Word of Mouth versus Word of Mouse: Speaking about a Brand Connects You to It More Than Writing Does

Word of Mouth versus Word of Mouse: Speaking about a Brand Connects You to It More Than Writing Does Abstract This research merges insights from the communications literature with that on the self-brand connection to examine a novel question: how does speaking versus writing about a liked brand influence the communicator’s own later reactions to that brand? Our conceptualization argues that because oral communication involves a greater focus on social interaction with the communication recipient than does written communication, oral communicators are more likely to express self-related thoughts than are writers, thereby increasing their self-brand connection (SBC). We also assess the implications of this conceptualization, including the identification of theoretically derived boundary conditions for the speech/writing difference, and the downstream effects of heightened SBC. Results from five studies provide support for our predictions, informing both the basic literature on communications, and the body of work on consumer word of mouth. word of mouth, communication, social interaction, self-expression, self-brand connection With the rise of the internet, more and more word of mouth is being generated online. Consumers often share their thoughts about products and brands by writing online reviews, posting comments on websites, sending text messages, and so on. These forms of online communication typically involve writing/typing, whereas the traditional face-to-face offline format involves speech. This distinction between written and oral word of mouth is at the heart of the current research. Do these two different forms of word of mouth have different implications not just for the marketer, but also for the communicating consumer? In particular, if a consumer shares her views about a favorite brand either orally or in writing, will her subsequent reactions to the brand differ because of her earlier use of these different communication channels? Although past research has examined the effect of using different communication modalities (such as speech vs. writing), this work primarily focuses on the recipient—for example, how the recipient processes information differently when it is conveyed by different channels (Chaiken and Eagly 1976, 1983; Fondacaro and Higgins 1985; Unnava, Burnkrant, and Erevelles 1994). Much less work has been done on how the communication can influence the communicator, one notable exception being Moore’s (2012) research on how explaining versus simply describing an experience can influence the communicator’s subsequent attitudes toward that experience. As far as we are aware, however, neither the communications literature nor consumer research has examined the focal question at the heart of the present inquiry: namely, whether the use of different communication channels (such as speech vs. writing) influences the communicator. The conceptualization we build to address this issue integrates insights from two different areas: the extant literature on communications (Jahandarie 1999; Tannen 1982), and consumer research on the self-brand connection, which refers to the extent to which the brand representation is integrated into one’s self-identity (Escalas 2004; Escalas and Bettman 2003; Fournier 1998). Briefly, our conceptualization argues that because oral communication involves a greater focus on social interaction with the communication recipient than does written communication (Rubin 1987; Tannen 1985), oral communicators are more likely to express self-related thoughts than are writers. When a consumer is communicating about a liked brand, the use of the brand for such self-expression builds a stronger connection between the consumer’s self-representation and the brand representation. The increased self-brand connection then manifests in downstream consequences that are favorable to the marketer, such as increased resistance to an attack on the brand, and also a greater willingness to wait for it in cases of stock-out. A series of studies finds support for these predictions, and also provides process insights by identifying theoretically derived moderators for the differences between talking and writing. For example, we show that the effect of communication channel on self-brand connection is mitigated either when the level of self-expression is controlled (studies 1 and 2) or when writers are induced to focus on the interaction aspect of the communication (studies 4 and 5). In examining these predictions, the current research makes the following contributions. First, we advance basic theoretical knowledge by illuminating the processes that underlie the differences between speech and writing. Of most importance, the current research is the first, we believe, to provide empirical support for one possible mechanism that has been posited to underlie the self-expression difference between speech and writing: namely, the greater focus on interacting with the recipient that is associated with typical speech versus typical writing (Chafe 1982; Chafe and Danielewicz 1987; Tannen 1985). Second, we inform consumer research—especially that dealing with word-of-mouth effects (Berger 2014; Chen and Lurie 2013; Moore 2012)—by providing an understanding of how different communication modalities can influence the communicating consumer’s own subsequent reactions to the brand. Third, we add to the extant consumer literature on the use of brands for self-expressive purposes by identifying a new antecedent of the self-brand connection: namely, talking about the brand (Belk 1988; Escalas 2004; Escalas and Bettman 2003). Finally, from an applied perspective, our findings carry implications for marketers who wish to know when and why it might benefit them to encourage consumers to engage in oral rather than written word of mouth about their brands. THEORETICAL BACKGROUND Our theoretical platform rests on two major premises, both of which derive from the extant communications literature. First, speech (vs. writing) is associated with a greater focus on interacting with the recipient of the communication; second, this heightened interaction focus promotes the discussion of self-relevant thoughts during the process of communication. We then build on these ideas to posit that when a communicator is discussing a favored brand this increased self-expression should increase her connection with that brand. Communication Channel and Interpersonal Involvement A rich body of work in the communications literature attests to the many differences between oral and written communication (Chafe 1982; Chafe and Danielewicz 1987; Fondacaro and Higgins 1985; Jahandarie 1999). For example, compared to written communication, oral communication is faster (Fondacaro and Higgins 1985), more transient and easily forgotten (Fondacaro and Higgins 1985; Jahandarie 1999), more redundant (Jahandarie 1999), and less precise (Jahandarie 1999). Several scholars have suggested, however, that perhaps the most critical distinction between these two modes of communication is that speakers, as compared to writers, are more focused on the interaction with the audience during the process of communication, while writers are primarily focused on the information to be conveyed (Chafe 1982; Fondacaro and Higgins 1985; Jahandarie 1999; Tannen 1985). This heightened focus of speakers (vs. writers) on the social interaction aspect of communication has sometimes been referred to as a difference in “interpersonal involvement” (Chafe and Danielewicz 1987; Jahandarie 1999; Tannen 1985); however, we use the term “interaction focus” in this article in order to avoid any confusion arising from the word “involvement,” which has specific connotations in the consumer literature. The difference in interaction focus stems from the audience typically being more salient for speakers as compared to writers, a major reason for which has to do with the very nature of speaking versus writing. For the most part, speaking involves the idea of talking “to someone”—it is relatively unusual for people to talk aloud without any audience (Wakefield 1992). Writing, in contrast, is often conducted without the intended recipient being present (Chafe 1982). Thus, as Milroy and Milroy (1985, 63) put it: “Speaking is a social activity whereas writing is solitary.” This difference between the two communication channels lends itself to a stronger interaction focus in speech than in writing. Relatedly, another reason for the difference has to do with the greater shared context between the communicator and recipient in speech (vs. writing) (Chafe 1985; Rubin 1987; Tannen 1985). Quite often (even if not always) speakers are face-to-face with the audience and are therefore together in time and space. This shared context itself naturally heightens the extent to which the audience is currently salient to the speaker—that is, it increases a focus on the interaction. Indeed, oral communication is more likely than written communication to involve simultaneous feedback from the audience (Fondacaro and Higgins 1985; Rubin 1987). This allows the speaker to monitor the effect of what she is saying on the listener; conversely, the listener can signal understanding and ask for clarification (Jahandarie 1999). Again, this increases the extent to which the speaker is focused on the interpersonal interaction. In contrast, with writing the recipient is often removed in time and space; the writer at times may not even know whom the reader will be. This lack of shared context inevitably creates a detachment between communicator and recipient, just as the overlap in context between speaker and listener heightens the salience of the interaction (Chafe 1982, 1985). Communication Channel, Interaction Focus, and Self-Expression The difference in interaction focus between speech and writing affects the extent to which self-relevant thoughts are expressed during the process of communication. In one illustrative study, Chafe (1982) found that first-person references, including “I,” “me,” and “us,” occur about 61.5 times per 1,000 words for spoken discourse, but only 4.6 times per 1,000 words for written discourse. A wealth of other research has found similar evidence for the greater use of personal pronouns and expressions of personal opinions and experiences in speech compared to writing (Biber 1988; DeVito 1966; Einhorn 1978; Tannen 1985). The communications literature identifies several reasons for the greater self-expression induced by the heightened interaction focus of speakers. One such reason is that the interaction aspect of the communication is typically associated with a social function: namely, the goal of establishing a degree of connectedness with the listener (Chafe 1982; Tannen 1985). The speaker can achieve greater connectedness by embedding the self into the speech—for example, by expressing personal opinions and experiences in the conversation (Chafe 1982, 1985; Chafe and Danielewicz 1987). Indeed, as Tannen (1985) points out, in many conversations, even more important than the subject matter of the speech is the act of speaking—because the speaking itself, by encouraging self-expression, fosters connectedness. In contrast, the writer, who has a lower degree of involvement with the interaction aspect, is primarily focused on conveying information rather than establishing connectedness (Fondacaro and Higgins 1985; Rubin 1987; Tannen 1985). As a result, prototypical writing tends to be relatively objective and detached, expressing logical and propositional content as much as possible rather than allowing the intrusion of self and subjective experience (Olson and Torrance 1981). A second feature of a heightened interaction focus, which leads to convergent conclusions regarding self-expression, is that it implies a two-way loop between the speaker and the listener (Chafe 1982, 1985; Fondacaro and Higgins 1985; Tannen 1985). Thus, not only are speakers focusing on the audience, but they are aware of the audience’s focus on them. This is particularly relevant to the self-expression argument because typical speech is ephemeral: it exists only so long as the speaker chooses to speak and therefore cannot be separated from the speaker (Jahandarie 1999; Rubin 1987). Accordingly, as Jahandarie puts it (1999, 139), the important question for the audience in deciphering speech is “What does the speaker mean?” as opposed to the focal question with writing of “What do the words mean?” (see also Coulmas 1989). Because speakers are likely to be aware of the audience’s focus on them as individuals, this not only gives them license to inject their selves into the speech, but actively encourages it (Chafe 1985; Rubin 1987). In contrast, writing exists as a permanent record, and has an existence of its own even after the writer is done writing or typing (we use the terms “typing” and “writing” interchangeably in this article). Therefore, unlike the interaction-focused speaker, the writer is unlikely to believe that the reader expects the communication to reflect the idiosyncrasies of the communicator; rather, the writer’s goal is to produce a communication that holds value even when the writer is no longer present (Jahandarie 1999). Thus, writing typically serves to “disembody the written text from the writer” (Rubin 1987, 10), lessening the role of the self. These related strands of reasoning converge on the conclusion that oral communication typically entails greater self-expression than writing, as captured in interrelated terms such as ego involvement (Chafe 1985; Jahandarie 1999) and self-activation (Chafe 1982; Rubin 1987). The heightened self-expression in speech manifests in a greater number of self-related thoughts, such as the inclusion of personal experiences and opinions rather than factual content (Biber 1988; Chafe 1982; Tannen 1985). In the current context of communicating about a brand, this suggests that speakers will offer more personal opinions about the brand and also will be more likely to discuss their personal experiences with it, whereas writers are more likely to maintain a relatively objective detachment in their brand communication. Two points of clarification are in order. First, we do not wish to claim that a heightened interaction focus is the only mechanism by which communication channel influences self-expression. Other mechanisms may also be at play. For example, people tend to spontaneously generate their thoughts during oral communication, but they have more time to deliberate on what they should write (Berger and Iyengar 2013; Fondacaro and Higgins 1985; Klesse, Levav, and Goukens 2015). Because the self tends to be highly accessible, spontaneous communication is likely to rely more on self-related thoughts than deliberative communication, with the latter offering communicators more opportunity to access nonself thoughts as well (Morewedge, Giblin, and Norton 2014). But although we believe the self-expression difference between speech and writing is multiply determined, this research delves into the role of interaction focus because of considerable convergence in the communications literature that it is the crucial factor underlying several speech/writing differences (Chafe 1982; Chafe and Danielewicz 1986; Tannen 1985). Note also that the posited difference in interaction focus between speech and writing, and the consequent difference in self-expression, is emblematic of prototypical speech and prototypical writing. We are in agreement with Tannen (1985) that not all forms of writing are detached from the recipient, nor do all forms of speech signify strong involvement with the interaction (e.g., memorizing and delivering a formal speech is likely associated with a relatively low focus on the interaction aspect). However, research on procedural knowledge has robustly shown that procedures that are initially learned to fulfill a particular purpose can become overlearned so that they are automatically reproduced even when that initial purpose is no longer salient (Schneider and Shiffrin 1977; Shen and Wyer 2008; Wood and Neal 2007; Wyer, Shen, and Xu 2013). Similarly, it seems likely that to begin with, people deliberately express the self more often when talking face-to-face with others (vs. writing to an unknown recipient), because doing so fulfills salient goals, such as facilitating social interaction. With frequent repetition, however, this self-expression procedure will become strongly associated with speech in general, to the point of being performed automatically even in nonprototypical situations. Speech, Writing, and Self-Brand Connection The difference in self-expression between speaking and writing contains clear implications, we believe, for the extent to which consumers develop a self-brand connection as a result of communicating about a favored brand. Originating in the premise that consumers often use favored brands and products to establish, maintain, and signal self-concept (Aaker 1999; Belk 1988; Chaplin and John 2005; Cheng, White, and Chaplin 2012; Escalas 2004; Fournier 1998; Sirgy 1982), a self-brand connection (SBC) is defined as the strength of the link between a consumer’s self-representation and brand representation (Moore and Homer 2008). Thus, SBC captures the extent to which the brand is integrated into the consumer’s idea of self (Escalas 2004; Ferraro, Kirmani, and Matherly 2013; Fournier 1998; Park, Macinnis, and Priester 2008). Note that the very nature of a self-brand connection means that this concept applies to the case of positively viewed brands, which is the focus of the current research. At the same time, it should be noted that SBC and attitude extremity are two different constructs. Thus, it is possible to hold positive attitudes toward brands with which one lacks a self-brand connection (e.g., brands that a consumer evaluates favorably on purely functional grounds, without necessarily forming a link to self-representation). As Escalas (2004; footnote 1) points out, “although the existence of a SBC implies a positive brand attitude, the inverse is not true. Consumers may have favorable attitudes towards many brands where no SBCs exist.” How, then, does a high SBC develop? According to the literature, SBC is strengthened when consumers develop connections between their brand thoughts and their self-representations (Cheng et al. 2012; Escalas and Bettman 2003; Park, MacInnis, and Priester 2006). For example, a brand that expresses a valued aspect of self-identity—such as an environmentally friendly lawnmower—will be associated with a high SBC because consumer cognitions about the brand are likely to be connected to thoughts of the self (Escalas and Bettman 2003). In a similar vein, SBC is likely to be high when the brand is used to “communicate the self-concept to others” (Escalas and Bettman 2005, 378). Merging these insights on SBC with the communications literature reviewed earlier, the current research argues that talking about a liked brand (as opposed to writing about it) should also enhance the self-brand connection. Both speaking and writing about a brand involve communicating about it to others, and in both channels, given the context of a liked brand, this communication is likely to be primarily of a positive nature. However, the greater self-expression involved in speech (itself arising from a heightened interaction focus) should manifest in speakers expressing a greater number of self-relevant positive thoughts about the brand. For example, speaking is more likely to involve statements such as “I like Apple because I am a creative kind of person” versus “Apple is a creative brand.” The former type of thought, by definition, should produce a stronger connection with the brand, because it explicitly integrates the brand with the self (see figure 1). FIGURE 1 View largeDownload slide CONCEPTUAL FRAMEWORK FIGURE 1 View largeDownload slide CONCEPTUAL FRAMEWORK Implications This conceptualization yields several testable implications relating to the distinction between speaking and writing about a brand. Our research focuses on two such implications: a) downstream consequences, and b) moderators of the effect. It is worth reiterating that the SBC construct—and therefore the current inquiry—focuses exclusively on positively viewed brands. Downstream Consequences A high self-brand connection has been shown to benefit the brand in multiple ways. Because high SBC reflects a sense of “oneness” with the brand, consumers with a high SBC are motivated to maintain a connection with the brand even if there is a cost involved in doing so (Park et al. 2010)—as manifested, for example, in a greater willingness to wait for the brand if they have to (Keller 2001; Sprott, Czellar, and Spangenberg 2009). Similarly, because strongly connected brands form a part of the consumer’s self-identity, they are also less susceptible to criticism, because consumers are more likely to defend such a brand from an attack (Cheng et al. 2012; Ferraro et al. 2013). Tying these findings to the current perspective, we predict that the enhanced SBC arising from consumers’ speaking (vs. writing) about a favored brand should yield positive consequences for the marketer, including an increased willingness to wait for the brand, and increased defense against brand criticism. Moderators of the Speak versus Write Effect Our theorizing implicates two types of moderators—one arising from the proximal influencer on SBC (extent of self-expression) and the other arising from the more distal influencer (extent of interaction focus). Regarding the former, we posit that speaking increases the extent to which consumers discuss self-related thoughts when discussing the brand, leading to a stronger connection between the two than in the case of writing. If true, this suggests that one can attenuate the predicted SBC difference between speaking and writing by increasing (decreasing) writers’ (speakers’) focus on the self during brand communication. Thus, explicitly encouraging writers to focus on self-relevant thoughts while discussing the brand should increase SBC for writers. Conversely, encouraging speakers to adopt another person’s perspective and recount what such a person might think about the brand should decrease SBC for speakers. We examine both of these predictions (studies 1 and 2). Second, a key reason for the greater self-expression in speech (vs. writing), we have argued, is the heightened interaction focus in the former: speakers are more focused on the interaction with the recipient than are writers. This argument suggests that exogenously inducing writers to focus more on the interaction aspect of the communication should also increase the expression of self-related brand thoughts, and thereby strengthen writers’ connection with the brand about which they are communicating. We examine those possibilities in studies 3–5. STUDY 1: THE MODERATING INFLUENCE OF SELF-PERSPECTIVE Study 1 had several objectives. First, it sought to provide initial evidence for the effect of communication channel on self-brand connection. Second, this study investigated a theoretically derived moderator of the SBC difference produced by speaking (vs. writing) about the brand. As discussed above, while the SBC difference is believed to arise because speakers spontaneously engage in self-expression during communication, encouraging writers to also engage in self-expression (e.g., by explicitly instructing them to adopt a personal perspective when discussing the brand) should enhance the strength of the link between the brand representation and the self-representation for writers as well, thereby increasing SBC. Third, a downstream consequence of the basic SBC effect was also measured—namely, consumers’ willingness to wait for the brand they had discussed, given an out-of-stock situation. Method This study comprised a 2 (communication channel: speaking vs. writing) × 2 (discussion focus: default vs. explicit self-focus) between-subjects design. Two hundred fifty undergraduate students at a Hong Kong university participated in this study for a cash payment of HK$40 (US$5). They were grouped into pairs and told that we were interested in students’ views of brands. Apple, a brand pretested to be viewed favorably across this student population (as were the brands used in later studies), was the focus of communication. Those in the speaking condition were randomly assigned to either be a communicator or a listener, with one participant in each pair asked to talk about his or her views of Apple to the other. In this condition, communicators were told that we would like them to share their views of Apple with their partner. Those in the writing condition were simply asked to share their views of Apple by writing them down on a piece of paper; the partner was asked to complete a different questionnaire irrelevant to this study. Note that in this case, the presence of the partners was not required; they were not described as being the recipient of the communicator’s thoughts. However, having a partner present even in the writing condition allowed us to check that our predicted effects would still hold even when the writer is not physically alone. We did this to rule out the possibility that the effects are driven simply by a possible increase in self-consciousness caused by the sheer physical presence of another person. We manipulated the focus factor by instructing participants in the default condition to discuss anything they wished to about the brand. In all conditions, participants were told that: a) they could freely share any of their thoughts about the brand; b) they could use any language; c) they could take as long as they liked. In the self-focus condition, they were additionally asked to adopt a first-person perspective when discussing the brand, and were also encouraged to think about their personal experiences when doing so. After they had provided their views of the brand, all communicators (n = 125) responded to a questionnaire. (Because this research focuses on the effect of communication channel on the communicator’s reaction toward the brand, the partner was not asked to complete this questionnaire.) Communicators first reported their attitude toward the Apple brand by rating it on a five-point scale (1 = bad to 5 = good). They were then exposed to the following scenario: “Imagine that you participate in a lottery and luckily win an Apple notebook computer. However, you are told that this computer is out of stock now. You need to wait for 1 month for the computer to be available at the store. Alternatively, you can get a Dell notebook computer immediately, which has the same price.” Then, they were asked to indicate the extent to which they were willing to wait for the Apple notebook from 1 (not likely at all) to 7 (very likely). Next, all participants