《社会心理学》课程教学资源（文献资料）social self——Self-Talk as a Regulatory Mechanism - How You Do It Matters

4 Vot 0 N 2109700173 Self-Talk as a Regulatory Mechanism:How You Do It Matters Jiyoung Park,Aleah Burson,Adrienne Dougherty. HolxshablackamgRaBeomncr Mic niveraty Uaen把handky i They demonsraed tht )h a m () e the .ruminaiosef-reuaionmtoution didn' F decision-making process in an interview noting."One thing that latter.Specifically.we hypothesize that using one's own name and Harnessing Language to Promote Self-Regulation berkeley.edu 304

Self-Talk as a Regulatory Mechanism: How You Do It Matters Ethan Kross University of Michigan, Ann Arbor Emma Bruehlman-Senecal University of California, Berkeley Jiyoung Park, Aleah Burson, Adrienne Dougherty, Holly Shablack, and Ryan Bremner University of Michigan, Ann Arbor Jason Moser Michigan State University Ozlem Ayduk University of California, Berkeley Does the language people use to refer to the self during introspection influence how they think, feel, and behave under social stress? If so, do these effects extend to socially anxious people who are particularly vulnerable to such stress? Seven studies explored these questions (total N
585). Studies 1a and 1b were proof-of-principle studies. They demonstrated that using non-first-person pronouns and one’s own name (rather than first-person pronouns) during introspection enhances self-distancing. Studies 2 and 3 examined the implications of these different types of self-talk for regulating stress surrounding making good first impressions (Study 2) and public speaking (Study 3). Compared with the first-person group, the non-first-person group performed better according to objective raters in both studies. They also displayed less distress (Studies 2 and 3) and engaged in less maladaptive postevent processing (Study 3). Studies 4 and 5 examined how these different forms of self-talk influence the way people appraise social-anxiety-provoking events. They demonstrated that non-first-person language use (compared with first-person language use) leads people to appraise future stressors in more challenging and less threatening terms. Finally, a meta-analysis (Study 6) indicated that none of these findings were moderated by trait social anxiety, highlighting their translational potential. Together, these findings demonstrate that small shifts in the language people use to refer to the self during introspection consequentially influence their ability to regulate their thoughts, feelings, and behavior under social stress, even for vulnerable individuals. Keywords: distancing, stress, social anxiety, rumination, self-regulation, emotion regulation During the summer of 2010, LeBron James, a future Hall-of￾Fame basketball player, faced a tough decision. Should he stay with the small market team that nurtured his career from its inception, or move to a larger city market? Shortly after making his choice (he joined the larger market team), he described his decision-making process in an interview noting, “One thing I didn’t want to do was make an emotional decision. I wanted to do what’s best for LeBron James and to do what makes LeBron James happy” (Greenberg, 2010). Notice how James begins by referring to himself using the pronoun I, but then quickly switches to using his own name after indicating that he does not want to make an emotional decision. Does this shift from I to James represent a mere quirk of speech? Or could it represent something more—a process, for example, that consequentially influences people’s capacity to control their thoughts, feelings, and behavior? Here we suggest that it is the latter. Specifically, we hypothesize that using one’s own name and other non-first-person pronouns to refer to the self during intro￾spection is a form of self-distancing that enhances self-regulation.1 Harnessing Language to Promote Self-Regulation Several lines of research motivate this prediction. First, con￾verging evidence suggests that enhancing psychological distance, 1 People can focus on a variety of stimuli from a psychologically distanced perspective (e.g., the self, other people, other objects; Trope & Liberman, 2010). We use the term self-distance to refer to instances in which people focus specifically on the self from a distanced perspective. Ethan Kross, Psychology Department, University of Michigan, Ann Arbor; Emma Bruehlman-Senecal, Psychology Department, University of California, Berkeley; Jiyoung Park, Aleah Burson, Adrienne Dougherty, Holly Shablack, and Ryan Bremner, Psychology Department, University of Michigan, Ann Arbor; Jason Moser, Psychology Department, Michigan State University, East Lansing; Ozlem Ayduk, Psychology Department, University of California, Berkeley. We thank the many research assistants at Michigan and Berkeley for their assistance conducting the studies, and Vivian Zayas, Robin Edelstein, Phoebe Ellsworth, and Oscar Ybarra for their feedback. Correspondence concerning this article should be addressed to Ethan Kross, Psychology Department, University of Michigan, Ann Arbor, MI 48109, or to Ozlem Ayduk, Psychology Department, University of Cali￾fornia, Berkeley, CA 94720-1650. E-mail: ekross@umich.edu or ayduk@ berkeley.edu This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology © 2014 American Psychological Association 2014, Vol. 106, No. 2, 304 –324 0022-3514/14/$12.00 DOI: 10.1037/a0035173 304

SELF-TALK.REGULATION.AND SOCIAL ANXIETY 305 broadly as the capaciry to transeend one's Exploring Generalizability to Vulnerable Populations: Social Anxiety ample,that psych mode pcople's when fac e.Li Levin-Sa ,06 Kober et al atory with a high degree of ccoloica uin painful past from a sef mulated in the laboratory relatively easily.in ways that lend on their bo The conc experience social anxiety and the concomitant negative psycho Segal,Bu Kennedy.2007:Ingram Hollo 1.1986.Th Rapee Heimb 1997).When these situations are possible gor"elf as contet.is also central to mny newer forms of worry exces ognitive the their ot their fee ings (eg Hayes,Luoma Hcim sing o 2008 in the age people use to refer tivebehavioral therapie social anxiety i s ha ed that the la ents(D.M.Cark).However.to the extent tha nguage people use to ret to the I dur ectio ouns that peopl bo dista cope with social threat &P Research Overview le to retl on (Gr Kro with Ibwere proof-of-principlstudies. ing fewer irst-person pro hereon)rather than first-person p cinaIns eral.peopl of these different forms of self-talk for egulating the thinking about. s of self-talk self-distanc ghts and fee s they have privileged access to he uoh 5 to exa ine whether trait social anxiety moderates the In co people virtually exclusively use first person pron self-regulatory effects of non-first-person language use Thus Studies la and 1b ospection should think about the self as they normally do an n sum.our a nalysis suggests that the language people use to stancing -distan ncing.One way t ate their thou hts.f ngs,and behavior under stress.Our first ont that people adopt when visua goal was to explore this questio on-first-person

