attachment anxiety, verbal immediacy, and blood pressure: results

Personal Relationships, 18 (2011), 285–301. Printed in the United States of America.Copyright © 2011 IARR; DOI: 10.1111/j.1475-6811.2011.01360.x

Attachment anxiety, verbal immediacy, and bloodpressure: Results from a laboratory analog studyfollowing marital separation

LAUREN A. LEE, DAVID A. SBARRA, ASHLEY E. MASON, AND RITA W. LAW

University of Arizona

AbstractMarital separation and divorce increase risk for all-cause morbidity and mortality. Using a laboratory analogparadigm, this study examined attachment anxiety, language use, and blood pressure (BP) reactivity among 119(n = 43 men, 76 women) recently separated adults who were asked to mentally reflect on their relationship historyand separation experience. A language use composite of verbal immediacy from participants’ stream-of-consciousnessrecordings about their separation experience as a behavioral index of attachment-related hyperactivation was created.Verbal immediacy moderated the association between attachment anxiety and BP at the beginning of adivorce-specific activation task. Participants reporting high attachment anxiety who discussed their separation in afirst-person, present-oriented, and highly engaged manner evidenced the highest levels of BP at the start of thedivorce-specific task. Results provide a deeper understanding of the association between marital dissolution andhealth and suggest that verbal immediacy may be a useful behavioral index of hyperactivating coping strategies.

The end of a marriage is associated with avariety of positive and negative outcomes.Although most people are resilient in theface of divorce (e.g., Hetherington & Kelly,2002), separation experiences are consis-tently rated among life’s most stressful events(Holmes & Rhae, 1967) and are associatedwith increased risk for a range of nega-tive emotional and physical health outcomes(Ben-Shlomo, Smith, & Shipley, 1993;

Lauren A. Lee, David A. Sbarra, Ashley E. Mason, andRita W. Law, Department of Psychology, University ofArizona.

This research was supported by grants from theNational Institute of Aging (AG 028454), the NationalInstitute of Mental Health (MH 074637), and the NationalScience Foundation (BCS 0919525) to D.A.S., and a F31National Research Service Award (HD059396) to L.A.L.The authors are grateful to Lindsey Bupp for her day-to-day management of the marital transitions study andto David Lozano for technical assistance with the bloodpressure calibration and data collection.

Correspondence should be addressed to Lauren A.Lee, Department of Psychology, University of Arizona,1503 East University Blvd., Rm 313, Tucson, AZ 85721-0068, e-mail: [email protected].

Kiecolt-Glaser et al., 1987; Lucas, 2005;Tucker, Friedman, Wingard, & Schwartz,1996). What is not yet known in greatdetail is why and how some people navi-gate divorce with minimal or transient dis-tress, whereas other people become mired inperiods of considerable emotional pain andstuck on trajectories toward poor health out-comes. Attachment theory (Bowlby, 1969,1973, 1980; Mikulincer & Shaver, 2007) pro-vides a useful vantage point for investigatingindividual differences in emotional respond-ing following relationship separations. Wealso believe that the theory is useful for devel-oping a more refined mechanistic account ofthe variables that connect social separationsto important health outcomes. The primarygoal of this investigation is to operationalizea potential behavioral index of attachment-related hyperactivation (see Mikulincer &Shaver, 2003, 2005; Shaver & Mikulincer,2002)—the behaviors highly anxious adultsuse to cope with real or perceived attachment

285

286 L. A. Lee et al.

threats—and to determine if this responsepattern is associated with increased bloodpressure (BP) when adults think about theirrecent separation experience.

Adult Attachment Anxiety, Divorce, andEmotion Regulation

One of the most robust and well-replicatedfindings in the literature on social separationsis that individual differences in attachmentstyles, which are presumed to be relativelystable person variables (Fraley, 2002), arehighly associated with divorce adjustment(Bakermans-Kranenburg & van IJzendoorn,1997; Berman, 1988; Birnbaum, Orr, Mikulin-cer, & Florian, 1997) and moderate theways in which people respond to nonmari-tal breakups (Davis, Shaver, & Vernon, 2003;Sbarra & Emery, 2005). Attachment stylesreflect how people view themselves and oth-ers in close relationships and play a criticalrole in regulating the experience of felt secu-rity (Mikulincer & Shaver, 2007; Sroufe &Waters, 1977). In the face of real or per-ceived threats to felt security, when the pri-mary strategy of attachment figure proximityseeking is not a viable option, people highin anxiety and avoidance engage in differ-ent secondary strategies to regulate distress.Attachment anxiety is associated with hyper-activating strategies, which are defined byrepetitive efforts to feel close to, or reunitewith, the attachment figure that render the sys-tem chronically activated. In contrast, highlyavoidant individuals tend to engage in deac-tivating strategies by dropping all attemptsto restore security and down-regulating theattachment system to minimize their distress(for a detailed account, see Mikulincer &Shaver, 2007).

Hyperactivating strategies, in particular,have clear, health-relevant biological corre-lates (Diamond, 2001), and, for this reason,are the focus of the current investigation. Forexample, Roisman, Tsai, and Chiang (2004)hypothesized that specific profiles of sym-pathetic nervous system responding wouldmap onto these individual difference dimen-sions during the Adult Attachment Interviewand found that hyperactivating strategies were

associated with increased heart rate reactivity(Roisman, 2007; Roisman et al., 2004). Dur-ing a conflict discussion with a romanticpartner, anxiously attached adults evidencedan increase in both diastolic blood pres-sure (DBP) and rate-pressure product (acomposite measure of pulse rate and sys-tolic blood pressure [SBP]; Kim, 2006).These findings are consistent with the the-ory that high attachment anxiety can pro-duce heightened physiological responding instressful situations (Feeney & Kirkpatrick,1996).

What behaviors define these hyperactivat-ing coping strategies? A specific and testablehypothesis derived from attachment theoryis that heightened physiological responses(upon an attachment threat) will be observedamong people higher in attachment anxietywhen they engage in hyperactivating copingstrategies. As described below, we believethat studying the specific words that peopleuse to describe their experiences can pro-vide a rich behavioral index of variabilityin attachment-related hyperactivation. In turn,these behaviors may illuminate the psychoso-cial context in which anxiety is related tohealth-relevant biological responses followingmarital separation.

Language Use and Emotion Regulation

The last decade has witnessed a surge ofresearch investigating the role of naturallanguage use as a behavioral indicator ofemotion regulatory strategies (Tausczik &Pennebaker, 2010). Language plays a pow-erful role in communicating psychologicalstates (Pennebaker & Graybeal, 2001) anddirecting attention to different aspects ofthe environment (Stapel & Semin, 2007).The study of language use reveals newinformation about how individuals organizeand experience life events, especially inthe context of social relationships (Chung& Pennebaker, 2007; Graham et al., 2009;Pressman & Cohen, 2007; Rohrbaugh, Mehl,Shoham, Reilly, & Ewy, 2008; Slatcher,Vazire, & Pennebaker, 2008). In addition,the words people use to describe their expe-riences can reflect individual differences in

Anxiety, immediacy, and blood pressure 287

response to stressful life events (Cohn, Mehl,& Pennebaker, 2004; Pennebaker, Mayne,& Francis, 1997). Specifically, psychologicaldistancing, a composite category that reflectsless personal and more detached language, hasemerged as one such variable (Cohn et al.,2004). Psychological distancing is implicatedin adaptive self-reflection and directed think-ing about a negative event from a third-person perspective, which are associated withless emotional and physiological reactivityin the short term and mitigates ruminationin the long term (Ayduk & Kross, 2010;Kross, 2009). A recent language use studydemonstrated that people used psychologicaldistancing to cope with the terrorist attacksof September 11, 2001. Cohn, Mehl, andPennebaker (2004) interpreted these data assuggesting that the 9/11 attacks may have cat-apulted people in to a state of shock thatnecessitated cognitive detachment from theevent to continue with daily life. Decreasesin psychological distancing language wereobserved as time passed following the attacks(Cohn et al., 2004).

