Integration and organization of trauma memories and posttraumatic symptoms

Abstract

To examine the connection between trauma memory integration in personal memory, memory organization, and posttraumatic symptom severity, 47 trauma-exposed adults undertook an event-cuing task for their trauma memory and for a memorable nontraumatic negative event. Measures of integration provided by self-endorsement, rated by naïve judges, or calculated from the language of the memories, did not significantly predict posttraumatic stress disorder symptom severity after adjusting for age, time since the event, anxiety when disclosing, familiarity of the memory, and integration of nontrauma memory. Less use of casual connectives in the trauma memory narrative was associated with higher trauma-related avoidance (r = .33; p = .03), whereas self-rating of the trauma memory as disorganized was associated with higher overall symptom severity (r = .42; p = .006).

Full text

Journal of Traumatic Stress, Vol. 24, No. 6, December 2011, pp. 716–725 (C
2011)
Integration and Organization of Trauma Memories and
Posttraumatic Symptoms
Richard O’Kearney and Aliza Hunt
Australian National University
Nancy Wallace
Connecticut College
To examine the connection between trauma memory integration in personal memory, memory organization,
and posttraumatic symptom severity, 47 trauma-exposed adults undertook an event-cuing task for their trauma
memory and for a memorable nontraumatic negative event. Measures of integration provided by self-endorsement,
rated by na¨
ıve judges, or calculated from the language of the memories, did not significantly predict posttraumatic
stress disorder symptom severity after adjusting for age, time since the event, anxiety when disclosing, familiarity
of the memory, and integration of nontrauma memory. Less use of casual connectives in the trauma memory
narrative was associated with higher trauma-related avoidance (r=.33; p=.03), whereas self-rating of the
trauma memory as disorganized was associated with higher overall symptom severity (r=.42; p=.006).
There are divergent views about the connection between post-
traumatic adjustment and how the memory for the trauma is in-
tegrated within autobiographical memory. Some (Ehlers & Clark,
2000; Foa & Rothbaum, 1998) propose that isolation of memo-
ries of the trauma from other personal memories is critical in the
maintenance of symptoms. Ehlers and Clark (2000) argue that in
posttraumatic stress disorder (PTSD) “one of the main problems
is that the trauma memory is poorly elaborated and inadequately
integrated into its context in time, place, subsequent and previous
information and other autobiographical memories” (p. 325). Oth-
ers (Berntsen & Rubin, 2007; Berntsen, Willert, & Rubin, 2003)
argue that trauma memories can act as landmarks in autobiograph-
ical memory and “rather than poorly integrated...traumatic mem-
ory stays highly accessible and may form a cognitive reference point
for the organisation of autobiographical knowledge” (Berntsen &
Rubin, 2007, p. 418). For people with PTSD, the trauma memo-
ries are central components of autobiographical memory.
Findings cited for the claim that inadequate integration re-
lates to poor adjustment include observations that memory for the
trauma in PTSD is dominated by perceptual and sensory impres-
sions (Hellawell & Brewin, 2002), is disorganized (Jones, Harvey,
Richard O’Kearney, Aliza Hunt, Department of Psychology, Australian National University;
Nancy Wallace, Department of Psychology, Connecticut College.
This study was supported by an Australian Research Council Discovery Grant (DP0664860
Trauma Memory and Narrative Structure in Post-Traumatic Distress).
Correspondence concerning this article should be addressed to Richard O’Kearney, Depart-
ment of Psychology, Australian National University, Canberra, ACT 0200, Australia. E-mail:
richard.okearney@anu.edu.au
C
2011 International Society for Traumatic Stress Studies. View this article online at
wileyonlinelibrary.com DOI: 10.1002/jts.20690
& Brewin, 2006; Jelinek, Randjbar, Seifert, Kellner, & Mortiz,
2009), lacks a temporal perspective (Hellawell & Brewin, 2002),
and is less conceptually connected (Ehlers, Hackmann, & Michael,
2004; Krans, N¨
aring, Holmes, & Becker, 2009). It has not, how-
ever, been shown that these characteristics of the trauma memory
itself means that it is isolated within autobiographical memory.
The evidence for the landmark view comes from several studies
showing that self-report Centrality of Events Scale scores are posi-
tively related to severity of PTSD symptoms (Berntsen & Rubin,
2007, 2006). Given this disparity in views, and the limitations
of memory self-reports (Kindt & van den Hout, 2003), we be-
lieve it is critical to develop multimethod approaches to assess the
link between trauma memory integration within autobiographical
memory and PTSD symptoms.
We know of only one experimental study that has examined this
connection. Kleim, Wallott, and Ehlers (2008) asked assault vic-
tims with and without PTSD questions from the Autobiographical
Memory Inventory while they undertook script-driven imagining
of the assault and another negative event. Kleim et al. (2008) found
that those with PTSD took significantly longer to retrieve auto-
biographical memories during the most disturbing parts in their
trauma script than during hot spots of the negative event script.
The participants with PTSD were also significantly slower in re-
sponding during trauma memory than those without PTSD. The
results indicated disruption in speed of memory responses during
trauma memory hot spots for those with PTSD. It is not certain,
however, that slower responses to the inventory imply poor integra-
tion of the trauma memory into autobiographical memory. Inven-
tory response slowing may reflect additional cognitive resources re-
quired by the PSTD group to inhibit their trauma remembering or
716

Organization of Trauma Memories and Symptoms 717
factors such as between-group differences “in physiological arousal
or transitory attention” (Klein et al., 2008, p. 232) during the hot
spots.