which we construed broadly as the capacity to transcend one’s egocentric viewpoint of a stimulus, facilitates self-regulation. Re￾search in children and adults indicates, for example, that psycho￾logical distancing strategies enhance people’s capacity to exert self-control when faced with tempting options in the short term (e.g., Fujita, Trope, Liberman, & Levin-Sagi, 2006; Kober et al., 2010; Mischel & Rodriguez, 1993; Sigel & McGillicuddy-De Lisi, 2003). In a similar vein, research on self-reflection indicates that cueing people to reflect on painful past experiences from a self￾distanced or “fly-on-the-wall” visual perspective helps them re￾flect on their experiences without ruminating (Gruber, Harvey, & Johnson, 2009; Kross & Ayduk, 2011; also see Ray, Wilhelm, & Gross, 2008; Wisco & Nolen-Hoeksema, 2011). The concept of self-distancing also factors prominently into several cognitive and behavioral treatment paradigms. Beck (1970) described distancing as a process that allows clients to think objectively about irrational thoughts and emphasized the impor￾tance of this process for effective cognitive therapy (pp. 189 – 190)—a view that many scholars have since echoed (e.g., Fresco, Segal, Buis, & Kennedy, 2007; Ingram & Hollon, 1986). The concept of distancing, although sometimes referred to as “decen￾tering” or “self as context,” is also central to many newer forms of cognitive therapy, which emphasize the importance of enhancing psychological distance from the self for allowing people to observe and accept their feelings (e.g., Hayes, Luoma, Bond, Masuda, & Lillis, 2006; Teasdale et al., 2002). These findings are noteworthy in the current context because researchers have speculated that the language people use to refer to the self may influence self-distancing. For example, research on expressive writing conceptualizes the degree of first-person pro￾nouns that people use when writing about emotional experiences as a marker of self-distancing—the fewer first-person pronouns people use, the more people attempt to distance themselves from their experiences (e.g., Cohn, Mehl, & Pennebaker, 2004; also see Pennebaker & King, 1999). Supporting this inference, cueing people to reflect on emotional (Grossmann & Kross, 2010; Kross & Ayduk, 2008) and nonemotional (Mcisaac & Eich, 2002) expe￾riences from the visual perspective of a psychologically distanced observer (compared with a first-person visual perspective) leads them to describe these experiences using fewer first-person pro￾nouns. Why might certain patterns of language use, in particular peo￾ple’s use of non-first-person pronouns and their own name to refer to the self, promote self-distancing? In general, people use these parts of speech when thinking about, referencing, or speaking to other people. Thus, if people use these parts of speech to refer to the self, this may enhance self-distancing by leading people to think about themselves as though they were someone else—albeit another self whose inner thoughts and feelings they have privileged access to. In contrast, people virtually exclusively use first person pronouns when thinking about or referring to the self from their egocentric point of view. Thus, people who use these parts of speech during introspection should think about the self as they normally do— from a self-immersed, first-person perspective. In sum, our analysis suggests that the language people use to refer to the self during introspection may influence self-distancing, and thus have consequential implications for their ability to regu￾late their thoughts, feelings, and behavior under stress. Our first goal was to explore this question. Exploring Generalizability to Vulnerable Populations: Social Anxiety Our second goal was to explore whether social anxiety moder￾ates the self-regulatory benefits of non-first-person language use during introspection. We focused on social anxiety because it can be studied in the laboratory with a high degree of ecological validity. That is, the situations that socially anxious people fear most— being evaluated by others in a social context— can be simulated in the laboratory relatively easily, in ways that lend themselves well to testing predictions about mechanisms underly￾ing self-regulation. Moreover, social anxiety is one of the most common forms of anxiety (e.g., Kessler, Berglund, Demler, Jin, & Walters, 2005). From the apprehensive partygoer to the anxious public speaker to the socially phobic patient, countless people experience social anxiety and the concomitant negative psycho￾logical and physical consequences that it generates. Cognitive models of social anxiety suggest that socially anxious individuals experience high levels of anxiety in response to the threat of future social interactions (D. M. Clark & Wells, 1995; Rapee & Heimberg, 1997). When these situations are possible, they worry excessively about the likelihood of performing poorly, which interferes with their performance. The perception and actu￾ality of their poor performance reinforce their negative beliefs about the self, giving rise to maladaptive postevent processing or rumination (Brozovich & Heimberg, 2008). In current cognitive behavioral therapies, social anxiety is treated by challenging negative expectations about upcoming events (e.g., D. M. Clark et al., 2003). However, to the extent that the language people use to refer to the self during introspection influences self-regulation, it might provide an additional (poten￾tially easily implemented) tool for helping vulnerable populations cope with social threat. Research Overview Seven studies examined whether language use during introspec￾tion influences people’s ability to self-regulate under social stress. Studies 1a and 1b were proof-of-principle studies. They examined whether using non-first-person pronouns and one’s own name (non-first-person group from hereon) rather than first-person pro￾nouns (first-person group from hereon) during introspection en￾hances self-distancing. Studies 2 and 3 then examined the impli￾cations of these different forms of self-talk for regulating the cognitive, emotional, and behavioral sequalae of social stress. Studies 4 and 5 examined how these different forms of self-talk influence the way people appraise future stressors. Finally, Study 6 consisted of a meta-analysis that combined data from Studies 2 through 5 to examine whether trait social anxiety moderates the self-regulatory effects of non-first-person language use. Studies 1a and 1b Our starting point is that using non-first-person pronouns and one’s own name to refer to the self during introspection promotes self-distancing. One way to test this prediction is to examine whether language use during self-reflection influences the vantage point that people adopt when visualizing emotional experiences. If non-first-person language use (i.e., “linguistic” self-distancing) This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. SELF-TALK, REGULATION, AND SOCIAL ANXIETY 305

06 KROSS ET AL anipula二28S时1hn-4 rod One of the things we're int sted in in this study is the b which suggests that sby thinking abo ve rson pro ould like We tested this prediction by asking pa icipants to analyze their unde an anger-provo ng (S y lb) or non-first-person pronounsand their ow Participants in the non-first-person group were told: Method do.ryd why you telt th 56 .SD. and rea s for Jane's feclings 05 If-distan rch(Mischowski White.129%Asian American.2.2%African American.and 5.5 man,2012;Park et al,2013).part tems to e re materials After providing informed conser partici they a ants n Study Ia:M=635.:suudy 7 1b:M=6.74,SD=1.46 ond.o ay from the scen ence: 1b:a=85.M=3.13.SD=1.66 we had another perso tion manipulation check at theend of ly to d a aly take the time read th k abo were then preser shopping" k c-m hem n16 24.72).Aside fr recal Results and Discussion n Study Ib.we n group and seven i ants in Study Ib 28-Smds Ib:n-49)or non-first-person pronouns and their own name Jane was replaced with Joe for male participants

