in line
with the flexible-boundary hypothesis) or more broadly (on the
basis of the uniform encoding of event shifts and in line with the
rigid-boundary hypothesis). In Experiments 1A and 1B, we exam-
ined whether explicit spatial statements generally have an impact
on the accessibility of spatial locations. These experiments pro-
vided a baseline for evaluating the interactive effects of other
dimensions on representations of spatial locations. In Experiment
2, we specifically examined whether statements conveying infor-
mation about the passage of time would influence the accessibility
of location information from memory in a similar way. In Exper-
iment 3, we evaluated whether these interactions actually matter by
examining whether results could be attributed to tracking only a
single, discrete temporal dimension. Our results demonstrate that
the interactive nature of text dimensions facilitates the construction
of rich representations in situation models.
Experiment 1A
To begin this set of studies, we evaluated whether explicit
information from a single, discrete dimension (in this case, space)
would influence the structure and accompanying accessibility of
spatial information from situation models. If so, these experiments
would provide a baseline for later evaluation of interactivity by
allowing us to compare those accessibility patterns with patterns
obtained within a single dimension. In Experiment 1A, participants
read a series of stories that described a character moving from a
start location to a final location. Each story included an explicit
statement that described characters traveling a long or short phys-
ical distance between locations (e.g., “Emily walked four miles.”
vs. “Emily walked four blocks.”). Following each story, partici-
pants completed a probe word recognition task. For critical exper-
imental items, the probe word was one of the two locations
mentioned in the story. We recorded recognition times and error
rates to probe words, as well as reading times for distance state-
ments. Our predictions for the probe task were based on the two
hypotheses outlined in the beginning of the present article. The
990
RAPP AND TAYLOR
rigid-boundary hypothesis predicts main effects of distance and
location and little in the way of an interaction. That is, recognition
times should decrease uniformly with an event shift regardless of
the magnitude of the shift. In contrast, the flexible-boundary
hypothesis predicts an interaction between distance and location
such that recognition times will differ as a function of the magni-
tude of the shift. That is, there should be a larger decrease in
accessibility for start location probes following long shifts com-
pared with the same start location probes following short shifts.
Both hypotheses predict a main effect of location: Participants
should take less time to recognize final locations than start loca-
tions. Because final locations represent the current “here-and-now”
of the story, they should remain in reader focus and accessible
from memory (Bower & Morrow, 1990). We did not have specific
predictions about error rates.
Method
Participants. Thirty-six Tufts University undergraduates participated
in this study for course credit. All participants were native English
speakers.
Apparatus. The experiment was run on two Macintosh G3 computers
using Superlab software. Participants were seated in front of a color
monitor with their hands resting on the keyboard. They used buttons on the
keyboard to make appropriate responses. The sentences were displayed in
the center of the screen in standard upper- and lower-case type.
Materials. To begin this experiment, we wrote 20 stories, each 12
sentences long (see Appendix A for examples). The first sentence intro-
duced a main character, and the second sentence introduced the character’s
starting location. The 3rd and 4th sentences provided a reason for the
character to travel to a final location. The 5th through 8th sentences served
as a transition for the character to prepare to travel to the final destination.
The actual start and final locations were not mentioned in the 5th through
8th sentences. The 9th sentence of each story explicitly described either a
short-distance statement (i.e., “Emily walked for four blocks.”)oralong-
distance statement (i.e., “Emily walked for four miles.”). Distance state-
ments always described a spatial distance of either 4 blocks or 4 miles, and
were therefore equated for length across stories (mean number of words for
long and short-distance statements ⫽ 5.25). The 10th through 12th sen-
tences described the character reaching and remaining at the final location,
without explicitly mentioning either the start or final location. Following
the final sentence of the story, participants saw a recognition probe for
either the start or final location (location probe). Across all 20 experimen-
tal passages the probes mentioned story locations and therefore required
yes responses.
In addition to the experimental stories, we also wrote 20 filler stories. All
filler stories were 12 sentences long. The filler stories described a similar
range of scenarios as the experimental stories. Although the stories did not
include character movement between two locations, several of the stories
described characters intending to travel to a location at a later time point.
Each filler story included a single recognition probe, and there was only
one version of each filler item. For the fillers, recognition probes never
occurred in the stories, and the correct response was always no. These
probes were not limited to locations, but also included objects and verbs.
We wrote three practice stories, none of which included distance informa-
tion. We also included a secondary task to ensure that participants carefully
read the stories (as well as to serve as baseline data for a future experi-
ment). After 10 randomly selected stories (5 experimental and 5 filler),
participants were instructed to write down a sentence to continue the story
by using a sheet of paper and a pen located next to the computer.
Design. Overall, there were four versions of each experimental story
varying as a function of distance statement (long vs. short) and location
probe (start vs. final). Using a Latin square design, we constructed four
story lists; thus, each story appeared in a different version on each list. The
20 filler stories were added to these lists. Each participant read one version
of each experimental story and all fillers in a different random order.
Procedure. Participants began with three practice stories to become
acquainted with the stimulus format and keyboard controls. Participants
were instructed to read each story line-by-line for comprehension. Each
story began with the prompt “Prepare for the next story” followed by the
first line of the story. Participants read through each story at their own
pace, pressing the space bar to advance. The last line of each story was
followed by a recognition probe in blue text presented in the center of the
screen and surrounded by six asterisks. Participants indicated yes (i.e.,
“This word appeared in the story.”)orno (i.e., “This word did not appear
in the story.”) by pressing either of the appropriately labeled (Q or P) keys
on the keyboard. Their response was followed by a 1,000-ms pause before
the prompt for the next story appeared. Participants had to respond to the
recognition probe within 3,000 ms, otherwise the words TOO SLOW
appeared for 2,500 ms. Their response, or lack of response, was followed
by a 1,000-ms pause before a prompt for the next story appeared on the
screen.
Results and Discussion
Table 1 presents the results of Experiment 1A. Participants
failed to respond before the deadline 1.5% of the time, and these
responses were not included in our analyses. We also eliminated
decision times falling more than three standard deviations above
the mean. This resulted in a loss of an additional 1.8% of the data.
None of the participants expressed awareness of the purpose of the
experiment. However, during the debriefing, 26 of the participants
reported noticing the preponderance of distance statements across
the experimental session. This is not surprising given the consis-
tency of the distance statements across stories and our goal of
presenting explicit spatial statements conveying distance informa-
tion. The remaining 10 participants expressed recognition only
after being told that 20 of the stories contained sentences with
explicit distance information.
To assess the reliability of our data, we carried out analyses with
both participants (F
1
) and items (F
2
) as random variables. We
analyzed recognition times for correct responses to probe loca-
tions. The results were consistent with the flexible-boundary hy-
pothesis. The interaction between location and distance was reli-
able: Participants were slower to recognize start locations after
reading long-distance statements in comparison to their recogni-
Table 1
Means and Standard Deviations for Correct Recognition Probe
Times and Distance Statement Reading Times for Start
Locations (L1) and Final Locations (L2) in Experiment 1A
Correct
recognition
probe times
Distance
statement
reading times
L1 L2 L1 L2
Long-distance statement
M 1,420 1,284 1,288 1,320
SD 291 314 618 709
Short-distance statement
M 1,347 1,368 1,393 1,373
SD 266 287 631 638
Note. All probe and reading times in ms.
991
INTERACTIVE DIMENSIONS IN TEXT
tion of start locations after reading short-distance statements, sig-
nificant by both participants and items, F
1
(1, 35) ⫽ 13.68, MSE ⫽
16,053, p ⬍ .005; F
2
(1, 19) ⫽ 6.62, MSE ⫽ 26,430, p ⬍ .05. The
main effect of location was also significant: Participants took
longer to recognize probe words for characters’ start locations
(M ⫽ 1,384 ms) than for final locations (M ⫽ 1,326 ms), signif-
icant by participants and marginal by items, F
1
(1, 35) ⫽ 4.86,
MSE ⫽ 24,558, p ⬍ .05; F
2
(1, 19) ⫽ 4.04, MSE ⫽ 15,645, p ⫽
.059. There was no main effect of distance (both Fs ⬍ 1).