The current study provides an additional direct examination
of the link between posttraumatic stress symptoms and trauma
memory integration. We use the method of event cueing, which
assesses whether specific event memories are embedded in struc-
tures that organize autobiographical information into clusters of
related events (Brown, 2005; Brown & Schopflocher, 1998). In
event cueing, participants first narrate a specific personal event
memory that is used to cue a second autobiographical memory.
Participants then indicate how the cuing and cued narratives are re-
lated by endorsing relationship domains, i.e., one event was part of
the other, both part of a larger story, both have same theme, or they
took place in the same location, involved the same people, describe
the same activity, or same period (Brown & Schopflocher, 1998).
The relationship endorsement task is used to classify memory pairs
into those which are subsumed or integrated within a larger auto-
biographical structure (general story, theme), and pairs thatare not
integrated (only share specific components, e.g., people, location,
time). The classification is consistent with approaches to personal
memory that propose organizing structures within autobiograph-
ical memory such as themes (Conway, 2005; Conway & Pleydell-
Pearce, 2000), self-narratives (Robinson & Taylor, 1998) or narra-
tive episodes (Radvansky, Copeland, & Zwaan, 2005). From this
perspective, integration is understood as how well the memory sys-
tem incorporates specific event memories into the larger structure
(Conway, 2005). An event memory that is integrated will be more
likely than a nonintegrated memory to cue another personal event
related by theme or story. The event-cuing task, therefore, pro-
vides a face-valid way to assess how the integration of the trauma
memory in autobiographical memory relates to PTSD symptoms.
Two other methods are added here to assess integration of the
memory pair generated from the cuing task. The first of these is
na¨
ıve judges’ ratings of the degree of relationship between the pair
of narratives in the relationship domains. The second is a measure
of the semantic similarity between the two narrative memories us-
ing latent semantic analysis. Latent semantic analysis is a statistical
characterization of narrative to narrative relations using partici-
pants’ word choice and provides an index of the strength of their
semantic similarity (Landauer, Foltz, & Laham, 1998). This tri-
angulation of methods provides a categorical measure of whether
the cued and cuing memories are integrated (self-endorsement), a
continuous measure of the degree to which the pairs are integrated
(judges’ ratings), and a continuous measure of the strength of the
semantic similarity between the two narratives.
Besides inadequate integration, the deficit view proposes that a
disorganized quality of the trauma memory itself predicts PTSD
symptom severity (Ehlers et al., 2004). Memory organization refers
to the internal, structural integrity of the memory itself regardless
of its integration within autobiographical memory. When critically
appraised, the evidence about the association of trauma memory
organization and PTSD symptoms is inconclusive (O’Kearney &
Perrott, 2006; Rubin, 2011). The inconsistency arises in part from
the considerable heterogeneity in the definition and measurement
of disorganization. Studies have used self-report as well as memory
narrative coding and judges’ ratings. The measures often combine
incongruent understandings of narrative (dis)organization. For ex-
ample, Evans, Ehlers, Mezey, and Clark’s (2007) global rating of
disorganization ask judges to combine how well the narrative is
organized as a sequence of events (temporal organization) with
how much detail the narrative has without separate considera-
tion of the role of detail in the overall measure of organization.
Similarly, some narrative coding schemes measure disorganization
by combining repetitions in the narrative with well-formed ex-
pressions of uncertainty (disorganized thoughts) and expressions
indicating lack of understanding (organized thoughts–reversed).
Repetitions in narratives may indicate discontinuity, but can also
be used to emphasize or to conceal thoughts or emotions. In addi-
tion, although expressions of uncertainty (“I don’t remember. . . ”;
“I don’t know. . . ”) indicate cognitive confusion (a semantic dis-
tinction) they can be well organized, i.e., a person can describe
their uncertainty about what happened in an organized way. A
deficit in organized thoughts on the other hand is judged by the
paucity of syntactic connection within the trauma narrative.
The current study uses measures of disorganization informed by
knowledge of the formal organization of personal event narratives
(Graesser, McNarama, Louwerse, & Cai, 2004; Petersen & Mc-
Cabe, 1983). From this vantage (dis)organization is understood as
reflected in two language domains: narrative cohesion and narrative
coherence. Cohesion focuses on objective linguistic structures or
connections between sentences or clauses and is a direct linguistic
measure of the narrative organization. The current study focuses
on linguistic devices (connectives) that establish spatiotemporal
and causal cohesion. These conjunctions organize sentences and
clauses within a narrative via addition (and, also), comparison (but,
however), temporality (after, before, when), and causality (so, be-
cause, that is). Coherence, on the other hand, focuses on discourse
about participants’ understanding of goals, actions, and outcomes,
or topics within the narrative and depends on the knowledge that
the participants bring to the situation (Graesser et al., 2004). It
is assessed here by judge’s rating. We have previously found that
the narrative cohesion measured by the proportion of temporal
and causal connectives within the trauma narrative was related to
higher levels of intrusive symptoms in children and youth after an
accident requiring hospitalization (O’Kearney, Speyer, & Kenardy,
2007).
The present study has two aims. First, it examines the rela-
tionship between PTSD symptom severity and the integration
of the trauma memory using three indices of integration from
the event-cuing task: self-endorsement of memory pairs as inte-
grated, judges’ integration rating of the pairs, and their seman-
tic similarity. Second, the study assesses the relationship between
PTSD symptom severity and trauma memory organization using
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

718 O’Kearney, Hunt, and Wallace
measures of narrative organization: participants’ self-report of
memory disorganization; the proportion of additive, comparative,
temporal, and causal connectives in the narratives; and the rating
of the narratives’ coherence by na¨
ıve judges.