enhances self-distancing, then people who use this type of lan￾guage during introspection should be more likely to report seeing themselves in their past experience from an observer’s visual perspective (i.e., “visual” self-distancing) than people who use first-person pronouns during introspection. This logic is broadly consistent with construal level theory, which suggests that poten￾tiating psychological distance in one domain enhances psycholog￾ical distance in other domains (Trope & Liberman, 2003, 2010). We tested this prediction by asking participants to analyze their feelings surrounding an anger-provoking (Study 1a) and anxiety￾provoking (Study 1b) negative autobiographical experience using first-person pronouns or non-first-person pronouns and their own name. We focused on two types of negative emotional experiences in these studies to examine the reliability and generalizability of our findings. After participants analyzed their feelings, we asked them to indicate the degree to which they adopted the visual perspective of an observer as they reflected on their feelings. Method Participants. In Study 1a, participants were 56 undergradu￾ates (Mage
18.95 years, SDage
4.74; 38 females; 52% White, 23% Asian American, 11% African American, and 14% other) who received course credit for their participation. In Study 1b, 93 participants were recruited through Amazon Mechanical Turk (MTurk; Mage
32.23 years, SDage
12.98; 50 females; 79.6% White, 12.9% Asian American, 2.2% African American, and 5.5% other). Procedure and materials. Baseline affect. After providing informed consent, partici￾pants in both studies rated how they felt “right now” (1
very negative, 9
very positive; Study 1a: M
6.35, SD
1.17; Study 1b: M
6.74, SD
1.46). Negative experience recall task. Next, we asked participants to recall for Study 1a an anger-related autobiographical experi￾ence: No matter how well two people get along there are times when they experience conflict . . . [T]ake a few moments right now to recall a time when you experienced such conflict with another person—a time when you became truly enraged at this person. For Study 1b, we asked participants to recall an anxiety-related autobiographical experience: No matter how satisfied people are with their lives, there are times when they worry and experience anxiety . . . [T]ake a few moments right now to think about a time from your past when you worried about something happening to you. They were given as much time as they needed to recall their experience (Study 1a: Msec
37.18, SDsec
19.36; Study 1b: Msec
45.16, SDsec
24.72). Aside from cueing people to recall different types of experiences, the only other difference between the two sets of recall instructions was that in Study 1b, we instructed participants to “close their eyes” before the memory recall instructions, whereas participants in Study 1a did not receive this instruction. Experimental manipulation. Participants were then randomly assigned to reflect on their feelings surrounding their recalled experience using first-person pronouns (Study 1a: n
28; Study 1b: n
49) or non-first-person pronouns and their own name (Study 1a: n
28; Study 1b: n
44). The instructions used to manipulate linguistic self-distancing were identical in both studies. Participants in the first person group were told: One of the things we’re interested in in this study is the language people use to understand their feelings. Some people try to understand their feelings by thinking about themselves using first-person pro￾nouns, so this is what we would like you to do. Please try to understand why you felt the way you did in the experience you just recalled using the pronouns “I” and “my” as much as possible. In other words, ask yourself, “Why did I feel this way? What were the underlying causes and reasons for my feelings?” Participants in the non-first-person group were told: One of the things we’re interested in in this study is the language people use to understand their feelings. Some people try to understand their feelings by thinking about themselves using their own name and other non-first-person pronouns, so this is what we would like you to do. Please try to understand why you felt the way you did in the experience you just recalled using the pronoun “you” and “[your own name]” as much as possible. In other words, if your name was Jane, you would ask yourself, “Why did Jane feel this way? What were the underlying causes and reasons for Jane’s feelings?”2 Participants had 60 s to reflect on their feelings following these instructions. Visual self-distance. Following prior research (Mischowski, Kross, & Bushman, 2012; Park et al., 2013), participants com￾pleted two items to measure visual self-distancing after the reflec￾tion period was over. First, participants rated the extent to which they saw the event replay through their own eyes versus watched the event unfold as an observer as they analyzed their feelings (1
predominantly immersed participant, 7
predominantly dis￾tanced observer). Second, they rated how far away from the scene they were as they analyzed their feelings (1
very close, saw it through my own eyes; 7
very far, saw it as if an observer). Ratings on these items were averaged to create a visual self￾distancing index, Study 1a:
.74, M
2.72, SD
1.40; Study 1b:
.85, M
3.13, SD
1.66. Instructional manipulation check. Because we had less con￾trol over the online environment in which Study 1b was performed, we administered an instructional manipulation check at the end of the study to ensure that participants paid attention to our instruc￾tions. Specifically, participants read, “We are interested in whether you actually take the time to read the directions . . . [I]n order to demonstrate that you have read these instructions, please select ‘online shopping’ and ‘check e-mail only’ below.” Participants were then presented with a series of Internet activity choices with “online shopping” and “check e-mail only” embedded among them. Results and Discussion Preliminary analyses. Twenty-one participants failed the in￾structional manipulation check: 14 in the first person group and seven in the non-first-person group. They were excluded from subsequent analyses on a priori grounds, leaving 72 participants in Study 1b, 35 in the first person group, and 37 in the non-first-person group. 2 Jane was replaced with Joe for male participants. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 306 KROSS ET AL.

SELF-TALK.REGULATION.AND SOCIAL ANXIETY 307 0).Exclusionsdid by Study 0.06 Method Study la;two in Study Ib)had missing values because of omitted Visual self-distaneine. As expected.participants in the non antly higher levels of vs only in this study beca they are disproportionatel 4.4,p Studies 2 and 3:Overview 41)befor language to refer e of me"and"Sometimes I think I am too concerned with 23) hey had similarproc of this process.participants that they would be havin M 230D=096 Study 2 ositive first impr other's perfor d b were I ha d h ning so St-pesol ■First Person■Non First Perso 4.5 was to examine how selves 38 al int tions wer n a roon menter retrieved the participant's condition assignment from an- ulation After the 2-min wait 1.0 Study self-prepara on has on performanc Fiure.The effec of condition on visual self-dista rd errors

This exclusion rate (23%) is consistent with prior studies that have used instructional manipulation checks (Oppenheimer, Meyvis, & Davidenko, 2009). Exclusions did not differ by condition, Study 1a: 2 (1)
0.06, p
.80; Study 1b: 2 (1)
2.13, p
.15. All analyses controlled for baseline affect, which the groups did not differ on, Fs  1. Gender was not related to the dependent variable in either study, Fs  1, and controlling for it did not substantively alter any of the results. Three participants (one in Study 1a; two in Study 1b) had missing values because of omitted responses. Visual self-distancing. As expected, participants in the non￾first-person groups displayed significantly higher levels of visual self-distancing than participants in the first-person groups in both Study 1a, F(1, 52)
5.96, p
.018, p 2
.103, and Study 1b, F(1, 67)
4.45, p
.039, p 2
.062 (see Figure 1). These findings demonstrate that using non-first-person pronouns and one’s own name to refer to the self enhances self-distancing. Studies 2 and 3: Overview Having established that using non-first-person language to refer to the self during introspection increases self-distancing, we ex￾amined in Studies 2 and 3 whether language use during introspec￾tion influences self-regulation under social stress. Both Studies 2 and 3 focused on these same core issues but were designed by different laboratories and implemented independently. As a result, they had similar procedures and focused on a partially overlapping set of dependent variables. Thus, they provide convergent evi￾dence for the role that language use plays in self-regulation across different situations, laboratories, and measures.3 Study 2 We recruited participants for a study on the psychology of first impressions. They were told that they would be asked to make a positive first impression on another person. Prior to engaging in this task, we asked them to reflect on their feelings concerning their upcoming social interaction using either first-person pro￾nouns or non-first-person pronouns and their own name. We then examined the effect of this manipulation on participants’ anxiety and performance during the social interaction. We predicted that participants in the non-first-person group would experience less anxiety and perform better during the impression formation task. Method Participants. Participants were 97 undergraduate women (Mage
20.05 years, SDage
1.86; 26.8% White, 49.5% Asian American, 5.1% African American, and 18.6% other) who re￾ceived course credit or payment for participating. We focused on women only in this study because they are disproportionately affected by anxiety-related problems, including social anxiety (Kessler et al., 1994; Spurr & Stopa, 2002; Weinstock, 1999). Procedure and materials. Phase 1: Trait social anxiety. Participants completed the 12- item Brief Fear of Negative Evaluation Scale (BFNE; Leary, 1983) online approximately 5 days (M
5.19 days, SD
3.41) before the study. Sample items include, “I am afraid that others will not approve of me” and “Sometimes I think I am too concerned with what other people think of me” (1
not at all characteristic of me, 5
extremely characteristic of me;
.80; M
37.03, SD
9.23). Phase 2: Cover story and premanipulation anxiety. Informed consent was obtained at the beginning of the experiment. As part of this process, we informed participants that they would be having a videotaped conversation with another participant. Next, partici￾pants rated how anxious they felt (1
not at all, 5
extremely; M
2.30, SD
0.96). Phase 3: Explanation of study. An experimenter told partic￾ipants that the study focused on impression formation. Specifi￾cally, they were told that their goal would be to make a good first impression during a short conversation with an opposite sex par￾ticipant after which they would be evaluating each other’s perfor￾mance. In addition, they were informed that their conversation would be videotaped and assessed by trained psychologists who would be rating their social skills. Such “getting to know you” tasks are commonly used to elicit social anxiety in the laboratory (e.g., J. V. Clark & Arkowitz, 1975; Glass, Merluzzi, Biever, & Larsen, 1982; Turner, Beidel, & Larkin, 1986). Next, the experimenter added that an additional goal of the study was to examine how people can effectively prepare themselves to make good first impressions. Thus, they were told that they would be receiving instructions about how to prepare themselves for their social interaction in a few moments. After these instructions were conveyed, the experimenter left the participant alone in a room without any distractions for 2 min. During this time, the experi￾menter retrieved the participant’s condition assignment from an￾other room. Phase 4: Experimental manipulation. After the 2-min wait￾ing period, the experimenter told participants, “We are interested in the different ways people go about preparing themselves psy￾chologically for meeting new people and what effect each type of self-preparation has on performance.” They were then randomly 3 In Studies 2 and 3, a number of additional measures were included for exploratory purposes. They are available upon request from the corre￾sponding authors. Figure 1. The effect of condition on visual self-distancing in Study 1a and Study 1b measured on scales ranging from 1 (predominantly immersed participant) to 7 (predominantly distanced observer). Error bars indicate standard errors. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. SELF-TALK, REGULATION, AND SOCIAL ANXIETY 307