filled out a standard measure of self-brand connection. This measure, taken from Escalas (2004), consists of seven items, such as: “This brand reflects who I am,” “I can identify with this brand,” and “I feel a personal connection to this brand.” Participants indicated their agreements with each item along a seven-point scale (1 = disagree, 7 = agree; α = .91). Finally, to examine whether the communications channel also leads to differential expression of feelings versus cognitions, we asked participants to indicate the extent to which their discussion of the brand comprised “mainly their thoughts about the brand” (1) versus “mainly their feelings about the brand” (7). Results and Discussion Self-Brand Connection We predicted that participants would feel more connected to Apple if they talked rather than wrote about the brand; further, this effect would be attenuated when they were explicitly instructed to adopt a personal perspective during the discussion. In support of this prediction, a 2 × 2 ANOVA revealed a marginally significant interaction effect of communication channel and focus of discussion (F(1, 121) = 3.75, p = .06). Those in the default condition felt more connected to the brand when they talked (Mtalk = 4.24) rather than wrote about it (Mwrite = 3.38, F(1, 121) = 6.69, p = .01). However, this effect disappeared if they were explicitly asked to adopt a personal perspective while discussing the brand (Mtalk = 4.05 vs. Mwrite = 4.10, F <1). Viewed differently, and also in line with our theorizing, SBC increased for writers in the self-focus condition (Mself-focus = 4.10 vs. Mdefault = 3.38, F(1, 121) = 4.56, p = .03); no such difference was observed for speakers (Mself-focus = 4.05 vs. Mdefault = 4.24, F < 1). Also, this pattern of self-brand connections was not driven by brand attitudes. Only a main effect of the focus factor was obtained on attitude toward the brand, such that participants reported liking the brand more when they focused on the self while discussing it than if not (Mself-focus = 4.25 vs. Mdefault = 4.00, F(1, 121) = 4.74, p = .03). No other effects were significant. Finally, no differences were observed as to whether participants mainly discussed thoughts or feelings about the brand (all ps > .20; see table 1 for details). TABLE 1 THE EFFECT OF COMMUNICATION CHANNEL AND DISCUSSION FOCUS ON SELF-BRAND CONNECTION AND WILLINGNESS TO WAIT (STUDY 1) Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 1 THE EFFECT OF COMMUNICATION CHANNEL AND DISCUSSION FOCUS ON SELF-BRAND CONNECTION AND WILLINGNESS TO WAIT (STUDY 1) Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 2 THE EFFECT OF COMMUNICATION CHANNEL AND COMMUNICATION PERSPECTIVE ON THOUGHTS, SELF-BRAND CONNECTION, AND ATTITUDE CHANGE (STUDY 2) Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 2 THE EFFECT OF COMMUNICATION CHANNEL AND COMMUNICATION PERSPECTIVE ON THOUGHTS, SELF-BRAND CONNECTION, AND ATTITUDE CHANGE (STUDY 2) Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 3 THE EFFECT OF COMMUNICATION CHANNEL AND INTERACTION FOCUS ON THOUGHTS AND SELF-BRAND CONNECTION (STUDY 4) Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 3 THE EFFECT OF COMMUNICATION CHANNEL AND INTERACTION FOCUS ON THOUGHTS AND SELF-BRAND CONNECTION (STUDY 4) Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 4 THE EFFECT OF COMMUNICATION CHANNEL AND PRIOR RECIPIENT INTERACTION ON ONGOING INTERACTION FOCUS, THOUGHTS, AND SELF-BRAND CONNECTION (STUDY 5) Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 4 THE EFFECT OF COMMUNICATION CHANNEL AND PRIOR RECIPIENT INTERACTION ON ONGOING INTERACTION FOCUS, THOUGHTS, AND SELF-BRAND CONNECTION (STUDY 5) Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. Willingness to Wait (WTW) Drawing on previous research (Park et al. 2006, 2008; Sprott et al. 2009), we have argued that higher SBC should produce a greater willingness to wait for the brand. In support of this argument, the two-way interaction of communication channel and discussion focus on willingness to wait was marginally significant (F(1, 121) = 3.35, p = .07). As with SBC, participants in the default condition were more willing to wait for the brand if they had talked rather than written about it (Mtalk = 6.30 vs. Mwrite = 5.26, F(1, 121) = 6.46, p = .01). Further, this effect disappeared in the explicit self-focus condition (Mtalk = 6.06 vs. Mwrite = 6.07, F <1; see table 1 for details). Discussion This study provides initial support for our major prediction: consumers feel more connected to a brand after they talk rather than write about it. It also informs the finding in two major ways. First, it explores a downstream consequence of the basic effect, showing that the increased SBC resulting from speaking about the brand also yields a greater willingness to wait for that brand. Second, consistent with crucial role posited for self-expression in driving SBC, we found that when writers were explicitly encouraged to focus on the self, the resultant SBC (and willingness to wait) was as high as in the speech condition. It is worth reiterating that the pattern of self-brand connection was not driven by brand attitudes. Thus, unlike with SBC, we did not observe an interaction effect of communication channel and focus of discussion on brand attitudes (p > .20). This is consistent with past research showing that brand attitude is a different construct from SBC (Escalas 2004; Park et al. 2008). In particular, not all routes to brand liking necessarily involve self-expression. For example, Escalas (2004) found that autobiographical recall (remembering past experiences of a brand) improved SBC compared to mental simulation (imagining future use of the brand) because autobiographical memories are more closely tied to the self than mental simulations; however, because the two processes do not differ in terms of the valence of thoughts expressed about the brand, there was no difference in brand attitudes. Similarly in the current context, both speaking and writing about a favored brand are likely to evoke positively valenced thoughts about the brand; thus, as observed, the two conditions do not differ in terms of brand attitudes. We posit, however, that the greater self-expression in the former case enhances SBC. Study 1 was not without limitations, however, the most important of which concerned the lack of an objective measure of self-expression (although see the web appendix for a self-report measure). This limitation is addressed in study 2, which analyzes the actual content of participants’ communications to analyze a brand-specific measure of self-expression. Another drawback of study 1 was that speakers, but not writers, were assigned a specific communication recipient. In order to minimize the possibility that the effects were driven simply by this difference, the next study relaxes one of the conditions associated with prototypical speech—namely, being face-to-face with the listener. Thus, participants in the speech condition of study 2 were asked to speak into a voice recorder rather than sharing their thoughts with a live person. It might be argued that the absence of a physical audience should lessen the focus on the interaction and therefore reduce self-expression, attenuating the SBC difference between speech and writing. As noted earlier, however, the principles of procedural accessibility suggest that because speakers have repeatedly practiced a procedure of expressing the self to others to facilitate social interaction, such self-expression is likely to obtain even in nonprototypical conditions. In particular, speech is so strongly associated with the idea of “talking to” someone, it is likely that speakers will imagine an audience even if none is physically present—an assumption we test—which would then cause them to express the self more, enhancing SBC. STUDY 2: THE MODERATING INFLUENCE OF OTHER-PERSPECTIVE In addition to generalizing the basic SBC effect to the context described above (i.e., speakers in this study spoke into a voice recorder), study 2 extended our previous findings in three important directions. First, drawing on prior research showing that consumers are more likely to refute criticism of a brand with which they feel a stronger connection (Ferraro et al. 2013), we examined the prediction that talking (vs. writing) about a brand should increase resistance to an attack, because of the higher SBC in the former case. Second, this study examined another moderator of the speech/writing SBC difference. If this difference is based, as we have argued, on greater self-expression when speaking, it should be attenuated if participants are explicitly instructed to discuss what they believe to be another person’s perceptions of the brand. Taking an other-perspective should dampen self-expression and therefore SBC for speakers, bringing it closer to the default level for writers. This boundary condition can be seen as the reverse of that tested in study 1, in which we increased SBC for writers (to the default level for speakers) by explicitly forcing a personal perspective in communication. Third, as noted above, we analyzed the communications provided by participants in this study in detail to obtain insights into the self-expression process by which speech versus writing affects SBC. Method This study comprised a 2 (communication channel: talking vs. writing) × 2 (communication perspective: default vs. other-perspective) between-subjects design. One hundred sixty-four undergraduate students participated in this study for a cash payment of HK$40 (US$5). All participants were initially told that we were interested in students’ views of various brands. To manipulate the perspective factor, we then told those in the default condition that we would like them to provide their views of Samsung (again, a positively viewed brand in this population). In the other-perspective condition, participants were asked to think of someone else who had used Samsung products, and to provide their views of how that person perceived the brand. With regard to the communications channel, those in the writing condition were asked to write their views on a piece of paper that they were given for that purpose. In the speaking condition, they were given a voice recorder and asked to talk about the brand into the recorder. Thus, the two conditions were made equivalent in that (unlike in study 1) neither speakers nor writers were asked to communicate to an externally specified recipient. All participants completed the study one at a time. Also, in all conditions, after giving the instructions, the experimenter exited the room and left the participant alone to write/speak about the brand and then complete the subsequent questionnaire. Therefore, any observed effects of communication channel on SBC and attitude resistance are unlikely to proceed from differences in the extent of public commitment to one’s position (Cialdini and Trost 1998; Gopinath and Nyer 2009). After participants had finished writing/talking about the brand, they provided initial attitudes toward it by rating it on a five-point scale (1 = bad to 5 = good). Next, they read the following purported comment by another customer of the brand: “I am using a Samsung mobile phone. It used to work just fine. But a couple of days ago, I had problems loading the map. Sometimes it took around 5 seconds to display the searching result. The calendar reminders also failed to work properly. For some events, the reminders go off a few minutes after the events. I had tried to fix the problems but they still occur. This causes a bit of inconvenience to my social and work life.” After they finished reading the scenario, participants once again reported their attitudes toward Samsung, this time along a 10-point scale from 1 (bad) to 10 (good). A different scale was used to prevent participants from simply retrieving and reporting their earlier evaluation. Scores on the second scale were divided by two to correspond to the initial five-point scale. The difference between the initial and the latter attitudes provided a measure of post-attack attitude change. We also assessed a self-report measure of susceptibility to attack by then asking participants to indicate the extent to which the other customer’s negative comment had influenced their final attitude toward Samsung, along a scale from 1 (not at all) to 7 (very much). Finally, all participants filled out the same standard measure of self-brand connection used in study 1. We excluded five out of 164 participants who either discussed nothing about the brand or failed to answer all the questions. Thus, the analyses used 159 data points. Results Self-Brand Connection We expected to replicate our earlier findings in the default condition—that is, greater SBC following spoken (vs. written) brand communication—but we predicted an attenuation of this effect when participants were explicitly instructed to discuss another person’s views of the brand. A 2 × 2 ANOVA revealed a significant interaction of communication channel and communications perspective (F(1, 155) = 5.08, p = .03). As predicted, participants in the default condition felt more connected to the brand after talking (Mtalk = 3.42) rather than writing about it (Mwrite = 2.71, F(1, 155) = 6.61, p = .01). However, this effect disappeared in the other-perspective condition (Mtalk = 2.86 vs. Mwrite = 3.02, F <1). Attitude Change As in the previous study, no effects were observed on initial brand attitude (ps > .10). However, we expected a systematic pattern for the attitude change measure (the difference between initial and post-attack attitudes). In particular, following the SBC results, we predicted lower attitude change—that is, greater resistance to attack—for speech (vs. writing) in the default conditions; however, this difference in attitude resistance should be attenuated in the other-perspective conditions. Results were consistent with these predictions. The two-way interaction between communication channel and communication perspective on attitude change was marginally significant (F(1, 155) = 3.13, p = .08). Those in the default condition displayed less attitude change when they had earlier talked (Mdiff = .55) rather than written about the brand (Mdiff =.80, F(1, 155) = 4.82, p = .03). However, this effect was no longer observed in the other-perspective conditions (Mdiff = .78 vs. Mdiff = .75, in talking vs. writing conditions, respectively, F <1). A reassuringly convergent pattern was observed on the self-reported measure of participants’ susceptibility to the attack. To elaborate, a significant interaction (F(1, 155) = 4.23, p = .04) was based on lower susceptibility to attack for speech (vs. writing) in the default condition (Mtalk = 3.70 vs. Mwrite = 4.59, F(1, 155) = 6.47, p = .01), while no such difference was observed in the other-perspective conditions (Mtalk = 4.64 vs. Mwrite = 4.52, F <1). Process Measures Our theorizing holds that in default conditions (but not in the other-perspective condition), talking versus writing about the brand increases the self-brand connection because the greater self-expression associated with speech induces speakers to generate a greater number of self-relevant brand thoughts, creating the positive link between the self and brand that is captured by heightened SBC. We assessed this self-expression tendency by examining the number of thoughts explicitly relating the self to the brand (e.g., personal opinions of the brand, and personal experiences with it). Furthermore, since our goal was to assess the extent to which brand-related self-expression heightens SBC, which refers to a positive connection with the brand, we focused on the net number of such self-brand thoughts that reflected a positive view of the brand. This measure thus specifically isolates the favorable use of the brand for self-expression, which we argue should mediate the effect of communication channel on SBC. Two independent coders, unaware of our experimental hypotheses and participant condition, coded each participant’s thoughts about the brand for the following: (a) total number of thoughts;1 (b) number of both positive and negative thoughts about the brand that include a reference both to the brand and to the self. This included personal experiences, defined as thoughts about the brand reflecting actual experience (e.g., “I have used a Samsung smartphone for a long time and find it very comfortable to use”; “I find my Samsung computer often gets problems”), and personal opinions, defined as subjective evaluations of a brand that do not explicitly refer to personal experience (e.g., personal beliefs, observations, or knowledge about a brand, such as “I like the Samsung brand; my personal view is that Samsung’s smartphone has a very good design”; “I don’t like the advertisement of Samsung”). The difference between the number of such positive self-brand thoughts and the number of negative self-brand thoughts (i.e., the net positive self-related brand thoughts) served as our key measure of self-expression. (c) number of both positive and negative thoughts about the brand that do not include mentions of self (e.g., “Samsung is a brand that is popular in Hong Kong”; “A lot of people are using Samsung smartphones”; “Samsung smartphones have a lower market share than iPhone in Hong Kong”). The difference between positive and negative thoughts in this case produced an index of net positive self-unrelated thoughts. Overall interrater agreement was high (more than 80% for any item). Differences between coder ratings were resolved through discussion. We expect that in the default conditions, speech (vs. writing) should be associated with higher scores on our index of brand-related self-expression (i.e., net positive self-brand thoughts) and, conversely, fewer thoughts about the brand that do not reflect such self-expression (i.e., net positive self-unrelated brand thoughts). Such thought patterns would correspond with the higher SBC observed for speech (vs. writing) in the default case. In contrast, we did not expect to observe differences in the thought indices in the other-perspective conditions, where speech and writing did not differ on SBC. Main effects of the two factors were obtained on both the self-expression index and also on the index of self-unrelated brand thoughts (ps < .001). More importantly, the two-way interaction between person involved and communication channel was also significant for both indices (ps < .01). In support of our conceptualization, speakers in the default condition mentioned a greater number of net positive self-related brand thoughts, indicating greater use of the brand for self-expression (Mdiff = 2.22) than writers (Mdiff = .25, F(1, 155) = 15.75, p < .001), while mentioning fewer net positive brand thoughts that were unrelated to the self (Mdiff = –.43 vs. 1.10, in talking vs. writing conditions, respectively, F(1, 155) = 6.01, p = .02). Further, consistent with the null effect on SBC given other-focus, speakers and writers did not differ on either of these thought indices when discussing another person’s perception of the brand (net positive self-related brand thoughts: Mdiff = .03 vs. –.09, in talking vs. writing conditions, respectively, F <1; net positive self-unrelated brand thoughts: Mdiff = 1.72 vs. 1.76, F <1; see table 2 for details). Finally, we assessed whether the thoughts-based measure of self-expression mediated the interactive effect of communication channel and person involved on SBC, as our theorizing would argue. To examine this possibility, we conducted moderated mediation analysis using PROCESS model 7 (Hayes 2013; Preacher, Rucker, and Hayes 2007). As reported previously, the effect of interaction between channel and person involved on SBC was significant. In addition, across conditions, the thoughts-based index of self-expression had a positive effect on SBC (b = .45, t(157) = 6.38, p < .001). Finally, the mean indirect effect of the interaction between communication channel and person involved on SBC through self-expression (based on 5,000 bootstrap samples) was significant, with a 95% confidence interval excluding zero (.05 to .43). More specifically, the indirect effect of communication channel on SBC through self-expression was significant only in the default conditions (point estimate = .23, 95% CI = [.08, .44]), but not when participants were asked to discuss another person’s perception of the brand (point estimate = .01, 95% CI = [−.04, .08]). These results provide good evidence for the role of self-expressive thoughts in mediating the interactive effect of communication channel and person involved on self-brand connection. Explaining versus Nonexplaining Language Recent research finds that the use of explaining language to discuss a hedonic experience (e.g., explaining why the person likes it) lowers post-discussion evaluations of the experience, as compared to the use of nonexplaining language (e.g., simply describing and providing an evaluation of the experience without including explanations) (Moore 2012). Accordingly, it was important to explore whether writers and speakers differ in their use of explaining language, and if this affects SBC. We note that post-communication brand attitudes did not differ for the speech and writing conditions in our studies, which in itself argues against this account (to reiterate, the use of explaining language has been found to dampen attitudes). To further test this explanation, two independent coders were asked to process each communication in the default conditions and code each sentence providing any positive view of the brand as either explaining or nonexplaining language, following Moore (2012): a) explaining (e.g., sentences including reasons for a positive evaluation, such as “The Samsung mobile phone is good because it is very comfortable to use”), and b) nonexplaining (e.g., sentences including a positive evaluation without a reason provided, such as “I like the Samsung brand” or “The Samsung brand is good”). The overall agreement was high (more than 80% for each item). We calculated the proportion of explaining language (PEL) by dividing the number of explaining sentences by the total number of explaining and nonexplaining sentences (Moore 2012). We did not observe any difference across conditions (talking conditions: PEL = .48, SD = .39 vs. writing conditions: PEL = .45, SD = .33, F < 1). Therefore, the effects observed are unlikely to be due to differences in the use of explaining language. Discussion Study 2 not only provided further evidence for our key premise that speaking (vs. writing) about a favored brand can increase the self-brand connection, it also provided an assessment of some of the other implications of our conceptualization. First, adding to the study 1 finding that talking (vs. writing) about a brand increases the willingness to wait for it, we found that talking about the brand has another benefit to the marketer: it produces a greater resistance to a subsequent attack on the brand. Second, study 2 provides process insight into this effect by documenting another theoretically consistent boundary condition. In reassuring symmetry with our earlier finding that the SBC difference between writing and talking is attenuated when writers are explicitly instructed to adopt a personal perspective (thereby increasing their SBC), study 2 found that this difference is also mitigated when speakers are explicitly instructed to describe another person’s views of the brand (thereby decreasing their SBC). Both of these boundary conditions are consistent with the basic premise that greater self-expression during brand communication leads to heightened SBC. Third, study 2 provided more direct support for this premise by assessing the extent to which the brand was used for self-expression, as measured by the thoughts that linked the brand with the self. An analysis of participants’ thoughts revealed, as predicted, that the default speech condition did indeed produce a relatively high number of net positive self-brand thoughts, which in turn led to increased SBC. Note that the overall level of self-expression can be assessed through another measure—namely, the frequency of use of the personal pronoun “I,” with such usage presumably being greater for speakers than writers (Chafe 1982, 1985). However, we did not rely on this gross measure because it does not capture the use of the brand for self-expression (e.g., a statement such as “I am feeling happy today” does not invoke the brand even though it uses “I”), nor does it necessarily reflect positive brand views, both of which are an integral aspect of the process by which brand-related self-expression heightens SBC. That said, across studies, we computed both the thoughts-based measure of self-expression used in study 2 as well as the simple frequency count of “I” mentions. The pattern of results on the two measures paralleled each other, as reported in the web appendix. Finally, it is worth noting that in the default conditions of this study, a greater number of self-brand thoughts, and accordingly an increased SBC, manifested for speakers (vs. writers) even though no audience was present (speakers spoke into a voice recorder). We argue that this is the case because speech is so strongly associated with the idea of an audience that speakers automatically imagine interacting with one even when none is actually present. In turn, this causes them to engage in greater self-expression, as manifested in the greater number of thoughts that explicitly referenced the self in relation to the object of discussion (i.e., the brand). A post-test was run to verify this assumption. As in the default conditions of study 2, participants were asked to discuss a liked brand either by talking into a voice recorder or by writing down their thoughts on a piece of paper. All participants were then asked to report the extent to which they thought about the communication recipient while discussing the brand. Results revealed that, as anticipated, speakers did indeed imagine an actual audience—even though none was present—to a significantly greater extent than writers (please see the web appendix for full details of the procedure and findings). STUDY 3: THE ROLE OF INTERACTION FOCUS Studies 1 and 2 provided consistent evidence that speaking about a liked brand increases the self-brand connection as compared to writing about it; further, study 2 showed that this effect is due to the greater expression of self-relevant brand thoughts by speakers (vs. writers). The next three studies provide further insight into the process. As discussed earlier, we argue that the reason speakers engage in greater self-expression than writers (which, in the brand communication context, manifests as a greater number of self-related brand thoughts) has to do with their heightened focus on the interaction. This increased interaction focus then induces