- Top of page
- Literature Review
This study analyzes requests made via email and voicemail for properties of politeness (Brown & Levinson, 1987). Voicemail users have less control over planning, composing, editing, and executing messages, and must manage more nonverbal cues than email users. Thus, it is predicted that email will enable users to create more polite speech than voicemail. A 2 (communication medium: email or voicemail) x 2 (imposition: low or high) factorial design was implemented to test this hypothesis. One hundred fifty-one participants created request messages that were subsequently analyzed for properties of politeness. Overall, the results indicate email requests were more polite than voicemail requests. These results are consistent with the hyperpersonal model (Walther, 1996).
- Top of page
- Literature Review
Two competing themes characterize the computer-mediated communication (CMC) literature concerning the ability of low bandwidth media to foster social and negotiation tasks. Short, Williams, and Christie (1976) suggested that effects on interaction can be predicted by the cues that are not transmitted via various media (p. 207). Lower bandwidth media, because they restrict more communication cues than face-to-face communication, are less suited for social and negotiation tasks than face-to-face media. CMC theories following this line of reasoning have been referred to as the cues filtered out model (Culnan & Markus, 1987). Walther and Parks (2002) describe the basic assumption of this model as follows:
the functions served by nonverbal cues in face-to-face interaction go unmet in computer-mediated interaction because the nonverbal cues are absent. If no other cues can perform the social functions that physical appearance, copresence and dynamic nonverbal behavior can, then … CMC must always be impersonal. (p. 532)
An alternative and more recent development is Walther’s (1996) hyperpersonal model, which predicts that low bandwidth media can facilitate social and negotiation tasks beyond what is possible in face-to-face interaction. According to this model, the filtering of nonverbal cues advantages communicators. Communicators are strategically enabled to manipulate their identity, time the transmission of their messages, and plan, organize, and edit their communication in pursuit of relational goals. Such strategic control in CMC can facilitate negotiation, relationship development, and social tasks.
The cues filtered out model has received a great deal of attention in the CMC literature by virtue of its tenure and face validity. In contrast, empirically-based research articles investigating Walther’s (1996) hyperpersonal model are only beginning to appear in the literature. The relative novelty of the model, the initial support it has received (Chester & Gwynne, 1998; Walther, 1997; Walther, Slovacek, & Tidwell, 2001), and the existence of alternative hypotheses concerning the effects of low bandwidth on social tasks justify its continued investigation.
To this end, a useful vehicle for investigating hyperpersonal communication is politeness theory (Brown & Levinson, 1987). This is a theory of language production based on a rational individual’s cognitive assessment of a social situation. It is a theory about the manner in which a person phrases “things” given an assessment of the social situation (Holtgraves, 2002, p. 38). A major component of politeness theory is an arrangement of politeness strategies along a continuum from least polite to most polite. Politeness theory thus outlines a system where speech can be evaluated according to its overall politeness.
Politeness theory has primarily been investigated in face-to-face situations, with some exceptions (Herring, 1994; Morand & Ocker, 2003; Sussman & Sproull, 1999). Yet variations in CMC technology identified by the cues filtered out and hyperpersonal models may significantly influence one’s ability to cognitively assess a social situation. In turn these variations may influence one’s ability to produce polite speech. For example, electronic mail allows communicators more control over planning, composing, editing, and delivering messages than face-to-face communication (Herring, 2002; Walther, 1996). This elevated control over message production and delivery suggest that email may help communicators create more polite speech than with the use of other CMC media. The purpose of the current research is to utilize politeness theory as a tool to investigate the effect of CMC technology on communication. Specifically, email requests are predicted to be more polite than voicemail requests. Literature relevant to politeness theory and CMC is reviewed below, followed by the theoretical development of the hypotheses, presentation of the nature and results of the empirical tests, and a discussion of the findings.
- Top of page
- Literature Review
A goal of the present study was to measure the effect of communication technology on the politeness of requests. Although many measures of politeness exist, Holtgraves and Yang’s (1992) measuring technique was utilized. To assess the impact of the politeness coding of request messages on the evaluation of overall politeness, a standard multiple regression analysis was performed using SPSS REGRESSION. Overall politeness served as the criterion variable. The politeness superstrategy for meeting request, superstrategy for request for reply, formality of address phrase, number of words, and number of adjuncts served as predictor variables in the analysis. Nearly half of the variance (R2= .46; adj R2= .442) in overall politeness was accounted for by the five predictor variables, F [5, 143] = 23.62, p < .001. Table 1 displays the unstandardized regression coefficients (β), standardized regression coefficients (β), t-value, and probability for each predictor variable. Each of the predictor variables contributed significantly to overall politeness. Unexpectedly, the number of adjunct phases was inversely related to overall politeness (β=−3.41, t=−4.404, p < .001).
