You Need to Stop Talking About This!’’ Verbal Rumination and the Costs of Social Support.pdf

Human Communication Research ISSN 0360-3989

ORIGINAL ARTICLE

‘‘You Need to Stop Talking About This!’’:
Verbal Rumination and the Costs of Social
Support
Tamara Afifi1 , Walid Afifi1 , Anne F. Merrill1 , Amanda Denes2 , & Sharde
Davis1
1 Department of Communication, University of California, Santa Barbara, CA 93106-4020, USA
2 Department of Communication Sciences, University of Connecticut, Storrs, CT 06269, USA

This study examined whether the type of support individuals receive when they are
verbally ruminating affects their cognitive rumination (brooding), anxiety, and relationship
satisfaction; 233 young adults were randomly assigned to be the subject, 233 others the
confederate. The confederate was trained to provide ‘‘good support’’ or ‘‘poor support’’ to the
subject who talked about a topic he/she had been verbally ruminating about recently. When
individuals verbally ruminated and received poor support, they became more anxious and
dissatisfied with the friendship. When individuals received good support, they were more
satisfied with their friendship, but their anxiety was not significantly reduced. In addition,
verbal rumination was directly associated with more brooding after the conversation,

regardless of the type of support provided.
doi:10.1111/hcre.12012

When people experience a stressful event and cannot stop thinking about it, they
frequently turn to friends for support (Rime, Philippot, Boca, & Mesquita, 1992).
Although social support generally makes people feel better, talking about a stressor
with others does not necessarily relieve one’s distress and can sometimes make it
worse (Holmstrom, Burleson, & Jones, 2005; Xu & Burleson, 2001). People are
not always supportive when someone comes to them with a problem, particularly
when this person talks about the same problem incessantly (Stroebe, Zech, Stoebe, &
Abakoumkin, 2005). Friends may not want to hear about the problem anymore, may
not know what to say, or may criticize the person for still thinking about it. Even if
support providers have good intentions, they might not be able to help the person
stop brooding about the stressor and could magnify his/her anxiety and ruminative
tendencies.
Research has shown that when people verbally ruminate, it can adversely affect
their social networks if their stress spills over onto others (see Saxbe & Repetti,
2010) or they communicate social depression cues that make it difficult for people
to be around them (e.g., Conway et al., 2011). Little is known, however, about how
Corresponding author: Tamara Afifi; e-mail: [email protected]

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

395

Verbal Rumination and Social Support

T. Afifi et al.

social support affects the individuals who are verbally ruminating and whether the
support reduces their brooding. Brooding rumination involves focusing passively and
repetitively on one’s negative emotions and problems, their causes and consequences,
and self-evaluations related to them (Nolen-Hoeksema & Morrow, 1991). Verbal
rumination is similar to brooding rumination, except that it focuses on repetitive
talk—continually talking about a problem and its potential consequences, negative
emotions surrounding the problem, and one’s role in it. These two processes often
occur concurrently (Tait & Silver, 1989), but tend to be analyzed separately. Yet, the
successful management of one most likely affects the other (see also Nolen-Hoeksema
& Davis, 1999).
Social support may act as an important moderator of the impact of verbal
rumination on brooding, as well as mental and relational health. If people offer

support that is validating, reassuring, and helps the support seeker reframe the
problem about which they are ruminating, it could minimize the endless cycle of
brooding and anxiety they are experiencing while also enhancing their relationship
with the support provider. To the contrary, when people receive poor support
when they are verbally ruminating, it likely exacerbates their anxiety and brooding
and strains the relationship. We test this argument in the context of young adult
friendships. Young adults face numerous stressors, such as financial challenges, risky
behaviors, and career aspirations, for the first time when they live independently
from their parents. These stressors may be sources of worry that become chronic
topics of discussion with close friends.
Verbal rumination and brooding rumination