the speaker to embed the self into the conversation (Chafe 1982, 1985; Fondacaro and Higgins 1985; Tannen 1985). In order to provide an initial test of this reasoning, study 3 measured interaction focus along with self-brand connection. For greater generalizability, study 3 also used a different manipulation to induce speech versus writing, with speakers (writers) being asked to make a phone call (write an online message) about the brand. Finally, because we have already demonstrated the downstream consequences of SBC in studies 1 and 2, we did not examine those consequences again in study 3 and subsequent studies. Instead, these remaining studies focused on replicating the basic SBC effect and providing insights into the mechanism underlying the effect. Method Seventy-nine undergraduate students participated in this study for a cash payment of HK$40 (US$5). Participants were told that the researchers were interested in students’ perception of another brand pretested to be well liked in this population: Nike. They were then asked to communicate anything they wished about this brand to a research assistant who (purportedly) could not come to the experiment room at the time, because she was in a different location. On this pretense, half of the participants were asked to communicate with the assistant via a phone call (on receiving the call, the assistant simply asked each participant to start talking about the brand, and did not give any feedback during the communication). The other participants were asked to communicate their message about Nike in an online survey and were informed that their communication would be sent to the assistant immediately after they clicked Submit. Participants were informed that their messages would not be preserved in either condition. Note that because messages were indeed not preserved in either condition (it would not have been possible to do for speakers since no voice recorder was used in this study), study 3 does not contain any measures of self-expression. Rather, the goal of this study is to provide an initial test of the role of interaction focus in influencing self-brand connection. To assess this, after participants in both conditions finished communicating their thoughts about Nike, they were asked to complete a questionnaire. They first filled out the seven-item SBC measure, which is the key dependent variable. Participants then indicated their level of agreement with three statements assessing the extent to which they were focused on the interaction aspect during communication: (a) when they discussed the brand, they imagined themselves talking/writing to the assistant that they were communicating with; (b) when they discussed the brand, they communicated similarly to the way they would during a social interaction; (c) when they discussed the brand, they talked/wrote as if they were socially interacting with the assistant they were sending the message to. Each item was anchored by 1 (disagree) to 7 (agree). The three items were highly correlated with one another (α = .89) and were averaged to create an index for interaction focus. Finally, all participants responded to two measures, each of which served to examine a possible alternate explanation for the SBC pattern observed for speakers versus writers. First, they were asked how much they felt they knew the research assistant with whom they had communicated: 1 (not at all) to 7 (very much). Note that the assistant had not been introduced to participants in either condition, nor was she physically present at the location; she was thus presumably equally unfamiliar to all participants. However, it is possible that the mere act of speaking creates greater familiarity, which then changes the content of the communication for speakers versus writers. This alternate account thus suggests that not only is speaking associated with a greater wish for connectedness than writing, but it actually achieves this goal and creates greater familiarity with the recipient—and it is the latter that drives heightened SBC. Note that the results in study 2 already seem inconsistent with this account. Participants in that study simply talked into a voice recorder; no recipient was specified, and therefore recipient familiarity could not have come into play. However, we directly measured recipient familiarity in study 3 to further assess this explanation. A second alternate account is that speakers, because they discuss more self-relevant thoughts, feel more committed to the attitude they express than do writers, and it is this difference in commitment that drives the greater SBC for the former. Thus, after the familiarity item above, participants filled out a measure of their attitude toward the focal brand, Nike (1 = bad to 7 = good), followed by a measure of the extent to which they felt committed to that attitude: 1 (not at all) to 7 (very committed). Results Self-Brand Connection This study again replicated our major finding on SBC: participants reported feeling more connected to the brand when they spoke about it (Mtalk = 4.22) than when they wrote about it (Mwrite =3.44, F(1, 77) = 6.70, p = .01). Interaction Focus As predicted, participants also focused more on the social interaction with the recipient if they talked to her (Mtalk = 4.58) than wrote to her (Mwrite = 3.30, F(1, 77) = 17.79, p < .001). Furthermore, this interaction focus influenced SBC (b = .45, t(1, 77) = 4.42, p < .001). Finally, the mean indirect effect of communication channel on SBC through interaction focus (based on 5,000 bootstrap samples) was significant, with a point estimate of .49 and a 95% confidence interval excluding zero (.21 to .87). Consistent with our theorizing, therefore, these findings provide evidence for the role of interaction focus in mediating the effect of communication channel on self-brand connection. Testing Alternative Accounts Arguing against the possibility that our effects are driven by speakers’ greater perceived familiarity with the communication recipient, no significant difference was obtained on the measure of how well participants thought they knew the recipient after speaking (vs. writing) to her (Mtalk = 2.59 vs. Mwrite = 2.05, F(1, 77) = 2.22, p = .14). Note that the context of study 3 provides a particularly strong test of this familiarity-based explanation, because speakers heard the voice of the recipient when the recipient answered the phone call, whereas writers did not. Thus, a difference in perceived recipient familiarity would have been somewhat understandable. That familiarity was still not significantly different for speakers and writers suggests that this factor is not critical to the observed SBC effect. It also bears repeating that in study 2, even when there was no recipient specified and speakers simply spoke into a voice recorder, thus minimizing any role of recipient familiarity, our predicted effects were nevertheless obtained. Finally, as in other studies, our results showed that participants again did not differ in terms of their brand attitude across conditions (Mtalk = 5.10 vs. Mwrite = 4.80, F(1, 77) = 1.24, p = .27). More important, their commitment to the expressed brand attitude did not differ either (Mtalk = 4.92 vs. Mwrite = 4.63, F <1), arguing against a commitment-based account of the difference in SBC. Discussion Study 3 replicated our key SBC finding (namely, a heightened self-brand connection for those who speak vs. those who write about a liked brand), and did so using a different manipulation of speaking and writing: talking to someone on the phone versus communicating a message online. More importantly, this study obtained evidence for an important link of our theorizing—namely, that speakers, as compared to writers, display a greater focus on the interaction aspect of the communication. We have argued that this heightened interaction focus is a critical antecedent for the greater SBC that we observe for those who speak (vs. write) about the brand, and the mediational evidence from study 3 supported this reasoning. Further, study 3 findings were inconsistent with two plausible alternate accounts for the SBC effect: speakers and writers did not differ in the extent to which they were committed to their brand attitudes, nor did they differ in perceived familiarity of the communication recipient. However, study 3 did not obtain a record of the communication in the speech condition. Thus, it was not possible to compare the content of communication for speakers and writers—in particular, the extent to which the communication featured self-relevant thoughts about the brand. Accordingly, an important aspect of our theorizing remains untested: the premise that heightened interaction focus for speakers increases SBC by enhancing the extent to which the self is embedded in the brand communication. The remaining two studies provide evidence for this aspect, with study 4 adopting a moderation-based approach and study 5 a mediation-based approach. STUDY 4: INTERACTION FOCUS AS MODERATOR If, as we have argued, the increased SBC observed throughout for speakers is due to their heightened interaction focus (and the consequent increase in self-relevant brand thoughts), then either reducing the focus on recipient interaction for speakers or increasing it for writers should attenuate the SBC difference between speakers and writers. As study 2 finds, however, speaking is so inextricably associated with the idea of speaking “to someone” that speakers are likely to imagine interacting with a recipient even when none is present or even specified (e.g., even when speakers simply speak into a voice recorder). Accordingly, study 4 seeks to moderate the SBC difference observed thus far by increasing writers’ focus on interacting with the communication recipient, via explicit instructions to do so. Further, unlike in study 3, study 4 recorded the content of communication for both speakers and writers, so that we were able to examine the pattern of self-relevant brand thoughts featured in the communication, as was done in study 2. We expected this pattern to be parallel to the pattern for SBC; thus, in the default condition, speakers’ (vs. writers’) thoughts should reflect greater use of the brand for self-expression, but this difference should be attenuated if writers are induced to focus more on the interaction. Method This study comprised a 2 (communication channel: talking vs. writing) × 2 (interaction focus: default vs. high) between-subjects design. One hundred fifty-eight undergraduate students participated in this study for a cash payment of HK$40 (US$5). Participants received a questionnaire, the first page of which informed them that a master’s student from their university, named Tom, was doing research on students’ perceptions of the Apple brand. On this pretense, participants were asked to discuss anything they wished to about this brand. As in study 2, they were either asked to talk into a voice recorder or write on a piece of paper. In order to control the use of language, we instructed participants in this study to discuss the brand in English only (unlike in studies 1–3, in which they could use any language). Until this point, instructions were identical in the high-interaction-focus and default-interaction-focus conditions. At this point, however, participants in the high-interaction-focus condition were instructed that during the discussion, they should keep in mind that they were talking (writing) to Tom and should try to think about interacting with him as if they were talking (writing) to him right then.2 In the default-interaction-focus case, this instruction was omitted from both talking and writing conditions (note, though, that the recipient—“a master’s student named Tom”—had been specified for these participants as well). All participants then filled out the same measure of self-brand connection used in our previous studies. We excluded seven out of 158 participants who either discussed nothing about the brand or did not discuss the brand in English. Thus, the analyses used 151 data points. Results Self-Brand Connection Our central prediction was that speakers would report greater SBC than writers in the default condition, with an attenuation of this difference in the condition featuring a high interaction focus. Consistent with expectations, the interaction between channel and interaction focus on SBC was significant (F(1, 147) = 4.09, p = .04). Participants in the default condition again reported feeling more connected to the brand when they had talked rather than written about the brand (Mtalk = 4.33 vs. Mwrite = 3.36, F(1, 147) = 10.23, p < .01). This difference, however, was mitigated if they were induced to focus on interacting with the recipient they were communicating with (Mtalk = 4.58 vs. Mwrite = 4.46, F < 1). Thoughts-Based Self-Expression As in study 2, two independent coders coded participants’ communication for both positive and negative self-brand thoughts; the difference between the two tapped into the positive use of the brand for self-expression, which we anticipate should directly influence self-brand connection. Overall interrater agreement was high (more than 80% for each item). Note also that for the sake of simplicity, we no longer coded thoughts unrelated to the self (study 2 had done so). We expected higher scores on this index (i.e., greater net positive self-brand thoughts) for speech (vs. writing) in the default condition, but an attenuation of this difference when participants are explicitly asked to focus on interacting with the recipient. Consistent with this expectation, the interaction between channel and interaction focus was significant (F(1, 147) = 5.03, p = .03). In the default condition, the thoughts-based index revealed higher self-expression for speakers (Mdiff = 2.14) than writers (Mdiff = .17, F(1, 147) = 14.30, p < .001). However, if they were instructed to focus on the recipient, this difference was mitigated (Mdiff = 2.18 vs. Mdiff = 1.85, F < 1, in talking and writing conditions, respectively; see table 3 for details). We again conducted moderated mediation analysis using PROCESS model 7 (Hayes 2013; Preacher et al. 2007). As in study 2, the thoughts-based index of self-expression influenced SBC (b = .32, t(149) = 4.12, p < .001). Finally, as anticipated, the indirect effect of communication channel on SBC via the self-expression index was significant only in the default conditions (point estimate = .17, 95% CI = [.08, .29]), but not when participants were asked to focus on interacting with the recipient (point estimate = .03, 95% CI = [−.07, .15]). Discussion Study 4 found that simply instructing communicators to focus on the interaction with the intended recipient mitigates the effect of communication channel on self-brand connection. This result builds on the findings of study 3 to provide convergent evidence for the critical role of interaction focus in driving the observed differences between speakers and writers in the context of brand communication. The thoughts-based results in this study also provided insight into why interaction focus has the effect it does. We found that simply instructing writers to focus on the interaction with the intended recipient makes them more likely to use the brand for self-expression, thus mitigating the effect of communication channel on SBC. This result is supportive of the thesis that a greater focus on the interaction is one of the reasons for the greater self-expression typically observed for speakers than for writers. While this link between interaction focus and self-expression has been articulated at the conceptual level in the communications literature (Chafe 1982; Chafe and Danielewicz 1987; Tannen 1985), to our knowledge this is the first time it has received experimental support. The final study sought to improve on studies 3 and 4 in two ways. First, rather than explicitly instructing participants to focus on the recipient, we manipulated the context of the communication to induce interaction focus in a more natural way. Second, a multiple mediation analysis was conducted to provide support for the entire chain of reasoning—namely, in default conditions (where interaction focus has not been externally increased), talking (vs. writing) about a brand increases the extent to which favorable thoughts about the brand reference the self; in turn, such increased self-expression enhances SBC. STUDY 5: MEDIATING INFLUENCES OF INTERACTION FOCUS AND SELF-BRAND THOUGHT Study 5 investigated the role of interaction focus by varying the context of the communication—specifically, by manipulating whether or not communicators had interacted with the recipient immediately prior to the communication. We assumed that such prior interaction should increase the extent to which communicators focus on the subsequent interaction, an assumption we checked. Further, in order to keep other aspects of the communication as equivalent as possible, participants in the study were asked to communicate via a social network app (WeChat) that allows for both written and spoken communication. In the default condition (no prior interaction with recipient), we expected to observe our usual effects: higher expression of self-brand thoughts, and correspondingly higher SBC for speech (vs. writing). However, in the heightened-interaction-focus condition (immediately prior interaction with the recipient), we expected an attenuation of these effects. Method This study used a 2 (communication channel: talk vs. write) × 2 (prior interaction with recipient: no vs. yes) between-subjects design. One hundred fifty-three undergraduate students participated in this study one at a time, for a cash payment of HK$40 (US$5) each. The research assistant first introduced herself to each participant and then told the participant that the researchers were interested in students’ perception of the Adidas brand (another positively viewed brand in this population). On this pretense, participants were asked to provide their thoughts about Adidas using WeChat, a social network app (similar to WhatsApp) popular in Hong Kong, which allows people to send both written and voice messages to others. The assistant helped each participant log in to WeChat using an experiment account. Participants were asked either to type messages via WeChat (writing condition) or to generate voice messages via WeChat (speech condition). As a way of manipulating the extent to which communicators would think in terms of interacting with the recipient during communication, the second factor varied whether or not they interacted with the recipient immediately prior to the communication. In the high-interaction-focus condition, participants were asked to direct their written or oral discussion of the brand to the assistant with whom they had just had a one-on-one interaction, and who had gone to a different room after giving all instructions. Our manipulation of interaction focus was based on research showing that an initial interaction can enhance later interaction (Berger and Calabrese 1975; Carley 1991; McPherson, Smith, and Cook 2001). Similarly, in the present context, engaging in an initial face-to-face interaction with the communication recipient should increase the salience of social interaction aspect during the subsequent communication with the same person, resulting in a heightened interaction focus. As described later, our manipulation check of interaction focus bore out this assumption. In the control condition (default level of interaction focus), participants were asked to direct their brand discussion to another assistant whom they had not met before and who was also in a different room. This control condition was thus similar to previous studies in which participants were asked to share their brand perceptions with unknown others. In both conditions, after the assistant left the room, participants got only a short message via WeChat that asked them to start when they were ready. They did not receive any other messages after that while they communicated their thoughts about the brand. Next, all participants were asked to complete a paper questionnaire on which they filled out the self-brand connection measure and the interaction-focus manipulation check, identical to those used in study 3. Results Self-Brand Connection The two-way effect of communication channel and prior interaction was significant (F(1, 149) = 4.18, p = .04). When participants communicated to an assistant they had not interacted with, we replicated earlier findings: SBC was higher for talking than for writing (Mtalk = 3.75 vs. Mwrite = 3.06, F(1, 149) = 6.97, p < .01). This difference, however, was no longer significant when participants communicated to an assistant they had interacted with immediately before (Mtalk = 3.89 vs. Mwrite = 3.97, F <1). As expected, this attenuation was caused by writers experiencing greater SBC when communicating with an assistant that they had earlier interacted with (Minteraction = 3.97 vs. Mno interaction = 3.06, F(1, 149) = 11.48, p < .01). Interaction Focus Consistent with our expectations, the two-way effect of communication channel and prior interaction on the extent to which participants focused on the ongoing recipient interaction was significant (F(1, 149) = 6.32, p = .01). In particular, when participants communicated with an assistant they had not interacted with before, they focused more on the interaction aspect of communication when talking (Mtalk = 5.27) rather than writing (Mwrite = 4.37, F(1, 149) = 10.71, p < .001) to her. This aligns with our past findings showing that speakers are more likely than writers to spontaneously think in terms of interacting with the recipient. Importantly, however, this difference in interaction focus was mitigated when participants communicated with an assistant they had interacted with in person immediately prior to the communication (Mtalk = 5.24 vs. Mwrite = 5.32, F < 1). Thoughts Index of Self-Expression As in previous studies, the difference between the positive and negative self-related brand thoughts assessed the favorable use of the brand for self-expression, which we anticipate should be a proximal influence on SBC. Overall interrater agreement on the thoughts coding was high (more than 80% for each item); disagreements were resolved via discussion. The two-way interaction on this index was significant (F(1, 149) = 4.74, p = .03). In particular, when participants communicated with an assistant they had not interacted with earlier, they engaged in greater use of the brand for self-expression when talking (Mdiff = 1.80) than when writing about the brand (Mdiff = .82, F(1, 149) = 7.24, p < .01). Importantly, however, this difference disappeared when they communicated with an assistant they had personally interacted with immediately before (Mdiff = 1.66 vs. Mdiff = 1.81, F < 1; see table 4 for details). The Mediating Roles of Interaction Focus and Self-Expression The control conditions—in which participants had not personally interacted with the recipient earlier—were used to provide insight into the entire chain of reasoning for the process by which speech (vs. writing) enhances SBC under typical conditions. To briefly reiterate our theorizing, we argue that speech in itself heightens the extent to which communicators focus on the interaction aspect of the communication; this in turn increases the extent to which the self is expressed in relation to the brand; in the context of a liked brand, such use of the brand for self-expression then improves SBC. This reasoning indicates a multiple mediation process for the effect of communication channel on self-brand connection, with the distal mediator being interaction focus, and the proximal mediator of SBC being the use of the brand for self-expression (Hayes, Preacher, and Myers 2011; Preacher and Hayes 2008). This multiple mediation model was analyzed in the control conditions, with interaction focus measured by the interaction index, and self-expression captured by our thoughts-based index (see figure 2). FIGURE 2 View largeDownload slide MULTIPLE MEDIATION ANALYSIS IN THE CONTROL CONDITION FIGURE 2 View largeDownload slide MULTIPLE MEDIATION ANALYSIS IN THE CONTROL CONDITION A 5,000 resamples bootstrap approach for this model generated a 95% CI of (.05 to .33) for the multiple mediators’ indirect effect, indicating a significant multiple-mediation effect at the p <.05 level. Analysis of individual paths in the model provided further information about this multiple-mediation effect. A separately run individual set of regressions indicated significant direct effects of communication channel on interaction focus (b = .36, t = 3.34, p = .001), self-expression index (b = .32, t = 2.96, p < .01), and SBC (b = .31, t = 2.84, p < .01). When both interaction focus and self-expression were included in the multistep multiple-mediator model, however, only the individual paths from communication channel to interaction focus (b = .36, t = 3.34, p = .001), from interaction focus to self-expression (b = .35, t = 3.21, p < .01), and from self-expression to SBC (b = .49, t = 4.70, p < .001) remained significant, while all other paths became nonsignificant. This result confirms that oral (vs. written) communication induced a higher focus on the interaction aspect, which then induced a higher level of self-related brand communication, leading to enhanced self-brand connection (see figure 2). On the other hand, but also as expected, when participants communicated with an assistant they had personally interacted with immediately before, neither interaction focus nor self-expression mediated the effect of communication channel on SBC (interaction focus: 95% CI = [−.06, .03]; self-expression: 95% CI = [−.18, .12]). Discussion Study 5 provided further evidence for the mechanism by which communication channel influences SBC. Conceptually replicating study 4, we obtained our usual effects (speech increases both the number of self-brand thoughts and SBC as compared to writing) in default conditions; however, a heightened focus on interacting with the recipient attenuated the differences between speech and writing. Of note, we managed to vary interaction focus in this study by simply varying the context (such that participants either did or did not engage in a personal interaction with the recipient prior to the communication) rather than explicitly requiring them to think about interacting with the recipient. Reassuringly, the same pattern of results was obtained as in study 4 despite this change in manipulation. Finally, the multiple-mediation analysis in the control conditions provided good support for the entire process by which communication channel affects SBC under typical conditions. Namely, results supported a mechanism whereby speech increases a focus on interaction (as compared to writing), which in turn increases the expression of the self in relation