The inverse of psychological distance isverbal immediacy (Biber, 1988). The immedi-acy/distance construct is a factor-analyticallyderived composite consisting of the stan-dardized language categories of first-personsingular pronouns (I, me, my), discrepancywords (e.g., should, would ), present tenseverbs and inverse scores for articles (a, the),and words of more than six letters that werevalidated on an adult sample (Mehrabian,1967; Pennebaker & King, 1999). Immedi-ate language is associated with a focus onthe here and now, reflected by greater useof concrete, personal, involved, and expe-riential language. Immediate language alsois positively correlated with neuroticism andnegatively correlated with openness, suggest-ing that the language construct is associatedwith other valid individual differences (Pen-nebaker & King, 1999). Borelli, Sbarra, Mehl,and David (in press) recently used transcriptsfrom child attachment interviews and foundthat attachment group membership is asso-ciated with the degree of experiential con-nectedness (assessed by verbal immediacy)

that children express when describing dif-ficult relationship experiences. Highly anx-ious children used more immediate languagein their attachment interviews relative tosecure and avoidant children. The authorssuggested that different levels of experientialconnectedness are associated with specificchild attachment styles. This behavioral find-ing provides important evidence for constructvalidity of the hyperactivating emotion-reg-ulation strategy used by anxiously attachedchildren. Furthermore, this finding suggeststhat at least one dimension of the hyperacti-vation construct may be overinvolvement withone’s own thoughts and feelings and that thisprocess may be indexed behaviorally throughverbally immediate language.

The Current Study

Attachment anxiety is associated with hyper-activating emotion-regulation strategies andhas clear implications for how individualscope with relationship disruptions. Little isknown, however, about the specific behaviorsthat index hyperactivation. Using a behavioralindex of hyperactivation may help illuminatethe extent to which anxiety is associated withphysiological responding following the end ofmarriage. In the case of divorce, we hypoth-esize that the magnitude of the anxiety traiteffect on BP responses will depend on thedegree to which adults evidence a high degreeof the hyperactivation state, which we oper-ationalize using verbal immediacy. In partic-ular, we expect that the greatest increases inBP responses during a divorce-specific men-tal activation task (DMAT) will be observedamong highly anxious adults who talk abouttheir separation experience in an immediate,present-oriented, and experiential way duringa stream-of-consciousness (SOC) task abouttheir separation. The term hyperactivationsuggests an affective amplification process(see Sbarra & Ferrer, 2006), and we thereforeexpect a three-way interaction whereby peo-ple who report greater attachment anxiety anduse more immediate language will evidencethe greatest increases in BP when asked tothink about their relationship history and sep-aration experience.


Method

Participants

Participants were 119 community-dwellingadults (n = 43 men, 76 women) recruitedthrough newspaper advertisements, divorcerecovery groups, and the local family andconciliation court. On average, participantswere 41.5 years old (SD = 9.75, range =19–63), reported having been married for14.33 years before the separation (SD = 8.5,range = 0.42–42.5), and having physicallyseparated from their former partner an aver-age of 3 months and 3 weeks before thestudy session (SD = 2 months, range = 2–46weeks). Twenty-two percent of the samplewere legally divorced, 38% were legally sep-arated, and 40% were physically separatedwithout any legal action (the remainder ofthe sample did not describe their separa-tion status). Seventy-nine percent of sam-ple described themselves as White (non-Hispanic), 13% Hispanic, 2% African Amer-ican, 1% Native American, 1% Asian Amer-ican, and 4% Other (the remainder of thesample chose not to provide ethnicity data).Participants reported spending an average of27% of each day (in the 2 weeks beforestudy intake) thinking about their ex-partnerand the demise of their marriage (SD = 26%,range = 0%–95%). Data for this study werecollected over a period of 40 months between2006 and 2010, and all aspects of this studywere approved by the University of ArizonaHuman Subjects Protection Program.

Procedures

Adults who responded to the study adver-tisements were screened along several dimen-sions: All participants who reported that theywere generally healthy, without a history ofa psychotic disorder, and, for women, thosenot pregnant, were deemed eligible to partic-ipate. Participants were told that the purposeof the study was to understand “how adultsadjust to marital separation and the ways inwhich your body responds when you thinkabout and reflect on your separation experi-ence.” Participants were asked to refrain fromusing tobacco and caffeine for at least 4 hr

before the study visit. At the laboratory visitand prior to the physiological measurementsection of the study, participants completeddemographic, health, and relationship disso-lution questionnaires in addition to an SOCspeech task in which they spoke about theirseparation experience. For the SOC task, par-ticipants were seated alone in a room witha digital voice recorder device and instructedto “describe any thoughts and feelings” aboutthe relationship or their former partner. Par-ticipants were asked to speak continuously fora 4-min period. Participants’ SOC narrativeswere transcribed and analyzed to derive verbalimmediacy scores (see below).

Following the SOC, a research assistantattached a BP device to a participant’s wrist.Participants were then asked to sit quietly andto mentally reflect (silently) on a series ofmundane scenarios presented on a computermonitor. This mundane events recall (MER)task served as a baseline period, and mundanescenarios included thinking about preparinga dinner, doing laundry, going to the gro-cery store, and mailing a letter. The MERtask was intended to capture physiologicaldemands associated with nonemotional men-tal reflection, and this period constituted thebaseline assessment.

Following the MER, participants com-pleted the 7-min DMAT. Participants wereinstructed to “spend some time thinking aboutyourself and your partner in a variety of dif-ferent situations.” A series of seven questionswas then presented on a computer screen andparticipants were asked to “concentrate on thequestion by letting any relevant thoughts, feel-ings, or images come to mind” for a 1-minperiod; after 1 min, the next question waspresented. The DMAT items are reported indetail elsewhere (Sbarra, Law, Lee, & Mason,2009), but, in general, this task asked partic-ipants to reflect on their relationship history,the first time they thought about or learnedabout the separation, and difficult momentsduring the separation. In the DMAT, partici-pants did not respond to the questions aloudbut mentally reflected on their answers tothe questions during each 1-min period. Ingeneral, participants reported a high degreeof engagement during the DMAT and found


the task to be representative (although notidentical) to how they think about the separa-tion experience outside of the laboratory (seeSbarra et al., 2009).

Following the DMAT and task engagementappraisal items, participants were asked tosit quietly for 4 min as we continued tomeasure their BP: This period constituted theDMAT recovery (DMATR) period. After theDMATR, a research assistant removed the BPequipment from the participants. Participantswere then debriefed about the overall natureof the study and paid $100 for the laboratoryvisit.

Psychological measures

Experiences in CloseRelationships–Revised (ECR–R)

The ECR–R (Fraley, Waller, & Brennan,2000) is a reliable, widely used 36-item ques-tionnaire that measures attachment-relatedanxiety in close relationships (see also Butzer& Campbell, 2008; Campa, Hazan, & Wolfe,2009; Domingue & Mollen, 2009; Fraley,Niedenthal, Marks, Brumbaugh, & Vicary,2006). The anxiety scale captures thoughtsand feelings related to approval and respon-siveness from partners, as well as worry aboutromantic relationships (e.g., “I get frustratedif romantic partners are not available whenI need them” and “I worry that romantic part-ners won’t care about me as much as I careabout them”). Reliability in our sample forthe anxiety scale of the measure was high(α = .93).

Linguistic Inquiry and Word Count(LIWC) program

The LIWC system (Pennebaker & Francis,1999) is a computer-based text analysis pro-gram that counts words in a given text file andreports the percentages of word use across 74psychological and grammatical categories. AllSOC recordings were transcribed and format-ted for LIWC analysis by a research assistant,then independently checked for accuracy byanother research assistant. The LIWC pro-gram matches target words in a given speechsample to words in its dictionary; the resulting

variables reflect the percentage of total wordsin the speech sample. The present analysisfocused on the verbal immediacy compositescore, which includes standardized languagecategories of first-person singular pronouns(I, me, my), discrepancy words (e.g., should,would ), present tense verbs and inverse scoresfor articles (a, the), and words composedof more than six letters (see Cohn et al.,2004; Pennebaker & King, 1999). Higherimmediacy scores reflect a focus on the hereand now by using more concrete, personal,involved, and experiential language, whereaslower immediacy scores are characterized byhaving a more detached and impersonal tone.The reliability of the items comprising thisscale was acceptable (α = .71).