Most of the work on self-disclosure has found that disclosing personal information
is health promoting. While there are numerous reasons why people disclose, much
of the research on self-disclosure suggests that people feel better after they disclose
because it is cathartic (see Afifi, Caughlin, & Afifi, 2007 for a review; Derlega,
Winstead, Greene, Serovich & Elwood, 2004). For example, Pennebaker’s work
(Pennebaker, 1989; 1995) shows that people often experience physical health benefits
when they disclose stressful events.
Research on the fever model (Stiles, 1987; Stiles, Shuster, & Harrigan, 1992) also

demonstrates that when people disclose what they are ruminating about, disclosure
tends to reduce their anxiety and ruminative tendencies. Moreover, self-disclosure
helps build and maintain relationships because people develop emotional bonds
through the sharing of private information (see Altman & Taylor, 1973; Pearce &
Cronen, 1980; Petronio, 1991, 2002; Wheeless, 1976). As the basic tenants of social
penetration theory (Altman & Taylor, 1973) suggest, people move from disclosing
non-intimate information (or ‘‘breadth’’) to intimate information (or ‘‘depth’’) as
their relationships progress, which fosters greater psychological closeness between
them.
Even though the vast majority of the research suggests that disclosure is beneficial
to people’s relational and personal health, research also shows that its benefits
396

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

T. Afifi et al.

Verbal Rumination and Social Support

depend heavily on the response (expected or actual) from the person to whom one

is disclosing (Checton & Greene, 2012; Derlega, Winstead and Folk-Barron, 2000;
Derlega et al., 2004, 2010; Petronio, 1991, 2000, 2002; Petronio, Reeder, Hecht, &
Ros-Mendoza, 1996). In general, when people perceive others as trusting and expect
them to respond positively to their disclosure, the benefits of disclosing increase and
the costs decrease (Afifi & Steuber, 2009; Petronio, 2002; Wheeless & Grotz, 1977).
Whether the disclosure of a stressor is beneficial depends upon numerous factors,
including the validating nature of the response (Afifi, Olson, & Armstrong, 2005;
Greene, 2009) and the extent to which the support enables the person to make sense
of a stressful event (Kelly & Macready, 2009).
Verbal rumination is also different than simply disclosing. Because verbal rumination focuses on repeated disclosures of personal problems and negative emotions,
it could have a more harmful effect on one’s mental health than self-disclosure in
general. For example, parents’ negative disclosures about the other parent to their
child tend to be associated with anxiety in both the parent and the child (Afifi
& McManus, 2010; Denes, Afifi, Granger, Joseph, & Aldeis, in press). Afifi and
McManus (2010) found that parents’ negative disclosures about the other parent
were associated with closer parent–child relationships, but increased anxiety in the
child. Even though disclosures, perhaps in some cases regardless of their valence, can
foster psychological closeness between individuals, they can simultaneously harm
people’s mental health if the negativity spills over onto others or begins to affect
one’s own psychological well-being. If these negative disclosures occur excessively,

they could create even greater anxiety.
Most of the recent research on verbal rumination focuses on corumination or
two people excessively discussing problems with each other (Rose, 2002), rather
than verbal rumination where one person discloses his/her problems to another.
In particular, that research examines corumination in adolescence in an attempt
to explain why girls tend to have more anxiety and depressive symptoms than
boys and how these effects materialize in friendships over time (e.g., SchwartzMette & Rose, 2012; Stone, Uhrlass, & Gibb, 2010). Research shows that girls
report being more satisfied with their friendships than boys, which should protect
against emotional problems, but ironically, they simultaneously report being more
anxious and depressed (Starr & Davila, 2009). The mutual disclosure likely brings
the friends closer emotionally, but also fosters anxiety due to the preoccupation
with negative emotions (Rose, 2002). Because self-disclosure is more common in
female friendships than male friendships, researchers argue that females tend to
experience the heightened benefits and consequences of corumination (Smith &
Rose, 2011). It is worth noting, however, that, the research supporting such a sex
effect is mixed, with an equal amount of research finding no sex differences (see
Hankin, Stone, & Wright, 2010; Smith & Rose, 2011; Stone, Hankin, Gibb, & Abela,
2011).
Developmental trends also suggest that corumination continues, and potentially
increases, with age. For example, Hankin et al. (2010) found that initial levels of