to the brand, en route to enhancing SBC. GENERAL DISCUSSION Before the advent of the internet, consumer-to-consumer communication about products was, quite literally, a case of word of mouth. Nowadays, however, the prevalence of social media, email, text messaging, and the like means that consumers can share their views not just orally, but also in writing. Do these different channels of communication shape the communicators themselves? That is, does the choice of oral versus written communication then influence how the consumer might subsequently react to the brand? The current research explores this novel question in the context of positively viewed brands, suggests that this is indeed the case, and provides a possible answer as to why such an effect obtains. We merge insights from the communications literature with the research on self-brand connection to argue that the heightened self-expression in speech, as compared to writing, leads the communicating consumer to subsequently feel more connected with the brand. In turn, this increases the consumer’s tendency to invest resources into the brand, as manifested in indicators such as lower susceptibility to brand criticism, and a greater willingness to wait for the brand in cases of stock-out. Results from five studies not only document the basic effect, but also provide support for associated implications. Aligned with our theorizing, we found that explicitly encouraging writers to adopt a personal perspective attenuated the speech/writing difference by increasing writers’ self-brand connection; conversely, instructing speakers to adopt an other-perspective attenuated the difference by reducing speakers’ connection with the brand (studies 1 and 2). Importantly, we also examined one possible mechanism that can explain why speech leads to greater self-expression than writing—namely, speakers are more likely to focus on the interaction aspect of the communication, which therefore induces them to share self-related brand thoughts (e.g., personal opinions and experiences) with the communication recipient. We found both mediation- and moderation-based support for this reasoning, with study 3 showing that communication channel influenced SBC via its effect on interaction focus, and studies 4 and 5 showing that the difference between talking and writing was mitigated when writers were also encouraged to focus on the interaction with the recipient. Reassuringly, additional process insights were also supportive of the conceptualization. Across multiple studies, we found that participants in the default speech conditions, as compared to those in the default writing conditions, did indeed engage in greater self-expression featuring the liked brand, as assessed via the number of net positive self-related brand thoughts. Finally, the last study examined the complete chain of reasoning via a multiple-mediation analysis that provided support for the distal influence of interaction focus and the proximal influence of self-expression in driving the obtained SBC difference between speech and writing. While the current research thus finds good support for the view that the increased self-brand connection obtained for speech (vs. writing) is driven by increased interaction focus and consequently the heightened self-expression obtaining in the former, it is important to reiterate that speech and writing vary in a number of ways. Some of these other differences might also contribute to the difference in self-brand connection, an avenue worth exploring in future research. However, a few possible explanations for the SBC effect have already been ruled out by the current set of studies. In particular, the SBC difference for speech versus writing in our studies is unlikely to have been driven by a greater reliance on emotions in the former (study 1), by a differential use of explaining and nonexplaining language (ancillary data in study 2), or by increased recipient familiarity for speakers versus writers (study 3). Theoretical Contributions This research makes several different contributions. Perhaps of most importance, our work advances knowledge in the communications literature regarding the distinction between speech and writing. In particular, we provide support for one possible mechanism that drives the heightened self-expression for speakers as compared to writers—namely, the greater focus on interacting with the recipient that is associated with typical speech versus typical writing. In addition, our work is among the first to provide experimental evidence of the basic difference in self-expression for speakers versus writers. Finally, we show that the difference in interaction focus and self-expression between speech and writing then shapes speakers’ versus writers’ subsequent reactions to the object of communication (in this case, a favored brand). Collectively, these findings help us to obtain a deeper understanding of the fundamental differences between speech and writing. A second major contribution pertains to consumer research on word-of-mouth communications. Research in our field has generated many rich insights into this broad area (Berger 2014; Chen and Lurie 2013; Moore 2012), yet the idea of distinguishing between spoken and written word of mouth, whether in terms of outcome or process differences, is still a relatively novel one. Given that the rise in social media has provided a wider platform for written communications, it is not surprising that in recent years consumer scientists have begun to systematically examine this important distinction in communication channel. For example, an interesting insight from recent research in this area is that people tend to mention more interesting products and brands for self-enhancement concerns during written (vs. oral) communication (Berger and Iyengar 2013). The current inquiry adds to knowledge in this emergent area, documenting both novel processes (heightened self-expression, increased self-brand connection) and outcomes (greater resistance to attack, greater willingness to wait) associated with speaking versus writing about a favored brand. Our work also provides insights into when and why these effects might be attenuated. A third contribution lies in extending the literature on self-brand connections (Cheng et al. 2012; Escalas 2004; Escalas and Bettman 2003; Ferraro et al. 2013). Research in this area, which derives from the premise that brands contain symbolic value (Belk 1988; Sirgy 1982), proposes that brands can be integrated into the consumer’s self-representation, leading to a heightened self-brand connection. Quite apart from being of theoretical interest in itself, the SBC construct takes on additional significance because it exerts an influence on meaningful consequences, such as resistance to attack and willingness to wait for a valued brand (Cheng et al. 2012; Ferraro et al. 2013; Keller 2001; Sprott et al. 2009). Identifying antecedents of the self-brand connection is therefore an issue of importance for consumer scholars. Adding to past work in this area, which has shown that manipulations such as autobiographical recall and narrative processing (Escalas 2004) increase SBC, the current research identifies a novel antecedent: namely, sharing one’s views of a favored brand orally rather than in writing. Managerial Implications While not the major focus of our research, the obtained findings also contain implications for practitioners. The finding that the choice of communication channel can influence consumers’ subsequent reactions to a favored brand is of particular significance. This is especially the case given the trend, noted earlier, for marketers to allow consumers a choice of channel—spoken or written—when asking for feedback on their products. Our results suggest that at least in the case of popular brands, rather than providing such a choice, it might actually be advantageous for marketers to actively encourage spoken feedback, since oral communication can strengthen the self-brand connection and also yield beneficial downstream consequences. Of relevance to this argument is recent research by Bhattacharya, Phan, and Goh (2016), which combined purchase data from 2,301 consumers over a year with 240 million pieces of textual content they generated on a popular social network website. The authors found that the frequency with which consumers mentioned the self (e.g., the number of “I” or “my” mentions) on brand pages significantly influenced their purchase quantity of those brands. Since, as the current research finds, speech is even more likely to induce self-mentions, allowing for voice comments on Facebook brand pages (for example) should further strengthen such effects. Future Research The current conceptualization of spoken versus written word of mouth provides several interesting avenues for future exploration. The most fruitful of these, we believe, consists of delving deeper into the notion that sharing one’s brand views orally (vs. in writing) will develop stronger connections between the consumer’s self-representation and brand representation. This premise contains implications that go beyond those studied in the present research. One particularly intriguing possibility is that the heightened self-brand connection induced by oral communication will cause the consumer’s self-views to be affected by his/her views of the brand. To specify, this suggests that consumers who talk about a “creative” brand such as Apple, as opposed to writing about it, are more likely to subsequently view themselves as being creative (Fitzsimons, Chartrand, and Fitzsimons 2008) because of the heightened self-brand connection induced by speech. Such a prediction finds a parallel in recent research on egocentric categorization (Weiss and Johar 2013, 2016), which shows that the brands consumers own can directly affect their self-views. Similar to the effect of ownership, we propose that simply talking (vs. writing) about the brand might exert a reflexive impact on self-identity, a proposition that we are currently testing in our lab. The current studies examine only positively viewed brands, because the theoretical construct at the heart of this research is the self-brand connection, which can be developed only for liked brands (Escalas 2004). Departing from a SBC perspective, however, it would be interesting to examine the other processes and outcomes that might ensue as a result of talking (vs. writing) about disliked brands. One possibility is a symmetric one: just as talking about a liked brand induces consumers to talk about their personal experiences, it might do the same with a disliked brand (with the experiences related now being negative); in turn, this might increase the tendency to avoid such brands in the future, even more than writing about the brand. Another likely possibility, of course, is that people may not connect to a disliked brand at all, thus drowning out any effect of communication channel. A final, more interesting possibility, which is especially relevant when the causes of product or service failure are relatively ambiguous (Pham et al. 2010), is that the ego-centric nature of speech induces speakers to accept greater blame for such failure, as opposed to finding solely the product at fault. This would represent another instance, therefore, when encouraging consumers to “speak up” rather than “write down” would benefit marketers. Our research suggests that communication channel can activate different levels of interaction focus that then influences self-expression during communication. In addition to communication channel, another relevant aspect that may influence self-expression is the goal salient at the time of communication. For example, writers might be more likely to share personal experiences or opinions with others if they have a goal to affiliate with others or to entertain others, whereas speakers might be less likely to do so if they simply want to inform others. It would be interesting for future work to examine the interactive aspects of communication channel and communication goal on self-expression and SBC. As these speculations suggest, there is clearly much room for inquiry into the differences between spoken and written brand communication. By taking a step in that direction, the current research hopes to spark further investigation into an area that is of clear relevance to consumer behavior, but which remains understudied at this point. DATA COLLECTION INFORMATION The data reported in our studies were collected over the past five years (starting from 2014) in behavioral research labs at the Chinese University of Hong Kong (CUHK) and the Hong Kong University of Science and Technology (HKUST), Hong Kong. These studies were run by the first author (Hao Shen), with the assistance of lab managers Future Cheung, Esther NIP, and Stephy Lau. Hao Shen conducted all the data analyses. Experiment 1: Winter 2014 Experiment 2: Summer 2015 Experiment 3: Autumn 2017 Experiment 4: Winter 2016 Experiment 5: Spring 2017 Post-tests in the web appendix: Winter 2016; Autumn 2017 We would like to thank Darren Dahl, Amy Dalton, Bob Wyer, and Dengfeng Yan for their insights on this research. The research for this project was supported by research grants CUHK 14502114, CUHK 14502517, and HKUST16503115, all from the Research Grants Council of Hong Kong. Supplementary materials are included in the web appendix accompanying the online version of this article. Footnotes 1 Neither the two-way interaction (F > .25) nor the main effect of communication channel (Mtalk = 7.08 vs. Mwrite = 6.66, F(1, 155) = 1.30, p > .25) on total number of thoughts was significant. These results were consistent with past research in the communications literature (Fondacaro and Higgins 1985), refuting the possibility that SBC differences in speech versus writing were driven by message elaboration differences between the two channels. 2 Note that because the manipulation of interaction focus was very directive, we did not include a manipulation check for interaction focus in the study. In hindsight, however, we ran a post-test to make sure the manipulation worked as intended. The post-test, which is described in detail in the web appendix, confirmed the efficacy of this interaction focus manipulation. References Aaker Jennifer L. ( 1999 ), “The Malleable Self: The Role of Self-Expression in Persuasion,” Journal of Marketing Research , 36 ( 1 ), 45 – 57 . Google Scholar CrossRef Search ADS Belk Russell W. ( 1988 ), “Possessions and the Extended Self,” Journal of Consumer Research , 15 ( 2 ), 139 – 68 . Google Scholar CrossRef Search ADS Berger Charles R. , Calabrese Richard J. ( 1975 ), “Some Explorations in Initial Interaction and Beyond: Toward a Developmental Theory of Interpersonal Communication,” Human Communication Research , 1 ( 2 ), 99 – 112 . Google Scholar CrossRef Search ADS Berger Jonah ( 2014 ), “Word of Mouth and Interpersonal Communication: A Review and Directions for Future Research,” Journal of Consumer Psychology , 24 ( 4 ), 586 – 607 . Google Scholar CrossRef Search ADS Berger Jonah , Iyengar Raghuram ( 2013 ), “Communication Channels and Word of Mouth: How the Medium Shapes the Message,” Journal of Consumer Research , 40 ( 3 ), 567 – 79 . Google Scholar CrossRef Search ADS Bhattacharya Prasanta , Phan Tuan Q. , Goh Khim Yong ( 2016 ), “The Value of Social Media Brand Pages: Investigating the Effects of Self Presentation at Online Social Network Sites on Offline Purchase,” paper presented at the 1st CEIBS Marketing Conference, Department of Marketing of China Europe International Business School (CEIBS), Shanghai, China. Biber Douglas ( 1988 ), Variation across Speech and Writing , Cambridge, UK : Cambridge University Press . Google Scholar CrossRef Search ADS Carley Kathleen ( 1991 ), “A Theory of Group Stability,” American Sociological Review , 56 ( 3 ), 331 – 54 . Google Scholar CrossRef Search ADS Chafe Wallace L. ( 1982 ), “Integration and Involvement in Speaking, Writing, and Oral Literature,” in Spoken and Written Language: Exploring Orality and Literacy , ed. Tannen Deborah , Norwood, NJ : Ablex , 35 – 53 . Chafe Wallace L. ( 1985 ), “Linguistic Differences Produced by Differences between Speaking and Writing,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 105 – 23 . Chafe Wallace L. , Danielewicz Jane ( 1987 ), “Properties of Spoken and Written Language,” in Comprehending Oral and Written Language , ed. Horowitz Rosalind , Jay Samuels S. , San Diego, CA : Academic Press , 83 – 113 . Chaiken Shelly , Eagly Alice H. ( 1976 ), “Communication Modality as a Determinant of Message Persuasiveness and Message Comprehensibility,” Journal of Personality and Social Psychology , 34 ( 4 ), 605 – 14 . Google Scholar CrossRef Search ADS Chaiken Shelly , Eagly Alice H. ( 1983 ), “Communication Modality as a Determinant of Persuasion: The Role of Communicator Salience,” Journal of Personality and Social Psychology , 45 ( 2 ), 241 – 56 . Google Scholar CrossRef Search ADS Chaplin Nguyen , John Deborah Roedder ( 2005 ), “The Development of Self-Brand Connections in Children and Adolescents,” Journal of Consumer Research , 32 ( 1 ), 119 – 29 . Google Scholar CrossRef Search ADS Chen Zoey , Lurie Nicholas ( 2013 ), “Temporal Contiguity and the Negativity Bias in the Impact of Online Word of Mouth,” Journal of Marketing Research , 50 ( 4 ), 463 – 76 . Google Scholar CrossRef Search ADS Cheng Shirley Y. Y. , White Tiffany Barnett , Chaplin Nguyen ( 2012 ), “The Effects of Self-Brand Connections on Responses to Brand Failure: A New Look at the Consumer–Brand Relationship,” Journal of Consumer Psychology , 22 ( 2 ), 280 – 8 . Google Scholar CrossRef Search ADS Cialdini Robert, B. , Trost Melanie R. ( 1998 ), “Social Influence: Social Norms, Conformity, and Compliance,” in The Handbook of Social Psychology , ed. Gilbert Daniel T. , Fiske Susan T. , Lindzey Gardner , New York : McGraw-Hill , 151 – 92 . Coulmas Florian ( 1989 ), The Writing Systems of the World , Oxford, UK : Blackwell . DeVito Joseph A. ( 1966 ), “Psychogrammatical Factors in Oral and Written Discourse by Skilled Communicators,” Speech Monographs , 33 ( 1 ), 73 – 6 . Google Scholar CrossRef Search ADS Einhorn Lois ( 1978 ), “Oral and Written Style: An Examination of Differences,” Southern Speech Communication Journal , 43 ( 3 ), 302 – 11 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson ( 2004 ), “Narrative Processing: Building Consumer Connections to Brands,” Journal of Consumer Psychology , 14 ( 1–2 ), 168 – 80 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson , Bettman James R. ( 2003 ), “You Are What They Eat: The Influence of Reference Groups on Consumers’ Connections to Brands,” Journal of Consumer Psychology , 13 ( 3 ), 339 – 48 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson , Bettman James R. ( 2005 ), “Self Construal, Reference Groups and Brand Meaning,” Journal of Consumer Research , 32 ( 3 ), 378 – 89 . Google Scholar CrossRef Search ADS Ferraro Rosellina , Kirmani Amna , Matherly Ted ( 2013 ), “Look at Me! Look at Me! Conspicuous Brand Usage, Self-Brand Connection and Dilution,” Journal of Marketing Research , 50 ( 4 ), 477 – 88 . Google Scholar CrossRef Search ADS Fitzsimons Grainne M. , Chartrand Tanya L. , Fitzsimons Gavan J. ( 2008 ), “Automatic Effects of Brand Exposure on Motivated Behavior: How Apple Makes You ‘Think Different,’” Journal of Consumer Research , 35 ( 1 ), 21 – 35 . Google Scholar CrossRef Search ADS Fondacaro Rocco A. , Tory Higgins E. ( 1985 ), “Cognitive Consequences of Communication Mode: A Social Psychological Perspective,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 73 – 104 . Fournier Susan ( 1998 ), “Consumers and Their Brands: Developing Relationship Theory in Consumer Research,” Journal of Consumer Research , 24 ( 4 ), 343 – 53 . Google Scholar CrossRef Search ADS Gopinath Mahesh , Nyer Prashanth U. ( 2009 ), “The Effect of Public Commitment on Resistance to Persuasion: The Influence of Attitude Certainty, Issue Importance, Susceptibility to Normative Influence, Preference for Consistency and Source Proximity,” International Journal of Research in Marketing , 26 ( 1 ), 60 – 8 . Google Scholar CrossRef Search ADS Hayes Andrew F. ( 2013 ), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach , New York : Guilford Press . Hayes Andrew F. , Preacher Kristopher J. , Myers Teresa A. ( 2011 ), “Mediation and the Estimation of Indirect Effects in Political Communication Research,” in Sourcebook for Political Communication Research: Methods, Measures, and Analytical Techniques , ed. Bucy Erik P. , Lance Holbert R. , New York : Routledge , 434 – 65 . Jahandarie Khosrow ( 1999 ), “Structural Differences between Speech and Writing: The Linguistic Evidence,” in Spoken and Written Discourse: A Multi-Disciplinary Perspective , ed. Jahandarie Khosrow , Stamford, CT : Ablex , 131 – 50 . Keller Kevin L. ( 2001 ), “Building Customer-Based Brand Equity: A Blueprint for Creating Strong Brands,” Marketing Management ( July/August ), 15 – 19 . Klesse Anne-Kathrin , Levav Jonathan , Goukens Caroline ( 2015 ), “The Effect of Preference Expression Modality on Self-Control,” Journal of Consumer Research , 42 ( 4 ), 535 – 50 . McPherson Miller , Smith-Lovin Lynn , Cook James M. ( 2001 ), “Birds of a Feather: Homophily in Social Networks,” Annual Review of Sociology , 27 ( 1 ), 415 – 44 . Google Scholar CrossRef Search ADS Milroy James , Milroy Lesley ( 1985 ), Authority in Language: Investigating Language Prescription and Standardization , London : Routledge & Kegan Paul . Moore David J. , Homer Pamela M. ( 2008 ), “Self-Brand Connections: The Role of Attitude Strength and Autobiographical Memory Primes,” Journal of Business Research , 61 ( 7 ), 707 – 14 . Google Scholar CrossRef Search ADS Moore Sarah G. ( 2012 ), “Some Things Are Better Left Unsaid: How Word of Mouth Influences the Storyteller,” Journal of Consumer Research , 38 ( 6 ), 1140 – 54 . Google Scholar CrossRef Search ADS Morewedge Carey K. , Giblin Colleen E. , Norton Michael I. ( 2014 ), “The (Perceived) Meaning of Spontaneous Thoughts,” Journal of Experimental Psychology: General , 143 ( 4 ), 1742 – 54 . Google Scholar CrossRef Search ADS PubMed Olson David R. , Torrance Nancy ( 1981 ), “Learning to Meet the Requirements of Written Text: Language Development in the School Years,” in Writing: The Nature, Development, and Teaching of Written Communication, Vol. 2: Process, Development, and Communication , ed. Frederiksen Carl H. , Dominic Joseph F. , Hillsdale, NJ : Erlbaum , 235 – 55 . Park C. W. , MacInnis Deborah J. , Priester Joseph R. ( 2006 ), “Beyond Attitudes: Attachment and Consumer Behavior,” Seoul Journal of Business , 12 ( 2 ), 3 – 36 . Park C. W. , MacInnis Deborah J. , Priester Joseph R. ( 2008 ), “Brand Attachment: Construct, Consequences and Causes,” Foundations and Trends in Marketing , 1 ( 3 ), 190 – 230 . Park Choong W. , MacInnis Deborah J. , Priester Joseph , Eisingerich Andreas B. , Iacobucci Dawn ( 2010 ), “Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers,” Journal of Marketing , 74 ( 6 ), 1 – 17 . Google Scholar CrossRef Search ADS Pham Michel Tuan , Goukens Caroline , Lehmann Donald R. , Stuart Jennifer Ames ( 2010 ), “Shaping Customer Satisfaction through Self-Awareness Cues,” Journal of Marketing Research , 47 ( 5 ), 920 – 32 . Google Scholar CrossRef Search ADS Preacher Kristopher J. , Hayes Andrew F. ( 2008 ), “Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models,” Behavior Research Methods , 40 ( 3 ), 879 – 91 . Google Scholar CrossRef Search ADS PubMed Preacher Kristopher J. , Rucker Derek D. , Hayes Andrew F. ( 2007 ), “Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions,” Multivariate Behavioral Research , 42 ( 1 ), 185 – 227 . Google Scholar CrossRef Search ADS PubMed Rubin Donald L. ( 1987 ), “Divergence and Convergence between Oral and Written Communication,” Topics in Language Disorders , 7 ( 4 ), 1 – 18 . Google Scholar CrossRef Search ADS Schneider Walter , Shiffrin Richard M. ( 1977 ), “Controlled and Automatic Human Information Processing: I. Detection, Search and Attention,” Psychological Review , 84 ( 1 ), 1 – 66 . Google Scholar CrossRef Search ADS Shen Hao , Wyer Robert S. ( 2008 ), “Procedural Priming and Consumer Judgments: Effects on the Impact of Positively and Negatively Valenced Information,” Journal of Consumer Research , 34 ( 5 ), 727 – 37 . Google Scholar CrossRef Search ADS Sirgy Joseph M. ( 1982 ), “Self-Concept in Consumer Behavior: A Critical Review,” Journal of Consumer Research , 9 ( 3 ), 287 – 300 . Google Scholar CrossRef Search ADS Sprott David , Czellar Sandor , Spangenberg Eric ( 2009 ), “The Importance of a General Measure of Brand Engagement on Market Behavior: Development and Validation of a Scale,” Journal of Marketing Research , 46 ( 1 ), 92 – 104 . Google Scholar CrossRef Search ADS Tannen Deborah ( 1982 ), Spoken and Written Language: Exploring Orality and Literacy , Norwood, NJ : Ablex . Tannen Deborah ( 1985 ), “Relative Focus on Involvement in Oral and Written Discourse,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 124 – 47 . Unnava H. R. , Burnkrant Robert E. , Erevelles Sunil ( 1994 ), “Effects of Presentation Order and Communication Modality on Recall and Attitude,” Journal of Consumer Research , 21 ( 3 ), 481 – 90 . Google Scholar CrossRef Search ADS Wakefield Jerome C. ( 1992 ), “The Concept of Mental Disorder: On the Boundary between Biological Facts and Social Values,” American Psychologist , 47 ( 3 ), 373 – 88 . Google Scholar CrossRef Search ADS PubMed Weiss Liad , Johar Gita ( 2013 ), “Egocentric Categorization and Product Judgment: Seeing Your Traits in What You Own (and Their Opposite in What You Don’t),” Journal of Consumer Research , 40 ( 1 ), 185 – 201 . Google Scholar CrossRef Search ADS Weiss Liad , Johar Gita ( 2016 ), “Products as Self-Evaluation Standards: When Owned and Unowned Products Have Opposite Effects on Self-Judgment,” Journal of Consumer Research , 42 ( 6 ), 915 – 30 . Google Scholar CrossRef Search ADS Wood Wendy , Neal David T. ( 2007 ), “A New Look at Habits and the Habit-Goal Interface,” Psychological Review , 114 ( 4 ), 843 – 63 . Google Scholar CrossRef Search ADS PubMed Wyer Robert S. , Shen Hao , Xu Jing A . ( 2013 ), “The Role of Procedural Knowledge in the Generalization of Social Behavior,” in Oxford Handbook of Social Cognition , ed. Carlston Donald E. , New York : Oxford University Press , 257 – 81 . © The Author(s) 2018. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Consumer Research Oxford University Press

Word of Mouth versus Word of Mouse: Speaking about a Brand Connects You to It More Than Writing Does

Loading next page...
 
/lp/ou_press/word-of-mouth-versus-word-of-mouse-speaking-about-a-brand-connects-you-kcs7atYVe4
Publisher
University of Chicago Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0093-5301
eISSN
1537-5277
D.O.I.