BP assessment

BP was assessed using a noninvasive (rela-tive to a standard oscillometric cuff) tonom-etry device over the radial artery to providefrequent, real-time updates of SBP and DBP(Vasotrac AMP205; Medwave Inc., ArdenHills, MN). SBP is the peak pressure in thearteries at the beginning of the cardiac cycle,whereas DBP is the lowest pressure at theresting phase of the cycle. The Vasotrac usesfrequent compression and decompression ofthe radial artery at the wrist to detect thezero-load state around which the pressure sig-nals are measured. This information is usedto detect and then display arterial pressureand wave form every 12 to 15 beats. TheVasotrac was calibrated against radial cathetermeasures of BP and demonstrated excellentconvergent validity (mean R2 for SBP andDBP = .95; Belani et al., 1999). The tonom-etry device was placed over the radial arteryon the participants’ nondominant arm, andthe participants placed their arm on a tablein front of them throughout the duration ofthe study. BP data were scored using Mind-ware Technology’s (Westerville, OH) BP 2.6postprocessing software module. Minute-by-minute mean values were computed for SBPand DBP across the DMAT and DMATR. Inthe DMAT, each minute of BP data includesthe amount of time the participant spent read-ing each question as well as the time spent


reflecting on each answer. In this sample, 87%(n = 104) of the participants had complete BPdata across the 7-min DMAT.

Covariates

To account for possible differences in demo-graphic and health status variables on BP atstudy entry several covariates were includedin these analyses: participants’ age, grossannual income, body mass index (BMI; ascalculated by weight in kg/height in meters),and self-reported information on participants’(a) history of physician-diagnosed high BP(yes–no) and (b) current tobacco use (yes–no). In addition, two relationship-specificvariables were included in the analyses: thelength of the relationship before physical sep-aration (described above in the Participantssection), and the amount of time since theseparation at entry into study (rounded to thenearest half-month).

Data analysis

The main study hypothesis was examinedusing multilevel modeling in SAS PROCMIXED (Singer, 1998). This approach takesinto account the nonindependent (observa-tions are nested under individual participants)nature of SBP and DBP across the 7 minof the DMAT and 4 min of the DMATRand allows us to examine BP changes overtime during the task. We used a model build-ing strategy (Singer & Willett, 2003) acrossthe analyses and began by examining anempty (unconditional means) model that con-tained only the covariates of interest. Next, wesought to characterize the functional form ofSBP and DBP across the 7 min of the DMATand 4 min of the DMATR (analyzing eachtask separately) by testing linear and quadraticmodels to represent changes over time. Oncewe characterized the temporal changes inBP, we added attachment anxiety and ver-bal immediacy as main effects in the model,in addition to their two-way interactions withtime. The final model, which includes anexplicit test of the main study hypothesis,examines the three-way interaction of timeAttachment Anxiety × Verbal Immediacy.

Results

Figure 1 illustrates the patterns of SBP andDBP across all 15 min of the study. Table 1displays the bivariate correlations (and de-scriptive statistics) for the predictors andcovariates of interest. The association betweenattachment anxiety and verbal immediacy waspositive and statistically significant (p < .01),indicating that participants reporting moreanxious attachment also spoke about their sep-aration experience in a present-oriented andexperiential manner during the SOC recordingsession.

To examine the possibility that attach-ment anxiety and verbal immediacy wereassociated with higher BP at entry into thestudy, we examined a single multilevel modelthat included the covariates of interest, themain effects of attachment anxiety and ver-bal immediacy, as well as their interactionterm on average BP responses during theMER. There were no significant main effectdifferences on BP for attachment anxiety,verbal immediacy, or the interaction term.

Figure 1. Systolic and diastolic blood pres-sure (BP) scores across the three study periods.Note. MER = mundane events recall task(4 min); DMAT = divorce-specific mentalactivation task (7 min); DMATR = mentalactivation task recovery (4 min).


Tab

le1.

Dem

ogra

phic

vari

able

s,re

lati

onsh

ip-s

peci

ficco

vari

ates

,an

xiou

sat

tach

men

t,an

dve

rbal

imm

edia

cy:

Cor

rela

tion

san

dde

scri

ptiv

est

atis

tics

(N=

119)

Var

iabl

es1

23

45

67

89

1.A

ge—

2.G

ross

inco

me

0.11

∗∗—

3.L

engt

h0.

60∗∗

0.20

∗∗—

4.Se

para

tion

−0.1

0∗∗0.

10∗∗

−0.0

7∗—

5.To

bacc

o0.

18∗∗

0.14

∗∗0.

20∗∗

0.03

—6.

Hig

hB

P0.

32∗∗

0.02

0.17

∗∗0.

03−0

.02

—7.

BM

I−0

.01

0.04

0.06

0.03

−0.0

40.

24∗∗

—8.

Anx

iety

−0.0

4−0

.04

−0.0

10.

10∗

−0.0

40.

19∗∗

0.24

∗∗—

9.Im

med

iacy

−0.1

0∗∗

−0.2

1∗∗

−0.0

10.

00−0

.11∗

∗0.

060.

19∗∗

0.30

∗∗M

41.5

01.

7317

2.25

3.88

1.27

1.14

25.6

13.

480

SD9.

600.

8410

1.73

1.90

0.45

0.35

5.33

1.30

0.70

Ran

ge19

–63

1–

45

–51

00.

5–

11.5

01

–2

1–

217

–47

1–

6.11

−1.5

0–

2.32

Not

e.G

ross

inco

me

=ap

prox

imat

egr

oss

inco

me

(1=

less

than

$30,

000,

2=

$30,

000

–$5

0,00

0,3

=$5

0,00

0–

$100

,000

,4=

grea

ter

than

$100

,000

);L

engt

h=

leng

thof

time

inm

onth

spa

rtic

ipan

tsw

ere

mar

ried

befo

reth

ese

para

tion;

Sepa

ratio

n=

leng

thof

time

roun

ded

toth

ene

ares

tha

lf-m

onth

asth

epa

rtic

ipan

tph

ysic

ally

sepa

rate

dfr

ompa

rtne

r;To

bacc

o=

Doe

sth

epa

rtic

ipan

tus

ean

yfo

rmof

toba

cco

incl

udin

gsm

okin

gci

gare

ttes,

pipe

sor

ciga

rs,

and

chew

ing

toba

cco

(1=

no,

2=

yes)

;H

igh

BP

=pa

rtic

ipan

tm

edic

aldi

agno

sis

ofhy

pert

ensi

on(1

=no

,2

=ye

s);

BM

I=

body

mas

sin

dex;

Anx

iety

=av

erag

eat

tach

men

tan

xiet

yite

m-s

core

;V

erba

lIm

med

iacy

=co

mpo

site

scor

eof

stan

dard

ized

LIW

Cva

riab

les

I,di

scre

panc

y,pr

esen

tte

nse

verb

s,an

dth

ein

vers

esc

ore

ofar

ticle

and

six-

lette

rw

ords

.∗ p

<.0

5.∗∗

p<

.01.


Changes in BP across the DMAT as afunction of anxiety and immediacy

The first series of multilevel models exam-ined our main hypothesis that the interac-tion of attachment anxiety and verbal imme-diacy would alter the course of BP acrosseach minute of the DMAT after accountingfor the covariates of interest. After estab-lishing a baseline model (Table 2, Model 1),we tested a series of alternative models todetermine if participants demonstrated signif-icant change over the DMAT period. Noneof the three specifications for time describedabove improved model fit over-and-above thebaseline model. On average, then, partici-pants did not exhibit significant (i.e., system-atic) increases or decreases in BP across theDMAT.