To further evaluate recognition times, we conducted planned
comparisons using paired t tests (Bonferroni corrected) with both
participants (t
1
) and items (t
2
) as random variables. Following
long-distance statements, participants were 136 ms slower to rec-
ognize probes for start locations than for final locations, significant
by participants and items, t
1
(35) ⫽ 4.39, p ⬍ .005; t
2
(19) ⫽ 3.09,
p ⬍ .01. Participants were 73 ms slower to recognize probes for
start locations following long-distance statements compared with
start locations following short-distance statements, marginally sig-
nificant by participants and items, t
1
(35) ⫽ 1.95, p ⫽ .059;
t
2
(19) ⫽ 2.02, p ⫽ .058. Participants were also 84 ms slower to
recognize probes for final locations that followed short-distance
statements as compared with cases in which final location probes
followed long-distance statements, marginal by participants only,
t
1
(35) ⫽ 1.96, p ⫽ .058; t
2
(19) ⫽ 1.41, p ⬎ .10. No other
comparisons were significant (all ts ⬍ 2).
Analyses of error rates for recognition responses showed no
main effects or interaction (M ⫽ 29% incorrect for start locations
following long distances, 29% incorrect for start locations follow-
ing short distances, 34% incorrect for final locations following
long distances, and 34% incorrect for final locations following
short distances; all Fs ⬍ 1.9). We also analyzed reading times for
distance statements (e.g., “Emily walked for four blocks.” vs.
“Emily walked for four miles.”) to evaluate whether distance
effects could be identified at encoding (see also Table 1). Reading
time analyses helped suggest whether readers were expending
special effort to encode a particular type of statement. Analyses
revealed that participants took 79 ms longer to read short-distance
statements (M ⫽ 1,383 ms) than long-distance statements (M ⫽
1,304 ms), significant by participants but not by items, F
1
(1, 35) ⫽
4.53, MSE ⫽ 233,904, p ⬍ .05; F
2
(1, 19) ⫽ 2.30, MSE ⫽ 424,925,
p ⬎ .10. No other effects were significant (all Fs ⬍ 1.4). These
results suggest that readers may have had more difficulty reading
short-distance statements than long-distance statements. More im-
portantly, the reading time pattern does not match the recognition
probe data, providing little direct evidence for the encoding view.
The pattern of recognition times is more consistent with a
flexible-boundary hypothesis (compared with a rigid-boundary
hypothesis) for stories including explicit distance information.
Most surprisingly, participants demonstrated shorter recognition
times to identify final locations after reading long-distance state-
ments than after reading short-distance statements. Recall that both
the rigid- and flexible-boundary hypotheses suggest that final
locations will be equally accessible following either short- or
long-distance statements. Although this comparison did not reach
significance, we address a potential reason for this contrary result.
One possibility is that the long-distance statements may have been
particularly effective at instantiating expectations that characters
had reached their final destinations, whereas short-distance state-
ments may have been less effective at conveying a sense of
completed movement. For participants, the spatial cue “four
blocks” may not have been an appropriate distance for a character
to travel before arriving at a final location. Note that this expla-
nation also fits the reading time data.
We conducted Experiment 1B to address this possibility di-
rectly. Therefore, the purpose of Experiment 1B was two-fold.
First, we examined whether the inclusion of distance statements
that explicitly described characters arriving at their final locations
would reduce the advantage for final location probes following
long, as compared with short, distance statements. Second, we
wished to provide additional support for the flexible-boundary
hypothesis by replicating the critical interaction. Recall that
whereas the rigid boundary hypothesis predicts little in the way of
an interaction between distance and location, the flexible boundary
hypothesis suggests that recognition times for start locations
should differ as a function of intervening spatial distances.
Experiment 1B
Method
Participants. Forty-two University of Minnesota undergraduates par-
ticipated in this study for course credit. All participants were native English
speakers. Four participants’ data were eliminated for a failure to follow
experimental instructions. Two participants’ data were eliminated because
they self-reported reading disabilities on completion of the experiment.
Apparatus. The apparatus was identical to that in Experiment 1A,
except the experiment was run on three Dell personal computers equipped
with Superlab software.
Materials. In Experiment 1A, it was never explicitly mentioned that
characters had arrived at their destinations (the final locations); this infor-
mation had to be inferred. In addition, characters were described as
traveling either 4 blocks or 4 miles: For all cases, characters traveled four
units of distance. In Experiment 1B, we changed these qualities of the
experimental stories. We modified the experimental stories from Experi-
ment 1A to include explicit mention of characters arriving at their final
locations (see Appendix A for examples). To do this, we revised the ninth
sentence of each story to include the words “to his/her location” (depend-
ing on the gender of the character) at the end of the sentence. We also
changed the numerical units for the distance statements to better differen-
tiate between short and long statements. Therefore, participants read a
description of a character that had traveled either a short or long physical
distance between locations (i.e., “Emily walked three blocks to her desti-
nation.” vs. “Emily walked five miles to her destination.”). Distance
statements always described a spatial distance of either 3 blocks or 5 miles
(mean number of words for long- and short-distance statements ⫽ 8.35). In
all other ways, the materials were identical to those presented in Experi-
ment 1A.
Design. The design was identical to Experiment 1A.
Procedure. The procedure was identical to Experiment 1A.
Results and Discussion
Table 2 presents the results of Experiment 1B. Participants
failed to respond before the deadline 1.3% of the time, and those
responses were not included in our analyses. We also eliminated
decision times falling more than three standard deviations above
the mean, resulting in a loss of an additional 1.1% of the data.
None of the participants expressed awareness of the purpose of the
experiment. However, during the debriefing, 24 of the participants
reported noticing the preponderance of distance statements across
the experimental session. Twelve participants expressed recogni-
992
RAPP AND TAYLOR
tion only after being told that 20 of the stories contained sentences
with explicit distance information.
Preliminary examination of the item means revealed that par-
ticipants took much longer to respond to one particular item in
comparison with the rest of the item list. Mean responses for that
item in two of the four conditions fell either three standard devi-
ations (in the long-distance statement by final location condition)
or two standard deviations (in the short-distance statement by start
location condition) above the mean for the respective condition.
This item described a scenario in which a character was described
walking through a shopping center. For this item, unlike the others,
the two locations (a cafeteria and florist) were subsumed under a
single larger location (the shopping center), effectively creating
three locations. As well, it is unclear what it means to walk several
blocks or miles through a shopping center. The term blocks is not
very informative because a block could refer to a set of cross-street
boundaries, aisles, or even shop kiosks. In addition, although it is
plausible that one could walk several miles inside a shopping
center (as in the case of exercising), it seems less likely when the
goal is to reach a particular store in that mall. None of the other
items described a character remaining inside a particular location
while traveling to two specific subsumed locations. This item did
not yield a similar outlier pattern in Experiment 1A (in which
stories did not contain the explicit completion “to his/her destina-
tion”). For consistency, we reran all analyses from Experiment 1A
with this item removed. Removal of the item did not change the
nature of any effects obtained in Experiment 1A. Therefore, we
analyzed the data in Experiment 1B after removing this problem-
atic item.
For Experiment 1B, we first analyzed recognition times for
correct responses to probe locations. The data were, again, sup-
portive of the flexible-boundary hypothesis. Participants were
slower to recognize start probes after reading long-distance state-
ments in comparison with recognizing start probes after reading
short-distance statements. This interaction was significant by par-
ticipants and marginal by items, F
1
(1, 35) ⫽ 8.78, MSE ⫽ 12,979,
p ⬍ .005; F
2
(1, 18) ⫽ 2.99, MSE ⫽ 24,501, p ⫽ .098. For the main
effect of location probe, participants took longer to recognize
probes for start locations (M ⫽ 1,387 ms) compared with final
locations (M ⫽ 1,256 ms), significant by participants and items,
F
1
(1, 35) ⫽ 33.39, MSE ⫽ 18,597, p ⬍ .001; F
2
(1, 18) ⫽ 16.26,
MSE ⫽ 19,901, p ⬍ .005. For the main effect of distance state-
ment, participants took longer to recognize probes following long-
distance statements (M ⫽ 1,354 ms) compared with short-distance
statements (M ⫽ 1,289 ms), significant by participants but not by
items, F
1
(1, 35) ⫽ 8.78, MSE ⫽ 17,402, p ⬍ .01; F
2
(1, 18) ⫽ 2.86,
MSE ⫽ 17,706, p ⬎ .10.