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

397

Verbal Rumination and Social Support

T. Afifi et al.

internalizing problems in adolescents (6th to 10th graders) predicted greater corumination with friends over time and that corumination, in turn, predicated greater
internalizing problems. These bidirectional effects escalated in severity through adolescence. A couple of studies have also found effects for corumination in young
adulthood (Byrd-Craven, Geary, Rose, & Ponzi, 2008; Calmes & Roberts, 2008).
For instance, Byrd-Craven et al. (2008) discovered that when college-age female
friends co-ruminated, their cortisol increased. These studies provide evidence that
excessively discussing problems in young adult friendships is linked to internalizing
problems.
Unlike with corumination, verbal rumination typically involves one person
disclosing problems and negative emotions while the other provides support. The
type of support individuals receive when they are verbally ruminating could affect
their relationship and their anxiety and brooding. The research on corumination
assumes that people receive comparable levels of support, despite this often not

being the case. Social support might provide a powerful theoretical explanation for
how and why verbal rumination affects anxiety and satisfaction, as well as brooding,
irrespective of biological sex.
Theoretical connections to social support

When people become preoccupied with a stressor, they often talk about it with
others to relieve their distress. The extent to which that social support is effective
for individuals who are verbally ruminating, however, is still unclear. Social support
tends to buffer the ill effects of stress on people’s mental and physical health
(see Afifi, Granger, Denes, Joseph, & Aldeis, 2011; Albrecht & Goldsmith, 2003;
Burleson, 2009). According to the stress buffering hypothesis, better quality support
and larger social support networks minimize the effect of stress on people’s health
(Cohen & Wills, 1985) and should make people feel better when they are verbally
ruminating. However, research on verbal rumination and traumas, natural disasters,
contagion effects, and depression provides evidence of both stress buffering and stress
deterioration (see Conway et al., 2011; Stone et al., 2010). People may be able to
provide good support initially, but are unable to sustain it because their emotional
resources have been drained (e.g., Afifi, Afifi, & Merrill, 2013). People might deny
social support to friends who talk about the same problem excessively because they
should be ‘‘over it’’ by now. Support providers can also become stressed themselves

as a result of the discloser’s stress spilling over onto them through repeated, negative
talks. Or, sometimes individuals provide poor support if they feel like the person
deserved his/her stress (Kaniasty & Norris, 1993).
The social support individuals receive when they are verbally ruminating most
likely affects their brooding and anxiety afterward. Rumination often fosters anxiety
because of the negative thoughts and emotions people experience when they brood.
Ruminators become anxious and uncertain about situations being under their
control (Nolen-Hoeksema, 2000). Rumination also involves a sense of hopelessness
and negative evaluations of the self (Zawadzki, Graham, & Gerin, 2012). As a result,
398

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

T. Afifi et al.

Verbal Rumination and Social Support

‘‘rumination and anxiety can create a reciprocally determinative cycle in which each
tends to promote and prolong the other’’ (Zawadzki et al., 2012, p. 2).
Social support could help disrupt this negative cycle. For example, NolenHoeksema and Davis (1999) found that high ruminators reported less distress over