METHOD
Participants and Procedure
The study was approved by the Australian National University
Human Research Ethics Committee. The sample comprised 47
fluent English speakers (9 men) who reported exposure to a trau-
matic event meeting Diagnostic and Statistical Manual of Mental
Disorders (4th ed., DSM-IV ; American Psychiatric Association,
1994) Criterion A and who did not have a history of psychoses,
head injury with loss of consciousness, drug or alcohol addiction,
or any neurological disorder. Trauma types included witnessing
the death of family member/friend (15/47; 31.9%), diagnosis of
a severe illness (8/47; 17.0%), accident or disaster (9/47; 19.1%),
sexual assault (7/47; 14.9%), nonsexual assault (5/47; 10.6%),
and exposure to war-like combat (1/47; 2.1%). The mean age of
participants was 22.83 (SD =8.19).
Participants were seen individually and asked to nominate a
significant traumatic event they have experienced and then com-
plete the Impact of Event Scale (IES) in regard to this event. They
then undertook the narrative generation and event-cueing tasks
for two types of memories: (a) the IES traumatic event, and (b) a
nontraumatic but negative personal event. Participants were given
the following instructions adapted from Jones et al. (2006): “In
a moment I’m going to ask you to recall the traumatic event.
Please describe the memories of the event as vividly as possible.
I’d like you to close your eyes and tell me what happened in as
much detail as you remember as if it were happening right now.
This includes details about the surroundings, your activities, how
you felt and what your thoughts were during the event.” For the
nontraumatic event, participants were given the same instructions
except introduced by: “In a moment I’m going to ask you to recall
a negative memorable event that happened to you.” An audiotape
of the narrative was replayed to the participant and participants
were asked, using the same narrative generation instructions, to
describe another memory “somehow related” to the cueing event.
After the administration of a WAIS-III vocabulary test, the event
generation and event-cuing procedure was repeated for the other
event type (trauma or nontrauma). The order of generation of
narrative event type was counterbalanced across participants.
After each event-cueing task, participants nominated whether
the cueing and cued events were related in regard to two categories:
integrated (one was part of the other; both were part of a broader
event, story, or theme); not integrated (events only share the same
people, activity, location, time). Participants were asked to date
each event and rate each on a 5-point scale of personal importance
(1 =not important at all to 5 =extremely important). Participants
then completed the remainder of the self-report measures.
Measures
Memory narratives were transcribed and total number of words
and utterances, defined as a language unit conveying meaning,
were calculated. A judge blind to the study’s aims and to the
PTSD symptom severity of participants rated the overall strength
of similarities (1 =not at all to 5 =extremely similar) between
narrative pairs for the integrated relationship (one was part of
the other; both were part of a broader event, story, or theme).
A second na¨
ıve judge rated 30% of narrative pairs. Agreement
between the judges was acceptable (one-way layout intraclass
coefficient =.79).
The measure of narrative pairs’ semantic relatedness was gener-
ated using pair wise latent semantic analysis (LSA). Term-to-term
semantic similarity for the two texts for each pair was estimated
using latent semantic analysis software (Latent Semantic Analysis
@ CU Boulder, 2010). Higher scores indicate stronger semantic
similarity.
The largest class of cohesive devices is connectives, which link
sentences and clauses via addition (e.g., and, also), comparison
(e.g., but, however), temporality (e.g., after, before, now), and
causality (e.g., so, because). The number of references in each cat-
egory (additive, comparative, temporal, causal) was converted to
a proportion of total number of connectives. Some words (e.g.,
and) serve additive or temporal functions. In such cases, the nar-
rative context was used to determine the function. A second judge
scored 25% of the narratives. There was agreement between the
two judges in the proportion of connectives identified: r=.82 for
additive, r=.94 for comparative, r=.91 for temporal, and r=
.90 for causal.
Narrative coherence was rated by a na¨
ıve judge on a 6-point
scale using standard criteria (Petersen & McCabe, 1983): (1)
Disoriented–the narrative is too disoriented for the listener to
understand; (2) Impoverished–the narrative consists of too few
sentences for a pattern to be recognized; (3) Chronological–the
narrative is a simple description of consecutive, successive events;
(4) Leap-frogging–the narrative moves from one event to another
connecting later segments to earlier ones within an integrated ex-
perience; (5) Ending-at-the-high-point–the narrative builds up to
a high point and then ends, but there is no resolution; (6) Mature–
the narrative builds to a high point, dwells on it, and then resolves
it. Agreement between judges for coherence ratings was acceptable,
one-way layout ICC =.72.
The Disorganization subscale of the Trauma Memory Ques-
tionnaire (TMQ; Halligan, Michael, Clark, & Ehlers, 2003) was
used to measure self-reported disorganization of the trauma mem-
ory. In the current study α=.93.
The severity of PTSD symptoms overall was assessed with the
Posttraumatic Stress Diagnostic Scale using the specific trauma
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

Organization of Trauma Memories and Symptoms 719
(PDS; Foa, 1995). In the current study, PDS had high inter-
nal consistency for the total severity score (α=.90). Severity of
trauma-specific intrusive and avoidant symptoms related to the
specific trauma were measured using the Impact of Event Scale
(IES; Horowitz, Wilner, & Alvarez, 1979). The IES in the current
study had acceptable internal consistency (α=.76 for Intrusion
and α=.75 for Avoidance).