10.1093/jcr/ucy011
Publisher site
See Article on Publisher Site

Abstract

Abstract This research merges insights from the communications literature with that on the self-brand connection to examine a novel question: how does speaking versus writing about a liked brand influence the communicator’s own later reactions to that brand? Our conceptualization argues that because oral communication involves a greater focus on social interaction with the communication recipient than does written communication, oral communicators are more likely to express self-related thoughts than are writers, thereby increasing their self-brand connection (SBC). We also assess the implications of this conceptualization, including the identification of theoretically derived boundary conditions for the speech/writing difference, and the downstream effects of heightened SBC. Results from five studies provide support for our predictions, informing both the basic literature on communications, and the body of work on consumer word of mouth. word of mouth, communication, social interaction, self-expression, self-brand connection With the rise of the internet, more and more word of mouth is being generated online. Consumers often share their thoughts about products and brands by writing online reviews, posting comments on websites, sending text messages, and so on. These forms of online communication typically involve writing/typing, whereas the traditional face-to-face offline format involves speech. This distinction between written and oral word of mouth is at the heart of the current research. Do these two different forms of word of mouth have different implications not just for the marketer, but also for the communicating consumer? In particular, if a consumer shares her views about a favorite brand either orally or in writing, will her subsequent reactions to the brand differ because of her earlier use of these different communication channels? Although past research has examined the effect of using different communication modalities (such as speech vs. writing), this work primarily focuses on the recipient—for example, how the recipient processes information differently when it is conveyed by different channels (Chaiken and Eagly 1976, 1983; Fondacaro and Higgins 1985; Unnava, Burnkrant, and Erevelles 1994). Much less work has been done on how the communication can influence the communicator, one notable exception being Moore’s (2012) research on how explaining versus simply describing an experience can influence the communicator’s subsequent attitudes toward that experience. As far as we are aware, however, neither the communications literature nor consumer research has examined the focal question at the heart of the present inquiry: namely, whether the use of different communication channels (such as speech vs. writing) influences the communicator. The conceptualization we build to address this issue integrates insights from two different areas: the extant literature on communications (Jahandarie 1999; Tannen 1982), and consumer research on the self-brand connection, which refers to the extent to which the brand representation is integrated into one’s self-identity (Escalas 2004; Escalas and Bettman 2003; Fournier 1998). Briefly, our conceptualization argues that because oral communication involves a greater focus on social interaction with the communication recipient than does written communication (Rubin 1987; Tannen 1985), oral communicators are more likely to express self-related thoughts than are writers. When a consumer is communicating about a liked brand, the use of the brand for such self-expression builds a stronger connection between the consumer’s self-representation and the brand representation. The increased self-brand connection then manifests in downstream consequences that are favorable to the marketer, such as increased resistance to an attack on the brand, and also a greater willingness to wait for it in cases of stock-out. A series of studies finds support for these predictions, and also provides process insights by identifying theoretically derived moderators for the differences between talking and writing. For example, we show that the effect of communication channel on self-brand connection is mitigated either when the level of self-expression is controlled (studies 1 and 2) or when writers are induced to focus on the interaction aspect of the communication (studies 4 and 5). In examining these predictions, the current research makes the following contributions. First, we advance basic theoretical knowledge by illuminating the processes that underlie the differences between speech and writing. Of most importance, the current research is the first, we believe, to provide empirical support for one possible mechanism that has been posited to underlie the self-expression difference between speech and writing: namely, the greater focus on interacting with the recipient that is associated with typical speech versus typical writing (Chafe 1982; Chafe and Danielewicz 1987; Tannen 1985). Second, we inform consumer research—especially that dealing with word-of-mouth effects (Berger 2014; Chen and Lurie 2013; Moore 2012)—by providing an understanding of how different communication modalities can influence the communicating consumer’s own subsequent reactions to the brand. Third, we add to the extant consumer literature on the use of brands for self-expressive purposes by identifying a new antecedent of the self-brand connection: namely, talking about the brand (Belk 1988; Escalas 2004; Escalas and Bettman 2003). Finally, from an applied perspective, our findings carry implications for marketers who wish to know when and why it might benefit them to encourage consumers to engage in oral rather than written word of mouth about their brands. THEORETICAL BACKGROUND Our theoretical platform rests on two major premises, both of which derive from the extant communications literature. First, speech (vs. writing) is associated with a greater focus on interacting with the recipient of the communication; second, this heightened interaction focus promotes the discussion of self-relevant thoughts during the process of communication. We then build on these ideas to posit that when a communicator is discussing a favored brand this increased self-expression should increase her connection with that brand. Communication Channel and Interpersonal Involvement A rich body of work in the communications literature attests to the many differences between oral and written communication (Chafe 1982; Chafe and Danielewicz 1987; Fondacaro and Higgins 1985; Jahandarie 1999). For example, compared to written communication, oral communication is faster (Fondacaro and Higgins 1985), more transient and easily forgotten (Fondacaro and Higgins 1985; Jahandarie 1999), more redundant (Jahandarie 1999), and less precise (Jahandarie 1999). Several scholars have suggested, however, that perhaps the most critical distinction between these two modes of communication is that speakers, as compared to writers, are more focused on the interaction with the audience during the process of communication, while writers are primarily focused on the information to be conveyed (Chafe 1982; Fondacaro and Higgins 1985; Jahandarie 1999; Tannen 1985). This heightened focus of speakers (vs. writers) on the social interaction aspect of communication has sometimes been referred to as a difference in “interpersonal involvement” (Chafe and Danielewicz 1987; Jahandarie 1999; Tannen 1985); however, we use the term “interaction focus” in this article in order to avoid any confusion arising from the word “involvement,” which has specific connotations in the consumer literature. The difference in interaction focus stems from the audience typically being more salient for speakers as compared to writers, a major reason for which has to do with the very nature of speaking versus writing. For the most part, speaking involves the idea of talking “to someone”—it is relatively unusual for people to talk aloud without any audience (Wakefield 1992). Writing, in contrast, is often conducted without the intended recipient being present (Chafe 1982). Thus, as Milroy and Milroy (1985, 63) put it: “Speaking is a social activity whereas writing is solitary.” This difference between the two communication channels lends itself to a stronger interaction focus in speech than in writing. Relatedly, another reason for the difference has to do with the greater shared context between the communicator and recipient in speech (vs. writing) (Chafe 1985; Rubin 1987; Tannen 1985). Quite often (even if not always) speakers are face-to-face with the audience and are therefore together in time and space. This shared context itself naturally heightens the extent to which the audience is currently salient to the speaker—that is, it increases a focus on the interaction. Indeed, oral communication is more likely than written communication to involve simultaneous feedback from the audience (Fondacaro and Higgins 1985; Rubin 1987). This allows the speaker to monitor the effect of what she is saying on the listener; conversely, the listener can signal understanding and ask for clarification (Jahandarie 1999). Again, this increases the extent to which the speaker is focused on the interpersonal interaction. In contrast, with writing the recipient is often removed in time and space; the writer at times may not even know whom the reader will be. This lack of shared context inevitably creates a detachment between communicator and recipient, just as the overlap in context between speaker and listener heightens the salience of the interaction (Chafe 1982, 1985). Communication Channel, Interaction Focus, and Self-Expression The difference in interaction focus between speech and writing affects the extent to which self-relevant thoughts are expressed during the process of communication. In one illustrative study, Chafe (1982) found that first-person references, including “I,” “me,” and “us,” occur about 61.5 times per 1,000 words for spoken discourse, but only 4.6 times per 1,000 words for written discourse. A wealth of other research has found similar evidence for the greater use of personal pronouns and expressions of personal opinions and experiences in speech compared to writing (Biber 1988; DeVito 1966; Einhorn 1978; Tannen 1985). The communications literature identifies several reasons for the greater self-expression induced by the heightened interaction focus of speakers. One such reason is that the interaction aspect of the communication is typically associated with a social function: namely, the goal of establishing a degree of connectedness with the listener (Chafe 1982; Tannen 1985). The speaker can achieve greater connectedness by embedding the self into the speech—for example, by expressing personal opinions and experiences in the conversation (Chafe 1982, 1985; Chafe and Danielewicz 1987). Indeed, as Tannen (1985) points out, in many conversations, even more important than the subject matter of the speech is the act of speaking—because the speaking itself, by encouraging self-expression, fosters connectedness. In contrast, the writer, who has a lower degree of involvement with the interaction aspect, is primarily focused on conveying information rather than establishing connectedness (Fondacaro and Higgins 1985; Rubin 1987; Tannen 1985). As a result, prototypical writing tends to be relatively objective and detached, expressing logical and propositional content as much as possible rather than allowing the intrusion of self and subjective experience (Olson and Torrance 1981). A second feature of a heightened interaction focus, which leads to convergent conclusions regarding self-expression, is that it implies a two-way loop between the speaker and the listener (Chafe 1982, 1985; Fondacaro and Higgins 1985; Tannen 1985). Thus, not only are speakers focusing on the audience, but they are aware of the audience’s focus on them. This is particularly relevant to the self-expression argument because typical speech is ephemeral: it exists only so long as the speaker chooses to speak and therefore cannot be separated from the speaker (Jahandarie 1999; Rubin 1987). Accordingly, as Jahandarie puts it (1999, 139), the important question for the audience in deciphering speech is “What does the speaker mean?” as opposed to the focal question with writing of “What do the words mean?” (see also Coulmas 1989). Because speakers are likely to be aware of the audience’s focus on them as individuals, this not only gives them license to inject their selves into the speech, but actively encourages it (Chafe 1985; Rubin 1987). In contrast, writing exists as a permanent record, and has an existence of its own even after the writer is done writing or typing (we use the terms “typing” and “writing” interchangeably in this article). Therefore, unlike the interaction-focused speaker, the writer is unlikely to believe that the reader expects the communication to reflect the idiosyncrasies of the communicator; rather, the writer’s goal is to produce a communication that holds value even when the writer is no longer present (Jahandarie 1999). Thus, writing typically serves to “disembody the written text from the writer” (Rubin 1987, 10), lessening the role of the self. These related strands of reasoning converge on the conclusion that oral communication typically entails greater self-expression than writing, as captured in interrelated terms such as ego involvement (Chafe 1985; Jahandarie 1999) and self-activation (Chafe 1982; Rubin 1987). The heightened self-expression in speech manifests in a greater number of self-related thoughts, such as the inclusion of personal experiences and opinions rather than factual content (Biber 1988; Chafe 1982; Tannen 1985). In the current context of communicating about a brand, this suggests that speakers will offer more personal opinions about the brand and also will be more likely to discuss their personal experiences with it, whereas writers are more likely to maintain a relatively objective detachment in their brand communication. Two points of clarification are in order. First, we do not wish to claim that a heightened interaction focus is the only mechanism by which communication channel influences self-expression. Other mechanisms may also be at play. For example, people tend to spontaneously generate their thoughts during oral communication, but they have more time to deliberate on what they should write (Berger and Iyengar 2013; Fondacaro and Higgins 1985; Klesse, Levav, and Goukens 2015). Because the self tends to be highly accessible, spontaneous communication is likely to rely more on self-related thoughts than deliberative communication, with the latter offering communicators more opportunity to access nonself thoughts as well (Morewedge, Giblin, and Norton 2014). But although we believe the self-expression difference between speech and writing is multiply determined, this research delves into the role of interaction focus because of considerable convergence in the communications literature that it is the crucial factor underlying several speech/writing differences (Chafe 1982; Chafe and Danielewicz 1986; Tannen 1985). Note also that the posited difference in interaction focus between speech and writing, and the consequent difference in self-expression, is emblematic of prototypical speech and prototypical writing. We are in agreement with Tannen (1985) that not all forms of writing are detached from the recipient, nor do all forms of speech signify strong involvement with the interaction (e.g., memorizing and delivering a formal speech is likely associated with a relatively low focus on the interaction aspect). However, research on procedural knowledge has robustly shown that procedures that are initially learned to fulfill a particular purpose can become overlearned so that they are automatically reproduced even when that initial purpose is no longer salient (Schneider and Shiffrin 1977; Shen and Wyer 2008; Wood and Neal 2007; Wyer, Shen, and Xu 2013). Similarly, it seems likely that to begin with, people deliberately express the self more often when talking face-to-face with others (vs. writing to an unknown recipient), because doing so fulfills salient goals, such as facilitating social interaction. With frequent repetition, however, this self-expression procedure will become strongly associated with speech in general, to the point of being performed automatically even in nonprototypical situations. Speech, Writing, and Self-Brand Connection The difference in self-expression between speaking and writing contains clear implications, we believe, for the extent to which consumers develop a self-brand connection as a result of communicating about a favored brand. Originating in the premise that consumers often use favored brands and products to establish, maintain, and signal self-concept (Aaker 1999; Belk 1988; Chaplin and John 2005; Cheng, White, and Chaplin 2012; Escalas 2004; Fournier 1998; Sirgy 1982), a self-brand connection (SBC) is defined as the strength of the link between a consumer’s self-representation and brand representation (Moore and Homer 2008). Thus, SBC captures the extent to which the brand is integrated into the consumer’s idea of self (Escalas 2004; Ferraro, Kirmani, and Matherly 2013; Fournier 1998; Park, Macinnis, and Priester 2008). Note that the very nature of a self-brand connection means that this concept applies to the case of positively viewed brands, which is the focus of the current research. At the same time, it should be noted that SBC and attitude extremity are two different constructs. Thus, it is possible to hold positive attitudes toward brands with which one lacks a self-brand connection (e.g., brands that a consumer evaluates favorably on purely functional grounds, without necessarily forming a link to self-representation). As Escalas (2004; footnote 1) points out, “although the existence of a SBC implies a positive brand attitude, the inverse is not true. Consumers may have favorable attitudes towards many brands where no SBCs exist.” How, then, does a high SBC develop? According to the literature, SBC is strengthened when consumers develop connections between their brand thoughts and their self-representations (Cheng et al. 2012; Escalas and Bettman 2003; Park, MacInnis, and Priester 2006). For example, a brand that expresses a valued aspect of self-identity—such as an environmentally friendly lawnmower—will be associated with a high SBC because consumer cognitions about the brand are likely to be connected to thoughts of the self (Escalas and Bettman 2003). In a similar vein, SBC is likely to be high when the brand is used to “communicate the self-concept to others” (Escalas and Bettman 2005, 378). Merging these insights on SBC with the communications literature reviewed earlier, the current research argues that talking about a liked brand (as opposed to writing about it) should also enhance the self-brand connection. Both speaking and writing about a brand involve communicating about it to others, and in both channels, given the context of a liked brand, this communication is likely to be primarily of a positive nature. However, the greater self-expression involved in speech (itself arising from a heightened interaction focus) should manifest in speakers expressing a greater number of self-relevant positive thoughts about the brand. For example, speaking is more likely to involve statements such as “I like Apple because I am a creative kind of person” versus “Apple is a creative brand.” The former type of thought, by definition, should produce a stronger connection with the brand, because it explicitly integrates the brand with the self (see figure 1). FIGURE 1 View largeDownload slide CONCEPTUAL FRAMEWORK FIGURE 1 View largeDownload slide CONCEPTUAL FRAMEWORK Implications This conceptualization yields several testable implications relating to the distinction between speaking and writing about a brand. Our research focuses on two such implications: a) downstream consequences, and b) moderators of the effect. It is worth reiterating that the SBC construct—and therefore the current inquiry—focuses exclusively on positively viewed brands. Downstream Consequences A high self-brand connection has been shown to benefit the brand in multiple ways. Because high SBC reflects a sense of “oneness” with the brand, consumers with a high SBC are motivated to maintain a connection with the brand even if there is a cost involved in doing so (Park et al. 2010)—as manifested, for example, in a greater willingness to wait for the brand if they have to (Keller 2001; Sprott, Czellar, and Spangenberg 2009). Similarly, because strongly connected brands form a part of the consumer’s self-identity, they are also less susceptible to criticism, because consumers are more likely to defend such a brand from an attack (Cheng et al. 2012; Ferraro et al. 2013). Tying these findings to the current perspective, we predict that the enhanced SBC arising from consumers’ speaking (vs. writing) about a favored brand should yield positive consequences for the marketer, including an increased willingness to wait for the brand, and increased defense against brand criticism. Moderators of the Speak versus Write Effect Our theorizing implicates two types of moderators—one arising from the proximal influencer on SBC (extent of self-expression) and the other arising from the more distal influencer (extent of interaction focus). Regarding the former, we posit that speaking increases the extent to which consumers discuss self-related thoughts when discussing the brand, leading to a stronger connection between the two than in the case of writing. If true, this suggests that one can attenuate the predicted SBC difference between speaking and writing by increasing (decreasing) writers’ (speakers’) focus on the self during brand communication. Thus, explicitly encouraging writers to focus on self-relevant thoughts while discussing the brand should increase SBC for writers. Conversely, encouraging speakers to adopt another person’s perspective and recount what such a person might think about the brand should decrease SBC for speakers. We examine both of these predictions (studies 1 and 2). Second, a key reason for the greater self-expression in speech (vs. writing), we have argued, is the heightened interaction focus in the former: speakers are more focused on the interaction with the recipient than are writers. This argument suggests that exogenously inducing writers to focus more on the interaction aspect of the communication should also increase the expression of self-related brand thoughts, and thereby strengthen writers’ connection with the brand about which they are communicating. We examine those possibilities in studies 3–5. STUDY 1: THE MODERATING INFLUENCE OF SELF-PERSPECTIVE Study 1 had several objectives. First, it sought to provide initial evidence for the effect of communication channel on self-brand connection. Second, this study investigated a theoretically derived moderator of the SBC difference produced by speaking (vs. writing) about the brand. As discussed above, while the SBC difference is believed to arise because speakers spontaneously engage in self-expression during communication, encouraging writers to also engage in self-expression (e.g., by explicitly instructing them to adopt a personal perspective when discussing the brand) should enhance the strength of the link between the brand representation and the self-representation for writers as well, thereby increasing SBC. Third, a downstream consequence of the basic SBC effect was also measured—namely, consumers’ willingness to wait for the brand they had discussed, given an out-of-stock situation. Method This study comprised a 2 (communication channel: speaking vs. writing) × 2 (discussion focus: default vs. explicit self-focus) between-subjects design. Two hundred fifty undergraduate students at a Hong Kong university participated in this study for a cash payment of HK$40 (US$5). They were grouped into pairs and told that we were interested in students’ views of brands. Apple, a brand pretested to be viewed favorably across this student population (as were the brands used in later studies), was the focus of communication. Those in the speaking condition were randomly assigned to either be a communicator or a listener, with one participant in each pair asked to talk about his or her views of Apple to the other. In this condition, communicators were told that we would like them to share their views of Apple with their partner. Those in the writing condition were simply asked to share their views of Apple by writing them down on a piece of paper; the partner was asked to complete a different questionnaire irrelevant to this study. Note that in this case, the presence of the partners was not required; they were not described as being the recipient of the communicator’s thoughts. However, having a partner present even in the writing condition allowed us to check that our predicted effects would still hold even when the writer is not physically alone. We did this to rule out the possibility that the effects are driven simply by a possible increase in self-consciousness caused by the sheer physical presence of another person. We manipulated the focus factor by instructing participants in the default condition to discuss anything they wished to about the brand. In all conditions, participants were told that: a) they could freely share any of their thoughts about the brand; b) they could use any language; c) they could take as long as they liked. In the self-focus condition, they were additionally asked to adopt a first-person perspective when discussing the brand, and were also encouraged to think about their personal experiences when doing so. After they had provided their views of the brand, all communicators (n = 125) responded to a questionnaire. (Because this research focuses on the effect of communication channel on the communicator’s reaction toward the brand, the partner was not asked to complete this questionnaire.) Communicators first reported their attitude toward the Apple brand by rating it on a five-point scale (1 = bad to 5 = good). They were then exposed to the following scenario: “Imagine that you participate in a lottery and luckily win an Apple notebook computer. However, you are told that this computer is out of stock now. You need to wait for 1 month for the computer to be available at the store. Alternatively, you can get a Dell notebook computer immediately, which has the same price.” Then, they were asked to indicate the extent to which they were willing to wait for the Apple notebook from 1 (not likely at all) to 7 (very likely). Next, all participants filled out a standard measure of self-brand connection. This measure, taken from Escalas (2004), consists of seven items, such as: “This brand reflects who I am,” “I can identify with this brand,” and “I feel a personal connection to this brand.” Participants indicated their agreements with each item along a seven-point scale (1 = disagree, 7 = agree; α = .91). Finally, to examine whether the communications channel also leads to differential expression of feelings versus cognitions, we asked participants to indicate the extent to which their discussion of the brand comprised “mainly their thoughts about the brand” (1) versus “mainly their feelings about the brand” (7). Results and Discussion Self-Brand Connection We predicted that participants would feel more