time if they had dense social networks, received good support, and felt comfortable
talking about their stress. In contrast, high ruminators reported more distress over
time if their social support network was critical of them, disagreed with the decisions
they made, or had high conflict. Even though the authors did not examine verbal
rumination, specifically, they were able to show that when people confide in others
about their stress, social support can serve an important moderating function.
But, what constitutes ‘‘good’’ support when someone is verbally ruminating?
Support providers who use high person-centered (HPC) messages with friends who
are verbally ruminating should reduce their friends’ anxiety and brooding because
of the comfort and assurance provided. HPC messages explicitly validate, comfort,
contextualize, and elaborate on the distressed person’s feelings and emotions about
the stressor (Burleson, 1982, 2008; Burleson & MacGeorge, 2002; Holmstrom et al.,
2005; Jones & Guerrero, 2001). HPC messages also involve esteem support where
the support provider attempts to enhance how the support receiver feels about
him or herself (Holmstrom & Burleson, 2011). This is in contrast to low-person
centered (LPC) messages that tend to deny the feelings of the distressed person
(Burleson, 1982, 2008). Support that is more person-centered has been associated
with greater emotional improvement compared to lower person-centered messages
and could provide a more hopeful interpretation of stressful situations (Jones, 2004;
Jones & Guerrero, 2001; Jones & Wirtz, 2006) and reduce rumination. Bippus (2000,

2001), for example, found that the skillfulness of the comforting behaviors provided
(i.e., less negativity, taking the perspective of the other person, problem solving,
and reframing) was predictive of less rumination, more positive mood and greater
empowerment.
Indeed, social support that helps people who are verbally ruminating reframe
their stressor could minimize the cascade of anxiety and intrusive thoughts they
experience. According to Lazarus and Folkman’s (1984) theory of stress and coping,
events become stressful when people perceive that the demands of the situation
outweigh their resources or ability to cope (Lazarus & Folkman, 1984). When a
stressor occurs, people make primary appraisals about the extent of the stress and
they make secondary appraisals where they weigh their available coping options and
decide what to do about it (Lazarus & Folkman, 1984).
As Burleson and Goldsmith (1998) argue in their theory of conversationallyinduced reappraisals, social support can result in positive reappraisal of a stressor,
which can improve people’s emotional states. The support providers cannot change
the stressful event, but they can change a person’s appraisal of the stressor (Burleson
& Goldsmith, 1998; Burleson & MacGeorge, 2002). Reappraisal enables people to
find meaning in stressful situations, which can induce more positive emotions
(Denson, Spanovic, & Miller, 2009). Good social support could help those who are
Human Communication Research 39 (2013) 395–421  2013 International Communication Association

399

Verbal Rumination and Social Support

T. Afifi et al.

verbally ruminating to appraise the stressor differently and provide alternative coping
strategies—allowing them to sooth their anxiety and brooding. In so doing, they
probably also feel better about their relationship with the person who is providing
the support.
In addition to verbal HPC messages and reframing of the stressor, good social
support should also be nonverbally immediate (Guerrero, Jones & Burgoon, 2000;
Jones & Guerrero, 2001). As Jones and Guerrero argue, even though most research
on person-centered messages focuses on the verbal messages, verbal comforting
messages, and nonverbal immediacy cues, such as head nods, touch, eye contact,
paralanguage, and body orientation, occur simultaneously and support receivers
interpret them as such. As such, the verbal and nonverbal messages should be used
in conjunction with one another when assessing the effectiveness of social support.
As Miller (2007) has also found with her work on compassionate communication in
the workplace, communicating compassion includes verbal and nonverbal indicators
of noticing another, connecting, and responding. Taken together, social support
that is high person centered, encourages positive reframing of the stressor, and is
nonverbally immediate (e.g., involved, pleasant, encouraging), should provide an
optimal, ‘‘safe’’ environment to disclose.
In contrast, when people are verbally ruminating and they are provided with poor
support, this combination likely magnifies their anxiety and brooding. Some aspects
of Burleson’s low person-centered messages (see Burleson, 1982, 2008) where the
support provider dismisses an individual’s irrational thoughts (e.g., ‘‘that person is
not worth your time . . . just stop thinking about it’’) could actually reduce brooding
(see Nolen-Hoeksema & Davis, 1999). Messages that are more disconfirming of the
individual, however, should have an adverse effect on ruminative tendencies. Disconfirming messages that criticize individuals’ choices, blame them for their own stress,
or invalidate their feelings and perspectives likely reinforce their negative view of the
self. The disconfirming messages may exacerbate their beliefs that they are the source
of the problem, increasing their anxiety and dissatisfaction with the relationship. For
example, a series of studies by McLaren, Priem, and Solomon (McLaren & Solomon,
2008; McLaren, Solomon, & Priem, 2011; Priem, McLaren, & Solomon, 2010; Priem
& Solomon, 2011) has shown that hurtful messages that are disconfirming are
physiologically stressful and relationally distancing. Likewise, immediacy cues where
a person appears to be distracted, uninvolved, and lacking in expected levels of
intimacy are anxiety-producing for support receivers (Guerrero et al., 2000).
In general, how one’s anxiety and relationship satisfaction are influenced as a
result of verbally ruminating should depend upon the type of support received when
one is verbally ruminating. When individuals are verbally ruminating and receive
poor support, regardless of gender, it should increase one’s anxiety and brooding
and decrease relationship satisfaction. The opposite effect should occur when good
support is provided. Therefore, we hypothesize that:
H1: Social support will moderate the association between verbal rumination and
relationship satisfaction such that being provided with good support when verbally
400