We conducted planned comparisons for the data as in Experi-
ment 1A. The comparisons for the most part replicated the findings
from the earlier experiment. After reading long-distance state-
ments, participants were 196 ms slower to recognize probes for
start locations than for final locations, significant by participants
and items, t
1
(35) ⫽ 5.82, p ⬍ .001; t
2
(18) ⫽ 3.78, p ⬍ .001.
Participants were 130 ms slower to recognize probes for start
locations following long-distance statements compared with cases
in which start probes followed short-distance statements, signifi-
cant by participants and items, t
1
(35) ⫽ 5.04, p ⬍ .001; t
2
(18) ⫽
2.58, p ⬍ .05. Participants were 196 ms slower to recognize probes
for start locations following long-distance statements compared
with final locations following short-distance statements, signifi-
cant by participants and items, t
1
(35) ⫽ 5.42, p ⬍ .001; t
2
(18) ⫽
3.85, p ⬍ .001. After reading short-distance statements, partici-
pants were 66 ms slower to recognize probes for start locations
compared with probes for final locations, significant by partici-
pants only, t
1
(35) ⫽ 2.68, p ⬍ .05; t
2
(18) ⫽ 1.50, p ⬎ .10.
Similarly, participants took 66 ms longer to recognize start loca-
tion probes following short-distance statements compared with
final location probes following long-distance statements, signifi-
cant by participants and marginal by items, t
1
(35) ⫽ 2.53, p ⬍ .05;
t
2
(18) ⫽ 1.90, p ⫽ .074. There was no significant difference
between recognition latencies for final locations following either
long- or short-distance statements (both ts ⬍ 1).
We analyzed the error rates for recognition responses. Neither of
the main effects nor the interaction were reliable (M ⫽ 22%
incorrect for start locations following long distances, 18% incor-
rect for start locations following short distances, 22% incorrect for
final locations following long distances, and 23% incorrect for
final locations following short distances; all Fs ⬍ 1). We also
evaluated reading times for the distance statements (see Table 2).
Participants took 24 ms longer to read long-distance statements
(2,146 ms) than short-distance statements (2,122 ms); however,
neither of the main effects nor the interaction were reliable (all
Fs ⬍ 1.92).
Experiment 1B demonstrates that explicit distance cues influ-
ence the structure of readers’ situation models for locations de-
scribed in texts. The accumulated evidence from Experiments 1A
and 1B support a flexible-boundary interpretation rather than a
rigid-boundary hypothesis. The data revealed that participants
were slower to recognize start locations, compared with final
locations, following both long- and short-distance statements. This
finding is qualified by the fact that participants were slower to
recognize start locations following long-distance statements as
compared with when those start locations followed short-distance
statements. This result is not entirely surprising because descrip-
tions of space are readily available text cues for structuring spatial
situation models. The next experiment examined whether dimen-
sions other than space would also influence those spatial
representations.
Table 2
Means and Standard Deviations for Correct Recognition Probe
Times and Distance Statement Reading Times for Start
Locations (L1) and Final Locations (L2) in Experiment 1B
Correct
recognition
probe times
Distance
statement
reading times
L1 L2 L1 L2
Long-distance statement
M 1,452 1,256 2,118 2,173
SD 249 232 471 566
Short-distance statement
M 1,322 1,256 2,080 2,164
SD 210 203 406 542
Note. All probe and reading times in ms.
993
INTERACTIVE DIMENSIONS IN TEXT
Experiment 2
In Experiment 2, we examined whether character activities,
implicitly cuing distance, would result in effects similar to those
demonstrated for explicit spatial cues on the recognition of loca-
tion probes. Instead of reading sentences describing explicit dis-
tance statements, participants read stories that contained activity
statements normally associated with specific temporal durations. A
pattern of data similar to that observed in Experiments 1A and 1B
would suggest that cues from nonspatial dimensions can also
influence the accessibility of spatial information from readers’
situation models.
Method
Participants. Thirty-six Tufts University undergraduates participated
in this study for course credit. All participants were native speakers of
English.
Apparatus. The apparatus was identical to that in Experiment 1A.
Materials. We modified the experimental stories from Experiment 1A
to include statements describing character activity rather than statements
about spatial distance. To do this, we changed the ninth sentence of each
story to an activity statement, describing an activity that the character
engaged in while moving from the start to the final location (see Appendix
B for examples). There were two versions of this activity statement.
Long-activity statements should take a long amount of time for characters
to complete (e.g., “Elizabeth read some articles on linguistic communica-
tion.”), whereas short-activity statements should take a short time to
complete (e.g., “Elizabeth perused the journal’s table of contents.”). These
statements were equated for length (mean number of words for long and
short-activity statements ⫽ 11.1). Following the final sentence of each
story, participants were presented with a recognition probe for either the
start or final location (location probe).
We conducted an off-line norming study to evaluate whether the activity
statements we wrote were associated with long and short temporal dura-
tions. We asked 28 native English-speaking Tufts University undergradu-
ates to read each activity statement. Participants were each given a ques-
tionnaire with 20 pairs of action sentences. One sentence in each pair
described a long version of a particular action (e.g., “Sid carved the stick
into a flute.”) and the other sentence described a short version of a similar
action (e.g., “Richard carved his initials right on the stick.”) such that each
participant evaluated both the long and short versions of a statement pair.
We changed the names of the characters for each sentence in such pairs;
thus, the same characters were not performing multiple actions across the
questionnaire and as a result, participants would be less likely to compare
activity statements in a pair. We randomly placed the 40 action sentences
into a questionnaire booklet. Each page contained 8 sentences for a total of
five pages, and the pages of each questionnaire were randomly shuffled for
each participant. The instructions read “On the following pages, please
indicate in the provided blanks how long you think it would take to engage
in and complete the described action. Write your answer in minutes.”
Participants reported that long-activity statements should take longer to
complete (M ⫽ 55.33 min) than short-activity statements (M ⫽ 5.97 min),
t
1
(27) ⫽ 12.09, p ⬍ .001; t
2
(19) ⫽ 5.06, p ⬍ .001. These results suggested
that the actions described by the long-activity statements established ex-
pectations for a longer passage of time than did the actions described by the
short-activity statements. On average, participants believed that long ac-
tivities would take almost 50 min longer to complete than short activities.
These norming data provided us with the materials to pursue the more
theoretically oriented goals of the project.
We included an additional 20 filler stories to our original set to decrease
the proportion of critical items over the course of the experiment. This
resulted in a total of 60 stories: 20 experimental stories and 40 filler stories.
For the experimental stories, the correct response was always yes. For the
filler stories, correct yes probes followed only 10 of the stories. For the
remaining 30 filler stories, the recognition probe was not included in the
story, and the correct response was no. This balanced the number of correct
yes and no responses across the entire experiment. We also included the
same secondary task as in Experiment 1A.
Design. The design was identical to that in Experiment 1A.
Procedure. The procedure was identical to that in Experiment 1A.
Results and Discussion
Table 3 presents the results of Experiment 2. Participants failed
to respond before the deadline 2.3% of the time, and those re-
sponses were not included in our analyses. We also eliminated
decision times falling more than three standard deviations above
the mean, resulting in a loss of an additional 1.4% of the data.
None of the participants expressed any awareness of the goals of
the experiment.
We analyzed recognition times for correct responses to probe
locations. We again obtained support for the interaction predicted
by the flexible-boundary hypothesis: Participants were slower to
recognize probe words for characters’ start locations after reading
long activities compared with recognition times for start locations
following short activities, significant by participants and marginal
by items, F
1
(1, 35) ⫽ 9.24, MSE ⫽ 14,473, p ⬍ .005; F
2
(1, 19) ⫽
3.02, MSE ⫽ 28,124, p ⫽ .098. For the main effect of location
probe, participants took longer to recognize probes for start loca-
tions (M ⫽ 1,436 ms) compared with final locations, significant by
participants and items (M ⫽ 1,378 ms), F
1
(1, 35) ⫽ 11.48, MSE ⫽
30,180, p ⬍ .005; F
2
(1, 19) ⫽ 13.04, MSE ⫽ 14,278, p ⬍ .005.