The next series of models examined bothattachment anxiety and verbal immediacyas main effects on BP (Table 2, Model 2).Attachment anxiety had significant maineffects on both the SBP (B = 2.43, SE =0.92, p = .01) and DBP (B = 2.20, SE =0.70, p = .002), whereas the verbal immedi-acy main effects were not significant. Afteraccounting for immediacy, participants report-ing a high degree of attachment anxiety evi-denced higher SBP and DBP at the beginningof the DMAT.

Model 3 (Table 2) examined the two-wayinteractions of (a) Time × Attachment Anx-iety and (b) Time × Verbal Immediacy toinvestigate the potential moderating effectsof anxiety or immediacy on SBP and DBPchanges across the DMAT. As shown in thetable, neither of these interactions was sig-nificant. A final time-based model (Table 2,Model 4) examined the three-way interac-tion of Time × Attachment Anxiety × Ver-bal Immediacy. As shown in the table, thethree-way interaction of attachment anxietyand verbal immediacy across each minute ofthe DMAT was not significant. Thus, we didnot find support for the main hypothesis ofthe study.

Given that the two-way Anxiety × Imme-diacy effect was observed for SBP and DBP inModel 4, we conducted a series of exploratoryanalyses to formally examine the possibility

that these variables might interact to predictBP at the start of the DMAT (rather thanaltering the course of BP responses acrossthe DMAT). As shown in Model 5, afteraccounting for both main effects and all othercovariates, the Attachment Anxiety × Ver-bal Immediacy interaction was significant forboth SBP (B = 3.38, SE = 1.26, p = .01)and DBP (B = 2.36, SE = 0.96, p = .01).Using a recently developed computationaltool to decompose simple slopes in multi-level models (Preacher, Curran, & Bauer,2006), we found that the effects of anxi-ety on BP responses differed as a functionof verbal immediacy (SBP: z = 4.43, p <

.001; DBP: z = 4.51, p < .001). The inter-action, displayed in Figure 2, operates at theBP intercept during the DMAT and indicatesthose participants who spoke in a more expe-rientially engaged manner during the SOC(greater verbal immediacy) evidenced signif-icantly higher BP at the beginning of theDMAT task only if they also reported hightrait attachment anxiety. Finally, we recen-tered the time parameter and treated thefinal minute of the DMAT as the intercept.We found that the Anxiety × Immediacyinteraction persisted across the entire 7-minDMAT, B = 3.65, SE = 1.28, p = .005, forSBP and for DBP, B = 2.42, SE = 0.97, p =.01. Thus, adults higher in attachment anxietyand verbal immediacy ended the DMAT taskwith significantly higher SBP and DBP.

The analysis described above indicates thatelevations in BP among highly anxious peo-ple who speak about their separation in animmediate way become apparent when thesepeople are asked to reflect on their separationin the DMAT. In the health psychology liter-ature, it is also common to address individualdifferences in reactivity. Therefore, to ascer-tain if people high in anxiety and immediacyshow a larger increase in BP from the MER tothe DMAT, we entered mean MER scores forSBP and DBP into the respective final models.Once we accounted for mean MER scores,the previously significant two-way interaction(in Model 5) was nonsignificant. We exploredwhether the lack of reactivity effects was dueto statistical (rather than substantive) issues byaltering the proposed models to account for


Tab

le2.

Uns

tand

ardi

zed

para

met

eres

tim

ates

from

four

mul

tile

velm

odel

sas

sess

ing

chan

gein

syst

olic

(S)

and

dias

toli

c(D

)bl

ood

pres

sure

(BP

)du

ring

the

divo

rce-

spec

ific

men

tal

acti

vati

onta

sk(D

MA

T;

N=

119)

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Mod

el5

Var

iabl

eB

SEp

BSE

pB

SEp

BSE

pB

SEp

SBP In

terc

ept

141.

911.

90.0

001

142.

191.

30.0

001

141.

071.

34.0

001

141.

071.

34.0

001

141.

071.

34.0

001

Age

0.40

0.22

.07

0.52

0.15

.001

0.51

0.15

.001

0.51

0.15

.001

0.51

0.15

.001

Hig

hB

P6.

103.

55.0

95.

483.

49.1

25.

483.

49.1

25.

483.

49.1

2A

nxie

ty—

——

2.43

0.92

.01

2.48

0.91

.01

2.48

0.91

.007

2.48

0.91

.01

Imm

edia

cy—

——

0.87

1.73

.61

−0.0

11.

73.9

9−0

.01

1.73

.99

−0.0

11.

73.9

9A

nxie

ty×

Imm

edia

cy—

——

——

—3.

381.

26.0

13.

381.

26.0

083.

381.

26.0

1T

ime

×A

nxie

ty—

——

——

—−0

.06

0.88

.95

−0.0

60.

88.9

5—

——

Tim

e×

Imm

edia

cy—

——

——

—0.

011.

58.9

9−0

.004

1.62

.99

——

—T

ime

×A

nxie

ty×

Imm

edia

cy—

——

——

——

——

0.04

1.21

.98

——

—D

BP Inte

rcep

t81

.08

1.45

.000

181

.16

0.98

.000

180

.37

1.02

.000

180

.37

1.02

.000

180

.37

1.02

.000

1A

ge0.

190.

17.2

60.

240.

12.0

40.

240.

11.0

40.

240.

11.0

40.

240.

11.0

4H

igh

BP

6.15

3.87

.12

6.53

2.70

.02

6.09

2.66

.02

6.09

2.66

.02

6.09

2.66

.02

Anx

iety

——

—2.

200.

70.0

022.

230.

69.0

012.

240.

69.0

012.

230.

69.0

01Im

med

iacy

——

—−0

.14

1.31

.92

−0.7

61.

31.5

6−0

.76

1.31

.56

−0.7

61.

31.5

6A

nxie

ty×

Imm

edia

cy—

——

——

—2.

360.

96.0

12.

360.

96.0

12.

360.

96.0

1T

ime

×A

nxie

ty—

——

——

—−0

.10

0.66

.88

−0.1

00.

66.8

8T

ime

×Im

med

iacy

——

——

——

0.08

1.19

.95

0.09

1.21

.94

——

—T

ime

×A

nxie

ty×

Imm

edia

cy—

——

——

——

——

−0.0

30.

91.9

7—

——

Not

e.C

ovar

iate

sno

tm

eetin

ga

min

imum

p<

.20

cuto

ffha

vebe

enre

mov

edfo

rcl

arity

.Hig

hB

P=

part

icip

ant

med

ical

diag

nosi

sof

hype

rten

sion

(1=

no,2

=ye

s);

Tim

e=

each

min

ute

ofth

eD

MA

Tw

hich

acco

unts

for

alin

ear

chan

gein

BP

acro

ssth

e7-

min

task

;A

nxie

ty=

aver

age

atta

chm

ent

anxi

ety

item

-sco

re;

Imm

edia

cy=

com

posi

tesc

ore

ofst

anda

rdiz

edL

IWC

vari

able

sI,

disc

repa

ncy,

pres

ent

tens

eve

rbs,

and

the

inve

rse

scor

eof

artic

lean

dsi

x-le

tter

wor

ds.