The Depression subscale (14 items) of the Depression Anxiety
Stress Scale (DASS; Lovibond & Lovibond, 1993) was used to as-
sess current depression severity. Internal consistency in the current
study was α=.92.
The Confidence, Familiarity, and Anxiety in Disclosing mea-
sure developed for this study required participants to endorse on
a 5-point Likert scale single items assessing level of confidence
in the completeness of their memories, frequency of previous re-
countings of the trauma (familiarity), and anxiety when disclosing
the memory during the event-cuing task. This measure is available
from the first author.
The Vocabulary subtest of the Wechsler Adult Intelligence
Scale-Third Edition (WAIS-III; Wechsler, 1997) was used to as-
sess verbal abilities. Participants were required to define a series of
orally presented words of ascending difficulty until all words in
the list were defined or until six successive words were answered
incorrectly or not at all. The Vocabulary subtest had a reliability
in this study of α=.83.
Data Analyses
Hierarchical multiple regression analyses were used to test whether
the measures of the integration of the trauma memory could pre-
dict PTSD symptom severity. Separate models were investigated
for overall severity and for severity of intrusive and avoidant symp-
toms. The three measures of trauma memory integration were
included in the model as well as variables with modest (p<.1)
first-order association with at least one PTSD symptom severity
measure were included. The first step of the model included time
since trauma, age, anxiety when disclosing the trauma, and famil-
iarity. In the second step, nontrauma narrative pair semantic simi-
larity score was added. In the final step, trauma memory integration
measures were entered. A second set of hierarchical multiple re-
gression analyses tested whether the measures of trauma memory
organization could predict severity of PTSD symptoms overall and
intrusive and avoidance symptoms. Measures of trauma memory
organization and other variables were included in the model if
they had modest (p<.1) first-order association with one PTSD
symptom measure. List-wise deletion of variables was used to deal
with missing variables on participants’ self-report measures.
RESULTS
Severity of PSTD symptoms on the PDS ranged from 1 to 45
with a mean score of 13.0 (SD =9.9). The mean score on the
Intrusion subscale for the trauma event on the IES was 13.96
(SD =7.29); the mean Avoidance subscale score was 15.70 (SD =
9.43). The participants had a mean DASS depression score of
7.98 (SD =6.44) and a mean TMQ Disorganization score of
7.38 (SD =5.98). The mean WAIS-III Vocabulary scaled score
for the participants was 13.51 (SD =2.40).
Table 1 presents participant and narrative characteristics. The
trauma event was significantly more distant in time than the non-
trauma event and had more words and utterances. Participants
rated their trauma memory as significantly more important, that
they had significantly less confidence in the completeness of their
trauma memory, and had significantly more anxiety when disclos-
ing the trauma event. There was no significant difference between
narratives on self-endorsement, but trauma narrative pairs were
judged as significantly more integrated and were significantly more
semantically similar. The narratives did not differ significantly on
judged coherence, but the trauma narrative had a statistically sig-
nificant lower proportion of additive cohesive markers and higher
proportion of comparative markers.
Correlations between posttraumatic stress symptom severity,
depression, participant factors, trauma narrative characteristics,
confidence, familiarity, and anxiety in disclosing scale, measures
of integration of the trauma pair, and measures of organization of
the trauma narrative are presented in Table 2. The PTSD symp-
tom measures show strong positive association with one another
and with depression. Age and anxiety when disclosing the trauma
had moderate to strong positive associations with symptom sever-
ity as well as with the proportion of temporal markers in the
narrative. Familiarity had a moderate negative correlation with
avoidant symptoms and confidence in the memory a moderate
negative correlation with depression. Confidence in reporting was
associated with lower self-reported disorganization (TMQ) and
moderately with judge’s rating of the coherence of the trauma nar-
rative. Self-reported trauma memory disorganization showed a sig-
nificant moderate positive association with the degree to which the
trauma pair was rated as integrated. TMQ Disorganized score was
also positively related to semantic similarity score for the trauma
pairs.
Judges’ rating of integration of the trauma pairs were moder-
ately and significantly correlated with the trauma pair’s semantic
similarity scores. In addition, trauma memory pairs endorsed by
the participants as integrated were rated by the judges as signifi-
cantly more integrated, Integrated M=3.79, SD =1.00; Non-
integrated M=2.37, SD =.67, t(44) =4.01, p<.001. They
were also significantly higher on semantic similarity, Integrated
M=.99, SD =.005; Nonintegrated M=.98, SD =.006,
t(44) =2.28, p=.01. The measures of integration showed
little association with PTSD symptom severity and depression.
The TMQ disorganization was moderately associated with overall
symptom severity and depression, but only weakly with severity
of avoidant symptoms and not with intrusive symptoms. Partici-
pants’ use of fewer casual connectives in their trauma narrative was
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

720 O’Kearney, Hunt, and Wallace
Tab l e 1. Characteristics of Trauma and Nontrauma Narratives
Trauma narrative Nontrauma narrative
Variable MSD M SD t
Months since narrative event 54.24 45.81 14.39 12.01 65.87∗∗
Importance of narrative event 4.81 0.50 4.09 0.97 4.89∗∗
Confidence in memory 3.83 1.01 4.28 0.90 −2.49∗
Familiarity 2.96 1.14 3.02 1.11 −0.25
Anxiety in disclosing event 3.23 0.98 1.23 0.52 14.35∗∗
Word count 670.32 417.37 426.28 312.71 4.06∗∗
Total number of utterances 69.04 46.09 42.85 28.12 3.91∗∗
Prop (%) integrated 36/47 (76.6) 29/47 (61.7%) χ2=1.76
Judge’s rating as integrated 3.51 1.10 2.99 1.30 2.69∗
Semantic similarity .99 0.005 .98 0.011 2.79∗
Judge’s coherence rating 5.06 0.92 4.87 1.58 0.912
Additive .57 0.14 .65 .11 −2.69∗
Comparative .12 0.09 .08 0.07 2.63∗
Temporal .13 0.08 .12 0.10 0.66
Casual .16 0.08 .15 0.07 0.34
∗Note. p <.05. ∗∗ p<.01.