308 KROSS ET AL asked to think about their feelings conce ning an n of t Following the experiment ma articipants answe he following ques do you gth re er the p ngpant spoke and feeln about the overly brie f nor long-v 09 created a measure of overall performand by sum Pha valuation task Subs quently.the 205 scores on these dimensions (a =.80:A tion partner.whowasin fact a confederate. analyses indicated that scores on the nerv Glass et al.,1982;Kashdan&Rob 2006)the confederate i 4 Phase 7:Po . Par The in action be when the new sat nally,participants were for Results explo whether the manipu analyses.Four participants in each condition debriefing,and four because of protoc s (eo the confed did not follo he room here today Thres sation during the first 30s did not differ by When the experimenter retumed,he or cted,related to several of th (participan aven't". What is F.0.00.73. 3.80h 0.0:o- 3.80 and dis ed on the san ample tests indicated the 中aa on a on two dimensions.First,they rated how nervous the participant

assigned to a first-person (n
48) or non-first-person (n
49) condition using instructions that were virtually identical to those used in Study 1. The main difference was that participants were asked to think about their feelings concerning an upcoming anxiety-provoking event (rather than a past experience). Following these instructions, participants reflected over their thoughts and feelings alone for 3 min. Phase 5: Manipulation check and preinteraction anxiety. Following the experimental manipulation, participants answered the following questions, “How well do you feel you followed the speech preparation instructions” (1
not at all well, 5
perfectly well) and “To what extent did you use the first-person pronouns I and me (or non-first-person pronouns you and your own name) to refer to yourself when you were working through your thoughts and feelings about the upcoming conversation (1
not at all, 5
exclusively). Scores on these questions were averaged to create a manipulation check index (
.68; M
3.79, SD
0.70). Next, participants rated their anxiety again, using the same question that was administered prior to the manipulation (M
2.84, SD
0.97). Phase 6: Social evaluation task. Subsequently, the experi￾menter introduced the participant to a new experimenter and their interaction partner, who was in fact a confederate. Prior research indicates that opposite sex interactions are more anxiety provoking than same sex interactions, especially for socially anxious people (e.g., Turner et al., 1986). Therefore, following prior research (e.g., Burgio, Merluzzi, & Pryor, 1986; J. Clark & Arkowitz, 1975; Glass et al., 1982; Kashdan & Roberts, 2006), the confederate in this study was always male. Both the new experimenter and confederate were blind to condition. Unstructured interaction. The interaction began when the new experimenter and confederate entered the room and sat across from the participant. The experimenter told the participant that he or she had to leave the room to calibrate the cameras and would return shortly. The participant was given no instructions about how to interact with the confederate during this period. We included this unstructured interaction period to explore whether the manipula￾tion influenced participants’ tendencies to spontaneously initiate conversation with their partner. Confederates were instructed not to initiate conversation for the first 30 s. After an initial silence of 30 s, the confederate initiated contact by saying, “It wasn’t easy to find the room here today.” Three condition-blind raters coded whether participants initiated conversation during the first 30 s (intraclass correlation [ICC]
1). Structured interaction. When the experimenter returned, he or she presented the participant and the confederate with instructions for the social interaction. Following the procedure used by Me￾leshko and Alden (1993), participants were given a list of topics to talk about (e.g., “What have you always wanted to try but haven’t?”; “What is your biggest pet peeve?”). The experimenter instructed them to take turns selecting and answering questions from this list until they had both answered four questions. The confederate always began the exchange and disclosed on the same four questions with all participants. Confederates’ responses were scripted so that the content and delivery of their disclosures was consistent. The experimenter left the room prior to the initiation of the conversation and returned after the conversation was over. Three judges rated participants’ performance during this phase on two dimensions. First, they rated how nervous the participant was (1
below average level of nervousness, 5
above average level of nervousness; ICC
.73; M
2.87, SD
0.78). Second, they rated participants’ overall performance using a modified version of the Social Performance Rating Scale (Fydrich, Chamb￾less, Perry, Buergener, & Beazley, 1998)—a behavioral assess￾ment of social performance designed for social phobia. Specifi￾cally, judges coded participants’ behavior along the following four dimensions using a 5-point scale (1
poor, 5
excellent): gaze, vocal quality, speech length, and discomfort (ICC
.73).4 Gaze refers to whether the participant made appropriate eye contact with the confederate. Vocal quality refers to whether the participant spoke clearly and varied her vocal tone in an engaging manner. Length refers to whether the participant spoke for an appropriate amount of time when answering the questions (i.e., neither being overly brief nor long-winded) and lacked awkward pauses. Dis￾comfort refers to whether the participant showed verbal (e.g., speech dysfluencies) and nonverbal (e.g., self-manipulative behav￾iors like hair twirling, facial touching, and so on) indicators of anxiety during the interaction. Following Fydrich et al. (1998), we created a measure of overall performance by summing partici￾pants’ scores on these dimensions (
.80; M
13.10, SD
2.05). Preliminary analyses indicated that scores on the nervousness and overall performance indexes were highly correlated. There￾fore, we collapsed them to create a composite behavioral index of social interaction performance after reverse scoring the nervous￾ness scale and then standardizing scores on each measure (
.84). Phase 7: Postinteraction anxiety. Participants rated their anx￾iety at the end of the interaction using the same question they completed earlier (M
1.81, SD
0.92). Phase 8: Debriefing. Finally, participants were debriefed for suspicion and compensated. Results Preliminary analyses. Four participants in each condition were excluded—four because they suspected that their partner was a confederate and/or inferred the study aims during the funneled debriefing, and four because of protocol errors (e.g., the confed￾erate did not follow protocol)—leaving 89 participants, 44 in the first-person group and 45 in the non-first-person group. Exclusions did not differ by condition, 2 (1)
0.00, p
.98. The groups did not differ on premanipulation anxiety, t(86)
0.54, p
.590, or trait social anxiety, t(87)
0.61, p
.541. Both of these variables were, as expected, related to several of the dependent variables (see Table 1). Therefore, they were included as covariates. Degrees of freedom vary slightly across analyses due to missing data. Manipulation check. There was no effect of condition on the manipulation check, F(1, 84)
0.00, p
.973, p 2
.000 (first-person: M
3.80, SE
0.10; non-first-person: M
3.80, SE
0.10), indicating that both groups followed the instructions equally well. It should be noted that one-sample t tests indicated that the mean manipulation check score for each group was sig￾nificantly greater than the midpoint of the manipulation check 4 We did not include the conversational flow category because the structured nature of the interaction did not allow us to code this dimension. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 308 KROSS ET AL.