A B

Figure 2. Simple slopes depicting the two-way interaction between attachment anxiety andverbal immediacy predicting blood pressure (BP; systolic BP and diastolic BP, Panels A andB, respectively) during the divorce-specific mental activation task (DMAT).Note. Only high immediacy individuals evidenced significant simple slopes. High attachmentanxiety individuals with high verbal immediacy (1 SD above the verbal immediacy mean)scores evidenced significantly higher BP during each minute of the DMAT. All variables inthe model are continuous; for graphical purposes, changes are plotted for participants 1 SDabove and below the mean on attachment anxiety and verbal immediacy.

shared variance among covariates and focalpredictors that may suppress the Anxiety ×Immediacy effects. The first model includedthe significant covariates from the originalanalysis (i.e., age and history of high BP) andmean MER BP scores, along with anxiety,immediacy, and their interaction. Consistentwith the results from Model 5, the attachmentanxiety and verbal immediacy interaction wassignificant for both SBP (B = 2.55, SE =0.74, p = .001) and DBP (B = 1.43, SE =0.53, p = .01), and an identical pattern ofsignificant and nonsignificant effects emergedas described for the simple slopes displayedin Figure 2. We then tested a more completemodel by including all covariates reported inModel 5 (with the exception of BMI), meanMER BP scores, and the predictors of interest.BMI was removed from the analysis due toits significant correlation with both attachment

anxiety (r = .24, p < .001) and verbal imme-diacy (r = .19, p < .05).1 Consistent with theless restrictive model, the Attachment Anxiety

1. Miller and Chapman (2001) assert that standard statis-tical control by inclusion of covariates is implementedto reduce variance associated with the outcome ofinterest and that removal of shared variance betweencovariates and predictors of interest artificially altersthe relationship between the focal predictors and theoutcome. In the case of our analyses, BMI is unas-sociated with either MER or DMAT BP scores butsignificantly associated with both anxiety and immedi-acy. Therefore, controlling for BMI scores in a modelexamining change (whereby more than 65% of thevariance in the outcome in both SBP an DBP dur-ing the DMAT is already accounted for by MER BPscores), alters the nature, the attachment, and immedi-acy variables, which eliminates the interaction effecton BP reactivity. Given that (a) BMI is not associatedwith individual differences in BP at rest during theMER or with changes from the MER to the DMAT,and (b) accounting for BMI alters the nature of attach-ment anxiety and immediacy measures, we believethat there is a reasonable rationale for excluding this


× Verbal Immediacy interaction was signifi-cant for both SBP (B = 2.12, SE = 0.71, p =.003) and DBP (B = 1.29, SE = 0.52, p =.02), and a simple slopes analysis revealedthat the Anxiety × Immediacy interactioneffect was significant only for adults who evi-dence high levels of verbal immediacy duringthe SOC (SBP: z = 4.47, p < .0001; DBP:z = 4.96, p < .0001).

Does the Attachment × Immediacyinteraction on BP persist in time?

To investigate whether the Anxiety × Imme-diacy interaction effect (on initial BP at thebeginning of the DMAT) observed in Model5 would persist after the DMAT, we tested afinal series of models that included the 4-minrecovery period following the DMAT (i.e.,DMATR).2 As shown in Table 3, the Attach-ment Anxiety × Immediacy interaction effectwas significant for both SBP (B = 3.13, SE =1.27, p = .02) and DBP (B = 2.70, SE = .98,p = .01), indicating that BP was higher atthe beginning of the DMATR depending onlevel of trait attachment anxiety and expe-riential engagement during the SOC. Thiseffect did not persist across the entire recoveryperiod. By the end of the DMATR, the attach-ment anxiety and verbal immediacy interac-tion effect was not significant, but the attach-ment anxiety main effect remained significant(SBP: B = 2.94, SE = 0.94, p = .05; DBP:B = 1.60, SE = 0.70, p = .02).

Discussion

This study examined BP responses amongrecently separated/divorced adults during alaboratory task that asked them to thinkabout their relationship history and recent

variable from the reactivity analyses. When BMI isremoved from the analyses reported as part of Model5, the same pattern of significant effects is observed.

2. Prior to testing the Attachment Anxiety × VerbalImmediacy interaction on the starting point of theDMATR, we examined the possibility that individualshigh in attachment anxiety and verbal immediacy mayevidence BP changes across DMATR. Neither thelinear or quadratic interaction effects were significantand we chose to focus our analysis on the AttachmentAnxiety × Verbal Immediacy interaction.

separation experience. Although anxiety andverbal immediacy (our behavioral index ofattachment-related hyperactivation) were pos-itively correlated, we did not find supportfor the hypothesized Anxiety × Immedi-acy interaction when predicting BP changesacross the divorce-specific laboratory task(the DMAT). We did, however, find that adultswho reported greater attachment anxiety evi-denced significantly higher BP at the onsetof DMAT when they also spoke in a morepersonal, present-oriented, and engaged man-ner about their divorce experience earlier inthe laboratory paradigm. The reactivity anal-yses indicated that people high in anxietyand immediacy also evidenced significantlygreater increases from the MER to the DMATonce we removed the nonsignificant effectof BMI from the model. Finally, within therecovery task (the DMATR), we observed thesame interaction effect at the beginning of thetask, but only a significant main effect forattachment–anxiety at the end of the recoveryperiod.

A clear strength of this study is its mul-timethod framework: The significant two-way interaction consists of self-reported andbehaviorally sampled language use data oper-ating together to predict a health-relevantphysiological outcome. This type of method-ological convergence across multiple domainsof assessing emotional experience is rela-tively rare in psychological science (cf. Mauss& Robinson, 2009) and offers potentiallynew ways of thinking about the nature ofattachment-related hyperactivation and theconstruct of immediacy.

The findings provide a conceptual repli-cation of prior work documenting a sig-nificant positive association between anxi-ety and heightened cardiovascular respond-ing (Feeney & Kirkpartick, 1996; Kim, 2006;Roisman, 2007; Roisman et al., 2004). Inthe nonemotion, nondivorce baseline task (theMER), participants high in anxiety, immedi-acy, or both anxiety and immediacy did notevidence significantly elevated BP (despitehaving already spoken about their separationduring the SOC task), yet at the start of theDMAT, we observed the significant two-wayinteraction effect after accounting for a wide


Table 3. Unstandardized parameter estimates from the multilevel model assessing level ofsystolic (S) and diastolic (D) blood pressure (BP) during the divorce-specific mental activationtask recovery (DMATR) (N = 119)

SBP DBP

Variable B SE p B SE p

Intercept 138.35 1.37 .0001 78.98 1.01 .0001Age 0.37 0.15 .02 0.12 0.12 .32High BP 11.62 3.60 .001 9.12 2.61 .001Linear effect of time 1.17 1.15 .31 0.99 0.87 .26Quadratic effect of time −0.36 0.17 .04 −0.28 0.68 .06Anxiety 2.76 0.93 .004 2.06 0.68 .003Immediacy −0.88 1.75 .62 −1.77 1.30 .18Anxiety × Immediacy 3.13 1.27 .02 2.70 0.98 .01

Note. Covariates not meeting a minimum p < .20 cutoff have been removed for clarity. High BP = participant medicaldiagnosis of hypertension (1 = no; 2 = yes); Linear time = each minute of the DMATR which accounts for a linearchange in BP across the 4-min task; Quadratic time = linear time of DMATR squared to reflect increase and decreasein BP across the 4-min task; Anxiety = average attachment anxiety item-score; Immediacy = composite score ofstandardized LIWC variables I, discrepancy, present tense verbs, and the inverse score of article and six-letter words.

range of health-relevant and relationship-specific variables. One way of interpretingthis finding is in terms of the differential taskdemands; it is not until we asked partici-pants to think about their relationship historyand separation experience that we observeddifferences in BP. In this respect, we viewthe DMAT as revealing the BP correlates ofattachment-related processes. This perspectiveis consistent with the Coan, Allen, and McK-night (2006) capability model of physiologi-cal responding, which posits that responses toemotionally salient stimuli are the result of theinteraction between an individual trait and theevoked emotional state. The capability theoryholds that physiological responses associatedwith trait-level propensities are best evokedusing state manipulations designed to assessthe emotional system in question. In this anal-ysis, we see the BP correlates of high anx-iety/high immediacy more completely whenparticipants are asked to think about theirrecent separation experience. We presume,and future research can test, that the DMATis especially difficult for high anxiety peo-ple who also engage in hyperactivating copingstrategies.

The capability model does not require thatthe DMAT itself reveals the trait effect, and

it is plausible that merely thinking about thetask was enough to prime these responsesfor the high anxiety/high immediacy people.The idea that an anticipation effect is oper-ating in the current study is consistent withother research in health psychology demon-strating that preparing to do a stressful lab-oratory task can elicit a stronger cardiovas-cular response than the task itself (Feldman,Cohen, Hamrick, & Lepore, 2004; Gendolla,1999; Wright, 1996; Wright, Martin, & Bland,2003).