associated with higher intrusive and avoidance scores, but not with
overall symptom severity and depression. None of the other mea-
sures of cohesion or the rating of coherence was notably associated
with PTSD symptom severity.
Of the measures of integration from the nontrauma pairs only
semantic similarity score showed a moderate correlation with in-
trusion score, r(46) =.32; p=.03 with neither of the other
measures of nontrauma narrative integration associated with any
symptom severity. Of the measures of organization of the non-
trauma narrative only the rating of narrative coherence showed
a moderate correlation with avoidance, r(47) =−.31; p=.04,
whereas none of the cohesion measures for the nontrauma narra-
tive were associated with any symptom severity.
Table 3 presents the outcomes for the first set of regression
analyses. Neither the overall model nor the steps accounted for a
significant proportion of the variance in overall symptom severity.
Variables in the first step accounted for a significant proportion
ofthevarianceinseverityofintrusions,F(4, 40) =2.74, p=
.04, and avoidance, F(4, 41) =4.26, p=.006, with anxiety
when disclosing for intrusions and anxiety when disclosing and
familiarity for avoidance significant individual contributors. Most
importantly, the addition of the trauma integration variables did
not significantly add to the prediction of the severity of intrusive
or avoidance symptoms after controlling for step one variables and
the relevant nontrauma pair integration measure.
Table 4 presents the results of the second set of regression anal-
yses. In the first step of the model, time since trauma, age, anxiety
when disclosing, and familiarity were entered. In the second step,
nontrauma narrative coherence rating was added. In the final step,
memory organization measures (TMQ; causal connectives) were
entered. Higher self-reported disorganization score significantly
added to the prediction of overall severity of symptoms, F(2,39)
=4.43, p=.02, but not severity of avoidance symptoms whereas a
lower proportion of causal connectives in the trauma memory nar-
rative predicted higher avoidance, F(2, 39) =3.25, p=.05, but
not overall symptom severity. Neither measure of disorganization
added significantly to the prediction of intrusion severity.
DISCUSSION
This study used an event cueing task to generate three measures
(self-endorsement, judges’ ratings, and linguistic) of the integration
of event memories into personal memory. The better integration
of trauma memories than nontrauma memories may be due to the
higher salience of trauma memories compared to emotional but
nontraumatic memories. We found no evidence, however, consis-
tent with the views that either inadequate (Ehlers et al., 2000) or
enhanced (Berntsen & Rubin, 2007) integration of trauma mem-
ories is associated with severity of posttraumatic stress symptoms
overall or intrusive or avoidant symptoms. Degree of anxiety ex-
perienced when disclosing the trauma event predicted intrusive
symptom severity and anxiety when disclosing, and fewer previous
recountings of the event predicted avoidance severity. The mea-
sures of integration did not add to the prediction of the severity of
posttraumatic stress symptoms.
Our results document interesting connections between mea-
sures of the organization of the trauma memory and post-
traumatic stress symptoms. First, participants’ self-report of the
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

Organization of Trauma Memories and Symptoms 721
Tab l e 2. Correlations of Participant, Symptom, Trauma Narrative Integration, and Organization Measures
Variable 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
1. Gender −.04 −.08 −.13 −.09 .09 .15 .05 −.14 −.08 .05 −.13 −.11 .10 −.01 .14 −.19 .05 .32 .13 .33
2. Age – −.22 −.20 −.23 −.28 .23 .31 .17 .33 −.23 .46 .10 .24 −.37 .03 .11 −.04 −.21 −.02 −.32
3. PDS – .62 .58 .66 −.24 −.08 .05 −.24 .31 −.07 .03 −.11 .22 −.12 .04 −.02 −.06 −.13 .33
4. Intrusion – .62 .45 −.20 −.15 .15 −.16 .40 −.13 .02 −.07 .11 .10 .13 .01 .00 −.30 .01
5. Avoidance – .45 −.08 −.01 −.20 −.16 .44 −.33 −.09 −.26 −.07 .03 .07 .04 .15 −.38 .17
6. Depression – −.03 −.06 .14 −.33 .23 −.14 .05 .13 .24 .04 .00 .07 .10 −.19 .45
7. Time since trauma – .14 −.21 −.19 −.04 −.15 .16 .13 −.09 .01 −.17 .08 .01 .27 .29
8. Vocab. – .05 .01 .07 .14 .10 .24 .06 .03 −.04 .04 .01 −.01 .12
9. Importance – .33 .14 .18 .09 .17 .34 .17 .09 −.28 .21 −.02 −.08
10. Confidence – −.05 −.01 −.15 −.29 −.13 .29 .21 −.25 −.07 .01 −.64
11. Anxiety –−.07 .03 −.13 .08 .22 −.07 −.09 .36 .01 .09
12. Familiarity – .11 .02 .10 −.02 −.19 .10 .08 .20 .06
13. Endorsed –.51.29−.07 −.19 .12 .11 .07 .25
14. Rating –.35−.19 −.13 .05 .06 .07 .37
15. LSA –−.06 −.14 .03 .13 .16 .26
16. Coherence –.09−.21 .16 −.04 −.21
17. Additive –−.71 −.42 −.51 −.37
18.Comparative –−.10 .09 .23
19. Temporal –.05.17
20. Causal –.15
21. TMQ Dis. –
Note. N =47. PDS =Posttraumatic Stress Diagnostic Scale severity score; Vocab. =Wechsler Adult Intelligence Scale-Third Edition Vocabulary score; LSA =latent semantic analysis similarity score;
TMQ Dis. =Trauma Memory Questionnaire Disorganization score.