For the main effect of activity statement, participants also took
longer to recognize probes following long activities (M ⫽ 1,417
ms) compared with short activities, significant by participants and
items (M ⫽ 1,357 ms), F
1
(1, 35) ⫽ 6.19, MSE ⫽ 21,069, p ⬍ .05;
F
2
(1, 19) ⫽ 6.61, MSE ⫽ 17,227, p ⬍ .05.
We conducted planned comparisons as in Experiments 1A and
1B. Following long activities, participants were 159 ms slower to
recognize probes for start locations than for final locations, sig-
nificant by participants and items, t
1
(35) ⫽ 4.30, p ⬍ .001;
t
2
(19) ⫽ 3.19, p ⬍ .005. Participants were 121 ms slower to
recognize probes for start locations following long activities com-
pared with those start locations following short activities, signifi-
cant by participants and items, t
1
(35) ⫽ 4.12, p ⬍ .001; t
2
(19) ⫽
Table 3
Means and Standard Deviations for Correct Recognition Probe
Times and Activity Statement Reading Times for Start Locations
(L1) and Final Locations (L2) in Experiment 2
Correct
recognition
probe times
Activity
statement
reading times
L1 L2 L1 L2
Long-activity statement
M 1,496 1,337 2,616 2,555
SD 249 257 867 738
Short-activity statement
M 1,375 1,338 2,566 2,452
SD 243 304 891 620
Note. All probe and reading times in ms.
994
RAPP AND TAYLOR
2.86, p ⬍ .01. Participants were also 158 ms slower to recognize
probes for start locations following long activities compared with
final locations following short activities, significant by participants
and items, t
1
(35) ⫽ 3.79, p ⬍ .001; t
2
(19) ⫽ 4.17, p ⬍ .001. No
other comparisons were significant (all ts ⬍ 1.15).
We analyzed the error rates for recognition responses. Neither of
the main effects nor the interaction were reliable (M ⫽ 25%
incorrect for start locations following long activities, 25% incor-
rect for start locations following short activities, 32% incorrect for
final locations following long activities, and 28% incorrect for
final locations following short activities; all Fs ⬍ 2.75). We also
evaluated reading times for the activity statements (see Table 3).
Participants took 77 ms longer to read long-activity statements (2,586
ms) than short-activity statements (2,509 ms); however, neither of the
main effects nor the interaction were reliable (all Fs ⬍ 1.05).
The overall pattern of recognition times from Experiment 2 is in
line with that of the previous experiments. This pattern is sugges-
tive that other dimensions (such as time and character activity)
may provide cues for encoding spatial representations, supporting
the notion that interactivity among text dimensions facilitates
situation model construction. However, an alternative interpreta-
tion of these results that does not place emphasis on time and space
is worthy of note. The differences in accessibility for probe words
may have been a function of the degree to which activity state-
ments indicated a shift in discourse topic (to a new subject,
situation, or event), rather than a shift in time or space. This
alternative interpretation is based on the notion that topic shifts can
lead to decreases in the accessibility of information preceding
those shifts (e.g., Haviland & Clark, 1974; Lorch, Lorch, & Mat-
thews, 1985). In principle, the data collected from Experiment 2
speak against this interpretation. If longer activity statements were
consistently associated with larger discourse shifts, this should
have resulted in a main effect of activity statement but little in the
way of an interaction between activity statements and location
probes. Nevertheless, it is entirely possible that the results of
Experiment 2 were at least partially a function of the degree to
which activity statements indicated a shift in overall discourse
topic rather than a more specific index-based shift in time or space.
To address this issue, we conducted an off-line study to assess
whether long-activity statements instantiated greater expectations
for topic shifts as compared with short-activity statements. We
asked 24 English-speaking University of Minnesota students to
read one version of each of the 20 experimental stories and to
indicate how much of a topic shift was conveyed by the activity
statement. There were two versions of each story (long-activity
statement vs. short-activity statement). Each story ended immedi-
ately following the activity statement, at which point participants
were asked to judge the degree of topic shift conveyed by the
activity statement. We randomly placed one version of each story
on one of two questionnaires; each questionnaire included 30
stories (20 experimental stories and 10 filler stories). Each ques-
tionnaire page included two experimental stories and one filler,
and the story pages of each questionnaire were randomly shuffled
for each participant. The final sentence of each filler story always
described an explicit shift (e.g., “On the other side of the world, a
similar event was occurring with a boy named Kakimoto and his
favorite baseball player, Tetsuo Matsui.”) to provide an assessment of
whether participants were correctly evaluating topic shifts. The in-
structions to the questionnaire described the concept of topic shifts in
detail and provided two examples (one story containing a topic shift
and one without a shift). Following the examples, the instructions
read, “At the conclusion of each story, you are going to be asked to
think about the final sentence in relation to the rest of the story.
Specifically, you are to decide how much of a topic shift is conveyed
in that final sentence.” Participants used a Likert scale, ranging from
1(no shift)to7(drastic shift), to indicate the amount of shift.
The data from these questionnaires demonstrated that partici-
pants reported no difference in the amount of shift conveyed by the
long- and short-activity statements. The mean shift rating for
long-activity statements was 2.42, and the mean shift rating for
short-activity statements was 2.48; these ratings were not signifi-
cantly different by participants or items (all ts ⬍ 1). The mean shift
rating for the fillers was 5.73, which suggests that these results
were not due to a floor effect for which participants were unwilling
to provide ratings that indicated the existence of shift statements
(and also suggests that participants indeed understood the notion of
a discourse shift). These data suggest that the results of Experiment
2 were unlikely to be due to qualitative differences in the degree to
which long- and short-activity statements cued reader expectations
about changes in discourse topics.
Although we ruled out the topic shift hypothesis as a potential
explanation for the experimental findings, at this point we cannot
make any definitive claims about the particular dimensions that led
to the obtained differences in recognition times. Our belief is that
activity statements conveying information about time and space
influenced the accessibility of locations, with time acting as a
correlate for spatial proximity and spatial distance. However, an-
other viable possibility is that participants encoded events solely
based on the passage of time associated with activities, without
reference to the spatial organizations implied in stories. The pas-
sage of time need not provide information about spatial distance to
influence the accessibility of location information. Evidence has
suggested that with long-activity statements, information preced-
ing a shift can become less accessible solely due to longer time
shifts (Rapp & Gerrig, 2002; Zwaan, 1996). With short-activity
statements, early locations remain accessible because less time has
passed, and they are still considered part of the current focus of the
story (Bower & Morrow, 1990). This calls into question our
interactive position, suggesting that text cues for particular dimen-
sions may operate along a unitary, rather than multidimensional,
set of indices. Were locations encoded along spatial and temporal
dimensions or encoded along a single temporal dimension? This
issue was addressed in Experiment 3.
Experiment 3
To test the interactive nature of text dimensions, we modified
our stories to reduce the spatial movement described in the narra-
tives. We rewrote the stories so that characters remained in start
locations while completing their activities. The characters intended
to travel to a final location but never actually left to reach that
destination. Instead, characters remained at the start location while
engaging in either a long or short activity. Revising our stories in
this manner maintained temporal differences but eliminated the
spatial components of those shifts. This allowed us to examine
whether activity statements influenced accessibility by operating
solely as temporal cues or if they operated through the integration
of both spatial and temporal cues. If readers’ representations are
995
INTERACTIVE DIMENSIONS IN TEXT
structured purely by the passage of time in the stories, we would
expect to obtain a pattern of data similar to that found in our
previous experiments: Participants should take longer to recognize
probes for start locations following long activities compared with
when start probes follow short activities. If, however, readers’
representations are also influenced by character movement, then
the removal of spatial cues should result in similar recognition
times for start locations regardless of activity length. Although this
prediction is contingent on a null effect, it is nevertheless infor-
mative in contrast to the consistent pattern obtained in our previous
three experiments.