To the extent that the immediacy con-struct can be used as a behavioral index ofattachment-related hyperactivation, elevatedBP at the start of the DMAT was observedonly for those people who have a trait-liketendency toward hyperactivation as a meansof coping with attachment threat (i.e., high-anxiety participants) and who exhibit state-specific indicators of a high degree of experi-ential engagement. This finding highlights theneed to take care in equating anxiety withhyperactivation; not all anxious individualsbecome highly engaged (or, overinvolved) intheir experiences, and, in this study, for theDMAT period, it was the synergy of the traitand state that revealed the highest elevationsin BP (see Coan et al., 2006).


When relationships dissolve, highly anx-ious adults engage in much more ruminationand brooding about the loss event, as wellas prolonged thoughts about reunion with anex-partner and behavioral attempts to rekin-dle the relationship (Davis et al., 2003; Saf-frey & Ehrenberg, 2007; Sbarra, 2006), andthese affective amplification tendencies arethe behavioral hallmarks of attachment hyper-activation (Mikulincer & Shaver, 2007). Thecurrent findings suggest that more immedi-ate speech, which reflects a greater focus onthe present and uses a more personal andinvolved tone, also is a useful index for cat-aloging ways in behavioral hyperactivationis associated with exaggerated physiologi-cal responding during potential instances ofattachment threats. (We consider the DMATa potential attachment threat because weexpected that it would not evoke a needfor emotion regulatory effort among low-anxiety/low-immediacy participants but thatit would be an emotion regulatory chal-lenge for high-anxiety/high-immediacy par-ticipants.) The low-anxiety/high-immediacygroup, without the trait propensity to engagein hyperactivating strategies, did not exhibitthis association even when using highlyimmediate language.

A primary question of interest is the extentto which the current findings are health rel-evant. It is clear that marital separation anddivorce increase risk for health-related mor-bidity and mortality (Ben-Shlomo et al., 1993;Sbarra & Nietert, 2009) and that attach-ment anxiety is associated with worse psy-chological outcomes following these events(Bakermans-Kranenburg & van IJzendoorn,1997; Berman, 1988; Birnbaum, et al., 1997)and with poor physical health outcomes invariety of other contexts (Kotler, Buzwell,Romeo, & Bowland, 1994; Picardi et al.,2005; Rossi et al., 2005). Within the currentstudy, the high-anxiety/high-immediacy par-ticipants evidenced SBP measurements thatwould be considered Stage I hypertensionthroughout both the activation and recoveryperiods, and it is notable that these effectswere observed after statistically accountingfor a variety of covariates or competing pre-dictors. Within Model 5 (Table 2), we see

that participants who score 1 SD above themean on both anxiety and immediacy beginthe DMAT with an almost 5 mmHg increasein SBP and, from Table 3, we see that thesesame participants evidence an almost 6 mmHgincrease in SBP at the start of the DMATR.Slow/incomplete recovery of cardiovascularresponses following stress is associated withnegative physical and mental health outcomes(Waugh, Fredrickson, & Taylor, 2008; Waugh,Panage, Mendes, & Gotlib, 2010), and the BPmetric calibrated in a manner that has clearhealth relevance (e.g., an SBP of 130 mmHgis clearly better than an SBP of 140 mmHg;see Whelton, Chin, Xin, & He, 2002).

It is important to distinguish between theDMAT and DMATR periods; in the latter, themain effect of anxiety was stronger (and itpersisted until the end of the study period)than the Anxiety × Immediacy interaction.One explanation for this difference across theperiods of study may be that the synergisticeffect of both variables initially activates a BPresponse, but that the main effect of anxietyis enough to sustain elevations across timeonce observed. Future research is needed toclarify exactly when anxiety effects dependon the extent of concomitant hyperactivationand when the main effect of the attachmenttrait is sufficient to evoke a physiologicalresponse.

Finally, although the emphasis of this arti-cle is on the health implications of attachmentanxiety (via BP responding), this work is alsothe first to demonstrate, with adults, a signifi-cant positive correlation between anxiety andverbal immediacy, which we have argued canbe used as a behavioral index of attachment-related hyperactivation. One element of ver-bal immediacy that may be most relevantto hyperactivation is self-focused attentionvia first-person singular words. Althoughit may be intuitive to think about hyper-activating strategies as other-focused (e.g.,bids for reunion, proximity and reassuranceseeking), ruminative processes are decidedlyself-focused and have clear negative conse-quences. For example, in the nonmarriage lit-erature, use of I statements in expressive writ-ing tasks typifies postbreakup writings andis correlated to grief and intrusion measures


postbreakup (Boals & Kline, 2005). I lan-guage is also greater in depressed individu-als (Bucci & Freedman, 1981) and is associ-ated with poorer adjustment when first writ-ing about a traumatic event but linked tobetter psychological adjustment following afull expressive writing intervention, whichmay reflect self-efficacy (Dunnack & Park,2009). In the present data, there is no evi-dence that I words alone interact with anxi-ety to yield the same BP results (as observedwhen using the immediacy composite), so anessential question for future research is whataspects of immediacy—self-focused rumina-tion, present-oriented discourse, short andengaged speech—best reflect the hyperacti-vation construct. It is important to note thatverbal immediacy is distinct from negativeemotion and affect. Although Buchheim andMergenthaler (2000) found that highly anx-ious individuals used greater negative emotionwords, negative emotion words and immedi-acy are uncorrelated our sample, and includ-ing negative emotion words in the presentanalyses did not alter the pattern of the results.

Limitiations

The findings reported here should be con-sidered in light of several limitations. First,because the DMAT is a mental-activation task,there is no way to be certain individualsspent the full time during this period think-ing about their relationship. This study repli-cates previous research demonstrating that theDMAT can produce cardiovascular changes(Sbarra et al., 2009) but cannot reveal thespecific thoughts associated with this reactiv-ity. Despite this limitation, most participantsreported that they were highly engaged inthe task and that the images provoked dur-ing the DMAT were vivid and similar to howthey think about their separation outside ofthe lab. Second, we do not have a nondi-vorced comparison group in this study and weare therefore unable to definitively concludethat the effects we found are the results ofthinking about relationship separation specif-ically. We did observe BP increases from theMER to the DMAT and also demonstratedthese reactivity effects in models controlling

for health-relevant covariates; thus, the MERtask serves as a critical within-person con-trol demonstrating that highly anxious adultsdo not show elevated BP at entry into thestudy but, specifically, when they begin tothink about their separation experience. Third,relative to the entire sample, there were fewmen, which create an unbalanced design andraises concern that the observed effects maybe driven by this oversampling of womenin the study. To the extent that women aremore versed at discussing emotionally diffi-cult events (e.g., Newman, Groom, Handel-man, & Pennebaker, 2008), the findings maynot generalize to all adults experiencing amarital separation. Future replications shouldinclude more men.

Conclusion

This study represents one of the first attemptsto investigate language use as a behavioralindex of attachment-related emotion regula-tion. After accounting for a range of health-related demographic and relationship-specificcovariates, participants who reported a highdegree of attachment anxiety and who spokeabout their former relationship/partner in amore personal, present-oriented, and engagedmanner evidenced significantly elevated SBPand DBP at the beginning and end of theDMAT. Acting in combination, high attach-ment anxiety and high immediacy were asso-ciated with a pattern of BP responses that,if maintained over time, have clear nega-tive health outcomes. These works point to apotential route through which specific individ-ual differences can increase risk for prolongedbiological dysregulation in the face of divorce.