Critical rfor p<.05 =0.29; for p<.01 =0.38; for p<.001 =.51.
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

722 O’Kearney, Hunt, and Wallace
Tab l e 3. Hierarchical Regression Analysis for Integration Measures Predicting PTSD Symptom Severity
Variable R2tBSEβ
Dependent variable: Overall severity
Step 1 .16
Time since trauma −0.14 −0.05 0.03 −.22
Age −0.69 −0.13 0.18 −.11
Anxiety disclosing 1.79 2.71 1.51 .26
Familiarity −0.53 −0.69 1.31 −.01
Step 2 .00
LSA nontrauma 0.37 56.80 154.45 .06
Step 3 .05
Self-endorsement −0.59 −0.82 1.39 −0.9
Rating integrated 0.38 58.76 156.40 .06
LSA trauma −0.29 −0.38 1.35 −0.4
Dependent variable: Intrusions
Step 1 .22∗
Time since trauma −1.36 −0.03 0.20 −.20
Age −0.39 −0.05 0.13 −.06
Anxiety disclosing 2.56∗2.77 1.08 .37
Familiarity −0.92 −0.85 0.93 −.13
Step 2 .06
LSA nontrauma 1.76 199.14 113.44 .28
Step 3 .01 −.10
Self-endorsement 1.73 199.65 115.32 −.07
Rating integrated −0.53 −0.05 0.93 −.01
Dependent variable: Avoidance
Step 1 .29∗∗
Time since trauma −0.67 −0.02 .03 −.09
Age −0.72 −0.12 .16 −.01
Anxiety disclosing 2.88∗∗ 3.77 1.3 .39
Familiarity −2.31∗∗ −2.62 1.13 −.31
Step 2 .01
LSA nontrauma 0.80 107.4 132.92 .12
Step 3 .05
Self-endorsement −2.44 −2.83 1.16 .09
Rating integrated 0.89 115.90 130.99 −.23
LSA trauma −1.50 −1.70 1.13 −.05
Note.N=47. PTSD =posttraumatic stress disorder; LSA =latent semantic analysis; LSA nontrauma =semantic similarity for nontrauma pair; LSA trauma =semantic
similarity for trauma pair.
∗p<.05. ∗∗ p<.01.
disorganization of their trauma memory significantly enhanced
the prediction of overall symptom severity after controlling for
age, time since the event, anxiety when disclosing, familiarity of
the memory, and the organization of the nontrauma narrative
memory. Self-reported disorganization was not, however, a signif-
icant predictor of severity of intrusive or avoidant symptoms. In
contrast, a linguistic measure of organization added to the spe-
cific prediction of severity of posttraumatic avoidance but not
overall symptom severity. In particular, trauma memory narratives
that contained a smaller proportion of causal connectives (be-
cause, therefore) were associated with higher levels of self-reported
trauma-related avoidant symptoms. These data suggest the poor
causal organization of the trauma memory is one way in which
avoidant strategies are enacted in effortful trauma remembering.
Constructing casually disconnected pieces of the memory may en-
able avoidant participants to satisfy goals related to task demands,
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

Organization of Trauma Memories and Symptoms 723
Tab l e 4. Hierarchical Regression Analysis for Organization Measures Predicting PTSD Symptom Severity
Variable R2tBSEβ
Dependent variable: Overall severity
Step 1 .16
Time since trauma −0.15 −0.05 0.03 −.22
Age −0.71 −0.13 0.18 −.11
Anxiety disclosing 1.84 2.70 1.47 .26
Familiarity −0.55 −0.68 1.25 −.01
Step 2 .01
Coherence nontrauma −0.71 −0.65 0.90 −.10
Step 3 .15∗
Causal Connectives −0.45 −0.08 0.19 −.07
Disorganization TMQ 2.92∗∗ 0.74 0.25 .45
Dependent variable: Intrusions
Step 1 .22∗
Time since trauma −1.36 −0.30 0.20 −.20
Age −0.39 −0.05 0.13 −.06
Anxiety disclosing 2.56∗2.76 1.05 .37
Familiarity −0.95 −0.84 0.89 −.13
Step 2 .01
Coherence nontrauma −0.49 −0.33 0.68 −.07
Step 3 .05
Causal connectives −0.16 −0.24 0.15 −.26
Disorganization TMQ 0.17 0.03 .19 −.03
Dependent variable: Avoidance
Step 1 .29∗∗
Time since trauma −0.67 −0.02 0.03 −.09
Age −0.75 −0.12 0.16 −.01
Anxiety disclosing 2.96∗∗ 3.74 1.27 .39
Familiarity −2.40∗−2.60 1.08 −.31
Step 2 .06
Coherence nontrauma −1.97 −1.49 0.76 −.25
Step 3 .09∗
Causal connectives −2.31∗−0.37 0.16 −.31
Disorganization TMQ 1.20 0.26 0.22 .17
Note. N =47. PTSD =Posttraumatic stress disorder; TMQ =Trauma Memory Questionnaire.