Method
Participants. Thirty-six Tufts University undergraduates participated
in this study for course credit. All participants were native speakers of
English.
Apparatus. The apparatus was identical to that in Experiment 1A.
Materials. We modified the experimental stories from Experiment 2
(see Appendix C for examples). These stories described the same basic
characters, scenarios, and objectives using a similar structure as the pre-
vious sets. Again, each story was 12 sentences long. The stories were
modified to describe the character as engaging in an activity at the start
location (without mentioning the start location more than once in the
overall narrative). The stories concluded as the character considered leav-
ing or prepared to leave for the final location. For the activity statements,
only four required modification to remove references to movement, and
again we equated these statements for length (mean number of words for
long- and short-activity statements ⫽ 11.2). For some of the items it was
also necessary to rewrite sentences to maintain local and global coherence,
given that in these modified stories the character was only planning or
preparing to leave (rather than actually doing so, as in the previous
experiments). These story sentences were rewritten to contain the same
number of words as their original versions from the previous experiments
so as to keep them close to the original story structure. The number of filler
stories and correct yes and no probes was the same as in Experiment 2.
Design. The design was identical to that in Experiment 1A.
Procedure. The procedure was identical to that in Experiment 1A.
Results and Discussion
Table 4 presents the results of Experiment 3. Participants failed
to respond before the deadline 1.1% of the time, and these re-
sponses were not included as part of our analyses. We also elim-
inated decision times falling more than three standard deviations
above the mean, resulting in a loss of an additional 1.7% of the
data. None of the participants expressed any awareness of the goals
of the study.
We again began by analyzing recognition times for correct
responses to probe locations. Neither of the main effects nor the
interaction were significant (all Fs ⬍ 1.1). We conducted planned
comparisons using paired t tests (Bonferroni corrected), with none
of the comparisons significant (all ts ⬍ 1). There was no evidence
of the earlier accessibility pattern with the removal of the spatial
cue.
Next, we analyzed the error rates for recognition responses.
Neither the main effect of activity statement nor the interaction
was significant (M ⫽ 27% incorrect for start locations following
long activities, 28% incorrect for start locations following short
activities, 35% incorrect for final locations following long activi-
ties, and 34% incorrect for final locations following short activi-
ties; all Fs ⬍ 1). Participants tended to provide a higher proportion
of incorrect responses for final locations (M ⫽ 35%) compared
with start locations (M ⫽ 28%), marginal by participants only,
F
1
(1, 35) ⫽ 4.07, MSE ⫽ .04, p ⫽ .051; F
2
(1, 19) ⫽ 1.92, MSE ⫽
.069, p ⬎ .10. Recall that the characters in these stories never
visited final locations, remaining in start locations at the conclu-
sion of the stories. We might expect participants to have had more
difficulty accessing these final locations because they did not
remain the focus of characters’ attention as the stories unfolded
(and had only been mentioned once, in the fourth sentence of each
story). Had our stories placed greater emphasis on characters’
intentions to visit their final locations, we might have expected
readers to demonstrate better recognition for those locations (Mor-
row et al., 1989). Finally, we evaluated reading times for the
activity statements (see Table 4). Participants took 72 ms longer to
read long-activity statements (2,587 ms) than short-activity state-
ments (2,515 ms); however, neither of the main effects nor the
interaction were reliable (all Fs ⬍ 1.1). Because there were slight
differences in the number of words composing the critical activity
statements in Experiments 2 and 3, we repeated the reading time
analyses based on the number of words in each statement. These
analyses were entirely consistent with whole-statement reading
times.
The data from Experiment 3 suggest that when spatial cues are
decoupled from statements implying temporal duration, there is a
change in the baseline pattern of accessibility for spatial locations.
We found no differences in reaction latencies for start or final
location probes. This null result should, of course, be interpreted
with caution, but we point out that the main effects and interaction
for recognition times were not significant with all p values ex-
ceeding .30. Experiment 3 used the same number of participants,
the same methodologies and procedures, and similar stimuli to that
from the previous three experiments. Therefore, we have reason to
believe that had there been effects of our location and distance
manipulations similar to those obtained in Experiments 1A, 1B,
and 2, we would have obtained similar results. Given the p values
in Experiment 2 for the overall activity statement by probe inter-
action (significant by participants and marginal by items) and the
critical planned comparison for start probes following long-
activity statements compared with final probes following short-
activity statements (significant by participants and items), it is
Table 4
Means and Standard Deviations for Correct Recognition Probe
Times and Activity Statement Reading Times for Start Locations
(L1) and Final Locations (L2) in Experiment 3
Correct
recognition
probe times
Activity
statement
reading times
L1 L2 L1 L2
Long-activity statement
M 1,384 1,405 2,623 2,551
SD 233 301 866 652
Short-activity statement
M 1,406 1,379 2,468 2,562
SD 255 301 770 747
Note. All probe and reading times in ms.
996
RAPP AND TAYLOR
unlikely that these findings were spurious. The statistical compar-
isons were run with Bonferroni corrections, which guard against
Type-1 error (incorrect rejections of the null hypothesis). Even
with these corrections to the p values required for rejection of the
null hypothesis, the critical findings in Experiment 2 remained
significant, reducing the probability that the null interaction and
critical comparison were incorrectly rejected in Experiment 2
(Rosenthal & Rosnow, 1991).
Unfortunately, a power analysis for the null effect in Experiment
3 is not possible because all power analyses assume that the null
hypothesis is false (Hoenig & Heisey, 2001). This presents a
challenge in assessing whether there was sufficient power to reject
the null in Experiment 3. To address this issue in a different way,
we conducted an explicit statistical comparison between Experi-
ments 2 and 3 using a 2 (activity statement: long vs. short) ⫻ 2
(location: start vs. final) analysis of variance (ANOVA) with
experiment (2 vs. 3) as a between-subjects variable. This allowed
us to evaluate whether the null result obtained for Experiment 3
was indeed due to differences in the experiments themselves.
Evidence for this would be obtained in a significant interaction
effect by experiment. We note that this analysis is based on two
different experiments and thus should be viewed carefully. Nev-
ertheless, the data from this cross-experiment analysis are infor-
mative with respect to the effects obtained in one experiment
compared with the other. Consistent with this view, we found
significant effects by experiment. There was an interaction of
activity statement by experiment, marginal by participants and
items, F
1
(1, 70) ⫽ 2.70, MSE ⫽ 22,999, p ⫽ .098; F
2
(1, 38) ⫽
3.97, MSE ⫽ 17,273, p ⫽ .054. There was also an interaction of
location probe by experiment, significant by participants and mar-
ginal by items, F
1
(1, 70) ⫽ 5.00, MSE ⫽ 34,428, p ⬍ .05; F
2
(1,
38) ⫽ 2.95, MSE ⫽ 25,227, p ⫽ .094. The three-way interaction
of Activity Statement ⫻ Location Probe ⫻ Experiment was sig-
nificant by participants and marginal by items, F
1
(1, 70) ⫽ 4.85,
MSE ⫽ 26,807, p ⬍ .05; F
2
(1, 19) ⫽ 3.83, MSE ⫽ 26,971, p ⫽
.058. The analysis also revealed a main effect of location, signif-
icant by both participants and items, F
1
(1, 70) ⫽ 5.69, MSE ⫽
32,428, p ⬍ .05; F
2
(1, 38) ⫽ 4.51, MSE ⫽ 25,227, p ⬍ .05, and
a main effect of distance, marginal by participants only, F
1
(1,
70) ⫽ 2.97, MSE ⫽ 22,999, p ⫽ .089; F
2
(1, 38) ⫽ 2.68, MSE ⫽
17,273, p ⬎ .10. There were no other effects (all Fs ⬍ 1). The
evidence from these analyses suggests that the between-subjects
variable of experiment indeed influenced performance. This sta-
tistical comparison, the Bonferroni corrections protecting against
Type-1 errors in Experiment 2, and the similarity between Exper-
iment 2 and Experiment 3 (in terms of methodologies, procedures,
participants, and stimuli) provide convergent evidence for the
validity of the null result obtained in Experiment 3.