Table 1. Regression of coding components onto overall politeness
| ||Overall Politeness|
|Meeting Request Superstrategy||.783||.494||7.792*|
|Request for Reply Superstrategy||.384||.365||5.558*|
|Formality of Address Phrase||.104||.252||3.966*|
|Number of Words||.028||.474||6.32*|
|Number of Adjuncts||−.311||−.341||−4.404*|
Presented in Table 2 are the descriptive statistics for all measures of politeness, as well as the F statistics and their associated probability levels resulting from a series of ANOVAs.
Table 2. The effect of medium and degree of imposition on measures of politeness
| ||Voicemail||Electronic Mail||F|
|Low IMP (n= 36)||High IMP (n= 36)||Low IMP (n= 37)||High IMP (n= 38)||IMP||CM||CM x IMP|
|Number of Adjuncts|
| SD||.83||.83||1.39||1.03|| |
|Number of Words|
| SD||16.89||11.78||14.64||20.76|| |
|Formality of Address Phrase|
| SD||2.06||2.16||2.08||3.12|| |
| SD||1.04||.90||1.00||1.10|| |
Hypothesis One (H1)
Hypothesis One (H1) predicted that imposing requests would be rated higher on politeness than unimposing requests. Hypothesis One (H1) received partial support based on factorial ANOVAs. First, as predicted, significantly more adjunct phrases were produced by those creating requests that were imposing (M= 3.81) than by those creating less imposing requests (M= 3.44), F [1, 143] = 4.56, p < .05. Second, as predicted, those making highly imposing requests used significantly more words (M= 62.32) than those making less imposing requests (M= 55.67), F [1, 143] = 5.84, p < .05. Third, the results indicate that formality of address phrase did differ between high-imposition (M= 5.87) and low-imposition requests (M= 6.72), F [1, 143] = 4.54, p= .035, though the difference was not in the hypothesized direction. Fourth, the results indicate the overall politeness of imposing requests (M= 5.39) was rated as nearly the same as the politeness of unimposing requests (M= 5.33; F[1, 143] = .12).
Hypothesis Two (H2)
Hypothesis Two (H2) predicted that email message requests would be rated higher on measures of politeness than voicemail requests. Based on the results of factorial ANOVAs, Hypothesis Two (H2) was partially supported. First, as predicted, email requests incorporated more adjunct phrases (M= 3.89) than voicemail requests (M= 3.36), F [1, 143] = 9.10, p < .05. In addition, the difference between email requests and voicemail requests for number of words approached significance (M= 61.17 vs. 56.82) F [1, 143] = 2.50, p= .116. The address phrase was not more formal in email (M= 6.28) than voicemail requests (M= 6.31), F [1, 143] = .004, p= .949. Moreover, the email requests were not significantly more polite overall (M= 5.24) than voicemail requests (M= 5.48) F [1, 143] = 2.04, p= .156.
Hypothesis Three (H3)
Hypothesis Three (H3) predicted that less imposing requests made via email or voicemail would not differ on ratings of politeness, but highly imposing requests via email would be rated higher on measures of politeness than voicemail requests (see Figure 1). As predicted, the pattern of results and statistical tests indicate that email requests were more polite than voicemail requests under conditions of high imposition.
Number of Adjuncts
The interaction hypothesis for the dependent variable number of adjuncts was supported (see Figure 2). As predicted, the number of adjunct phrases in email requests and voicemail requests were similar under conditions of low imposition (M= 3.51 vs. 3.36). However, email requests included more adjunct phrases (M= 4.26) than voicemail requests (M= 3.36) under conditions of high imposition, F [1, 143] = 4.70, p < .05. A two-tailed independent samples t-test confirmed that highly imposing email requests included significantly more adjunct phrases (t=−4.124, p < .05) than highly imposing voicemail requests. As predicted, less imposing email requests and voicemail requests produced a similar number of adjunct phrases (t= .567, p= .572).