References

Ayduk, O., & Kross, E. (2010). Analyzing negativeexperiences without ruminating: The role of self-distancing in enabling adaptive self-reflection. Socialand Personality Psychology Compass, 4, 841–854.doi:10.1111/j.1751-90042010.00301.x

Bakermans-Kranenburg, M., & van IJzendoorn, M. H.(1997). Adult attachment and the break-up of roman-tic relationships. Journal of Divorce & Remarriage,27, 121–139. doi:10.1300/J087v27n03

Belani, K., Ozaki, M., Hynson, J., Hartmann, T., Rey-ford, H., Martino, J. M., . . . Miller, R. (1999). A new


noninvasive method to measure blood pressure. Anes-thesiology, 91, 686–692.

Ben-Shlomo, Y., Smith, G. D., & Shipley, M. (1993).Magnitude and causes of mortality differencesbetween married and unmarried men. Journal of Epi-demiology and Community Health, 47, 200–205.

Berman, W. H. (1988). The relationship of ex-spouseattachment to adjustment following divorce. Jour-nal of Family Psychology, 1, 312–328. doi:10.1037/h0080455

Biber, D. (1988). Variation across speech and writing.Cambridge, UK: Cambridge University Press.

Birnbaum, G. E., Orr, I., Mikulincer, M., & Florian, V.(1997). When marriage breaks up: Does attachmentstyle contribute to coping and mental health? Journalof Social and Personal Relationships, 14, 643–654.doi:10.1177/0265407597145004

Boals, A., & Klein, K. (2005). Word use in emotionalnarratives about failed romantic relationships andsubsequent mental health. Journal of Language andSocial Psychology, 24, 252–268. doi:10.1177/0261927X05278386

Borelli, J. L., Sbarra, D. A., Mehl, M., & David, D. H.(in press). Experiential connectedness in children’s at-tachment interviews: An examination of natural worduse. Personal Relationships. doi:10.1111/j.1475-6811.2010.01294.x

Bowlby, J. (1969). Attachment and loss: Vol. 1. Attach-ment (2nd ed.). New York: Basic Books.

Bowlby, J. (1973). Attachment and loss: Vol. 2. Separa-tion. New York: Basic Books.

Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss:Sadness and depression. New York: Basic Books.

Bucci, W., & Freedman, N. (1981). The language ofdepression. Bulletin of the Menninger Clinic, 45,334–358.

Buchheim, A., & Mergenthaler, E. (2000). Therelationship among attachment representation, emo-tion-abstraction patterns, and narrative style: A com-puter-based text analysis of the adult attachmentinterview. Psychotherapy Research, 10, 390–407.doi:10.1093/ptr/10.4.390

Butzer, B., & Campbell, L. (2008). Adult attachment,sexual satisfaction, and relationship satisfaction: Astudy of married couples. Personal Relationships, 15,141–154. doi:10.1111/j.1475-6811.2007.00189.x

Campa, M. I., Hazan, C., & Wolfe, J. E. (2009). Theform and function of attachment behavior in thedaily lives of young adults. Social Development, 18,288–304. doi:10.1111/j.1467-9507.2008.00466.x

Chung, C., & Pennebaker, J. (2007). The psychologi-cal functions of function words. In K. Fiedler (Ed.),Social communication (pp. 343–359). New York:Psychology Press.

Coan, J. A., Allen, J. J. B., & McKnight, P. E. (2006).A capability model of individual differences infrontal EEG asymmetry. Biological Psychology, 72,198–207. doi:10.1016/j.biopsycho.2005.10.003

Cohn, M. A., Mehl, M. R., & Pennebaker, J. W. (2004).Linguistic markers of psychological change surround-ing September 11, 2001. Psychological Science, 15,687–693. doi:10.1111/j.0956-7976.2004.00741.x

Davis, D., Shaver, P. R., & Vernon, M. L. (2003). Phys-ical, emotional, and behavioral reactions to breakingup: The roles of gender, age, emotional involvement,and attachment style. Personality and Social Psychol-ogy Bulletin, 29, 871–884. doi:10.1177/0146167203029007006

Diamond, L. M. (2001). Contributions of psychophysi-ology to research on adult attachment: Review andrecommendations. Personality and Social Psychol-ogy Review, 5, 276–295. doi:10.1207/S15327957PSPR0504

Domingue, R., & Mollen, D. (2009). Attachment andconflict communication in adult romantic relation-ships. Journal of Social and Personal Relationships,26, 678–696. doi:10.1177/0265407509347932

Dunnack, E. S., & Park, C. L. (2009). The effect of anexpressive writing intervention on pronouns: Thesurprising case of I. Journal of Loss and Trauma, 14,436–446. doi:10.1080/15325020902925084

Feeney, B. C., & Kirkpatrick, L. A. (1996). Effects ofadult attachment and presence of romantic part-ners on physiological responses to stress. Journalof Personality and Social Psychology, 70, 255–270.doi:10.1037/0022-3514.70.2.255

Feldman, P. J., Cohen, S., Hamrick, N., & Lepore, S. J.(2004). Psychological stress, appraisal, emotion andcardiovascular response in a public speaking task.Psychology & Health, 19, 353–368. doi:10.1080/0887044042000193497

Fraley, R. C. (2002). Attachment stability from infancyto adulthood: Meta-analysis and dynamic modelingof developmental mechanisms. Personality and SocialPsychology Review, 6, 123–151. doi:10.1207/S15327957PSPR0602

Fraley, R. C., Niedenthal, P. M., Marks, M.,Brumbaugh, C., & Vicary, A. (2006). Adult attach-ment and the perception of emotional expressions:Probing the hyperactivating strategies underlying anx-ious attachment. Journal of Personality, 74,1163–1190. doi:10.1111/j.1467-6494.2006.00406.x

Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000).An item response theory analysis of self-report mea-sures of adult attachment. Journal of Personality andSocial Psychology, 78, 350–365. doi:10.1037/0022-3514.78.2.350

Gendolla, G. H. E. (1999). Self-relevance of perfor-mance, task difficulty, and task engagement assessedas cardiovascular response. Motivation and Emotion,23(1), 45–66.

Graham, J. E., Glaser, R., Loving, T. J., Malarkey, W. B.,Stowell, J. R., & Kiecolt-Glaser, J. (2009). Cognitiveword use during marital conflict and increases inproinflammatory cytokines. Health Psychology, 28,621–630. doi:10.1037/a0015208

Hetherington, E. M., & Kelly, J. (2002). For better orfor worse: Divorce reconsidered. New York: W.W.Norton.


Holmes, T. H., & Rhae, R. H. (1967). The social read-justment rating scale. Journal of Psychosomatic Re-search, 11, 213–218.

Kiecolt-Glaser, J., Fisher L. D., Ogrocki, P., Stout, J. C.,Speicher, C. E., & Glaser, R. (1987). Marital quality,marital disruption, and immune function. Psychoso-matic Medicine, 49, 13–34.

Kim, Y. (2006). Gender, attachment, and relationshipduration on cardiovascular reactivity to stress in alaboratory study of dating couples. Personal Relation-ships, 13, 103–114. doi:10.1111/j.1475-6811.2006.00107.x

Kotler, T., Buzwell, S., Romeo, Y., & Bowland, J. (1994).Avoidant attachment as a risk factor for health. BritishJournal of Medical Psychology, 67, 237–245.

Kross, E. (2009). When the self becomes other: Towardan integrative understanding of the processes distin-guishing adaptive self-reflection from rumination. InA. Tobena (Ed.), Values, empathy, and fairness acrosssocial barriers. (pp. 35–40). New York: New YorkAcademy of Sciences.

Lucas, R. E. (2005). Time does not heal all wounds:A longitudinal study of reaction and adaptationto divorce. Psychological Science, 16, 945–950.doi:10.1111/j.1467-9280.2005.01642.x

Mauss, I. B., & Robinson, M. D. (2009). Measures ofemotion: A review. Cognition and Emotion, 23,209–237.

Mehrabian, A. (1967). Attitudes inferred from non-immediacy of verbal communications. Journal ofVerbal Learning & Verbal Behavior, 6, 294–295.doi:10.1016/S0022-5371(67)80113-0

Mikulincer, M., & Shaver, P. R. (2003). The attach-ment behavioral system in adulthood: Activation, psy-chodynamics, and interpersonal processes. In M. P.Zanna (Ed.), Advances in experimental social psychol-ogy (Vol. 35, pp. 53–152). San Diego, CA: Elsevier.