SELF-TALK.REGULATION.AND SOCIAL ANXIETY 309 1 2 3 4 cial anxiety ety (T) sD5s.“ps01“pss 045..thatt wasnot the case that people felt repcated-n analysis of covariance (ANCOVA) stat was the within-parica 348.P=035 stand th coCspnomsoainNeson,ndtoclae en-participants var ndtion 02,butasiemificantasociaionbetweentrmits0cial interaction.w performed univariate ANCOVAs on the change in nent F(l 86)565n=020n 062 The latter resul t(postintcraction the in prein con 10n on wa These tests dem ated that condition did not inf Behavior. We performed log istic egression on initiating con m t 012 hase of the soci xiety from the F( 2 the third anxiety asses en-subjects predictor and trait and premanipu be latte sult indicated that particina s in the edicted co Ps>13.However.this likeh eroup displayed a eiling effect as84%of participants initiated conversation with- Panel日 First Person--Non First Persor 3.4 0.4 06 First Person Non First

scale: first-person: t(43)
13.95, p  .001, d
4.25; and non-first-person: t(44)
12.14, p  .001, d
3.66, indicating that both groups also implemented the instructions successfully. Anxiety. We examined the effect of condition on state anxiety by performing a repeated-measures analysis of covariance (ANCOVA). Time (t) of state anxiety assessment was the within-participant variable (three: premanipulation [t1] vs. preinteraction [t2] vs. postinteraction [t3]), condition was the between-participants vari￾able (two: first-person vs. non-first-person), and trait social anxiety (continuous) was the covariate. This analysis revealed two effects. First, condition interacted marginally with time, F(2, 84)
2.93, p
.059, p 2
.065. To understand the meaning of this interaction, we performed univariate ANCOVAs on the change in anxiety from the (a) first to the second anxiety assessment (prein￾teraction anxiety [t2] MINUS premanipulation anxiety [t1]), and (b) the second to third anxiety assessment (postinteraction anxiety [t3] MINUS preinteraction anxiety [t2]) with condition as the between-subjects predictor and trait social anxiety as the covariate. These tests demonstrated that condition did not influence change in anxiety from the first to the second anxiety assessment, F(1, 85)
0.12, p
.726, p 2
.001, but did influence change in anxiety from the second to the third anxiety assessment, F(1, 86)
4.12, p
.045, p 2
.046. As Figure 2, Panel A illustrates, the latter result indicated that participants in the non-first-person group displayed a sharper decrease in anxiety after the social interaction than participants in the first-person group. This effect remained significant when performance during the structured phase of the interaction was controlled for, F(1, 83)
4.14, p
.045, p 2
.048, suggesting that it was not the case that people felt better simply because they performed better. Second, trait social anxiety interacted with time to predict state anxiety, F(2, 84)
3.48, p
.035, p 2
.077. To understand the meaning of this interaction, we conducted parallel analyses as described previously. These analyses revealed no relationship be￾tween trait social anxiety and change in state anxiety from the first to the second anxiety assessment, F(1, 85)
0.18, p
.670, p 2
.002, but a significant association between trait social anxiety and change in state anxiety from the second to the third anxiety assessment, F(1, 86)
5.65, p
.020, p 2
.062. The latter result demonstrated that the more apprehensive participants were about social evaluation, the more their anxiety levels declined following the interaction, pr
–.25, p
.020, possibly reflecting the relief socially anxious individuals experienced after the interaction was over. Behavior. We performed logistic regression on initiating con￾tact (yes
0 vs. no
1) with the confederate during the unstruc￾tured phase of the social interaction task with condition as the between-subjects predictor and trait social anxiety and premanipu￾lation anxiety as covariates. None of these variables significantly predicted contact initiation, ps  .13. However, this likely reflects a ceiling effect as 84% of participants initiated conversation with￾out prompting. Table 1 Study 2 Zero-Order Correlations Between Measured Variables Variable 1 2 3 4 5 6 1. Trait social anxiety — .18† .23
.04 .09 .05 2. Premanipulation anxiety (T1) — .33


.43


.16 .08 3. Preinteraction anxiety (T2) — .29

.04 .04 4. Postinteraction anxiety (T3) — .01 .04 5. Conversation initiation — .04 6. Performance — Note. T
Time. † p
.10.
p
.05.

p
.01.


p
.005. Figure 2. The effect of condition on anxiety over time (Panel A) and adaptive social interaction performance (Panel B) in Study 2. Error bars indicate standard errors. Time 1
t1. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. SELF-TALK, REGULATION, AND SOCIAL ANXIETY 309

310 KROSS ET AL adminis trait social FL.82=78 included measures of both global affect and shame.Shame is th ey discrete by public speech tasks(Dickern Summary and Discussion Finally.prior research indicates that people often persever orocessing (Brozovich&Hcimberg. 2008: also see Nole ndoalrsteneraedreekyknfnemciaghc event p self-talk manipula whether the manipulations this variable ospect ns and one's own name enhances cople's ability regu onditi ety Method son lan uage to reter to th tyetnmentespetionmyieadpepteo Procedure and mat Finally,the rimental manipulation influen We the pan hobia Inventory and performed their first-perso S L ehavioral implica "am bothered by blushing fron of other stered a the e ment (M 36 N were standard Study3 Phase 2:Baseline affect. w"(1 very negative elf thaaned en peoplc toYe We induced social stress using a nt social stress induction was used.the icipants deliver a public speech in front of an valuati y am erson .).the tod that they would hav )Building prior indicating that visual for their "dream"job t to pow 20121. using non-fiC and ne's owr edurineiniros Due to of affect that the two groups did not differ on.this measure was