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

T. Afifi et al.

Verbal Rumination and Social Support

ruminating will increase one’s satisfaction whereas poor support will decrease one’s
satisfaction.
H2: Social support will moderate the association between verbal rumination and (H2a)
state anxiety and (H2b) brooding rumination, such that being provided with good
support when verbally ruminating will decrease one’s anxiety and brooding
rumination, whereas poor support will increase one’s anxiety and brooding
rumination.

Verbal rumination and brooding rumination likely work together to affect negatively people’s anxiety and relationship satisfaction. Individuals’ verbal rumination
should be especially anxiety-producing when they are provided with poor support
and they brood about it afterward. When people brood about what transpired in
their interaction with their friend, they likely become even more anxious because they
have had the time to mull about their problem and their inability to solve it. They
are also probably more dissatisfied with their friendship because they have brooded
about the fact that their friend was disconfirming toward them and their situation.
Consequently, a third hypothesis is presented:
H3: The degree to which one’s verbal rumination affects one’s (H3a) anxiety and (H3b)
satisfaction with the friendship will depend on the support that is provided
(good/poor) and the extent of the brooding rumination afterward. Being provided with
poor support when one verbally ruminates and engaging in higher levels of brooding
rumination afterward should increase anxiety and relationship dissatisfaction, whereas
being provided with good support and engaging in lower levels of brooding
rumination afterward should decrease anxiety and increase relationship satisfaction.

Method
Participants

Two hundred thirty-one young adults and one of their friends came to the laboratory
on one occasion. The average age of the participants was 19.84 (range = 18 to 26).
Most of the participants were female (n = 159 or 68%; males = 72 or 31%). When
taking the friend into account, there were 112 (48%) female–female dyads, 35 (15%)
male–male dyads, and 86 (37%) female–male dyads. The sample was a diverse
representation of White (n = 81; 35%), Latino (n = 46 or 20%), Black (n = 61 or
26%), Asian (n = 30 or 13%) and ‘‘other’’ (n = 15 or 6%) ethnicities. Eighty-one
(35%) of the participants were first year undergraduates, 61 (26%) were sophomores,
55 (24%) were juniors, 30 (13%) were seniors, two were in graduate school, and four
were not in college.
Procedures

After the study’s approval through an institutional review board, the participants
were recruited from communication courses at a university on the West Coast of the
United States. When participants registered online for the study, they were told to
bring a good friend (male or female) with them to the laboratory. They were also
Human Communication Research 39 (2013) 395–421  2013 International Communication Association

