Five minutes vs. an hour ago: Processing flashbacks in narratives

Five minutes vs. an hour ago: Processing flashbacks in narratives
Berry Claus & Stephanie Kelter | Humboldt-Universität zu Berlin/Technische Universität Berlin
Introduction
A flashback refers to an event that occurred prior to the narrative now in the described world (see examples below)
Is the processing of a flashback sentence affected by how long ago the flashback event occurred in the described world?
Forward time shifts: no effect of temporal distance (e.g. An hour later vs. A day later) on processing time (Zwaan, 1996; see also Kelter et al., 2004)
The situation might be different with flashbacks: studies addressing the representation of non-chronological descriptions indicate that
comprehenders chronologically integrate flashback information (Claus & Kelter, 2006; see also Gennari, 2004)
General method
Experiment 3
In four experiments (conducted in German), participants read narrative texts, sentence by sentence, self-paced
Each experimental text contained a flashback; reading times (RTs)
for the flashback sentences were measured
The temporal distance between the flashback event and the narrative now was manipulated: small distance vs. large distance
Across experiments, the way in which the temporal distance was
specified, differed
The time of the flashback event was specified by referring to an
event mentioned before. The temporal distance was manipulated by
ascribing a short vs. long duration to that reference event
Sample text (excerpt): Panos Natsopoulos is Greek. …
… Panos has been asked to see about the flock of sheep today.
He rounds up the sheep in the meadow in front of the house.
Then he shears sheep for an hour/for eight hours. |reference event|
Then two tourists walk across the meadow.
Before he started sheering, they had already passed by. |flashback| …
In each experiment: 40 participants, 24 experimental texts and 27 filler texts
2400
(error bars represent 95% within-subject CIs)
Experiment 1
2350
(n.s.)
The temporal distance between the narrative now and the flashback
event was specified in the flashback sentence by a numerical temporal expression
2300
Example: Five minutes ago/An hour ago her son had called her.
2200
2100
2050
Mean RT for the flashback sentence (in ms)
(error bars represent
95% within-subject CIs)
1950
1900
1850
small
temporal
distance
large
temporal
distance
1800
The temporal distance was specified as in Expt 1 by a numerical
temporal expression, but relative to an event mentioned before
Example: Five minutes before/An hour before the meeting
her son had called her.
3100
Mean RT for the flashback sentence (in ms)
(error bars represent
95% within-subject CIs)
*
2950
2900
2850
2800
small
temporal
distance
large
temporal
distance
Temporal-distance effect: two accounts
(i) Representation search: gradual backward search through the
representation to the specified time of the flashback event
(ii) “Measuring”: determining the time of the flashback event requires a “translation” of numerical statements into time intervals
in the described world
Goal of Expt. 3 and 4: to contrast the two accounts
Conclusion
large
temporal
distance
Null effect of Expt 3 was replicated in three additional experiments
 It is difficult to reconcile with the representation-search account,
according to which a temporal-distance effect should occur irrespective of how the time of the flashback event is specified
 It is consistent with the measuring account, which implies that a
temporal-distance effect only occurs when it is necessary to measure a time interval
The flashback sentences were identical to those in Expt 3. However,
rather than describing the reference event from its beginning, the
text jumped to the end of the event, and stated that it had been
lasting for a certain time, such that determining the time when the
flashback event occurred, required establishing the beginning of the
reference event by using the duration information
Sample text (excerpt): … Panos … to see about the flock of sheep today.
He stands in the meadow in front of the house.
He has been shearing sheep for an hour/for eight hours*. |reference event|
Then two tourists walk across the meadow.
Before he started sheering, they had already passed by. |flashback| …
3050
3000
2150
small
temporal
distance
Experiment 4
Experiment 2
3150
2250
2100
*
2000
Mean RT for the flashback sentence (in ms)
*In the original German version, a seit-duration-adverbial was used,
literally: since an hour/eight hours
2400
2350
Mean RT for the flashback sentence (in ms)
(error bars represent
95% within-subject CIs)
2300
*
2250
2200
2150
2100
2050
small
temporal
distance
large
temporal
distance
The fact that the temporal-distance effect did occur in Expt 4 but not
(for identical flashback sentences ) in Expt 3 supports the measuring
account
Our results indicate that it is crucial how the time when the flashback event occurred, is specified. A temporal-distance effect was observed only
when a time interval needed to be measured in the representation of the described world in order to determine the time of the flashback event.
References Claus, B., & Kelter, S. (2006). Comprehending narratives containing flashbacks: Evidence for temporally organized representations. JEP: LMC, 32, 1031-1044. | Gennari, S. P. (2004). Temporal references and temporal
relations in sentence comprehension. JEP: LMC, 30, 877–890. | Kelter, S., Kaup, B., & Claus, B. (2004). Representing a described sequence of events: A dynamic view of narrative comprehension. JEP: LMC, 30, 451-464. | Zwaan, R. A.
(1996). Processing narrative time shifts. JEP: LMC, 22, 1196-1207
Many thanks to Phil Büttner and Alex Richter for their assistance in collecting the data
AMLaP 2013, Marseille
[email protected]