Mikulincer, M., & Shaver, P. R. (2005). Attachment the-ory and emotions in close relationships: Exploringthe attachment-related dynamics of emotional reac-tions to relational events. Personal Relationships, 12,149–168. doi:10.1111/j.1350-4126.2005.00108.x

Mikulincer, M., & Shaver, P. R. (2007). Attachment inadulthood: Structure, dynamics, and change. NewYork: Guilford.

Newman, M. L., Groom, C. J., Handelman, L. D., &Pennebaker, J. W. (2008). Gender differences in lan-guage use: An analysis of 14,000 text samples. Dis-course Processes, 45, 211–236.

Pennebaker, J. W., & Graybeal, A. (2001). Patterns ofnatural language use: Disclosure, personality, andsocial integration. Current Directions in Psycho-logical Science, 10(3), 90–93. doi:10.1111/1467-8721.00123

Pennebaker, J. W., & King, L. A. (1999). Linguisticstyles: Language use as an individual difference.Journal of Personality and Social Psychology, 77,1296–1312. doi:10.1037/0022-3514.77.6.1296

Pennebaker, J. W., Mayne, T. J., & Francis, M. E.(1997). Linguistic predictors of adaptive bereavement.

Journal of Personality and Social Psychology, 72,863–871. doi:10.1037/0022-3514.72.4.863

Picardi, A., Mazzotti, E., Gaetano, P., Cattaruzza, M. S.,Baliva, G., Melchi, C. F., . . . Pasquini, P. (2005).Stress, social support, emotional regulation, and exac-erbation of diffuse plaque psoriasis. Psychosomat-ics: Journal of Consultation Liaison Psychiatry, 46,556–564. doi:10.1176/appi.psy.46.6.556

Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006).Computational tools for probing interaction effectsin multiple linear regression, multilevel modeling,and latent curve analysis. Journal of Educational andBehavioral Statistics, 31, 437–448.

Pressman, S. D., & Cohen, S. (2007). Use of socialwords in autobiographies and longevity. Psychoso-matic Medicine, 69, 262–269. doi:10.1097/PSY.0b013e31803cb919

Rohrbaugh, M. J., Mehl, M. R., Shoham, V., Reilly,E. S., & Ewy, G. A. (2008). Prognostic significanceof spouse we talk in couples coping with heart failure.Journal of Consulting and Clinical Psychology, 76,781–789. doi:10.1037/a0013238

Roisman, G. I. (2007). The psychophysiology of adultattachment relationships: Autonomic reactivity inmarital and premarital interactions. DevelopmentalPsychology, 43, 39–53. doi:10.1037/0012-1649.43.1.39

Roisman, G. I., Tsai, J. L., & Chiang, K. S. (2004).The emotional integration of childhood experience:Physiological, facial expressive, and self-reportedemotional response during the adult attachmentinterview. Developmental Psychology, 40, 776–789.doi:10.1037/0012-1649.40.5.776

Rossi, P., Di Lorenzo, G., Malpezzi, M. G., Di Lorenzo,C., Cesarino, F., Faroni, J., . . . Troisi, A. (2005).Depressive symptoms and insecure attachment as pre-dictors of disability in a clinical population of patientswith episodic and chronic migraine. Headache: TheJournal of Head and Face Pain, 45, 561–570.doi:10.1111/j.1526-4610.2005.05110.x

Saffrey, C., & Ehrenberg, M. (2007). When thinkinghurts: Attachment, rumination, and postrelationshipadjustment. Personal Relationships, 14, 351–368.doi:10.1111/j.1475-6811.2007.00160.x

Sbarra, D. A. (2006). Predicting the onset of emotionalrecovery following nonmarital relationship dissolu-tion: Survival analyses of sadness and anger. Person-ality and Social Psychology Bulletin, 32, 298–312.doi:10.1177/0146167205280913

Sbarra, D. A., & Emery, R. E. (2005). The emotionalsequelae of nonmarital relationship dissolution: Anal-ysis of change and intraindividual variability overtime. Personal Relationships, 12, 213–232. doi:10.1111/j.1350-4126.2005.00112.x

Sbarra, D. A., & Ferrer, E. (2006). The structure andprocess of emotional experience following nonmar-ital relationship dissolution: Dynamic factor analysesof love, anger, and sadness. Emotion, 6, 224–238.doi:2006-07383-006 [pii]; 10.1037/1528-3542.6.2.224


Sbarra, D. A., Law, R. W., Lee, L. A., & Mason, A. E.(2009). Marital dissolution and blood pressure reac-tivity: Evidence for the specificity of emotionalintrusion-hyperarousal and task-related emotionaldifficulty. Psychosomatic Medicine, 71, 532–540.doi:10.1097/PSY.0b013e3181a23eee

Sbarra, D. A., & Nietert, P. J. (2009). Divorce and death:Forty years of the Charleston heart study. Psycholog-ical Science, 20, 107–113.

Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment & Human Devel-opment, 4, 133–161. doi:10.1080/14616730210154171

Singer, J. D. (1998). Using SAS PROC MIXED to fitmultilevel models, hierarchical models, and individualgrowth models. Journal of Educational and Behav-ioral Statistics, 23, 323–355.

Singer, J. D., & Willett, J. B. (2003). Applied longitudi-nal data analysis: Modeling change and event occur-rence. New York: Oxford University Press.

Slatcher, R. B., Vazire, S., & Pennebaker, J. W. (2008).Am “I” more important than “we”? couples’ worduse in instant messages. Personal Relationships, 15,407–424. doi:10.1111/j.1475-6811.2008.00207.x

Sroufe, L. A., & Waters, E. (1977). Attachment as anorganizational construct. Child Development, 48,1184–1199. doi:10.2307/1128475

Stapel, D. A., & Semin, G. (2007). The magic spellof language: Linguistic categories and their percep-tual consequences. Journal of Personality and SocialPsychology, 93, 23–33. doi:10.1037/0022-3514.93.1.23

Tausczik, Y. R., & Pennebaker, J. W. (2010). The psy-chological meaning of words: LIWC and computer-ized text analysis methods. Journal of Language andSocial Psychology, 29, 24–54. doi:10.1177/0261927X09351676

Tucker, J. S., Friedman, H. S., Wingard, D. L., &Schwartz, J. E. (1996). Marital history at midlife asa predictor of longevity: Alternative explanations tothe protective effect of marriage. Health Psychology,15(2), 94–101. doi:10.1037/0278-6133.15.2.94

Waugh, C. E., Fredrickson, B. L., & Taylor, S. F. (2008).Adapting to life’s slings and arrows: Individual dif-ferences in resilience when recovering from an antic-ipated threat. Journal of Research in Personality, 42,1031–1046. doi:10.1016/j.jrp.2008.02.005

Waugh, C. E., Panage, S., Mendes, W. B., & Gotlib,I. H. (2010). Cardiovascular and affective recoveryfrom anticipatory threat. Biological Psychology, 84,169–175. doi:10.1016/j.biopsycho.2010.01.010

Whelton, S. P., Chin, A., Xin, X., & He, J. (2002). Effectof aerobic exercise on blood pressure: A meta-analysis of randomized, controlled trials. Annals ofInternal Medicine, 136, 493–503.

Wright, R. A. (1996). Brehm’s theory of motivation asa model of effort and cardiovascular response. InP. M. Gollwitzer & J. A. Bargh (Eds.), The psychol-ogy of action: Linking cognition and motivation tobehavior. (pp. 424–453). New York: Guilford.

Wright, R. A., Martin, R. E., & Bland, J. L. (2003). En-ergy resource depletion, task difficulty, and cardio-vascular response to a mental arithmetic challenge.Psychophysiology, 40(1), 98–105.

attachment anxiety, verbal immediacy, and blood pressure: results

Documents