401

Verbal Rumination and Social Support

T. Afifi et al.

told that they could not be dating this person and that the friendship needed to be
platonic. Before arriving at the laboratory, one of the friends was randomly assigned
to be the subject (support recipient) and the other a confederate (support provider).
At the same time, the confederate was randomly assigned to the ‘‘good support’’
(n = 107) or ‘‘poor support’’ (n = 124) condition (see below).
To enhance the believability of the study, both the subject and the confederate
were briefly informed about what would happen in the study (including that their
interaction would be videotaped) and provided topic listing sheets. They were asked
to write down up to three topics they currently found stressful, that they could not
seem to get off their mind, and that they could not stop talking about. They were
also told that they should be topics that they already talked about with their friend
who came with them to the laboratory, but that the topics could not involve their
friend or their relationship. After each topic, they also completed four Likert-type
items that asked how much they currently mulled over the topic and the extent to
which they brought it up as a topic of conversation. The researcher also informed the
participants that they would complete all of the surveys in different rooms to ensure
privacy. The researcher chose the topic that the subject verbally ruminated about
the most as the topic for the friends to discuss (but informed the subject that it was
randomly chosen). The researcher refrained from choosing topics that were deemed
particularly sensitive (e.g., severe illness, death, body image).
After being placed in separate rooms, the subject began completing the preinteraction survey while the confederate was informed of the true purpose of the study and
trained to provide either good or poor support. Good and poor support were based
on Burleson’s (1982) person-centered messages (HPC messages for the good-support
condition and low person-centered messages for the poor-support condition) and
Jones and Guerrero’s (Guerrero et al., 2000; Jones & Guerrero, 2001) nonverbal
immediacy cues.
The confederates were trained to maintain eye contact, have their body oriented
toward their friend, use head nods, listen actively, be involved in the conversation
(e.g., vocal utterances that emphasize listening), generally reciprocate nonverbal
behaviors, and avoid distractions. In addition, in terms of the verbal channel, the
confederates were trained to ask their friend how he/she was feeling (e.g., ‘‘how are
you feeling about what happened?’’) and then validate those feelings (e.g., ‘‘I’m so
sorry that you experienced that’’) and provide comfort, probe for more information
or details about the stressor, elaborate in detail on the situation and the friend’s
feelings (e.g., ‘‘I know how deeply you cared for each other and how much time you
spent together, that must be really painful’’), and eventually try to help the friend
cognitively reframe the problem. They were also asked not to talk about their own
problems, but to keep the focus on their friend’s problem.
In the poor-support condition, the confederates were trained to appear distracted
and uninterested in what their friend was saying by leaning back, acting bored or
anxious to finish the conversation, reducing eye contact, playing with their shoes or
pants, or checking their cell phone. They were trained to be verbally nonsupportive
402

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

T. Afifi et al.

Verbal Rumination and Social Support

of their friend’s discussion of his/her problem, to be disconfirming of their friend’s
situation and ideas, and to blame him/her for being the source of the stress. Emphasis
was placed on engaging in these behaviors while appearing natural and maintaining
behavioral plausibility. Confederates in this condition were trained to act as they
typically would for the first minute, then to engage in some of the aforementioned
nonverbal behaviors, and gradually introduce the verbally disconfirming behaviors
that were low person centered, invalidated the person’s feelings, and blamed the
person for his/her stress.
Because the researcher knew the topic of the discussion and the confederate had
already talked about it before with the subject, the researcher then engaged in a role
play about the topic with the confederate as part of the training (in the good- and
poor-support conditions). For example, whether the topic was a messy roommate
(e.g., ‘‘Maybe you’re just too perfectionistic?’’ ‘‘I’ve never had a problem with this
person’’) or fear of doing poorly on an upcoming exam (e.g., ‘‘Everyone gets a
good grade on her exams, if you don’t, there is seriously something wrong,’’ ‘‘The
exams are so much harder in my major’’), the confederate was trained to provide
disconfirming messages geared toward that situation. Together, the researcher and
confederate thought of messages that would be disconfirming but natural for that
relationship.
When the subject was finished with the preinteraction survey, the confederate
was brought back into the room and the researcher explained that out of both
topic listing sheets, the selected subject’s topic was randomly chosen as the topic
for discussion. They were told to talk about the topic for as long as they wanted
and then to tell the researcher when they were finished. All of the interactions
were videotaped. When the conversation was finished, the friends were placed into
separate rooms again. The subjects completed a five minute survey that asked about
the support their friend provided, anxiety, and satisfaction with the friendship. The
subject was then left alone in the room for a 15-minute period to promote brooding.
At the end of the 15 minutes, the subjects completed the final survey, which included
their rumination after the interaction, anxiety, and satisfaction. The friends were
then brought together, debriefed thoroughly, gave their permission for us to use
their data, and were provided the opportunity to have their videotapes destroyed.
Information for free counseling services was provided, however, none of the dyads
appeared or reported being distressed after the debriefing.
Measures
Relationship satisfaction