connected to Apple if they talked rather than wrote about the brand; further, this effect would be attenuated when they were explicitly instructed to adopt a personal perspective during the discussion. In support of this prediction, a 2 × 2 ANOVA revealed a marginally significant interaction effect of communication channel and focus of discussion (F(1, 121) = 3.75, p = .06). Those in the default condition felt more connected to the brand when they talked (Mtalk = 4.24) rather than wrote about it (Mwrite = 3.38, F(1, 121) = 6.69, p = .01). However, this effect disappeared if they were explicitly asked to adopt a personal perspective while discussing the brand (Mtalk = 4.05 vs. Mwrite = 4.10, F <1). Viewed differently, and also in line with our theorizing, SBC increased for writers in the self-focus condition (Mself-focus = 4.10 vs. Mdefault = 3.38, F(1, 121) = 4.56, p = .03); no such difference was observed for speakers (Mself-focus = 4.05 vs. Mdefault = 4.24, F < 1). Also, this pattern of self-brand connections was not driven by brand attitudes. Only a main effect of the focus factor was obtained on attitude toward the brand, such that participants reported liking the brand more when they focused on the self while discussing it than if not (Mself-focus = 4.25 vs. Mdefault = 4.00, F(1, 121) = 4.74, p = .03). No other effects were significant. Finally, no differences were observed as to whether participants mainly discussed thoughts or feelings about the brand (all ps > .20; see table 1 for details). TABLE 1 THE EFFECT OF COMMUNICATION CHANNEL AND DISCUSSION FOCUS ON SELF-BRAND CONNECTION AND WILLINGNESS TO WAIT (STUDY 1) Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 1 THE EFFECT OF COMMUNICATION CHANNEL AND DISCUSSION FOCUS ON SELF-BRAND CONNECTION AND WILLINGNESS TO WAIT (STUDY 1) Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a Default Self-focus Talking Writing Talking Writing (N = 31) (N = 30) (N = 34) (N =30) Self-brand connection 4.24 (1.19)a 3.38 (1.59)b 4.05 (1.22)a 4.10 (1.19)a Brand attitude 4.13 (.67)ab 3.87 (.81)a 4.24 (.55)b 4.27 (.52)b Willingness to wait 6.30 (1.24)a 5.26 (2.14)b 6.06 (1.30)a 6.07 (1.57)a Relative focus on thoughts versus feelings 3.74 (1.86)a 3.37 (1.96)a 3.91 (1.56)a 3.97 (1.61)a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 2 THE EFFECT OF COMMUNICATION CHANNEL AND COMMUNICATION PERSPECTIVE ON THOUGHTS, SELF-BRAND CONNECTION, AND ATTITUDE CHANGE (STUDY 2) Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 2 THE EFFECT OF COMMUNICATION CHANNEL AND COMMUNICATION PERSPECTIVE ON THOUGHTS, SELF-BRAND CONNECTION, AND ATTITUDE CHANGE (STUDY 2) Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc Default Other-perspective Talking Writing Talking Writing (N = 37) (N = 41) (N = 39) (N = 42) Self-brand connection 3.42 (1.39)a 2.71 (1.04)b 2.86 (1.03)b 3.02 (1.32)ab Initial attitude 3.51 (.73)a 3.31 (.99)a 3.54 (.68)a 3.36 (.82)a Attitude after attack 2.96 (.87)a 2.51 (1.02)b 2.76 (.79)ab 2.61 (.87)b Attitude change .55 (.40)a .80 (.55)b .78 (.43)b .75 (.60)b Susceptibility to attack 3.70 (1.45)a 4.59 (1.40)b 4.64 (1.53)b 4.52 (1.71)b Total number of thoughts 8.05 (2.94)a 7.22 (2.39)a 6.15 (2.15)b 6.12 (2.06)b Number of positive brand thoughts related to self 4.08 (2.87)a 1.05 (1.41)b .41 (.99)c .12 (.50)c Number of negative brand thoughts related to self 1.86 (1.70)a .80 (1.08)b .38 (.99)c .21 (.90)c Mdiff (net positive self-related brand thoughts) 2.22 (3.79)a .25 (1.71)b .03 (1.31)b –.09 (1.03)b Number of positive brand thoughts unrelated to self .22 (.67)a 2.27 (1.87)b 3.08 (2.24)c 3.38 (2.10)c Number of negative brand thoughts unrelated to self .65 (1.27)a 1.17 (1.32)b 1.36 (1.51)bc 1.62 (1.68)c Mdiff (net positive self-unrelated brand thoughts) –.43 (1.46)a 1.10 (2.60)b 1.72 (3.06)bc 1.76 (3.39)bc NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 3 THE EFFECT OF COMMUNICATION CHANNEL AND INTERACTION FOCUS ON THOUGHTS AND SELF-BRAND CONNECTION (STUDY 4) Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 3 THE EFFECT OF COMMUNICATION CHANNEL AND INTERACTION FOCUS ON THOUGHTS AND SELF-BRAND CONNECTION (STUDY 4) Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a Default Interaction focus Talking Writing Talking Writing (N =37) (N = 36) (N =39) (N =39) Self-brand connection 4.33 (1.22)a 3.36 (1.37)b 4.58 (1.39)a 4.46 (1.21)a Number of positive brand thoughts related to self 2.60 (1.92)a .47 (1.08)b 3.00 (2.07)a 2.80 (2.28)a Number of negative brand thoughts related to self .46 (.73)a .30 (.71)a .82 (1.04)ab .95 (1.57)b Mdiff (net positive self-related brand thoughts) 2.14a .17b 2.18a 1.85a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 4 THE EFFECT OF COMMUNICATION CHANNEL AND PRIOR RECIPIENT INTERACTION ON ONGOING INTERACTION FOCUS, THOUGHTS, AND SELF-BRAND CONNECTION (STUDY 5) Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. TABLE 4 THE EFFECT OF COMMUNICATION CHANNEL AND PRIOR RECIPIENT INTERACTION ON ONGOING INTERACTION FOCUS, THOUGHTS, AND SELF-BRAND CONNECTION (STUDY 5) Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a Control Prior recipient interaction Talking Writing Talking Writing (N = 40) (N =38) (N =38) (N =37) Self-brand connection 3.75 (1.12)a 3.06 (1.05)b 3.89 (1.32)a 3.97 (1.17)a Ongoing interaction focus 5.27 (1.16)a 4.37 (1.22)b 5.24 (1.33)a 5.32 (1.13)a Number of positive brand thoughts related to self 2.58 (1.38)a 1.00 (1.12)b 2.40 (1.52)a 2.46 (1.48)a Number of negative brand thoughts related to self .78 (.89)a .18 (.51)b .74 (1.16)a .65 (.92)ab Mdiff (net positive self-related brand thoughts) 1.80a .82b 1.66a 1.81a NOTE.—Standard deviations are shown in parentheses. Cells with no overlapping alphabets in superscripts differ at p < .05. Willingness to Wait (WTW) Drawing on previous research (Park et al. 2006, 2008; Sprott et al. 2009), we have argued that higher SBC should produce a greater willingness to wait for the brand. In support of this argument, the two-way interaction of communication channel and discussion focus on willingness to wait was marginally significant (F(1, 121) = 3.35, p = .07). As with SBC, participants in the default condition were more willing to wait for the brand if they had talked rather than written about it (Mtalk = 6.30 vs. Mwrite = 5.26, F(1, 121) = 6.46, p = .01). Further, this effect disappeared in the explicit self-focus condition (Mtalk = 6.06 vs. Mwrite = 6.07, F <1; see table 1 for details). Discussion This study provides initial support for our major prediction: consumers feel more connected to a brand after they talk rather than write about it. It also informs the finding in two major ways. First, it explores a downstream consequence of the basic effect, showing that the increased SBC resulting from speaking about the brand also yields a greater willingness to wait for that brand. Second, consistent with crucial role posited for self-expression in driving SBC, we found that when writers were explicitly encouraged to focus on the self, the resultant SBC (and willingness to wait) was as high as in the speech condition. It is worth reiterating that the pattern of self-brand connection was not driven by brand attitudes. Thus, unlike with SBC, we did not observe an interaction effect of communication channel and focus of discussion on brand attitudes (p > .20). This is consistent with past research showing that brand attitude is a different construct from SBC (Escalas 2004; Park et al. 2008). In particular, not all routes to brand liking necessarily involve self-expression. For example, Escalas (2004) found that autobiographical recall (remembering past experiences of a brand) improved SBC compared to mental simulation (imagining future use of the brand) because autobiographical memories are more closely tied to the self than mental simulations; however, because the two processes do not differ in terms of the valence of thoughts expressed about the brand, there was no difference in brand attitudes. Similarly in the current context, both speaking and writing about a favored brand are likely to evoke positively valenced thoughts about the brand; thus, as observed, the two conditions do not differ in terms of brand attitudes. We posit, however, that the greater self-expression in the former case enhances SBC. Study 1 was not without limitations, however, the most important of which concerned the lack of an objective measure of self-expression (although see the web appendix for a self-report measure). This limitation is addressed in study 2, which analyzes the actual content of participants’ communications to analyze a brand-specific measure of self-expression. Another drawback of study 1 was that speakers, but not writers, were assigned a specific communication recipient. In order to minimize the possibility that the effects were driven simply by this difference, the next study relaxes one of the conditions associated with prototypical speech—namely, being face-to-face with the listener. Thus, participants in the speech condition of study 2 were asked to speak into a voice recorder rather than sharing their thoughts with a live person. It might be argued that the absence of a physical audience should lessen the focus on the interaction and therefore reduce self-expression, attenuating the SBC difference between speech and writing. As noted earlier, however, the principles of procedural accessibility suggest that because speakers have repeatedly practiced a procedure of expressing the self to others to facilitate social interaction, such self-expression is likely to obtain even in nonprototypical conditions. In particular, speech is so strongly associated with the idea of “talking to” someone, it is likely that speakers will imagine an audience even if none is physically present—an assumption we test—which would then cause them to express the self more, enhancing SBC. STUDY 2: THE MODERATING INFLUENCE OF OTHER-PERSPECTIVE In addition to generalizing the basic SBC effect to the context described above (i.e., speakers in this study spoke into a voice recorder), study 2 extended our previous findings in three important directions. First, drawing on prior research showing that consumers are more likely to refute criticism of a brand with which they feel a stronger connection (Ferraro et al. 2013), we examined the prediction that talking (vs. writing) about a brand should increase resistance to an attack, because of the higher SBC in the former case. Second, this study examined another moderator of the speech/writing SBC difference. If this difference is based, as we have argued, on greater self-expression when speaking, it should be attenuated if participants are explicitly instructed to discuss what they believe to be another person’s perceptions of the brand. Taking an other-perspective should dampen self-expression and therefore SBC for speakers, bringing it closer to the default level for writers. This boundary condition can be seen as the reverse of that tested in study 1, in which we increased SBC for writers (to the default level for speakers) by explicitly forcing a personal perspective in communication. Third, as noted above, we analyzed the communications provided by participants in this study in detail to obtain insights into the self-expression process by which speech versus writing affects SBC. Method This study comprised a 2 (communication channel: talking vs. writing) × 2 (communication perspective: default vs. other-perspective) between-subjects design. One hundred sixty-four undergraduate students participated in this study for a cash payment of HK$40 (US$5). All participants were initially told that we were interested in students’ views of various brands. To manipulate the perspective factor, we then told those in the default condition that we would like them to provide their views of Samsung (again, a positively viewed brand in this population). In the other-perspective condition, participants were asked to think of someone else who had used Samsung products, and to provide their views of how that person perceived the brand. With regard to the communications channel, those in the writing condition were asked to write their views on a piece of paper that they were given for that purpose. In the speaking condition, they were given a voice recorder and asked to talk about the brand into the recorder. Thus, the two conditions were made equivalent in that (unlike in study 1) neither speakers nor writers were asked to communicate to an externally specified recipient. All participants completed the study one at a time. Also, in all conditions, after giving the instructions, the experimenter exited the room and left the participant alone to write/speak about the brand and then complete the subsequent questionnaire. Therefore, any observed effects of communication channel on SBC and attitude resistance are unlikely to proceed from differences in the extent of public commitment to one’s position (Cialdini and Trost 1998; Gopinath and Nyer 2009). After participants had finished writing/talking about the brand, they provided initial attitudes toward it by rating it on a five-point scale (1 = bad to 5 = good). Next, they read the following purported comment by another customer of the brand: “I am using a Samsung mobile phone. It used to work just fine. But a couple of days ago, I had problems loading the map. Sometimes it took around 5 seconds to display the searching result. The calendar reminders also failed to work properly. For some events, the reminders go off a few minutes after the events. I had tried to fix the problems but they still occur. This causes a bit of inconvenience to my social and work life.” After they finished reading the scenario, participants once again reported their attitudes toward Samsung, this time along a 10-point scale from 1 (bad) to 10 (good). A different scale was used to prevent participants from simply retrieving and reporting their earlier evaluation. Scores on the second scale were divided by two to correspond to the initial five-point scale. The difference between the initial and the latter attitudes provided a measure of post-attack attitude change. We also assessed a self-report measure of susceptibility to attack by then asking participants to indicate the extent to which the other customer’s negative comment had influenced their final attitude toward Samsung, along a scale from 1 (not at all) to 7 (very much). Finally, all participants filled out the same standard measure of self-brand connection used in study 1. We excluded five out of 164 participants who either discussed nothing about the brand or failed to answer all the questions. Thus, the analyses used 159 data points. Results Self-Brand Connection We expected to replicate our earlier findings in the default condition—that is, greater SBC following spoken (vs. written) brand communication—but we predicted an attenuation of this effect when participants were explicitly instructed to discuss another person’s views of the brand. A 2 × 2 ANOVA revealed a significant interaction of communication channel and communications perspective (F(1, 155) = 5.08, p = .03). As predicted, participants in the default condition felt more connected to the brand after talking (Mtalk = 3.42) rather than writing about it (Mwrite = 2.71, F(1, 155) = 6.61, p = .01). However, this effect disappeared in the other-perspective condition (Mtalk = 2.86 vs. Mwrite = 3.02, F <1). Attitude Change As in the previous study, no effects were observed on initial brand attitude (ps > .10). However, we expected a systematic pattern for the attitude change measure (the difference between initial and post-attack attitudes). In particular, following the SBC results, we predicted lower attitude change—that is, greater resistance to attack—for speech (vs. writing) in the default conditions; however, this difference in attitude resistance should be attenuated in the other-perspective conditions. Results were consistent with these predictions. The two-way interaction between communication channel and communication perspective on attitude change was marginally significant (F(1, 155) = 3.13, p = .08). Those in the default condition displayed less attitude change when they had earlier talked (Mdiff = .55) rather than written about the brand (Mdiff =.80, F(1, 155) = 4.82, p = .03). However, this effect was no longer observed in the other-perspective conditions (Mdiff = .78 vs. Mdiff = .75, in talking vs. writing conditions, respectively, F <1). A reassuringly convergent pattern was observed on the self-reported measure of participants’ susceptibility to the attack. To elaborate, a significant interaction (F(1, 155) = 4.23, p = .04) was based on lower susceptibility to attack for speech (vs. writing) in the default condition (Mtalk = 3.70 vs. Mwrite = 4.59, F(1, 155) = 6.47, p = .01), while no such difference was observed in the other-perspective conditions (Mtalk = 4.64 vs. Mwrite = 4.52, F <1). Process Measures Our theorizing holds that in default conditions (but not in the other-perspective condition), talking versus writing about the brand increases the self-brand connection because the greater self-expression associated with speech induces speakers to generate a greater number of self-relevant brand thoughts, creating the positive link between the self and brand that is captured by heightened SBC. We assessed this self-expression tendency by examining the number of thoughts explicitly relating the self to the brand (e.g., personal opinions of the brand, and personal experiences with it). Furthermore, since our goal was to assess the extent to which brand-related self-expression heightens SBC, which refers to a positive connection with the brand, we focused on the net number of such self-brand thoughts that reflected a positive view of the brand. This measure thus specifically isolates the favorable use of the brand for self-expression, which we argue should mediate the effect of communication channel on SBC. Two independent coders, unaware of our experimental hypotheses and participant condition, coded each participant’s thoughts about the brand for the following: (a) total number of thoughts;1 (b) number of both positive and negative thoughts about the brand that include a reference both to the brand and to the self. This included personal experiences, defined as thoughts about the brand reflecting actual experience (e.g., “I have used a Samsung smartphone for a long time and find it very comfortable to use”; “I find my Samsung computer often gets problems”), and personal opinions, defined as subjective evaluations of a brand that do not explicitly refer to personal experience (e.g., personal beliefs, observations, or knowledge about a brand, such as “I like the Samsung brand; my personal view is that Samsung’s smartphone has a very good design”; “I don’t like the advertisement of Samsung”). The difference between the number of such positive self-brand thoughts and the number of negative self-brand thoughts (i.e., the net positive self-related brand thoughts) served as our key measure of self-expression. (c) number of both positive and negative thoughts about the brand that do not include mentions of self (e.g., “Samsung is a brand that is popular in Hong Kong”; “A lot of people are using Samsung smartphones”; “Samsung smartphones have a lower market share than iPhone in Hong Kong”). The difference between positive and negative thoughts in this case produced an index of net positive self-unrelated thoughts. Overall interrater agreement was high (more than 80% for any item). Differences between coder ratings were resolved through discussion. We expect that in the default conditions, speech (vs. writing) should be associated with higher scores on our index of brand-related self-expression (i.e., net positive self-brand thoughts) and, conversely, fewer thoughts about the brand that do not reflect such self-expression (i.e., net positive self-unrelated brand thoughts). Such thought patterns would correspond with the higher SBC observed for speech (vs. writing) in the default case. In contrast, we did not expect to observe differences in the thought indices in the other-perspective conditions, where speech and writing did not differ on SBC. Main effects of the two factors were obtained on both the self-expression index and also on the index of self-unrelated brand thoughts (ps < .001). More importantly, the two-way interaction between person involved and communication channel was also significant for both indices (ps < .01). In support of our conceptualization, speakers in the default condition mentioned a greater number of net positive self-related brand thoughts, indicating greater use of the brand for self-expression (Mdiff = 2.22) than writers (Mdiff = .25, F(1, 155) = 15.75, p < .001), while mentioning fewer net positive brand thoughts that were unrelated to the self (Mdiff = –.43 vs. 1.10, in talking vs. writing conditions, respectively, F(1, 155) = 6.01, p = .02). Further, consistent with the null effect on SBC given other-focus, speakers and writers did not differ on either of these thought indices when discussing another person’s perception of the brand (net positive self-related brand thoughts: Mdiff = .03 vs. –.09, in talking vs. writing conditions, respectively, F <1; net positive self-unrelated brand thoughts: Mdiff = 1.72 vs. 1.76, F <1; see table 2 for details). Finally, we assessed whether the thoughts-based measure of self-expression mediated the interactive effect of communication channel and person involved on SBC, as our theorizing would argue. To examine this possibility, we conducted moderated mediation analysis using PROCESS model 7 (Hayes 2013; Preacher, Rucker, and Hayes 2007). As reported previously, the effect of interaction between channel and person involved on SBC was significant. In addition, across conditions, the thoughts-based index of self-expression had a positive effect on SBC (b = .45, t(157) = 6.38, p < .001). Finally, the mean indirect effect of the interaction between communication channel and person involved on SBC through self-expression (based on 5,000 bootstrap samples) was significant, with a 95% confidence interval excluding zero (.05 to .43). More specifically, the indirect effect of communication channel on SBC through self-expression was significant only in the default conditions (point estimate = .23, 95% CI = [.08, .44]), but not when participants were asked to discuss another person’s perception of the brand (point estimate = .01, 95% CI = [−.04, .08]). These results provide good evidence for the role of self-expressive thoughts in mediating the interactive effect of communication channel and person involved on self-brand connection. Explaining versus Nonexplaining Language Recent research finds that the use of explaining language to discuss a hedonic experience (e.g., explaining why the person likes it) lowers post-discussion evaluations of the experience, as compared to the use of nonexplaining language (e.g., simply describing and providing an evaluation of the experience without including explanations) (Moore 2012). Accordingly, it was important to explore whether writers and speakers differ in their use of explaining language, and if this affects SBC. We note that post-communication brand attitudes did not differ for the speech and writing conditions in our studies, which in itself argues against this account (to reiterate, the use of explaining language has been found to dampen attitudes). To further test this explanation, two independent coders were asked to process each communication in the default conditions and code each sentence providing any positive view of the brand as either explaining or nonexplaining language, following Moore (2012): a) explaining (e.g., sentences including reasons for a positive evaluation, such as “The Samsung mobile phone is good because it is very comfortable to use”), and b) nonexplaining (e.g., sentences including a positive evaluation without a reason provided, such as “I like the Samsung brand” or “The Samsung brand is good”). The overall agreement was high (more than 80% for each item). We calculated the proportion of explaining language (PEL) by dividing the number of explaining sentences by the total number of explaining and nonexplaining sentences (Moore 2012). We did not observe any difference across conditions (talking conditions: PEL = .48, SD = .39 vs. writing conditions: PEL = .45, SD = .33, F < 1). Therefore, the effects observed are unlikely to be due to differences in the use of explaining language. Discussion Study 2 not only provided further evidence for our key premise that speaking (vs. writing) about a favored brand can increase the self-brand connection, it also provided an assessment of some of the other implications of our conceptualization. First, adding to the study 1 finding that talking (vs. writing) about a brand increases the willingness to wait for it, we found that talking about the brand has another benefit to the marketer: it produces a greater resistance to a subsequent attack on the brand. Second, study 2 provides process insight into this effect by documenting another theoretically consistent boundary condition. In reassuring symmetry with our earlier finding that the SBC difference between writing and talking is attenuated when writers are explicitly instructed to adopt a personal perspective (thereby increasing their SBC), study 2 found that this difference is also mitigated when speakers are explicitly instructed to describe another person’s views of the brand (thereby decreasing their SBC). Both of these boundary conditions are consistent with the basic premise that greater self-expression during brand communication leads to heightened SBC. Third, study 2 provided more direct support for this premise by assessing the extent to which the brand was used for self-expression, as measured by the thoughts that linked the brand with the self. An analysis of participants’ thoughts revealed, as predicted, that the default speech condition did indeed produce a relatively high number of net positive self-brand thoughts, which in turn led to increased SBC. Note that the overall level of self-expression can be assessed through another measure—namely, the frequency of use of the personal pronoun “I,” with such usage presumably being greater for speakers than writers (Chafe 1982, 1985). However, we did not rely on this gross measure because it does not capture the use of the brand for self-expression (e.g., a statement such as “I am feeling happy today” does not invoke the brand even though it uses “I”), nor does it necessarily reflect positive brand views, both of which are an integral aspect of the process by which brand-related self-expression heightens SBC. That said, across studies, we computed both the thoughts-based measure of self-expression used in study 2 as well as the simple frequency count of “I” mentions. The pattern of results on the two measures paralleled each other, as reported in the web appendix. Finally, it is worth noting that in the default conditions of this study, a greater number of self-brand thoughts, and accordingly an increased SBC, manifested for speakers (vs. writers) even though no audience was present (speakers spoke into a voice recorder). We argue that this is the case because speech is so strongly associated with the idea of an audience that speakers automatically imagine interacting with one even when none is actually present. In turn, this causes them to engage in greater self-expression, as manifested in the greater number of thoughts that explicitly referenced the self in relation to