Next, we performed an ANCOVA on judges’ ratings of perfor￾mance during the structured phase of the interaction with condi￾tion as the between-subjects predictor and trait social anxiety and premanipulation anxiety as covariates. Judges’ ratings indicated that the non-first-person group performed better on the social interaction task than the first-person group, F(1, 82)
7.18, p
.009, p 2
.081 (see Figure 2, Panel B). Neither premanipulation anxiety, F(1, 82)
0.51 p
.478, p 2
.006, nor trait social anxiety, F(1, 82)
0.08, p
.774, p 2
.001, predicted this variable. Summary and Discussion Study 2 examined whether the language people use to refer to the self during introspection influences how they feel and behave under social stress. It generated three key findings. First, both groups were equally capable of implementing the self-talk manipulations. This suggests that using non-first-person language during introspection is a feasible alternative to using first-person language. Second, contrary to our expectations, condition did not influ￾ence the increase in anxiety participants reported from the first to the second assessment. However, condition did predict changes in anxiety from the second to the third anxiety assessment, indicating that participants in the non-first-person group (compared with the first-person group) displayed a sharper decrease in anxiety from before to after the social interaction. This finding is particularly noteworthy because it suggests that using non-first-person lan￾guage to refer to the self during introspection may lead people to recover more quickly from social stressors. Finally, the experimental manipulation influenced participants’ behavior during the social interaction. Judges indicated that par￾ticipants in the non-first-person group were less nervous during the interaction and performed better than their first-person counter￾parts. These findings highlight the adaptive behavioral implica￾tions that using non-first-person pronouns and one’s own name during introspection have for people when they engage in tasks that elicit social stress. Study 3 Study 3 examined the implications of the same two types of self-talk that Study 2 focused on for allowing people to regulate their thoughts, feelings, and behavior under social evaluative stress. However, a different social stress induction was used, the focus was on different dependent variables, and men as well as women were included. In so doing, we aimed in Study 3 to extend the Study 2 findings in four ways. First, in Study 3, we examined whether the Study 2 results would generalize to an arguably more powerful social stress in￾duction—a public speech task rather than an impression formation task (Beazley, Glass, Chambless, & Arnkoff, 2001; Turner et al., 1986). Building on prior research indicating that visual self￾distancing manipulations generalize to powerful negative affect inductions (Mischowski et al., 2012), we predicted that the benefits associated with using non-first-person pronouns and one’s own name during introspection would generalize to this novel context. Second, although Study 2 included a premanipulation measure of affect that the two groups did not differ on, this measure was administered after participants were informed about the nature of the study. Thus, one could argue that the premanipulation anxiety measure used in Study 2 did not constitute a true baseline measure because anxiety was likely induced to a certain degree among all participants. Study 3 included a true baseline affect measure that was administered before participants learned about the nature of the study. Third, whereas Study 2 focused specifically on anxiety, Study 3 included measures of both global affect and shame. Shame is the key discrete emotion elicited by public speech tasks (Dickerson & Kemeny, 2004). Finally, prior research indicates that people often perseverate over their performance in social-anxiety-provoking tasks, which fuels social anxiety—a process that is often referred to as postevent processing (Brozovich & Heimberg, 2008; also see Nolen￾Hoeksema, Wisco, & Lyubomirsky, 2008). Therefore, Study 3 included multiple measures of postevent processing to examine whether the experimental manipulations influenced this variable. To the extent that reflecting on the self using non-first-person pronouns and one’s own name enhances people’s ability to regu￾late their social anxiety, we predicted that participants who en￾gaged in this process should also display less postevent processing. Method Participants. Participants were 89 undergraduates (60 fe￾males; Mage
19.01 years, SDage
1.04; 73.0% White, 12.4% Asian American, 6.7% African American, and 7.9% other) who received course credit or $20 for their participation. Procedure and materials. Phase 1: Trait social anxiety. We assessed trait social anxiety using two measures to enhance reliability: the BFNE (
.88; M
36.02, SD
8.13) and the 17-item Social Phobia Inventory (SPIN; Connor et al., 2000; scale: 0
not at all, 4
extremely;
.88; M
21.19, SD
9.91).5 Sample items for the SPIN include, “I am bothered by blushing in front of other people,” and “Parties and social events scare me.” Both measures were admin￾istered approximately 4 days before the experiment (Mdays
3.66, SDdays
1.76). Scores on the BFNE and SPIN were standardized and collapsed to form a single trait social anxiety index (
.69). Phase 2: Baseline affect. After providing informed consent, participants rated how they felt “right now” (1
very negative, 7
very positive; M
4.93, SD
1.07). Phase 3: Stress induction. We induced social stress using a modified version of the Trier Social Stress Task (TSST; Kirsch￾baum, Pirke, & Hellhammer, 1993). This task involves having participants deliver a public speech in front of an evaluative audience without receiving sufficient time to prepare; it is one of the most powerful ways of inducing stress in the laboratory among humans (Dickerson, Gruenewald, & Kemeny, 2004; Dickerson & Kemeny, 2004). Following established procedures (Kirschbaum et al., 1993), the experimenter told participants that they would have to give a speech on why they are qualified for their “dream” job to a panel of interviewers trained to evaluate speech performance. They were also told that their performance would be videotaped. 5 Due to a protocol error, SPIN Item 17 was not administered. We replaced this missing value with each participant’s mean rating of the first 16 items. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 310 KROSS ET AL.

SELF-TALK.REGULATION.AND SOCIAL ANXIETY 311 cewodn-blindatesyfor otes during this time nate over negative exp swith highe levels of re tin Gros zed as stat ents in which particinants alizations about their experience during the task (e.only st set up to n used in Study 2.The key differ- 79M=1.13,sD=134)and rece hat (a)in this study.participants were d wing prior res (C (e.g .20)-wcoilaped aero ot me rom reco ale th Phase 5:Speech task Next.the experimenter ret rned and lec M=0.63.SD Alder d in their 0).which w participants' and rated ther n or .to which they brooded over their sp coring the nervousnes As exp thes wo postevent processing measures wer foeach coder.Judge' debriefed at the end o ace index (a=86;M=3.39.SD 0.83 affeet and s oiced suspicio o allow m Results ext.they into the floo not f anxicty,87=0.70,p s no d on how the n anipulation influenced emo that fal ted to any of the c dent variabl vant to public on et al 2004).Re. ems arison n global affect and shame.we fir r room and asked them to sit quietly until they sistent with scores on the Sh ne scale.We then examined the &Zaldivar,2008).At the end of this 5-min period,postevent s non-first-pe on)and trait social anxiety as the covariate Thi gondionby,timcintcraction ehts that flowed th h their mind as theywaited for the person ved signifi experimenter to retur.Using a(not at all)to 4(completely)