Number of Words
The interaction hypothesis for the dependent variable number of words was supported (see Figure 3). As predicted, under conditions of low imposition, the number of words in email requests was similar to the number of words utilized for voicemail requests (M= 56.6 vs. 54.7). However, under conditions of high imposition, email requests included more words (M= 67.6) than voicemail requests (M= 57.0), F [1, 143] = 5.13, p < .05. A two-tailed independent samples t-test indicated that the number of words incorporated into highly imposing email requests was significantly different than the same measure for voicemail (t= 2.676, p < .05). Further supporting the hypothesis, an independent samples t-test revealed that less imposing email and voicemail requests did not differ significantly in the number of words needed to construct the message (t= .409, p= .684).
Formality of Address Phrase
The interaction hypothesis for the dependent variable formality of address phrase was supported by a Factorial ANOVA (see Figure 4). Under conditions of low imposition the mean of formality of address phrase for email requests was higher (M= 7.09) than the mean of formality of address phrase for voicemail (M= 6.35). Unexpectedly, under conditions of high imposition, the formality of address phrase mean of email requests was lower (M= 5.47) than the formality of address phrase mean of voicemail requests (M= 6.26), F [1, 143] = 3.67, p < .05.
Figure 4. Degree of imposition by communication medium: H3 results for formality address phrase interaction diagram.
Download figure to PowerPoint
The interaction hypothesis for the dependent variable overall politeness was not significant according to the factorial ANOVA, F [1, 143] = .765, p= .383. Under conditions of low imposition the overall politeness of email requests (M= 5.29) was slightly lower than voicemail requests (M= 5.38), and under conditions of high imposition, email requests were also rated slightly lower in overall politeness (M= 5.20) than voicemail requests (M= 5.58).
- Top of page
- Literature Review
The results demonstrate that communicators using electronic mail produced more adjunct phrases and more words than those using voicemail. In addition, email requests varied in the number of adjunct phrases, number of words, and the formality of address phrase according to imposition, while voicemail requests were stable and static across these same variables. This support for the hypotheses leads to the conclusion that email facilitates politeness strategies in support of hyperpersonal communication.
As predicted in Hypothesis Two (H2), politeness varied by communication medium. Electronic mail requests resulted in a significantly greater number of adjunct phrases than voicemail requests. The email requests also contained more words than voicemail requests, although not significantly so. Additionally, consistent with Hypothesis Three (H3), as imposition varied, so too did email requests. Yet voicemail requests remained stable and static. This pattern of results across most dependent measures suggests that email enabled users to customize their messages. For example, the average email request contained 3.52 adjunct phrases and 54.7 words when imposition was low, but when imposition was high email requests contained an average of 4.26 adjunct phrases and 67.6 words—a statistically significant difference between experimental conditions (see Figures 3 and 4, respectively). However, voicemail messages varied little across these same conditions. For example, in the voicemail condition there was no difference in the number of adjunct phrases between low and high imposition or the number of words.
This same pattern of the adaptability of email versus the stasis of voicemail was further evident for the dependent variable formality of address phrase. The address phrase in voicemail messages varied little in formality between low and high imposition, an outcome consistent with the hypothesis advanced in this study. However, the formality of the address phrase in email messages varied widely between low imposition and high imposition requests. The statistically significant interaction lends further support to the argument that voicemail limits and email facilitates hyperpersonal communication.
Unexpectedly, and contrary to the hypothesis, for those participants creating electronic mail messages, unimposing requests were accompanied by more formal address phrases than imposing requests, and according to a post-hoc independent samples t-test these means were significantly different. Politeness theorists would expect imposing requests to be accompanied by a higher degree of formality of address phrase. The use of less formal/polite address phrases when imposition was high might be a strategic tactic to redefine the nature of the student-professor relationship. Sensing a disparate relational bond, communicators may attempt to redefine the relationship by utilizing less formal address phrases. Since attempting to bolster the feeling of solidarity in a lean/low bandwidth context is consistent with the hyperpersonal model, this proposition warrants further research.
Significant main effects for imposition on measures of politeness emerged to support Hypothesis One (H1), even though the manipulation check for imposition proved insignificant. For most dependent measures (not overall politeness) there were significant main effects for imposition (see Table 2), thus supporting the conclusion that as imposition increased, so too did the measures of politeness. Participants assigned to the high imposition condition utilized more adjunct phrases than those assigned to the low imposition condition. Additionally, high imposition participants utilized significantly more words to make requests than did low imposition requesters. Also apparent was a significant difference between formality of address phrase under conditions of low imposition and high imposition. These results reflect a successful manipulation of imposition, and thus constitute support for politeness theory.