∗p<.05. ∗∗ p<.01.
i.e., to “tell me what happened as if it were happening right now,”
but also to maintain distance from the threat that the trauma mem-
ory represents. This account suggests that trauma information is
encoded self-referentially within personal memory, but when re-
trieved the causal organization of the trauma memory will be an
outcome of approach or avoidant strategies.
The inconsistency between our results and Kleim et al.’s (2008)
may be explained by the differences between the studies in how
memory integration is understood. The current study used the
whole trauma event memory to cue a related personal memory,
whereas Kleim et al. (2008) focused on specific, highly distressing
segments of the memory. Undoubtedly, particular segments of the
trauma memory may lack connection within the narrative. It is not
clear, however, to what degree such disjunction impacts on the in-
tegration of the trauma memory with the rest of autobiographical
memory. Even though our integration measures refer to the trauma
event memory, we would expect that hot spots in the remember-
ing of the event would impact on one of the measures and its
association with severity of posttraumatic stress symptoms. Nev-
ertheless, we acknowledge that because we measured integration
of the event memory our results do not directly address proposals
about the link between posttraumatic symptoms and integration
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

724 O’Kearney, Hunt, and Wallace
of specific segments of the memory. One of the major challenges
for future research is to develop valid conceptualization and mea-
sures of trauma memory integration to allow this distinction to be
investigated empirically.
The moderate correlations between our two continuous mea-
sures of integration together with the significantly higher inte-
grated rating and LSA score for trauma memory pairs endorsed
as integrated compared to those not endorsed indicates com-
monality in the construct assessed by the three measures. In
addition, using the event-cuing procedure in the standard way
provides a link between clinical models of trauma memory and
current theoretical approaches to examining the hierarchical struc-
ture of autobiographical remembering (Brown, 2005; Conway,
2005). We believe that such links are critical to the validity of
empirical investigation of proposals about the nature of trauma
memories in PTSD. At the same time, the procedure of event
cueing has not been used in PTSD memory research before and
the measures of integration are novel and require additional val-
idation in this context. In addition, the narrative generation in-
structions used here, although similar to some studies (Jones et al.,
2006) differs from others as they did not specify that the partic-
ipants recount the event in sequence chronologically. For these
reasons, our results need to be compared to those of other studies
cautiously.
There were a number of other limitations of the study. Inclu-
sion of the Centrality of Events Scale as a self-report measure of
integration would have helped in drawing inference about the en-
hancement of trauma memory integration and symptom severity.
Similarly, inclusion of a self-report measure of organization of the
nontrauma memory would have completed the multimethod ap-
proach. Although all the participants had experienced a Criterion
A trauma they were not currently seeking help and the intensity
of their symptoms may not have been as high as in other studies.
The participants described a mixture of types of traumatic events.
It is possible that self-relevant information from certain types of
trauma such as interpersonal abuse may be particularly difficult to
reconcile with existing autobiographical information. Using a mix
of types of traumas may have reduced the chances of observing
symptom-specific problems or enhancements in memory integra-
tion. The small sample of our study presents statistical limitations.
Although we reduced the number of predictors in standard ways,
the regression analyses may have still been underpowered to detect
real increments in prediction.
Overall, the results of the current study do not support claims
that inhibition or enhancement of the integration of the trauma
memory within autobiographical memory is associated with sever-
ity of posttraumatic symptoms. The results show, however, that
overall posttraumatic stress symptom severity is associated with
self-reported disorganization of the trauma memory and that sever-
ity of avoidant symptoms is associated with memories for the
trauma, which are impoverished in regard to attempts to provide
a casual account of the traumatic event.
REFERENCES
American Psychiatric Association. (1994). Diagnostic and statistical manual of men-
tal disorders (4th ed.). Washington, DC: Author.
Berntsen, D., & Rubin, D. C. (2007). When a trauma becomes a key
to identity: Enhanced integration of trauma memories predicts posttrau-
matic stress disorder symptoms. Applied Cognitive Psychology, 21, 417–431.
doi:10.1002/acp.1290
Berntsen, D., & Rubin, D. C. (2006). The centrality of Events Scale: A mea-
sure of integrating a trauma into one’s identity and its relationship to post-
traumatic stress disorder symptoms. Behaviour Research & Therapy,44,219–
231. doi:10.1016/j.brat.2005.01.009
Berntsen, D., Willert, M., & Rubin, D. C. (2003). Splintered memories or vivid
landmarks? Qualities and organization of traumatic memories with and without
PTSD. Applied Cognitive Psychology, 17, 675–693. doi:10.1002/acp.894
Brown, N. R. (2005). On the prevalence of event clusters in autobiographical
memory. Social Cognition, 23, 35–69. doi:10.1521/soco.23.1.35.59194
Brown, N. R., & Schopflocher, D. (1998). Event cueing, event clusters, and the
temporal distribution of autobiographical memories. Applied Cognitive Psychol-
ogy, 12, 305–319. doi:10.1002/(SICI)1099-0720(199808)12:4<305::AID-
ACP569>3.0.CO;2-5
Conway, M. A. (2005). Memory and the self. Journal of Memory and Language, 53,
594–628. doi:10.1016/j.jml.2005.08.005
Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobi-
ographical memories in the self-memory system. Psychological Review, 107,
261–288. doi:10.1037/0033-295X.107.2.261
Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic stress
disorder. Behaviour Research and Therapy, 38, 319–345. doi:10.1016/S0005-
7967(99)00123-0
Ehlers, A., Hackmann, A., & Michael, T. (2004). Intrusive re-experiencing in
post-traumatic stress disorder: Phenomenology, theory, and therapy. Memory,
12, 403–415. doi:10.1080/09658210444000025
Evans, C., Mezey, G., Ehlers, A., & Clark, D. (2007). Intrusive memories in
perpetrators of violent crimes: Emotions and cognitions. Journal of Consulting
and Clinical Psychology, 75,113–144. doi:10.1037/0022-006X.75.1.134
Foa, E. B. (1995). Posttraumatic Stress Diagnostic Scale manual.Minneapolis,MN:
National Computer Systems.