We take these findings as suggestive that, at least for our
previous experiments, readers’ representations of text events rely
on the interactivity of movement (space) and character activities
(time) in the construction of a situation model. This interactivity
results in differential accessibility for narrative locations.
General Discussion
The purpose of this set of experiments was two-fold. First, we
wished to evaluate the interactive nature of the cues that influence
readers’ construction of multidimensional situation models. Sec-
ond, we wanted to evaluate whether the mechanisms guiding the
structure of narrative representations involve general, shift-based
segmentation processes or more precise, expectation-based seg-
mentation processes. Our experiments therefore extend previous
work on situation models and narrative representations. In Exper-
iments 1A and 1B, explicit spatial distance statements influenced
the representations and accompanying accessibility of story loca-
tions. This pattern served as a baseline for evaluating the interac-
tivity of narrative indices. Participants in Experiment 2 demon-
strated a similar pattern of accessibility for stories that included
activity statements implying the passage of time but providing no
explicit mention of spatial distance. Experiment 3 eliminated char-
acter movement, suggesting that our earlier effects were likely a
result of the interactivity of multiple text dimensions.
Our results speak to the utility of the event-indexing model as an
account of the features encoded by readers during their narrative
experiences. Although research has described the dimensions that
are tracked during reading, less work has focused on the interactive
nature of those dimensions during the encoding and retrieval of
text representations. There has been considerable interest in the
notion of interactive event dimensions outside of the text compre-
hension domain (e.g., Boroditsky, 2000; McGlone & Harding,
1998). The experiments in this article embrace this view by de-
scribing how the construction and application of text dimensions,
such as time and space, are mutually determined. As such, the
results call into question the notion of discrete representational
dimensions in event-indexing model accounts of text processing.
By evaluating the interactive nature of construction processes, we
can provide a more comprehensive account of how situation mod-
els are dynamically updated during narrative experiences (see van
den Broek, Young, Tzeng, & Linderholm, 1999, for a review).
Of course, a process-based explanation for these results need not
rely solely on the integration of multiple dimensions but rather
should focus more directly on the construction of multiple mental
substructures. For instance, information in the here-and-now of the
text tends to remain more accessible than information from earlier
text events (see Morrow, 1994, for a review). With each new
event, a new substructure is built. The accessibility of earlier
information is reduced as a function of being represented in a prior
mental substructure. For the stories in our experiments, start loca-
tions may have become less accessible than final locations as a
function of being represented in earlier substructures. When dis-
tances were large, either based on larger temporal or spatial shifts,
start locations were less accessible. When distances were short,
readers did not need to construct a new mental substructure and
information remained readily accessible. Such a framework pro-
vides an indication of the potential processes guiding the construc-
tion of event representations during text comprehension. However,
this explanation on its own fails to precisely outline when a new
substructure may or may not be built. What determines whether a
particular distance or shift is of sufficient size, duration, or salience
to cue the construction of a new mental representation?
We argue that expectations about the range or duration of events
are largely a function of readers’ beliefs about the concomitants of
space and time. Thus, reader expectations and background knowl-
edge provide organizing frameworks that influence processes of
situation model construction. This view, along with our results,
suggests a necessary updating of structure-building theory to ac-
count for data supporting the flexible-boundary hypothesis.
997
INTERACTIVE DIMENSIONS IN TEXT
Structure-building theory proposes that particular events are seg-
mented in memory representations, resulting in differential acces-
sibility for mental substructures depending on reader or text focus.
The theory has had less to say about the malleability of that
segmentation process. Our data suggest that the construction pro-
cesses by which events are segmented, and substructures are
constructed, may be more flexible than originally described. The
underlying mechanisms for this segmentation process likely rely
on readers’ knowledge or beliefs about the duration and nature of
events, directly influencing the structure of resulting text
representations.
Our results argue against the rigid-boundary hypothesis as a
general theory of situation model construction. According to the
rigid-boundary hypothesis, event shifts should lead to broad, cat-
egorical reductions in the accessibility of text information from
memory. Our data suggest that for cases in which time and space
guide event shifts, the rigid-boundary hypothesis fails to capture
the types of sophisticated processing that readers rely on for
comprehension (Zwaan et al., 2000). We note, though, that it may
be the case that the appropriateness of either the flexible-boundary
or rigid-boundary hypothesis for describing model construction
may depend on the particular dimensions studied, as well as the
medium of interest. For instance, studies examining movie pre-
sentations have suggested that real-world interactions between
time and space are necessary during film comprehension (Magli-
ano, Miller, & Zwaan, 2001).
Related to this issue, continued investigation is necessary to
delineate the specific contributions of space and time in an inter-
active model of text comprehension. This is particularly important
because the findings of Experiment 3 are based largely on a null
result and should be interpreted with caution. Future research
should investigate the specific contributions of each index on
reader memory for locations and events. For example, stories
might be interrupted during character movement, such as during
travel from start to final locations, at which point characters could
engage in activities of variable duration. Probe latencies could
reveal the time course of accessibility for text information as
readers dynamically update their situation models.
Such work might also be designed to address whether text
dimensions such as time and space are nested within one another
or whether these dimensions are truly independent. Traditional
studies have tended to treat these dimensions as separate indices
(Zwaan, Langston, et al., 1995, Zwann, Magliano, et al., 1995).
The alternative view, that space and time are subsumed under a
larger dimension, has also been of interest, although it has received
less empirical support. These opposing views may represent a
continuum of representational possibilities based on the nature of
a particular task. In our studies, participants were asked to recog-
nize locations following shifts in both time and space, which may
have led participants to treat the two dimensions as a unitary
dimension. However, for studies that require participants to con-
struct and apply situation models based on particular task-specific
indices, text dimensions may be separated in a mental representa-
tion (e.g., Rapp, Gerrig, & Prentice, 2001; Trabasso, van den
Broek, & Suh, 1989; van den Broek, 1988; Zwaan, 1996).
As previously mentioned, a growing body of research has ex-
amined the notion of separate dimensions almost exclusively,
without attempting to account for the interactive nature of text
dimensions (Zwaan & Radvansky, 1998). One body of research
along these lines has evaluated the dominance of particular text
dimensions (Magliano et al., 2001; Rich & Taylor, 2000; Taylor &
Tversky, 1997). Another body of work has demonstrated that
reader preferences and beliefs can influence the likelihood that
readers encode specific dimensions into their situation models
(Rapp et al., 2001; Rapp & Gerrig, 2002; Taylor, Rapp, & Klug,
2002). Both sets of studies have outlined how readers’ predilec-
tions for encoding particular dimensions can influence their recall
of narrative events. The current project builds on these studies by
including the notion that readers’ prior knowledge about the in-
teractive nature of text dimensions can facilitate the degree to
which they build their narrative representations. Research in deci-
sion making has also supported this view, demonstrating that
judgments of causality are based not only on event descriptions,
but also on participants’ expectations about the temporal intervals
normally associated with those events (Hagmayer & Waldmann,
2002).
One potential question that can be raised is whether the inter-
active effects we have described commonly occur during general
discourse experiences. Reading tasks and goals can directly influ-
ence the nature of the processes that readers rely on as they read
(Magliano, Trabasso, & Graesser, 1999; Zwaan & van Oosten-
dorp, 1993). One could argue that the movement cues in our stories
may have been sufficient to signal that readers should track par-
ticular text dimensions more closely than usual. We do note,
however, that participants in Experiment 2 did not report explicitly
noticing such cues, yet movement effects were still obtained. In
addition, although extensive work was conducted to make our
experimental stories naturalistic (e.g., more similar to real-world
texts), they clearly differentiate themselves from true narrative
fiction and nonfiction. We contend, however, that the materials
and results described in these experiments have direct analogues in
everyday discourse, and as such, these effects should extend be-
yond our stimuli.
For example, in everyday conversation, it is not uncommon to
provide directions by using activity cues to indicate distances.