In summary, for the dependent measures of politeness, only email messages varied according to imposition. Voicemail messages remained static and unchanged between low and high imposition requests, and there was evidence of a main effect for communication medium and imposition. Overall, messages created with email included more adjunct phrases and words than voicemail messages. Consistent with the predictions of the hyperpersonal model, users of email were able to construct more polite messages. These results support the claim that control over editing, plan and performance, and the reduced cue properties of email, provide users with the capacity to create socially desirable messages. In contrast, voicemail, with its lack of editability, discrepancy between plan and performance, and multichanneled demands, appears to constrain users’ ability to produce similar messages.
At the same time, it is possible that the study design limited the findings in certain ways. A multiple regression analysis of the dependent variable overall politeness revealed a significant inverse relationship between it and the number of adjunct phrases (see Table 1). This inverse relationship accounts for the lack of significant differences for the variable of overall politeness in this study. Compared to a message containing few adjunct phrases, a request message containing many adjunct phrases resulted in a decrease in overall politeness. In the current study, judges were instructed that the use of positive politeness (expressions of solidarity) was less polite than negative politeness (admissions of impingement) no matter the number of positively polite or negatively polite adjunct phrases. The majority of adjunct phrases in the current study were positively polite. Moreover, email requests included significantly more adjunct phrases than voicemail requests. Because of this, the rating of overall politeness decreased disproportionately for email requests. Had the measure of overall politeness been more sensitive to the impact of the number of adjuncts on politeness, the measure of overall politeness for email would very likely have been rated higher than that of voicemail.
This research was designed to be generalizable to the population that was sampled, in a context in which that population actually uses email and voicemail. First, participants were asked to respond, using familiar communication technology, to a common situation. Students were asked to do what students do often: email or leave a voicemail for a professor. Second, politeness was observed in an asynchronous communication context. By definition, a message creator’s request is separated from the recipient’s reply by time and space; copresence is unnecessary. Therefore, politeness evaluations of the entire conversational turn (sender and receiver) are not strictly necessary. In this study, message creators focused entirely on the creation and delivery of a request message. Traditionally, politeness behavior (Ambady et al., 1996; Craig et al., 1986; Greene & Lindsey, 1989; Holtgraves & Yang, 1992; Lee, 1993, 1999; Leichty & Applegate, 1991; Sussman & Sproull, 1999) has been observed from either a message generation or receipt perspective, and this has been a limitation. Future research focusing on synchronous media and politeness should consider the entire conversational turn.
Future research into synchronous text-based CMC and politeness may find that real-time communication does not facilitate the creation of polite speech. Sussman and Sproull (1999) found synchronous CMC to be more direct and less polite than synchronous telephone or face-to-face interactions. Although Sussman and Sproull attribute “straight-talk” in synchronous CMC to low social presence and deindividuation, participants in their study reported greater satisfaction, comfort, and liking of their conversational partners compared to phone or face-to-face discussants. If deindividuation were the best explanation for these results, one would expect to observe a positive relationship between politeness and satisfaction, comfort, and liking.
The hyperpersonal model may be a more suitable explanation for the conflicting findings reported in Sussman and Sproull (1999). Because of the demanding time constraints of instantaneous interaction, communicating with synchronous text-based chat systems requires efficiency, utilization of short sentences, abbreviations, and straight-talk (Herring, 2002). Synchronous communication in a highly interactive environment demands simultaneous concentration on the mechanics of typing and the pursuit of maintaining face, consequently dividing one’s available cognitive resources (Greene & Lindsey, 1989). Senders may also consciously optimize their self-presentation by adopting visual and aural anonymity, while receivers formulate idealized perceptions of the sender on the basis of limited cues (Walther, 1995, 1996). According to the hyperpersonal model, synchronous text-based communication messages may be less polite than asynchronous messages, yet participants may still rate one another positively. Future research should test these propositions.
Future research into the politeness of CMC might also proceed by considering the work of a group of researchers studying the politeness strategies of nonverbal communication. The researchers (Ambady et al., 1996) found that nonverbal cues carry significant and meaningful politeness information. Politeness strategies were encoded through kinesics, facial expressions, and vocalic and linguistic cues. Such an approach in the current study might have revealed politeness strategies encoded in the vocal characteristics of voicemail messages. Politeness strategies not encoded in the content of the request might be given off through the vocal characteristics of the spoken words, the greater politeness in the email messages functioning as compensation for the lack of such nonverbal cues. This proposition would be important to test in future research.