the object of discussion (i.e., the brand). A post-test was run to verify this assumption. As in the default conditions of study 2, participants were asked to discuss a liked brand either by talking into a voice recorder or by writing down their thoughts on a piece of paper. All participants were then asked to report the extent to which they thought about the communication recipient while discussing the brand. Results revealed that, as anticipated, speakers did indeed imagine an actual audience—even though none was present—to a significantly greater extent than writers (please see the web appendix for full details of the procedure and findings). STUDY 3: THE ROLE OF INTERACTION FOCUS Studies 1 and 2 provided consistent evidence that speaking about a liked brand increases the self-brand connection as compared to writing about it; further, study 2 showed that this effect is due to the greater expression of self-relevant brand thoughts by speakers (vs. writers). The next three studies provide further insight into the process. As discussed earlier, we argue that the reason speakers engage in greater self-expression than writers (which, in the brand communication context, manifests as a greater number of self-related brand thoughts) has to do with their heightened focus on the interaction. This increased interaction focus then induces the speaker to embed the self into the conversation (Chafe 1982, 1985; Fondacaro and Higgins 1985; Tannen 1985). In order to provide an initial test of this reasoning, study 3 measured interaction focus along with self-brand connection. For greater generalizability, study 3 also used a different manipulation to induce speech versus writing, with speakers (writers) being asked to make a phone call (write an online message) about the brand. Finally, because we have already demonstrated the downstream consequences of SBC in studies 1 and 2, we did not examine those consequences again in study 3 and subsequent studies. Instead, these remaining studies focused on replicating the basic SBC effect and providing insights into the mechanism underlying the effect. Method Seventy-nine undergraduate students participated in this study for a cash payment of HK$40 (US$5). Participants were told that the researchers were interested in students’ perception of another brand pretested to be well liked in this population: Nike. They were then asked to communicate anything they wished about this brand to a research assistant who (purportedly) could not come to the experiment room at the time, because she was in a different location. On this pretense, half of the participants were asked to communicate with the assistant via a phone call (on receiving the call, the assistant simply asked each participant to start talking about the brand, and did not give any feedback during the communication). The other participants were asked to communicate their message about Nike in an online survey and were informed that their communication would be sent to the assistant immediately after they clicked Submit. Participants were informed that their messages would not be preserved in either condition. Note that because messages were indeed not preserved in either condition (it would not have been possible to do for speakers since no voice recorder was used in this study), study 3 does not contain any measures of self-expression. Rather, the goal of this study is to provide an initial test of the role of interaction focus in influencing self-brand connection. To assess this, after participants in both conditions finished communicating their thoughts about Nike, they were asked to complete a questionnaire. They first filled out the seven-item SBC measure, which is the key dependent variable. Participants then indicated their level of agreement with three statements assessing the extent to which they were focused on the interaction aspect during communication: (a) when they discussed the brand, they imagined themselves talking/writing to the assistant that they were communicating with; (b) when they discussed the brand, they communicated similarly to the way they would during a social interaction; (c) when they discussed the brand, they talked/wrote as if they were socially interacting with the assistant they were sending the message to. Each item was anchored by 1 (disagree) to 7 (agree). The three items were highly correlated with one another (α = .89) and were averaged to create an index for interaction focus. Finally, all participants responded to two measures, each of which served to examine a possible alternate explanation for the SBC pattern observed for speakers versus writers. First, they were asked how much they felt they knew the research assistant with whom they had communicated: 1 (not at all) to 7 (very much). Note that the assistant had not been introduced to participants in either condition, nor was she physically present at the location; she was thus presumably equally unfamiliar to all participants. However, it is possible that the mere act of speaking creates greater familiarity, which then changes the content of the communication for speakers versus writers. This alternate account thus suggests that not only is speaking associated with a greater wish for connectedness than writing, but it actually achieves this goal and creates greater familiarity with the recipient—and it is the latter that drives heightened SBC. Note that the results in study 2 already seem inconsistent with this account. Participants in that study simply talked into a voice recorder; no recipient was specified, and therefore recipient familiarity could not have come into play. However, we directly measured recipient familiarity in study 3 to further assess this explanation. A second alternate account is that speakers, because they discuss more self-relevant thoughts, feel more committed to the attitude they express than do writers, and it is this difference in commitment that drives the greater SBC for the former. Thus, after the familiarity item above, participants filled out a measure of their attitude toward the focal brand, Nike (1 = bad to 7 = good), followed by a measure of the extent to which they felt committed to that attitude: 1 (not at all) to 7 (very committed). Results Self-Brand Connection This study again replicated our major finding on SBC: participants reported feeling more connected to the brand when they spoke about it (Mtalk = 4.22) than when they wrote about it (Mwrite =3.44, F(1, 77) = 6.70, p = .01). Interaction Focus As predicted, participants also focused more on the social interaction with the recipient if they talked to her (Mtalk = 4.58) than wrote to her (Mwrite = 3.30, F(1, 77) = 17.79, p < .001). Furthermore, this interaction focus influenced SBC (b = .45, t(1, 77) = 4.42, p < .001). Finally, the mean indirect effect of communication channel on SBC through interaction focus (based on 5,000 bootstrap samples) was significant, with a point estimate of .49 and a 95% confidence interval excluding zero (.21 to .87). Consistent with our theorizing, therefore, these findings provide evidence for the role of interaction focus in mediating the effect of communication channel on self-brand connection. Testing Alternative Accounts Arguing against the possibility that our effects are driven by speakers’ greater perceived familiarity with the communication recipient, no significant difference was obtained on the measure of how well participants thought they knew the recipient after speaking (vs. writing) to her (Mtalk = 2.59 vs. Mwrite = 2.05, F(1, 77) = 2.22, p = .14). Note that the context of study 3 provides a particularly strong test of this familiarity-based explanation, because speakers heard the voice of the recipient when the recipient answered the phone call, whereas writers did not. Thus, a difference in perceived recipient familiarity would have been somewhat understandable. That familiarity was still not significantly different for speakers and writers suggests that this factor is not critical to the observed SBC effect. It also bears repeating that in study 2, even when there was no recipient specified and speakers simply spoke into a voice recorder, thus minimizing any role of recipient familiarity, our predicted effects were nevertheless obtained. Finally, as in other studies, our results showed that participants again did not differ in terms of their brand attitude across conditions (Mtalk = 5.10 vs. Mwrite = 4.80, F(1, 77) = 1.24, p = .27). More important, their commitment to the expressed brand attitude did not differ either (Mtalk = 4.92 vs. Mwrite = 4.63, F <1), arguing against a commitment-based account of the difference in SBC. Discussion Study 3 replicated our key SBC finding (namely, a heightened self-brand connection for those who speak vs. those who write about a liked brand), and did so using a different manipulation of speaking and writing: talking to someone on the phone versus communicating a message online. More importantly, this study obtained evidence for an important link of our theorizing—namely, that speakers, as compared to writers, display a greater focus on the interaction aspect of the communication. We have argued that this heightened interaction focus is a critical antecedent for the greater SBC that we observe for those who speak (vs. write) about the brand, and the mediational evidence from study 3 supported this reasoning. Further, study 3 findings were inconsistent with two plausible alternate accounts for the SBC effect: speakers and writers did not differ in the extent to which they were committed to their brand attitudes, nor did they differ in perceived familiarity of the communication recipient. However, study 3 did not obtain a record of the communication in the speech condition. Thus, it was not possible to compare the content of communication for speakers and writers—in particular, the extent to which the communication featured self-relevant thoughts about the brand. Accordingly, an important aspect of our theorizing remains untested: the premise that heightened interaction focus for speakers increases SBC by enhancing the extent to which the self is embedded in the brand communication. The remaining two studies provide evidence for this aspect, with study 4 adopting a moderation-based approach and study 5 a mediation-based approach. STUDY 4: INTERACTION FOCUS AS MODERATOR If, as we have argued, the increased SBC observed throughout for speakers is due to their heightened interaction focus (and the consequent increase in self-relevant brand thoughts), then either reducing the focus on recipient interaction for speakers or increasing it for writers should attenuate the SBC difference between speakers and writers. As study 2 finds, however, speaking is so inextricably associated with the idea of speaking “to someone” that speakers are likely to imagine interacting with a recipient even when none is present or even specified (e.g., even when speakers simply speak into a voice recorder). Accordingly, study 4 seeks to moderate the SBC difference observed thus far by increasing writers’ focus on interacting with the communication recipient, via explicit instructions to do so. Further, unlike in study 3, study 4 recorded the content of communication for both speakers and writers, so that we were able to examine the pattern of self-relevant brand thoughts featured in the communication, as was done in study 2. We expected this pattern to be parallel to the pattern for SBC; thus, in the default condition, speakers’ (vs. writers’) thoughts should reflect greater use of the brand for self-expression, but this difference should be attenuated if writers are induced to focus more on the interaction. Method This study comprised a 2 (communication channel: talking vs. writing) × 2 (interaction focus: default vs. high) between-subjects design. One hundred fifty-eight undergraduate students participated in this study for a cash payment of HK$40 (US$5). Participants received a questionnaire, the first page of which informed them that a master’s student from their university, named Tom, was doing research on students’ perceptions of the Apple brand. On this pretense, participants were asked to discuss anything they wished to about this brand. As in study 2, they were either asked to talk into a voice recorder or write on a piece of paper. In order to control the use of language, we instructed participants in this study to discuss the brand in English only (unlike in studies 1–3, in which they could use any language). Until this point, instructions were identical in the high-interaction-focus and default-interaction-focus conditions. At this point, however, participants in the high-interaction-focus condition were instructed that during the discussion, they should keep in mind that they were talking (writing) to Tom and should try to think about interacting with him as if they were talking (writing) to him right then.2 In the default-interaction-focus case, this instruction was omitted from both talking and writing conditions (note, though, that the recipient—“a master’s student named Tom”—had been specified for these participants as well). All participants then filled out the same measure of self-brand connection used in our previous studies. We excluded seven out of 158 participants who either discussed nothing about the brand or did not discuss the brand in English. Thus, the analyses used 151 data points. Results Self-Brand Connection Our central prediction was that speakers would report greater SBC than writers in the default condition, with an attenuation of this difference in the condition featuring a high interaction focus. Consistent with expectations, the interaction between channel and interaction focus on SBC was significant (F(1, 147) = 4.09, p = .04). Participants in the default condition again reported feeling more connected to the brand when they had talked rather than written about the brand (Mtalk = 4.33 vs. Mwrite = 3.36, F(1, 147) = 10.23, p < .01). This difference, however, was mitigated if they were induced to focus on interacting with the recipient they were communicating with (Mtalk = 4.58 vs. Mwrite = 4.46, F < 1). Thoughts-Based Self-Expression As in study 2, two independent coders coded participants’ communication for both positive and negative self-brand thoughts; the difference between the two tapped into the positive use of the brand for self-expression, which we anticipate should directly influence self-brand connection. Overall interrater agreement was high (more than 80% for each item). Note also that for the sake of simplicity, we no longer coded thoughts unrelated to the self (study 2 had done so). We expected higher scores on this index (i.e., greater net positive self-brand thoughts) for speech (vs. writing) in the default condition, but an attenuation of this difference when participants are explicitly asked to focus on interacting with the recipient. Consistent with this expectation, the interaction between channel and interaction focus was significant (F(1, 147) = 5.03, p = .03). In the default condition, the thoughts-based index revealed higher self-expression for speakers (Mdiff = 2.14) than writers (Mdiff = .17, F(1, 147) = 14.30, p < .001). However, if they were instructed to focus on the recipient, this difference was mitigated (Mdiff = 2.18 vs. Mdiff = 1.85, F < 1, in talking and writing conditions, respectively; see table 3 for details). We again conducted moderated mediation analysis using PROCESS model 7 (Hayes 2013; Preacher et al. 2007). As in study 2, the thoughts-based index of self-expression influenced SBC (b = .32, t(149) = 4.12, p < .001). Finally, as anticipated, the indirect effect of communication channel on SBC via the self-expression index was significant only in the default conditions (point estimate = .17, 95% CI = [.08, .29]), but not when participants were asked to focus on interacting with the recipient (point estimate = .03, 95% CI = [−.07, .15]). Discussion Study 4 found that simply instructing communicators to focus on the interaction with the intended recipient mitigates the effect of communication channel on self-brand connection. This result builds on the findings of study 3 to provide convergent evidence for the critical role of interaction focus in driving the observed differences between speakers and writers in the context of brand communication. The thoughts-based results in this study also provided insight into why interaction focus has the effect it does. We found that simply instructing writers to focus on the interaction with the intended recipient makes them more likely to use the brand for self-expression, thus mitigating the effect of communication channel on SBC. This result is supportive of the thesis that a greater focus on the interaction is one of the reasons for the greater self-expression typically observed for speakers than for writers. While this link between interaction focus and self-expression has been articulated at the conceptual level in the communications literature (Chafe 1982; Chafe and Danielewicz 1987; Tannen 1985), to our knowledge this is the first time it has received experimental support. The final study sought to improve on studies 3 and 4 in two ways. First, rather than explicitly instructing participants to focus on the recipient, we manipulated the context of the communication to induce interaction focus in a more natural way. Second, a multiple mediation analysis was conducted to provide support for the entire chain of reasoning—namely, in default conditions (where interaction focus has not been externally increased), talking (vs. writing) about a brand increases the extent to which favorable thoughts about the brand reference the self; in turn, such increased self-expression enhances SBC. STUDY 5: MEDIATING INFLUENCES OF INTERACTION FOCUS AND SELF-BRAND THOUGHT Study 5 investigated the role of interaction focus by varying the context of the communication—specifically, by manipulating whether or not communicators had interacted with the recipient immediately prior to the communication. We assumed that such prior interaction should increase the extent to which communicators focus on the subsequent interaction, an assumption we checked. Further, in order to keep other aspects of the communication as equivalent as possible, participants in the study were asked to communicate via a social network app (WeChat) that allows for both written and spoken communication. In the default condition (no prior interaction with recipient), we expected to observe our usual effects: higher expression of self-brand thoughts, and correspondingly higher SBC for speech (vs. writing). However, in the heightened-interaction-focus condition (immediately prior interaction with the recipient), we expected an attenuation of these effects. Method This study used a 2 (communication channel: talk vs. write) × 2 (prior interaction with recipient: no vs. yes) between-subjects design. One hundred fifty-three undergraduate students participated in this study one at a time, for a cash payment of HK$40 (US$5) each. The research assistant first introduced herself to each participant and then told the participant that the researchers were interested in students’ perception of the Adidas brand (another positively viewed brand in this population). On this pretense, participants were asked to provide their thoughts about Adidas using WeChat, a social network app (similar to WhatsApp) popular in Hong Kong, which allows people to send both written and voice messages to others. The assistant helped each participant log in to WeChat using an experiment account. Participants were asked either to type messages via WeChat (writing condition) or to generate voice messages via WeChat (speech condition). As a way of manipulating the extent to which communicators would think in terms of interacting with the recipient during communication, the second factor varied whether or not they interacted with the recipient immediately prior to the communication. In the high-interaction-focus condition, participants were asked to direct their written or oral discussion of the brand to the assistant with whom they had just had a one-on-one interaction, and who had gone to a different room after giving all instructions. Our manipulation of interaction focus was based on research showing that an initial interaction can enhance later interaction (Berger and Calabrese 1975; Carley 1991; McPherson, Smith, and Cook 2001). Similarly, in the present context, engaging in an initial face-to-face interaction with the communication recipient should increase the salience of social interaction aspect during the subsequent communication with the same person, resulting in a heightened interaction focus. As described later, our manipulation check of interaction focus bore out this assumption. In the control condition (default level of interaction focus), participants were asked to direct their brand discussion to another assistant whom they had not met before and who was also in a different room. This control condition was thus similar to previous studies in which participants were asked to share their brand perceptions with unknown others. In both conditions, after the assistant left the room, participants got only a short message via WeChat that asked them to start when they were ready. They did not receive any other messages after that while they communicated their thoughts about the brand. Next, all participants were asked to complete a paper questionnaire on which they filled out the self-brand connection measure and the interaction-focus manipulation check, identical to those used in study 3. Results Self-Brand Connection The two-way effect of communication channel and prior interaction was significant (F(1, 149) = 4.18, p = .04). When participants communicated to an assistant they had not interacted with, we replicated earlier findings: SBC was higher for talking than for writing (Mtalk = 3.75 vs. Mwrite = 3.06, F(1, 149) = 6.97, p < .01). This difference, however, was no longer significant when participants communicated to an assistant they had interacted with immediately before (Mtalk = 3.89 vs. Mwrite = 3.97, F <1). As expected, this attenuation was caused by writers experiencing greater SBC when communicating with an assistant that they had earlier interacted with (Minteraction = 3.97 vs. Mno interaction = 3.06, F(1, 149) = 11.48, p < .01). Interaction Focus Consistent with our expectations, the two-way effect of communication channel and prior interaction on the extent to which participants focused on the ongoing recipient interaction was significant (F(1, 149) = 6.32, p = .01). In particular, when participants communicated with an assistant they had not interacted with before, they focused more on the interaction aspect of communication when talking (Mtalk = 5.27) rather than writing (Mwrite = 4.37, F(1, 149) = 10.71, p < .001) to her. This aligns with our past findings showing that speakers are more likely than writers to spontaneously think in terms of interacting with the recipient. Importantly, however, this difference in interaction focus was mitigated when participants communicated with an assistant they had interacted with in person immediately prior to the communication (Mtalk = 5.24 vs. Mwrite = 5.32, F < 1). Thoughts Index of Self-Expression As in previous studies, the difference between the positive and negative self-related brand thoughts assessed the favorable use of the brand for self-expression, which we anticipate should be a proximal influence on SBC. Overall interrater agreement on the thoughts coding was high (more than 80% for each item); disagreements were resolved via discussion. The two-way interaction on this index was significant (F(1, 149) = 4.74, p = .03). In particular, when participants communicated with an assistant they had not interacted with earlier, they engaged in greater use of the brand for self-expression when talking (Mdiff = 1.80) than when writing about the brand (Mdiff = .82, F(1, 149) = 7.24, p < .01). Importantly, however, this difference disappeared when they communicated with an assistant they had personally interacted with immediately before (Mdiff = 1.66 vs. Mdiff = 1.81, F < 1; see table 4 for details). The Mediating Roles of Interaction Focus and Self-Expression The control conditions—in which participants had not personally interacted with the recipient earlier—were used to provide insight into the entire chain of reasoning for the process by which speech (vs. writing) enhances SBC under typical conditions. To briefly reiterate our theorizing, we argue that speech in itself heightens the extent to which communicators focus on the interaction aspect of the communication; this in turn increases the extent to which the self is expressed in relation to the brand; in the context of a liked brand, such use of the brand for self-expression then improves SBC. This reasoning indicates a multiple mediation process for the effect of communication channel on self-brand connection, with the distal mediator being interaction focus, and the proximal mediator of SBC being the use of the brand for self-expression (Hayes, Preacher, and Myers 2011; Preacher and Hayes 2008). This multiple mediation model was analyzed in the control conditions, with interaction focus measured by the interaction index, and self-expression captured by our thoughts-based index (see figure 2). FIGURE 2 View largeDownload slide MULTIPLE MEDIATION ANALYSIS IN THE CONTROL CONDITION FIGURE 2 View largeDownload slide MULTIPLE MEDIATION ANALYSIS IN THE CONTROL CONDITION A 5,000 resamples bootstrap approach for this model generated a 95% CI of (.05 to .33) for the multiple mediators’ indirect effect, indicating a significant multiple-mediation effect at the p <.05 level. Analysis of individual paths in the model provided further information about this multiple-mediation effect. A separately run individual set of regressions indicated significant direct effects of communication channel on interaction focus (b = .36, t = 3.34, p = .001), self-expression index (b = .32, t = 2.96, p < .01), and SBC (b = .31, t = 2.84, p < .01). When both interaction focus and self-expression were included in the multistep multiple-mediator model, however, only the individual paths from communication channel to interaction focus (b = .36, t = 3.34, p = .001), from interaction focus to self-expression (b = .35, t = 3.21, p < .01), and from self-expression to SBC (b = .49, t = 4.70, p < .001) remained significant, while all other paths became nonsignificant. This result confirms that oral (vs. written) communication induced a higher focus on the interaction aspect, which then induced a higher level of self-related brand communication, leading to enhanced self-brand connection (see figure 2). On the other hand, but also as expected, when participants communicated with an assistant they had personally interacted with immediately before, neither interaction focus nor self-expression mediated the effect of communication channel on SBC (interaction focus: 95% CI = [−.06, .03]; self-expression: 95% CI = [−.18, .12]). Discussion Study 5 provided further evidence for the mechanism by which communication channel influences SBC. Conceptually replicating study 4, we obtained our usual effects (speech increases both the number of self-brand thoughts and SBC as compared to writing) in default conditions; however, a heightened focus on interacting with the recipient attenuated the differences between speech and writing. Of note, we managed to vary interaction focus in this