Consider the case in which a colleague provides driving directions
for a cocktail party: “After you’ve passed University Avenue, keep
going, keep going, keep going, keep going, until you’ve reached
the pizzeria.” These multiple instances of “keep going” are un-
likely to be classified as purely spatial or temporal cues but rather
provide a multidimensional cue for the amount of time and space
that needs to be experienced before reaching the pizzeria. In this
example, the “keep going” cue functions in a similar manner to our
distance and activity statements. One could even imagine that the
notion of “keep going” might generate diverse expectations in
listeners depending on their propensities and willingness to drive
for different distances or their expectations for how suggestive
four instances of “keep going” are for actual movement. Narratives
use similar techniques, and we wished to evaluate some of the
underlying mechanisms involved in comprehending these
situations.
We have remained agnostic with respect to the issue of auto-
maticity in the use of linguistic cues for constructing situation
models. The situation model literature has presented evidence on
both sides of the argument (e.g., Clifton & Duffy, 2001; De Vega,
1995; Hakala, 1999; McKoon & Ratcliff, 1992; Morrow, 1994;
O’Brien, 1995; Zwaan, Radvansky, Hilliard, & Curiel, 1998). The
strong form of the automaticity argument suggests that readers
998
RAPP AND TAYLOR
construct representations that encode event dimensions on-line.
For example, evidence from research on causal structure (e.g., van
den Broek & Trabasso, 1986), spatial situation models (Levine &
Klin, 2001), and trait attributions (e.g., Uleman, Hon, Roman, &
Moskowitz, 1996) suggests that readers may be predisposed to
construct these representations on a moment-by-moment basis.
The competing perspective suggests that readers construct repre-
sentations if they conform to their strategies (e.g., Hakala, 1999) or
help them achieve their comprehension goals for the text experi-
ence (e.g., van den Broek, Lorch, Linderholm, & Gustafson,
2001). Our studies do not allow us to make claims concerning the
automaticity of construction processes. Given our results, both of
the aforementioned views need to take into account the subtle
types of linguistic cues (such as the activity statements presented
in our stories) that can influence text processing.
As we suggested in the introduction to this article, readers rely
on linguistic cues and background knowledge to help them “fill in”
the missing information from text descriptions. For example, many
of the details of Seabiscuit’s race for the Hollywood Gold Cup
were only partially available or even entirely left out from the text.
However, readers can rely on the interactive nature of text descrip-
tions and their background knowledge to engage in those “filling
in” processes. Cues for space or time, for example, can inform
expectations about space, time, characters, objects, causality, and
story settings. We propose that readers’ use of these cues is a
natural process in comprehending texts during everyday reading.
By defining these interactive cues and their impact on the com-
prehension of texts, we can further outline the underlying interac-
tive nature of mental representations and clarify how those repre-
sentations are constructed and updated during narrative
experiences.
References
Barsalou, L. W. (1999a). Language comprehension: Archival memory or
preparation for situated action? Discourse Processes, 28, 61–80.
Barsalou, L. W. (1999b). Perceptual symbol systems. Behavioral & Brain
Sciences, 22, 577–660.
Barsalou, L. W., Huttenlocher, J., & Lamberts, K. (1998). Basing catego-
rization on individuals and events. Cognitive Psychology, 36, 203–272.
Bestgen, Y., & Vonk, W. (2000). Temporal adverbials as segmentation
markers in discourse. Journal of Memory and Language, 42, 74–87.
Boroditsky, L. (2000). Metaphoric structuring: Understanding time
through spatial metaphors. Cognition, 75, 1–28.
Bower, G. H., & Morrow, D. G. (1990). Mental models in narrative
comprehension. Science, 247, 44–48.
Carreiras, M., Carriedo, N., Alonso, M. A., & Fernandez, A. (1997). The
role of verb tense and verb aspect in the foregrounding of information
during reading. Memory & Cognition, 3, 193–209.
Clifton, C., Jr., & Duffy, S. A. (2001). Sentence and text comprehension:
Roles of linguistic structure. Annual Review of Psychology, 52, 167–196.
De Vega, M. (1995). Backward updating of mental models during contin-
uous reading of narratives. Journal of Experimental Psychology: Learn-
ing, Memory, and Cognition, 21, 373–385.
Gernsbacher, M. A. (1990). Language comprehension as structure build-
ing. Hillsdale, NJ: Erlbaum.
Glenberg, A. M., Meyer, M., & Lindem, K. (1987). Mental models con-
tribute to foregrounding during text comprehension. Journal of Memory
and Language, 26, 69–83.
Hagmayer, Y., & Waldmann, M. R. (2002). How temporal assumptions
influence causal judgments. Memory & Cognition, 30, 1128–1137.
Hakala, C. (1999). Accessibility of spatial information in a situation model.
Discourse Processes, 27, 261–280.
Haviland, S. E., & Clark, H. H. (1974). What’s new? Acquiring new
information as a process in comprehension. Journal of Verbal Learning
and Verbal Behavior, 19, 635–650.
Hillenbrand, L. (2001). Seabiscuit: An American legend. New York: Bal-
lantine Books.
Hoenig, J. M., & Heisey, D. M. (2001). The abuse of power: The pervasive
fallacy of power calculations for data analysis. The American Statisti-
cian, 55, 19–24.
Kahneman, D., & Tversky, A. (1982). The simulation heuristic. In D.
Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty:
Heuristics and biases (pp. 201–208). New York: Cambridge University
Press.
Kaschak, M. P., & Glenberg, A. M. (2000). Constructing meaning: The
role of affordances and grammatical constructions in sentence compre-
hension. Journal of Memory and Language, 43, 508–529.
Levine, W. H., & Klin, C. M. (2001). Tracking of spatial information in
narratives. Memory & Cognition, 29, 327–335.
Lorch, R. F., Jr., Lorch, E. P., & Matthews, P. D. (1985). On-line process-
ing of the topic structure of a text. Journal of Memory and Language, 24,
350–362.
Madden, C. J., & Zwaan, R. A. (2003). How does verb aspect constrain
event representations? Memory & Cognition, 31, 663–672.
Magliano, J. P., Miller, J., & Zwaan, R. A. (2001). Indexing space and time
in film understanding. Applied Cognitive Psychology, 15, 533–545.
Magliano, J. P., & Schleich, M. C. (2000). Verb aspect and situation
models. Discourse Processes, 29, 83–112.
Magliano, J. P., Trabasso, T., & Graesser, A. C. (1999). Strategic processes
during comprehension. Journal of Educational Psychology, 91, 615–
629.
McGlone, M. S., & Harding, J. L. (1998). Back (or forward?) to the future:
The role of perspective in temporal language comprehension. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 24, 1211–
1233.
McKoon, G., & Ratcliff, R. (1992). Inference during reading. Psycholog-
ical Review, 99, 440–466.
Morrow, D. G. (1994). Spatial models created from text. In H. van
Oostendorp & R. A. Zwaan (Eds.), Naturalistic text comprehension (pp.
57–78). Norwood, NJ: Ablex Publishing.
Morrow, D. G., Bower, G. H., & Greenspan, S. L. (1989). Updating
situation models during narrative comprehension. Journal of Memory
and Language, 28, 292–312.
Morrow, D. G., Greenspan, S. L., & Bower, G. H. (1987). Accessibility
and situation models in narrative comprehension. Journal of Memory
and Language, 26, 165–187.
O’Brien, E. J. (1995). Automatic components of discourse comprehension.
In R. F. Lorch, Jr., & E. J. O’Brien (Eds.), Sources of coherence in
reading (pp. 159–176). Hillsdale, NJ: Lawrence Erlbaum.
Rapp, D. N., & Gerrig, R. J. (1999). Eponymous verb phrases and ambi-
guity resolution. Memory & Cognition, 27, 612–618.
Rapp, D. N., & Gerrig, R. J. (2002). Readers’ reality-driven and plot-driven
analyses in narrative comprehension. Memory & Cognition, 30, 779–
788.
Rapp, D. N., Gerrig, R. J., & Prentice, D. A. (2001). Readers’ trait-based
models of characters in narrative comprehension. Journal of Memory
and Language, 45, 737–750.