Satisfaction with the friendship was assessed in the pre-interaction survey (T1),
immediately after the interaction (T2) and at the end of the 15 minute wait period
(T3). It was measured with a modified version of the marital opinion questionnaire
(Huston, McHale, & Crouter, 1986), which has been successfully adapted for a
variety of relationships. Five items used 7-point semantic differential scales (e.g.,
‘‘miserable-enjoyable’’) and an additional item measured global satisfaction from 1,
Human Communication Research 39 (2013) 395–421  2013 International Communication Association

403

Verbal Rumination and Social Support

T. Afifi et al.

Completely dissatisfied to 7, Completely satisfied. All of the items were averaged and
larger scores indicated greater satisfaction. The alpha reliabilities were .91 for T1, .93
for T2, and .92 for T3.
Closeness

The subjects’ perception of closeness with the friend was assessed at time 1 with four
items (e.g., how psychologically and emotionally close they feel toward the friend,
how well the friend knows him/her). The Likert-type scale ranged from 1, Not at all
to 7, Very, with higher scores indicating greater closeness (α = .91).
State anxiety

Anxiety was measured with Spielberger, Gorsuch, Lushene, Vagg, and Jacob’s (1983)
state anxiety scale at time 1, 2, and 3. Six items asked the extent (on a 4-point
Likert-type scale with 1 being Not at all and 4 being Very much) to which the subject
currently felt calm, tense, upset, relaxed, content, and worried. Larger numbers
indicated greater state anxiety. The reliabilities were .82 (T1), .89 (T2), and .79 (T3).
Brooding rumination

The extent to which the participants engaged in brooding rumination about the
problem at time 1 and 3 was measured with items taken from the reflective
ruminative response scale (RRS; Nolen-Hoeksema, 1987). The subjects were asked
three items along a 5-point Likert-type scale (1 Strongly agree to 5 Strongly disagree)
how much they mulled about the problem and their management of it since the
conversation ended (e.g., ‘‘Mulled over the problem and thought why do I have
problems other people don’t have.’’). These items were reworded to reflect brooding
about the problem in general for the T1 assessment. The reliabilities were .77 (T1)
and .85 (T3). These items were also revised to measure participants’ brooding about
the support the friend provided in the conversation (T3) (e.g., ‘‘Thought over and
over about how my friend responded to my problem and wondered ‘Why does my
friend always react this way?’’’) (α = .80). Larger numbers indicate greater brooding.
Verbal rumination

A modified version of Rose’s (2002) corumination scale was used to measure the
subjects’ perspective of how much they verbally ruminated during the interaction
(T2). Fourteen items from the original scale that were relevant to the subject’s
verbal rumination to the friend, rather than corumination, were used. The items
ranged from 1, Not at all true, to 5, Very true, and focused on the following: (1)
frequency of discussing the problem, (2) discussing the same problem repeatedly, (3)
speculation about parts of the problem that were not understood, and (4) focusing on
negative feelings. These items were averaged to create one score, with larger numbers
indicating greater verbal rumination (α = .92).
Realism

The subjects were asked about the typicality of the conversation that transpired with
two items on a scale that ranged from 1 to 7, with higher numbers representing
404