They were then taken to a small room, which contained a desk and chair, and given 5 min to prepare. They were not permitted to take notes during this time. Phase 4: Experimental manipulation. After the 5-min prep￾aration period, participants were told: Besides preparing the content of a speech, people also need to prepare themselves psychologically before giving a speech, so we are inter￾ested in learning about the different ways people go about preparing themselves to give a speech, and what effect each type of self￾preparation has on performance. They were then randomly assigned to the first-person (n
45) or the non-first-person (n
44) condition using instructions that were virtually identical to those used in Study 2. The key differ￾ences were that (a) in this study, participants were directed to analyze their emotions before giving a speech (rather than before engaging in a social interaction task) and (b) “he or she” was given as an example of the type of non-first-person pronoun participants in the non-first-person condition could use in addition to “you.’ Phase 5: Speech task. Next, the experimenter returned and led the participant to another room where they delivered their 5-min speech to two confederates posing as evaluators. A video camera positioned in their field of vision recorded their performance. Phase 6: Speech performance. Two condition-blind coders watched videotapes of participants’ speeches and rated them on three dimensions: confidence, nervousness, and overall perfor￾mance (1
below average, 5
above average). After reverse scoring the nervousness scores, coders’ ratings were found to be consistent across these dimensions (Rater 1: ICC
.75; Rater 2: ICC
.78). Therefore, we collapsed across all three dimensions to create a single speech performance index for each coder. Judge’s ratings on these indexes were collapsed to form a single speech performance index (
.86; M
3.39, SD
0.83). Phase 7: Postspeech global affect and shame. After partici￾pants delivered their speeches, they again rated how they felt “right now” to allow measurement of global affect (1
very negative, 7
very positive; M
4.72, SD
1.16). Next, they completed 10 items that compose the Shame and Pride subscales (e.g., “I want to sink into the floor and disappear,” “I feel proud”; 1
not feeling this way at all, 5
feeling this way very strongly) of the State Shame and Guilt Scale (Marschall, Sanftner, & Tangney, 1994). We focused on how the manipulation influenced emotions that fall on the shame–pride dimension because they are particularly rele￾vant to public speech challenges (Dickerson et al., 2004). Re￾sponses to these items were averaged after reverse scoring the pride ratings, so that higher scores reflected greater shame (
.90; M
20.25, SD
7.16). Phase 8: Postevent processing. Next, the experimenters in￾formed participants that they had to set up the next phase of the study in another room and asked them to sit quietly until they returned in 5 min. This provided participants with an opportunity to ruminate over their speech performance (for a similar approach to assessing rumination, see Ayduk & Kross, 2008; Gerin, David￾son, Christenfeld, Goyal, & Schwartz, 2006; Zoccola, Dickerson, & Zaldivar, 2008). At the end of this 5-min period, postevent processing was assessed in two ways. First, we asked participants to describe in writing the stream of thoughts that flowed through their mind as they waited for the experimenter to return. Usinga0(not at all) to 4 (completely) scale, two condition-blind raters coded these essays for recounting and reconstruing. Prior research has linked the tendency to rumi￾nate over negative experiences with higher levels of recounting and lower levels of reconstruing (Ayduk & Kross, 2010b; Gross￾mann & Kross, 2010; Kross, Ayduk, & Mischel, 2005). Recount￾ing was operationalized as statements in which participants re￾hearsed the specific chain of events and emotions that they experienced during the speech task (e.g., “I was feeling nervous and fidgeted a lot while I was speaking”). Reconstruing was operationalized as statements in which participants described re￾alizations about their experience during the task (e.g., “I was only given 5 min to prepare my speech and was thus almost set up to not do well”). Interrater reliability was high for both recounting (ICC
.79; M
1.13, SD
1.34) and reconstruing (ICC
.82; M
0.49, SD
0.81). Following prior research (e.g., Ayduk & Kross, 2010b; Kross & Ayduk, 2008; Kross, Duckworth, Ayduk, Tsukayama, & Mischel, 2011), we collapsed across both measures by subtracting reconstruing scores from recounting scores to create a single thought content index. Higher scores reflected a greater tendency to recount versus reconstrue (M
0.63, SD
1.21). Second, after participants wrote their essay, they completed the five-item Rumination Questionnaire (RQ; Mellings & Alden, 2000), which we modified to apply to a speech task. Specifically, participants rated (1
not at all, 7
a lot) five items that assessed the degree to which they brooded over their speech task perfor￾mance (e.g., “To what extent did you criticize yourself about not handling the speech task well?”;
.70; M
3.65, SD
1.11). As expected, these two postevent processing measures were closely related (
.71). Therefore, we created a single index of maladaptive postevent processing by collapsing the data after standardizing scores on each measure. Phase 9: Debriefing. Participants were debriefed at the end of the study for suspicion and then compensated. No participant voiced suspicion about the study. Results Preliminary analyses. The groups did not differ on baseline affect, t(85)
0.26, p
.798, trait social anxiety, t(87)
0.70, p
.485, or gender, 2 (1)
0.09, p
.764. As in Study 2, baseline affect and trait social anxiety were included as covariates. Gender was not related to any of the dependent variables, and controlling for it did not substantively alter any of the results (for zero-order correlations, see Table 2). Therefore, it is not discussed further. Degrees of freedom vary across analyses due to missing data. Affect: Global affect and shame. To facilitate comparisons between global affect and shame, we first reverse scored partici￾pants’ baseline and postmanipulation global affect scores so that higher numbers on these scales reflected more negative affect, consistent with scores on the Shame scale. We then examined the effect of condition on global affect by performing a repeated￾measures ANCOVA with time of global affect measurement as the within-participants variable (two: baseline vs. postspeech task), condition as the between-participants variable (two: first-person vs. non-first-person) and trait social anxiety as the covariate. This analysis revealed a significant condition by time interaction, F(1, 83)
8.39, p
.005, p 2
.092, indicating that participants in the first-person group displayed a significant increase in negative affect over time, F(1, 42)
9.80, p
.003, p 2
.189. As Figure This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. SELF-TALK, REGULATION, AND SOCIAL ANXIETY 311

312 KROSS ET AL Table 2 Study 3 Zero-Order Correlations Between Measured Variables Postevent processing ote.To facilitate Panel A.ill ffered against this increase.)120.p 1g4=013 719. Behavior during the specch task.The effects of trait social anxiety.F( 83)=0.21.p=.652.n2 =.002.and baseline affect.F(1.83) 1,82) .70.p 007..As Figure 3.P e .19.p =.278.=014.were not significant the effect of baselin 20. .025. ocial anxiety and shame corre- F(l.3 P tual m-.026.and baseline affect.F(1.83)0.01.p-941.n of condition on global affect,F(1.81)=9.47.p=003.=.105. 000.were not significant.The effect of condition on posteven irst Person -Non First Pers 40 20 37 18 3 12 10 Post Speech T 06 37 20. 34 0.0 28 5-0.2 25 First Petson Non First Person Non Firs Person

3, Panel A, illustrates, however, participants in the non-first-person group were buffered against this increase, F(1, 40)
1.20, p
.280, p 2
.029. The interaction between trait social anxiety and time was not significant, F(1, 84)
0.13, p
.719, p 2
.002. We also performed an ANCOVA on shame with condition as the between-subjects predictor, controlling for trait social anxiety and baseline affect. This analysis revealed an effect of condition, F(1, 82)
7.70, p
.007, p 2
.086. As Figure 3, Panel B, illustrates, non-first-person participants felt less shame after the speech task. Although the effect of baseline affect was not signif￾icant, F(1, 82)
0.60, p
.441, p 2
.007, there was a significant effect of trait social anxiety in the predicted direction, F(1, 82)
5.20, p
.025, p 2
.060—trait social anxiety and shame corre￾lated positively, pr
.24, p
.025. Finally, conceptually replicating the Study 2 results, the effect of condition on global affect, F(1, 81)
9.47, p
.003, p 2
.105, and shame, F(1, 81)
4.91, p
.030, p 2
.057, remained significant when controlling for performance. Behavior. The effect of condition on speech performance was significant, F(1, 83)
5.43, p
.022, p 2
.061. As Figure 3, Panel C, illustrates, non-first-person participants performed better during the speech task. The effects of trait social anxiety, F(1, 83)
0.21, p
.652, p 2
.002, and baseline affect, F(1, 83)
1.19, p
.278, p 2
.014, were not significant. Postevent processing. The effect of condition on postevent processing was significant, F(1, 83)
5.47, p
.022, p 2
.062, indicating that non-first-person participants engaged in less poste￾vent processing than first-person participants (see Figure 3, Panel D). The effects of trait social anxiety, F(1, 83)
2.21, p
.141, p 2
.026, and baseline affect, F(1, 83)
0.01, p
.941, p 2
.000, were not significant. The effect of condition on postevent Table 2 Study 3 Zero-Order Correlations Between Measured Variables Variable 1 2 3 4 5 6 7 1. Trait social anxiety — .12 .01 .07 .09 .26
.21
2. Gender — .01 .00 .02 .06 .06 3. Baseline affect — .11 .01 .09 .00 4. Performance — .35