Rich, S. S., & Taylor, H. A. (2000). Not all narrative shifts function
equally. Memory & Cognition, 28, 1257–1266.
Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research:
Methods and data analysis (2nd ed.). New York: McGraw-Hill.
Taylor, H. A., Rapp, D. N., & Klug, J. K. (2002, November). Skipping
through space: Updating the unexpected in situation models. Paper
999
INTERACTIVE DIMENSIONS IN TEXT
presented at the 43rd Annual Meeting of the Psychonomic Society,
Kansas City, MO.
Taylor, H. A., & Tversky, B. (1997). Indexing events in memory: Evidence
for index dominance. Memory, 5, 509–542.
Trabasso, T., van den Broek, P., & Suh, S. (1989). Logical necessity and
transitivity of causal relations in stories. Discourse Processes, 12, 1–
25.
Uleman, J. S., Hon, A., Roman, R. J., & Moskowitz, G. B. (1996). On-line
evidence for spontaneous trait inferences at encoding. Personality and
Social Psychology Bulletin, 22, 377–394.
van den Broek, P. (1988). The effects of causal relations and hierarchical
position on the importance of story statements. Journal of Memory and
Language, 27, 1–22.
van den Broek, P., Lorch, R. F., Jr., Linderholm, T., & Gustafson, M.
(2001). The effects of readers’ goals on inference generation and mem-
ory for texts. Memory & Cognition, 29, 1081–1087.
van den Broek, P., & Trabasso, T. (1986). Causal networks versus goal
hierarchies in summarizing text. Discourse Processes, 9, 1–15.
van den Broek, P., Young, M., Tzeng, Y., & Linderholm, T. (1999). The
landscape model of reading: Inferences and the online construction of a
memory representation. In H. van Oostendorp & S. R. Goldman (Eds.),
The construction of mental representations during reading (pp. 71–98).
Mahwah, NJ: Erlbaum.
van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse compre-
hension. New York, NY: Academic Press.
Zwaan, R. A. (1996). Processing narrative time shifts. Journal of Experi-
mental Psychology: Learning, Memory, and Cognition, 22, 1196–1207.
Zwaan, R. A. (1999). Embodied cognition, perceptual symbols, and situ-
ation models. Discourse Processes, 28, 81–88.
Zwaan, R. A., Langston, M. C., & Graesser, A. C. (1995). The construction
of situation models in narrative comprehension: An event-indexing
model. Psychological Science, 6, 292–297.
Zwaan, R. A., Madden, C. J., & Whitten, S. N. (2000). The presence of an
event in the narrated situation affects its availability in the compre-
hender. Memory & Cognition, 28, 1022–1028.
Zwaan, R. A., Magliano, J. P., & Graesser, A. C. (1995). Dimensions of
situation model construction in narrative comprehension. Journal of Exper-
imental Psychology: Learning, Memory, and Cognition, 21, 386–397.
Zwaan, R. A., & Radvansky, G. A. (1998). Situation models in language
comprehension and memory. Psychological Bulletin, 123, 162–185.
Zwaan, R. A., Radvansky, G. A., Hilliard, A. E., & Curiel, J. M. (1998).
Constructing multidimensional situation models during reading. Scien-
tific Studies of Reading, 2, 199–220.
Zwaan, R. A., & van Oostendorp, H. (1993). Do readers construct spatial
representations in naturalistic story comprehension? Discourse Pro-
cesses, 16, 125–143.
Appendix A
Sample Stories From Experiment 1A and 1B (Including Recognition Probes)
Joe was working diligently on his term paper.
Joe had been laboring for quite some time in the library.
He began to feel kind of hungry.
He decided to get something to eat from the diner.
Joe gathered his things and left.
Outside, he noticed a stick on the ground.
He pulled out a small pocket knife from his bookbag.
Joe began to whittle away at the stick while he walked.
Joe walked for four miles. (long-distance statement, Experiment 1A)
Joe walked for four blocks. (short-distance statement, Experiment 1A)
Joe walked for five miles to his destination. (long-distance statement, Experiment 1B)
Joe walked for three blocks to his destination. (short-distance statement, Experiment 1B)
He put the finishing touches on it just as he arrived.
The smells of cooking burgers and french fries made his mouth water.
He stepped up to the counter and ordered a grilled cheese sandwich.
Location probes: library [start location], diner [final location]
Nick had a big test coming up.
But Nick was spending too much time at the gym.
He had finished playing basketball and it was time to work.
He planned to walk over to his dorm room to study.
Nick’s stomach was grumbling.
He carried his materials, including his textbook, notes, and audio cassettes of lectures.
He decided it might be worth listening to one of the lectures on the way.
He put the walkman headset on and listened.
Nick walked for four miles. (long-distance statement, Experiment 1A)
Nick walked for four blocks. (short-distance statement, Experiment 1A)
Nick walked for five miles to his destination. (long-distance statement, Experiment 1B)
Nick walked for three blocks to his destination. (short-distance statement, Experiment 1B)
He hit the stop button on his walkman and pulled open the door.
He smelled some food cooking in the building.
His stomach seemed to gurgle in response to the smells.
Location probes: gym [start location], dorm [final location]
1000
RAPP AND TAYLOR
Appendix B
Sample Stories From Experiment 2 (Including Recognition Probes)
Joe was working diligently on his term paper.
Joe had been laboring for quite some time in the library.
He began to feel kind of hungry.
He decided to get something to eat from the diner.
Joe gathered his things and left.
Outside, he noticed a stick on the ground.
He pulled out a small pocket knife from his bookbag.
Joe began to whittle away at the stick while he walked.
He carved the stick into a small flute. (long-activity statement)
He carved his initials right on the stick. (short-activity statement)
He put the finishing touches on it just as he arrived.
The smells of cooking burgers and french fries made his mouth water.
He stepped up to the counter and ordered a grilled cheese sandwich.
Location probes: library [start location], diner [final location]
Nick had a big test coming up.
But Nick was spending too much time at the gym.
He had finished playing basketball and it was time to work.
He planned to walk over to his dorm room to study.
Nick’s stomach was grumbling.
He carried his materials, including his textbook, notes, and audio cassettes of lectures.
He decided it might be worth listening to one of the lectures on the way.
He put the walkman headset on and listened.
Nick listened to almost half of a lecture. (long-activity statement)
Nick listened to the introduction of a lecture. (short-activity statement)
He hit the stop button on his walkman and pulled open the door.
He smelled some food cooking in the building.
His stomach seemed to gurgle in response to the smells.
Location probes: gym [start location], dorm [final location]
Appendix C
Sample Stories From Experiment 3 (Including Recognition Probes)
Joe was working diligently on his term paper.
Joe had been laboring for quite some time in the library.
He began to feel kind of hungry.
He thought about getting something to eat from the diner.
He was still on his diet.
From his bag he retrieved a small stick.
He also removed a pocket knife from his back pocket.
Joe began to whittle away at the stick while he sat.
He carved the stick into a small flute. (long-activity statement)
He carved his initials right on the stick. (short-activity statement)
He put the finishing touches on it and then stood up.
Thinking about eating a burger and french fries made his mouth water.
He decided he would go and order himself a grilled cheese sandwich.
Location probes: library [start location], diner [final location]
Nick had a big test coming up.
But Nick was spending too much time at the gym.
He had finished playing basketball and it was time to work.
He planned to walk over to his dorm room to study.
Nick opened his locker.
He picked up his materials, including textbooks, notes, and audio cassettes of lectures.
He sat on a bench and listened to one of the lectures before he left.
He put the walkman headset on and listened.
Nick listened to almost half of a lecture. (long-activity statement)
Nick listened to the introduction of a lecture. (short-activity statement)
He hit the stop button on his walkman and pressed the eject button.
He smelled another person’s lunch in the building.
His stomach seemed to gurgle in response to the smell.
Location probes: gym [start location], dorm [final location]
Received March 17, 2003
Revision received February 3, 2004
Accepted February 5, 2004 䡲
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Experimental Psychology Learning Memory and Cognition
This content is subject to copyright. Terms and conditions apply.