Human Communication Research 39 (2013) 395–421  2013 International Communication Association

T. Afifi et al.

Verbal Rumination and Social Support

more normal and typical. They found the interaction to be rather normal (M = 4.28,
SD = 2.04, α = .80).
Results
Manipulation checks

The means, standard deviations, and correlations for the variables are provided in
Table 1. On average, the participants reported moderate levels of anxiety, relatively
high levels of relationship satisfaction, and low levels of brooding rumination and
verbal rumination. As a manipulation check of the support conditions, participants
completed 21 items that were adapted from Xu and Burleson’s (2001) received
support scale and Ellis’ (2002) perceived parental confirmation scale immediately
after the conversation (T2). The items measured high and low person-centered
messages and confirming and disconfirming messages, which ranged from 1, Strongly
disagree, to 7, Strongly agree (e.g., ‘‘Expressed understanding of the situation that was
bothering me,’’ ‘‘Discounted my feelings,’’ α = .90).
These items were then adapted to allow for observer ratings of the support messages. Coders were trained on the measure until satisfactory reliability was achieved
(intercoder reliability = .95). Separate analyses of variance comparing positive and
negative support conditions on both the self-report measure and the coder ratings
confirmed the success of the manipulation: Participants in the good-support condition perceived their friends as providing better support (M = 6.20, SD = .64) than the
participants in the poor-support condition (M = 3.90, SD = 1.14, F (1, 230) = 250.01,
p < .001, η2 = .37), and coders rated the confederates in the good-support condition
as providing better support (M = 6.75, SD = .68) than those in the poor-support
condition (M = 2.56, SD = 1.32, F(1, 230) = 762.05, p < .001, η2 = .79).
Evidence of the manipulation’s success also comes from statistically significant
differences across the two conditions along several dimensions (see Table 2). Specifically, the participants in the good-support condition reported being more satisfied (at
T2 and T3), less anxious (at T2 and T3), and having less brooding rumination (T3)
after the conversation. Participants in the good-support condition did not report
verbally ruminating more during the discussion than those in the poor-support
condition. There were also no significant differences in verbal rumination for males
and females or for any combination of the friendship dyads.
Preliminary analyses

The subjects’ reports of interaction realism, their ratings of closeness with the friend,
and the dyad’s sex composition were included as control variables in all of the
regression analyses. In addition, the score for the relevant dependent variable at Time
1 (prior to the interaction) was included in the first step of all of the regressions
along with the other covariates. Because of the complexity of fully accounting
for all four levels of the dyadic sex compositions (i.e., sex of confederate × sex of
subject) and the primary interest in the corumination literature on female–female
Human Communication Research 39 (2013) 395–421  2013 International Communication Association

405

Variable

a

SD

1

2.26 .78



2.17 .84 .31***

2


.60
.71
.73
.84
1.12
1.07
1.17
.77

–.01
.12a
.18**
.11a
.04
–.01
.02
.17**
.25***

— — –.04
4.28 2.04 .14*

4

5

6

7

8

9

10

11

12


–.08

–.35*** .61***

–.41*** .63*** .84***

–.03
.66*** .42*** .44*** —
.44*** –.07
–.47*** –.52*** .09



13 14



1.78 .83 .36*** .42***

1.77
1.88
1.76
6.04
5.69
5.79
5.41
1.65

3



–.05
–.15*
***
.27
.25***
***
.22
.37***
*
.17
.33***
–.06
.01
–.12
–.08
–.17
–.10
–.07
.10
.22**
.42***
.002
–.04

–.02
–.06


–.12a
–.43***
–.30***
–.02
.37***
.33***
–.08
–.48***
.07
.52


.50***

.42*** .63***
–.08
–.12a
–.22** –.49***
–.19** –.38***
–.11a
.01
.16*
.46***
.02
–.08

–.02
–.04
–.35*** –.20**

–.33**
.01

–.21** –.15*
.31*** .19*

–.31*** –.04

.02 –.23** .08 —