Foa, E. B., & Rothbaum, B. O. (1998). Treating the trauma of rape: Cognitive
behavioral therapy for PTSD.NewYork,NY:GuilfordPress.
Graesser, A., McNamara, D. S., Louwerse, M., & Cai, Z. (2004). Coh-Metrix:
Analysis of text on cohesion and language. Behavioral Research Methods, Instru-
ments, and Computers, 36, 193–202. doi:10.3758/bf03195564
Halligan, S. L., Michael, T., Clark, D. M., & Ehlers, A. (2003). Posttraumatic stress
disorder following assault: The role of cognitive processing, trauma memory,
and appraisals. Journal of Consulting and Clinical Psychology, 71, 419–431.
doi:10.1037/0022-006X.71.3.419
Hellawell, S., & Brewin, C. (2002). A comparison of flashbacks and ordinary auto-
biographical memories of trauma: Cognitive resources and behavioural obser-
vations. Behaviour Research and Therapy, 40, 1139–1152. doi:10.1016/S0005-
7967(01)00080-8
Horowitz, M., Wilner, N., & Alvarez, W. (1979). Impact of Event Scale: A measure
of subjective stress. Psychosomatic Medicine, 41, 209–218.
Jelinek, L., Randjbar, S., Seifert, D., Kellner, M., & Mortiz, S. (2009). The
organisation of autobiographical and nonautobiographical memory in post-
traumatic stress disorder (PTSD). Journal of Abnormal Psychology, 118,288–
298. doi:10.1037/a0015633
Jones, C., Harvey, A. G., & Brewin, C. R. (2006). The organisation
and content of trauma memories in survivors of road traffic accidents.
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.

Organization of Trauma Memories and Symptoms 725
Behaviour Research and Therapy, 45, 151–162. doi:10.1016/j.brat.2006.02
.004
Kindt, M., & van den Hout, M. (2003). Dissociation and memory fragmenta-
tion: Experimental effects on meta-memory but not on actual memory per-
formance. Behaviour Research and Therapy, 41, 167–178. doi:10.1016/S0005-
7967(01)00135-8
Kleim, B., Wallott, F., & Ehlers, A. (2008). Are trauma memories disjointed
from other autobiographical memories in posttraumatic stress disorder? An
experimental investigation. Behavioural and Cognitive Psychotherapy, 36, 221–
234. doi:10.1017/S1352465807004080
Krans, J., N¨
aring, G., Holmes, E. A., & Becker E. S. (2009). Can a memory
test reduce analogue traumatic intrusions? Behaviour Research and Therapy, 47,
426–430. doi:doi:10.1016/j.brat.2009.01.009
Landauer, T. K., Foltz, P. W., & Laham, D. (1998). An introduc-
tion to latent semantic analysis. Discourse Processes, 25, 259–284.
doi:10.1080/01638539809545028
Latent Semantic Analysis @ CU Boulder. (n.d.). Latent semantic analysis. Retrieved
from http://lsa.colorado.edu/
Lovibond, S. H., & Lovibond, P. F. (1993). Manual for the Depression Anxiety Stress
Scales (DASS). Sydney, Australia: Psychology Foundation of Australia.
O’Kearney, R., & Perrott, K. (2006). Trauma narratives in posttraumatic stress
disorder: A review. Journal of Traumatic Stress, 19, 81–93. doi:10.1002/jts
.20099
O’Kearney, R., Speyer, J., & Kenardy, J. (2006). Children’s narrative memory for
accidents and their post-traumatic distress. Applied Cognitive Psychology, 21,
821–838. doi:10.1002/acp.1294
Peterson, C., & McCabe, A. (1983). Developmental psycholinguistics: Three ways of
looking at a narrative.NewYork,NY:Plenum.
Radvansky, G. A., Copeland, D. E., & Zwaan, R. A. (2005). A novel study: In-
vestigating the structure of narrative and autobiographical memories. Memory,
13, 796–814. doi:10.1080/09658210444000412
Robinson, J. A., & Taylor, L. R. (1998). Autobiographical memory
and self-narratives: A tale of two stories. In Autobiographical mem-
ory: Theoretical and applied perspectives (pp. 125–143). Mahwah, NJ:
Erlbaum.
Rubin, D. C. (2011). The coherence of memories for trauma: Evidence from
posttraumatic stress disorder. Consciousness and Cognition, 20, 857–865.
doi:10.1016/j.concog.2010.03.018
Wechsler, D. (1997). Wechsler Adult Intelligence Scale (3rd ed.). San Antonio,
TX: The Psychological Corporation.
Journal of Traumatic Stress DOI 10.1002/jts. Published on behalf of the International Society for Traumatic Stress Studies.