study by simply varying the context (such that participants either did or did not engage in a personal interaction with the recipient prior to the communication) rather than explicitly requiring them to think about interacting with the recipient. Reassuringly, the same pattern of results was obtained as in study 4 despite this change in manipulation. Finally, the multiple-mediation analysis in the control conditions provided good support for the entire process by which communication channel affects SBC under typical conditions. Namely, results supported a mechanism whereby speech increases a focus on interaction (as compared to writing), which in turn increases the expression of the self in relation to the brand, en route to enhancing SBC. GENERAL DISCUSSION Before the advent of the internet, consumer-to-consumer communication about products was, quite literally, a case of word of mouth. Nowadays, however, the prevalence of social media, email, text messaging, and the like means that consumers can share their views not just orally, but also in writing. Do these different channels of communication shape the communicators themselves? That is, does the choice of oral versus written communication then influence how the consumer might subsequently react to the brand? The current research explores this novel question in the context of positively viewed brands, suggests that this is indeed the case, and provides a possible answer as to why such an effect obtains. We merge insights from the communications literature with the research on self-brand connection to argue that the heightened self-expression in speech, as compared to writing, leads the communicating consumer to subsequently feel more connected with the brand. In turn, this increases the consumer’s tendency to invest resources into the brand, as manifested in indicators such as lower susceptibility to brand criticism, and a greater willingness to wait for the brand in cases of stock-out. Results from five studies not only document the basic effect, but also provide support for associated implications. Aligned with our theorizing, we found that explicitly encouraging writers to adopt a personal perspective attenuated the speech/writing difference by increasing writers’ self-brand connection; conversely, instructing speakers to adopt an other-perspective attenuated the difference by reducing speakers’ connection with the brand (studies 1 and 2). Importantly, we also examined one possible mechanism that can explain why speech leads to greater self-expression than writing—namely, speakers are more likely to focus on the interaction aspect of the communication, which therefore induces them to share self-related brand thoughts (e.g., personal opinions and experiences) with the communication recipient. We found both mediation- and moderation-based support for this reasoning, with study 3 showing that communication channel influenced SBC via its effect on interaction focus, and studies 4 and 5 showing that the difference between talking and writing was mitigated when writers were also encouraged to focus on the interaction with the recipient. Reassuringly, additional process insights were also supportive of the conceptualization. Across multiple studies, we found that participants in the default speech conditions, as compared to those in the default writing conditions, did indeed engage in greater self-expression featuring the liked brand, as assessed via the number of net positive self-related brand thoughts. Finally, the last study examined the complete chain of reasoning via a multiple-mediation analysis that provided support for the distal influence of interaction focus and the proximal influence of self-expression in driving the obtained SBC difference between speech and writing. While the current research thus finds good support for the view that the increased self-brand connection obtained for speech (vs. writing) is driven by increased interaction focus and consequently the heightened self-expression obtaining in the former, it is important to reiterate that speech and writing vary in a number of ways. Some of these other differences might also contribute to the difference in self-brand connection, an avenue worth exploring in future research. However, a few possible explanations for the SBC effect have already been ruled out by the current set of studies. In particular, the SBC difference for speech versus writing in our studies is unlikely to have been driven by a greater reliance on emotions in the former (study 1), by a differential use of explaining and nonexplaining language (ancillary data in study 2), or by increased recipient familiarity for speakers versus writers (study 3). Theoretical Contributions This research makes several different contributions. Perhaps of most importance, our work advances knowledge in the communications literature regarding the distinction between speech and writing. In particular, we provide support for one possible mechanism that drives the heightened self-expression for speakers as compared to writers—namely, the greater focus on interacting with the recipient that is associated with typical speech versus typical writing. In addition, our work is among the first to provide experimental evidence of the basic difference in self-expression for speakers versus writers. Finally, we show that the difference in interaction focus and self-expression between speech and writing then shapes speakers’ versus writers’ subsequent reactions to the object of communication (in this case, a favored brand). Collectively, these findings help us to obtain a deeper understanding of the fundamental differences between speech and writing. A second major contribution pertains to consumer research on word-of-mouth communications. Research in our field has generated many rich insights into this broad area (Berger 2014; Chen and Lurie 2013; Moore 2012), yet the idea of distinguishing between spoken and written word of mouth, whether in terms of outcome or process differences, is still a relatively novel one. Given that the rise in social media has provided a wider platform for written communications, it is not surprising that in recent years consumer scientists have begun to systematically examine this important distinction in communication channel. For example, an interesting insight from recent research in this area is that people tend to mention more interesting products and brands for self-enhancement concerns during written (vs. oral) communication (Berger and Iyengar 2013). The current inquiry adds to knowledge in this emergent area, documenting both novel processes (heightened self-expression, increased self-brand connection) and outcomes (greater resistance to attack, greater willingness to wait) associated with speaking versus writing about a favored brand. Our work also provides insights into when and why these effects might be attenuated. A third contribution lies in extending the literature on self-brand connections (Cheng et al. 2012; Escalas 2004; Escalas and Bettman 2003; Ferraro et al. 2013). Research in this area, which derives from the premise that brands contain symbolic value (Belk 1988; Sirgy 1982), proposes that brands can be integrated into the consumer’s self-representation, leading to a heightened self-brand connection. Quite apart from being of theoretical interest in itself, the SBC construct takes on additional significance because it exerts an influence on meaningful consequences, such as resistance to attack and willingness to wait for a valued brand (Cheng et al. 2012; Ferraro et al. 2013; Keller 2001; Sprott et al. 2009). Identifying antecedents of the self-brand connection is therefore an issue of importance for consumer scholars. Adding to past work in this area, which has shown that manipulations such as autobiographical recall and narrative processing (Escalas 2004) increase SBC, the current research identifies a novel antecedent: namely, sharing one’s views of a favored brand orally rather than in writing. Managerial Implications While not the major focus of our research, the obtained findings also contain implications for practitioners. The finding that the choice of communication channel can influence consumers’ subsequent reactions to a favored brand is of particular significance. This is especially the case given the trend, noted earlier, for marketers to allow consumers a choice of channel—spoken or written—when asking for feedback on their products. Our results suggest that at least in the case of popular brands, rather than providing such a choice, it might actually be advantageous for marketers to actively encourage spoken feedback, since oral communication can strengthen the self-brand connection and also yield beneficial downstream consequences. Of relevance to this argument is recent research by Bhattacharya, Phan, and Goh (2016), which combined purchase data from 2,301 consumers over a year with 240 million pieces of textual content they generated on a popular social network website. The authors found that the frequency with which consumers mentioned the self (e.g., the number of “I” or “my” mentions) on brand pages significantly influenced their purchase quantity of those brands. Since, as the current research finds, speech is even more likely to induce self-mentions, allowing for voice comments on Facebook brand pages (for example) should further strengthen such effects. Future Research The current conceptualization of spoken versus written word of mouth provides several interesting avenues for future exploration. The most fruitful of these, we believe, consists of delving deeper into the notion that sharing one’s brand views orally (vs. in writing) will develop stronger connections between the consumer’s self-representation and brand representation. This premise contains implications that go beyond those studied in the present research. One particularly intriguing possibility is that the heightened self-brand connection induced by oral communication will cause the consumer’s self-views to be affected by his/her views of the brand. To specify, this suggests that consumers who talk about a “creative” brand such as Apple, as opposed to writing about it, are more likely to subsequently view themselves as being creative (Fitzsimons, Chartrand, and Fitzsimons 2008) because of the heightened self-brand connection induced by speech. Such a prediction finds a parallel in recent research on egocentric categorization (Weiss and Johar 2013, 2016), which shows that the brands consumers own can directly affect their self-views. Similar to the effect of ownership, we propose that simply talking (vs. writing) about the brand might exert a reflexive impact on self-identity, a proposition that we are currently testing in our lab. The current studies examine only positively viewed brands, because the theoretical construct at the heart of this research is the self-brand connection, which can be developed only for liked brands (Escalas 2004). Departing from a SBC perspective, however, it would be interesting to examine the other processes and outcomes that might ensue as a result of talking (vs. writing) about disliked brands. One possibility is a symmetric one: just as talking about a liked brand induces consumers to talk about their personal experiences, it might do the same with a disliked brand (with the experiences related now being negative); in turn, this might increase the tendency to avoid such brands in the future, even more than writing about the brand. Another likely possibility, of course, is that people may not connect to a disliked brand at all, thus drowning out any effect of communication channel. A final, more interesting possibility, which is especially relevant when the causes of product or service failure are relatively ambiguous (Pham et al. 2010), is that the ego-centric nature of speech induces speakers to accept greater blame for such failure, as opposed to finding solely the product at fault. This would represent another instance, therefore, when encouraging consumers to “speak up” rather than “write down” would benefit marketers. Our research suggests that communication channel can activate different levels of interaction focus that then influences self-expression during communication. In addition to communication channel, another relevant aspect that may influence self-expression is the goal salient at the time of communication. For example, writers might be more likely to share personal experiences or opinions with others if they have a goal to affiliate with others or to entertain others, whereas speakers might be less likely to do so if they simply want to inform others. It would be interesting for future work to examine the interactive aspects of communication channel and communication goal on self-expression and SBC. As these speculations suggest, there is clearly much room for inquiry into the differences between spoken and written brand communication. By taking a step in that direction, the current research hopes to spark further investigation into an area that is of clear relevance to consumer behavior, but which remains understudied at this point. DATA COLLECTION INFORMATION The data reported in our studies were collected over the past five years (starting from 2014) in behavioral research labs at the Chinese University of Hong Kong (CUHK) and the Hong Kong University of Science and Technology (HKUST), Hong Kong. These studies were run by the first author (Hao Shen), with the assistance of lab managers Future Cheung, Esther NIP, and Stephy Lau. Hao Shen conducted all the data analyses. Experiment 1: Winter 2014 Experiment 2: Summer 2015 Experiment 3: Autumn 2017 Experiment 4: Winter 2016 Experiment 5: Spring 2017 Post-tests in the web appendix: Winter 2016; Autumn 2017 We would like to thank Darren Dahl, Amy Dalton, Bob Wyer, and Dengfeng Yan for their insights on this research. The research for this project was supported by research grants CUHK 14502114, CUHK 14502517, and HKUST16503115, all from the Research Grants Council of Hong Kong. Supplementary materials are included in the web appendix accompanying the online version of this article. Footnotes 1 Neither the two-way interaction (F > .25) nor the main effect of communication channel (Mtalk = 7.08 vs. Mwrite = 6.66, F(1, 155) = 1.30, p > .25) on total number of thoughts was significant. These results were consistent with past research in the communications literature (Fondacaro and Higgins 1985), refuting the possibility that SBC differences in speech versus writing were driven by message elaboration differences between the two channels. 2 Note that because the manipulation of interaction focus was very directive, we did not include a manipulation check for interaction focus in the study. In hindsight, however, we ran a post-test to make sure the manipulation worked as intended. The post-test, which is described in detail in the web appendix, confirmed the efficacy of this interaction focus manipulation. References Aaker Jennifer L. ( 1999 ), “The Malleable Self: The Role of Self-Expression in Persuasion,” Journal of Marketing Research , 36 ( 1 ), 45 – 57 . Google Scholar CrossRef Search ADS Belk Russell W. ( 1988 ), “Possessions and the Extended Self,” Journal of Consumer Research , 15 ( 2 ), 139 – 68 . Google Scholar CrossRef Search ADS Berger Charles R. , Calabrese Richard J. ( 1975 ), “Some Explorations in Initial Interaction and Beyond: Toward a Developmental Theory of Interpersonal Communication,” Human Communication Research , 1 ( 2 ), 99 – 112 . Google Scholar CrossRef Search ADS Berger Jonah ( 2014 ), “Word of Mouth and Interpersonal Communication: A Review and Directions for Future Research,” Journal of Consumer Psychology , 24 ( 4 ), 586 – 607 . Google Scholar CrossRef Search ADS Berger Jonah , Iyengar Raghuram ( 2013 ), “Communication Channels and Word of Mouth: How the Medium Shapes the Message,” Journal of Consumer Research , 40 ( 3 ), 567 – 79 . Google Scholar CrossRef Search ADS Bhattacharya Prasanta , Phan Tuan Q. , Goh Khim Yong ( 2016 ), “The Value of Social Media Brand Pages: Investigating the Effects of Self Presentation at Online Social Network Sites on Offline Purchase,” paper presented at the 1st CEIBS Marketing Conference, Department of Marketing of China Europe International Business School (CEIBS), Shanghai, China. Biber Douglas ( 1988 ), Variation across Speech and Writing , Cambridge, UK : Cambridge University Press . Google Scholar CrossRef Search ADS Carley Kathleen ( 1991 ), “A Theory of Group Stability,” American Sociological Review , 56 ( 3 ), 331 – 54 . Google Scholar CrossRef Search ADS Chafe Wallace L. ( 1982 ), “Integration and Involvement in Speaking, Writing, and Oral Literature,” in Spoken and Written Language: Exploring Orality and Literacy , ed. Tannen Deborah , Norwood, NJ : Ablex , 35 – 53 . Chafe Wallace L. ( 1985 ), “Linguistic Differences Produced by Differences between Speaking and Writing,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 105 – 23 . Chafe Wallace L. , Danielewicz Jane ( 1987 ), “Properties of Spoken and Written Language,” in Comprehending Oral and Written Language , ed. Horowitz Rosalind , Jay Samuels S. , San Diego, CA : Academic Press , 83 – 113 . Chaiken Shelly , Eagly Alice H. ( 1976 ), “Communication Modality as a Determinant of Message Persuasiveness and Message Comprehensibility,” Journal of Personality and Social Psychology , 34 ( 4 ), 605 – 14 . Google Scholar CrossRef Search ADS Chaiken Shelly , Eagly Alice H. ( 1983 ), “Communication Modality as a Determinant of Persuasion: The Role of Communicator Salience,” Journal of Personality and Social Psychology , 45 ( 2 ), 241 – 56 . Google Scholar CrossRef Search ADS Chaplin Nguyen , John Deborah Roedder ( 2005 ), “The Development of Self-Brand Connections in Children and Adolescents,” Journal of Consumer Research , 32 ( 1 ), 119 – 29 . Google Scholar CrossRef Search ADS Chen Zoey , Lurie Nicholas ( 2013 ), “Temporal Contiguity and the Negativity Bias in the Impact of Online Word of Mouth,” Journal of Marketing Research , 50 ( 4 ), 463 – 76 . Google Scholar CrossRef Search ADS Cheng Shirley Y. Y. , White Tiffany Barnett , Chaplin Nguyen ( 2012 ), “The Effects of Self-Brand Connections on Responses to Brand Failure: A New Look at the Consumer–Brand Relationship,” Journal of Consumer Psychology , 22 ( 2 ), 280 – 8 . Google Scholar CrossRef Search ADS Cialdini Robert, B. , Trost Melanie R. ( 1998 ), “Social Influence: Social Norms, Conformity, and Compliance,” in The Handbook of Social Psychology , ed. Gilbert Daniel T. , Fiske Susan T. , Lindzey Gardner , New York : McGraw-Hill , 151 – 92 . Coulmas Florian ( 1989 ), The Writing Systems of the World , Oxford, UK : Blackwell . DeVito Joseph A. ( 1966 ), “Psychogrammatical Factors in Oral and Written Discourse by Skilled Communicators,” Speech Monographs , 33 ( 1 ), 73 – 6 . Google Scholar CrossRef Search ADS Einhorn Lois ( 1978 ), “Oral and Written Style: An Examination of Differences,” Southern Speech Communication Journal , 43 ( 3 ), 302 – 11 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson ( 2004 ), “Narrative Processing: Building Consumer Connections to Brands,” Journal of Consumer Psychology , 14 ( 1–2 ), 168 – 80 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson , Bettman James R. ( 2003 ), “You Are What They Eat: The Influence of Reference Groups on Consumers’ Connections to Brands,” Journal of Consumer Psychology , 13 ( 3 ), 339 – 48 . Google Scholar CrossRef Search ADS Escalas Jennifer Edson , Bettman James R. ( 2005 ), “Self Construal, Reference Groups and Brand Meaning,” Journal of Consumer Research , 32 ( 3 ), 378 – 89 . Google Scholar CrossRef Search ADS Ferraro Rosellina , Kirmani Amna , Matherly Ted ( 2013 ), “Look at Me! Look at Me! Conspicuous Brand Usage, Self-Brand Connection and Dilution,” Journal of Marketing Research , 50 ( 4 ), 477 – 88 . Google Scholar CrossRef Search ADS Fitzsimons Grainne M. , Chartrand Tanya L. , Fitzsimons Gavan J. ( 2008 ), “Automatic Effects of Brand Exposure on Motivated Behavior: How Apple Makes You ‘Think Different,’” Journal of Consumer Research , 35 ( 1 ), 21 – 35 . Google Scholar CrossRef Search ADS Fondacaro Rocco A. , Tory Higgins E. ( 1985 ), “Cognitive Consequences of Communication Mode: A Social Psychological Perspective,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 73 – 104 . Fournier Susan ( 1998 ), “Consumers and Their Brands: Developing Relationship Theory in Consumer Research,” Journal of Consumer Research , 24 ( 4 ), 343 – 53 . Google Scholar CrossRef Search ADS Gopinath Mahesh , Nyer Prashanth U. ( 2009 ), “The Effect of Public Commitment on Resistance to Persuasion: The Influence of Attitude Certainty, Issue Importance, Susceptibility to Normative Influence, Preference for Consistency and Source Proximity,” International Journal of Research in Marketing , 26 ( 1 ), 60 – 8 . Google Scholar CrossRef Search ADS Hayes Andrew F. ( 2013 ), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach , New York : Guilford Press . Hayes Andrew F. , Preacher Kristopher J. , Myers Teresa A. ( 2011 ), “Mediation and the Estimation of Indirect Effects in Political Communication Research,” in Sourcebook for Political Communication Research: Methods, Measures, and Analytical Techniques , ed. Bucy Erik P. , Lance Holbert R. , New York : Routledge , 434 – 65 . Jahandarie Khosrow ( 1999 ), “Structural Differences between Speech and Writing: The Linguistic Evidence,” in Spoken and Written Discourse: A Multi-Disciplinary Perspective , ed. Jahandarie Khosrow , Stamford, CT : Ablex , 131 – 50 . Keller Kevin L. ( 2001 ), “Building Customer-Based Brand Equity: A Blueprint for Creating Strong Brands,” Marketing Management ( July/August ), 15 – 19 . Klesse Anne-Kathrin , Levav Jonathan , Goukens Caroline ( 2015 ), “The Effect of Preference Expression Modality on Self-Control,” Journal of Consumer Research , 42 ( 4 ), 535 – 50 . McPherson Miller , Smith-Lovin Lynn , Cook James M. ( 2001 ), “Birds of a Feather: Homophily in Social Networks,” Annual Review of Sociology , 27 ( 1 ), 415 – 44 . Google Scholar CrossRef Search ADS Milroy James , Milroy Lesley ( 1985 ), Authority in Language: Investigating Language Prescription and Standardization , London : Routledge & Kegan Paul . Moore David J. , Homer Pamela M. ( 2008 ), “Self-Brand Connections: The Role of Attitude Strength and Autobiographical Memory Primes,” Journal of Business Research , 61 ( 7 ), 707 – 14 . Google Scholar CrossRef Search ADS Moore Sarah G. ( 2012 ), “Some Things Are Better Left Unsaid: How Word of Mouth Influences the Storyteller,” Journal of Consumer Research , 38 ( 6 ), 1140 – 54 . Google Scholar CrossRef Search ADS Morewedge Carey K. , Giblin Colleen E. , Norton Michael I. ( 2014 ), “The (Perceived) Meaning of Spontaneous Thoughts,” Journal of Experimental Psychology: General , 143 ( 4 ), 1742 – 54 . Google Scholar CrossRef Search ADS PubMed Olson David R. , Torrance Nancy ( 1981 ), “Learning to Meet the Requirements of Written Text: Language Development in the School Years,” in Writing: The Nature, Development, and Teaching of Written Communication, Vol. 2: Process, Development, and Communication , ed. Frederiksen Carl H. , Dominic Joseph F. , Hillsdale, NJ : Erlbaum , 235 – 55 . Park C. W. , MacInnis Deborah J. , Priester Joseph R. ( 2006 ), “Beyond Attitudes: Attachment and Consumer Behavior,” Seoul Journal of Business , 12 ( 2 ), 3 – 36 . Park C. W. , MacInnis Deborah J. , Priester Joseph R. ( 2008 ), “Brand Attachment: Construct, Consequences and Causes,” Foundations and Trends in Marketing , 1 ( 3 ), 190 – 230 . Park Choong W. , MacInnis Deborah J. , Priester Joseph , Eisingerich Andreas B. , Iacobucci Dawn ( 2010 ), “Brand Attachment and Brand Attitude Strength: Conceptual and Empirical Differentiation of Two Critical Brand Equity Drivers,” Journal of Marketing , 74 ( 6 ), 1 – 17 . Google Scholar CrossRef Search ADS Pham Michel Tuan , Goukens Caroline , Lehmann Donald R. , Stuart Jennifer Ames ( 2010 ), “Shaping Customer Satisfaction through Self-Awareness Cues,” Journal of Marketing Research , 47 ( 5 ), 920 – 32 . Google Scholar CrossRef Search ADS Preacher Kristopher J. , Hayes Andrew F. ( 2008 ), “Asymptotic and Resampling Strategies for Assessing and Comparing Indirect Effects in Multiple Mediator Models,” Behavior Research Methods , 40 ( 3 ), 879 – 91 . Google Scholar CrossRef Search ADS PubMed Preacher Kristopher J. , Rucker Derek D. , Hayes Andrew F. ( 2007 ), “Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions,” Multivariate Behavioral Research , 42 ( 1 ), 185 – 227 . Google Scholar CrossRef Search ADS PubMed Rubin Donald L. ( 1987 ), “Divergence and Convergence between Oral and Written Communication,” Topics in Language Disorders , 7 ( 4 ), 1 – 18 . Google Scholar CrossRef Search ADS Schneider Walter , Shiffrin Richard M. ( 1977 ), “Controlled and Automatic Human Information Processing: I. Detection, Search and Attention,” Psychological Review , 84 ( 1 ), 1 – 66 . Google Scholar CrossRef Search ADS Shen Hao , Wyer Robert S. ( 2008 ), “Procedural Priming and Consumer Judgments: Effects on the Impact of Positively and Negatively Valenced Information,” Journal of Consumer Research , 34 ( 5 ), 727 – 37 . Google Scholar CrossRef Search ADS Sirgy Joseph M. ( 1982 ), “Self-Concept in Consumer Behavior: A Critical Review,” Journal of Consumer Research , 9 ( 3 ), 287 – 300 . Google Scholar CrossRef Search ADS Sprott David , Czellar Sandor , Spangenberg Eric ( 2009 ), “The Importance of a General Measure of Brand Engagement on Market Behavior: Development and Validation of a Scale,” Journal of Marketing Research , 46 ( 1 ), 92 – 104 . Google Scholar CrossRef Search ADS Tannen Deborah ( 1982 ), Spoken and Written Language: Exploring Orality and Literacy , Norwood, NJ : Ablex . Tannen Deborah ( 1985 ), “Relative Focus on Involvement in Oral and Written Discourse,” in Literacy, Language, and Learning , ed. Olson David R. , Torrance Nancy , Hildyard Angela , Cambridge, UK : Cambridge University Press , 124 – 47 . Unnava H. R. , Burnkrant Robert E. , Erevelles Sunil ( 1994 ), “Effects of Presentation Order and Communication Modality on Recall and Attitude,” Journal of Consumer Research , 21 ( 3 ), 481 – 90 . Google Scholar CrossRef Search ADS Wakefield Jerome C. ( 1992 ), “The Concept of Mental Disorder: On the Boundary between Biological Facts and Social Values,” American Psychologist , 47 ( 3 ), 373 – 88 . Google Scholar CrossRef Search ADS PubMed Weiss Liad , Johar Gita ( 2013 ), “Egocentric Categorization and Product Judgment: Seeing Your Traits in What You Own (and Their Opposite in What You Don’t),” Journal of Consumer Research , 40 ( 1 ), 185 – 201 . Google Scholar CrossRef Search ADS Weiss Liad , Johar Gita ( 2016 ), “Products as Self-Evaluation Standards: When Owned and Unowned Products Have Opposite Effects on Self-Judgment,” Journal of Consumer Research , 42 ( 6 ), 915 – 30 . Google Scholar CrossRef Search ADS Wood Wendy , Neal David T. ( 2007 ), “A New Look at Habits and the Habit-Goal Interface,” Psychological Review , 114 ( 4 ), 843 – 63 . Google Scholar CrossRef Search ADS PubMed Wyer Robert S. , Shen Hao , Xu Jing A . ( 2013 ), “The Role of Procedural Knowledge in the Generalization of Social Behavior,” in Oxford Handbook of Social Cognition , ed. Carlston Donald E. , New York : Oxford University Press , 257 – 81 . © The Author(s) 2018. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Consumer ResearchOxford University Press

Published: Feb 21, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off