.30


.07 5. Global affect — .69


.26
6. Shame — .35


7. Postevent processing — Note. To facilitate comparisons between global affect and shame, we reverse scored participants’ baseline and postmanipulation global affect scores so that higher numbers on the scales reflected more negative affect.
p
.05.


p
.005. Figure 3. The effect of condition on global affect (Panel A), shame (Panel B), adaptive speech task performance (Panel C), and postevent processing (Panel D) in Study 3. Error bars indicate standard errors. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 312 KROSS ET AL

SELF-TALK.REGULATION.AND SOCIAL ANXIETY 313 2010).In the current conet.we reasoned that focusingn the nts of the Summary and Discussion e that t pri study 5 was to test this hypoherh A sec ing how lan was to u we examined the rla First they demonstrate that non-fir erson lang e use dur oidanc that is.avoiding nitions that give rise to anxiet inospctionpromotcssCf-rguionnd onditions a number of re rche Second demo ny et also for thei cy to engage in postevent proc essing.This han Nolen-Hock etal)and physical perspectiv takin (c.g..Eyal ctive takin e i alter any of the rch tha on-first-person lan use c nces au s larger samples of men is ne ded,however.to confirm thi this proces nhe he final goal of this study was to examine the reliability of ou Studies 4 and 5 Overview raethato-rero otes self-regulation under social stress.However,they do no modified ersion of th ss how t ici we asked them to rat th Study 4 Study 4 had three goals.Its first goal was to examine how am o anguage use on challenge-threat appraisals ve taking alon at( n people per eipants deliver bec ewe felt that asking par cribe in writing how they 心 Spencer.Zanna. ple to focus relatively less on the concrete emotionally aro Method 2010a:Kro an,8%African American,and 6%other)who receiv course creditor$20 for participating in this two-session study

processing, F(1, 82)
5.12, p
.026, p 2
.059, remained significant when controlling for performance. Summary and Discussion In sum, participants who used non-first-person pronouns and their own name during introspection performed better on the speech task, experienced less global negative affect and shame after delivering their speech, and engaged in less postevent pro￾cessing. These findings extend the Study 2 results in three ways. First, they demonstrate that non-first-person language use dur￾ing introspection promotes self-regulation under conditions that are well known to arouse both psychological and physiological stress—an anxiety-provoking public-speaking challenge. Second, they demonstrate that language use during self-reflection has im￾plications not simply for how people think and feel in the moment but also for their tendency to engage in postevent processing. This finding is particularly noteworthy given the role that postevent processing plays as a risk factor for poor psychological (e.g., Nolen-Hoeksema et al., 2008) and physical (Brosschot, Gerin, & Thayer, 2006) health, and its prevalence in social phobia (e.g., Brozovich & Heimberg, 2008). Finally, controlling for gender did not alter any of the results we observed. This suggests that the benefits associated with language use as a tool for promoting self-regulation may extend to both genders. Future research that includes larger samples of men is needed, however, to confirm this finding. Studies 4 and 5 Overview The aforementioned studies demonstrate that non-first-person language use during introspection is a form of self-distancing that promotes self-regulation under social stress. However, they do not address how this process influences people’s appraisals of future anxiety-provoking events. Studies 4 and 5 addressed this issue. Study 4 Study 4 had three goals. Its first goal was to examine how language use during introspection influences people’s appraisals of future stressors. In particular, it sought to examine the effect of language use on challenge-threat appraisals. People naturally appraise future stressors along a challenge– threat continuum (Blascovich & Tomaka, 1996; also see Lazarus & Folkman, 1984). Challenge appraisals occur when people per￾ceive their ability to cope (i.e., their resources) with a situation as exceeding the demands of the situation; threat appraisals occur when people perceive the demands of the situation as exceeding their ability to cope (e.g., Blascovich & Tomaka, 1996; Lazarus & Folkman, 1984). In the current study, we hypothesized that non-first-person lan￾guage use would lead people to appraise future stressors in more challenging and less threatening terms. This prediction was moti￾vated by research indicating that visual self-distancing leads peo￾ple to focus relatively less on the concrete, emotionally arousing aspects of negative past experiences and relatively more on recon￾struing their experiences in ways that provide insight and closure (for review, see Ayduk & Kross, 2010a; Kross, 2009; Kross & Ayduk, 2011; cf. Fujita et al., 2006; Trope & Liberman, 2003, 2010). In the current context, we reasoned that focusing on the concrete, emotionally arousing aspects of a future stressful expe￾rience are precisely those elements of the situation that should lead a person to conclude that the demands of the situation outweigh his or her ability to cope—i.e., less challenge and more threat. The primary goal of Study 5 was to test this hypothesis. A secondary goal was to rule out two alternative predictions concerning how language use influences the way people cogni￾tively represent future stressors. First, we examined the relation￾ship between language use during introspection and cognitive avoidance—that is, avoiding cognitions that give rise to anxiety. We focused on this relationship because a number of researchers have speculated that self-distancing blunts emotional reactions via a cognitive avoidance mechanism (e.g., Kenny & Bryant, 2007; Kenny et al., 2009; Kuyken & Moulds, 2009). Another possibility is that non-first-person language use en￾hances people’s tendency to imagine how they appear from the perspective of the audience evaluating them. This prediction is motivated by research indicating that psychological-distancing manipulations can enhance perspective taking (e.g., Eyal & Epley, 2010; Schultz & Heimberg, 2008). Although there is conflicting evidence on whether such perspective taking is helpful (e.g., Galinsky, Maddux, Gilin, & White, 2008) or harmful (e.g., Schultz & Heimberg, 2008), we nevertheless examined whether non-first-person language use enhances au￾dience perspective taking to examine the role that this process plays in the phenomenon at hand. The final goal of this study was to examine the reliability of our failure to observe an effect of language use on anticipatory anxiety in Study 2. As noted previously, this null effect was unexpected. Thus, we measured anticipatory anxiety again in this study to examine whether it would replicate. We examined these issues using a modified version of the speech task paradigm used in Study 3. Specifically, we replicated the Study 3 procedure with the following exceptions. After par￾ticipants reflected on their feelings, we asked them to rate the extent to which they appraised the upcoming speech task as a challenge or threat, and their current level of anticipatory anxiety. We then asked participants to describe in writing the stream of thoughts that flowed through their mind during the reflection period and coded these essays for challenge–threat appraisals, avoidance, and audience perspective taking. After participants completed the writing task, they were in￾formed that the study was over and debriefed. We did not have participants deliver their speech because we felt that asking par￾ticipants to answer each of the aforementioned questions and then describe in writing how they thought about their experience would compromise the downstream effects of the manipulations, muddy￾ing the conclusions we could draw about outcomes that were measured subsequently (for similar argument, see Ellsworth & Gonzalez, 2003; Kassam & Mendes, 2013; Spencer, Zanna, & Fong, 2005). Method Participants. Participants were 97 undergraduates (66 fe￾males; Mage
20.39 years, SDage
5.16; 51% White, 35% Asian American, 8% African American, and 6% other) who received course credit or $20 for participating in this two-session study. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. SELF-TALK, REGULATION, AND SOCIAL ANXIETY 313