Journal of Pragmatics 42 (2010) 2531–2548

Contents lists available at ScienceDirect

Journal of Pragmatics journal homepage: www.elsevier.com/locate/pragma

Effects of positive politeness strategies in business letters Frank Jansen a,*,1, Daniel Janssen a,b,1 a b

Utrecht University, Utrecht Institute for Linguistics OTS, Trans 10, 3512 JK, Utrecht, Netherlands University of Antwerp, Belgium

A R T I C L E I N F O

A B S T R A C T

Article history: Received 11 October 2007 Received in revised form 23 February 2010 Accepted 27 February 2010

Brown and Levinson’s (1987) typology of politeness strategies, derived from the basic wants of a model person, leaves the question unanswered as to what extent these strategies are perceived and evaluated as contributions to the quality of communication. In this paper, we will discuss the effects of adding and combining positive politeness strategies to letters denying claims to policy holders. The combined results of two experiments offer a clear picture: ‘Give Reasons’ has a positive effect on the evaluation of the letter, while two other strategies have no effect at all. Theoretical implications are discussed, as is the use of experimentation in politeness research. ß 2010 Elsevier B.V. All rights reserved.

Keywords: Positive politeness Effects of strategies Give reasons

1. Introduction For more than thirty years, Politeness by Brown and Levinson (1987, first published in 1978) has been a source of inspiration for theoretical and empirical research in pragmatics. In retrospect it is not so difficult to understand why politeness theory has become so popular. Its success can be partly attributed to their extensive typology of politeness strategies. This typology, covering more than 130 pages, is carefully deducted from the presumed wants of a Model Person that feels the need to preserve one’s ‘face’, his or her2 own and that of his conversation partners. In their discussion of the strategies, Brown and Levinson provide numerous and extensive examples of the implementation of the strategies in conversations in different languages and cultures (English, Tamil and Mayan), which make it easier for the reader to recognize and accept the described phenomena as strategies. Furthermore, their typology is appealing to many researchers because it unifies phenomena in language use that were once considered to be disparate. Politeness has evoked criticism as well, both from a theoretical and an empirical point of view (see Eelen, 2001 for a comprehensive discussion). To cite just a few examples from recent years, Bargiela-Chiappini (2003) criticizes the way Brown and Levinson interpret Goffman’s (1967) concept of ‘face’, and use it instrumentally to distinguish negative and positive politeness strategies. She traces Goffman’s ‘face’ concept back to Chinese scholars, and demonstrates how they conceptualized (especially) ‘negative face’ as a form of deference—the way an individual adheres to social and situational determined rules—rather than the desire of an individual to be unimpeded by others. Furthermore, Bargiela-Chiappini claims that this original face concept is more suitable to deal with politeness phenomena in other cultures than English, for example Japanese and Chinese. This view is in line with what has been demonstrated in empirical studies by many other linguists (for example Matsumoto, 1988; Blum-Kulka and House, 1989; Holtgraves and Yang, 1992; Mao, 1994; Yeung, 1997). Although Chen (2001) shows that at least some of these criticisms are misdirected and Fukada and Asato (2004) demonstrate that the data on Japanese honorifics cannot be adduced as evidence against Brown and Levinson’s theory, we do think that Brown and Levinson’s universality assumption is questionable (see hereafter and section 6 for further discussion). * Corresponding author. Tel.: +31 302538331; fax: +31 302536000. E-mail address: [email protected] (F. Jansen). 1 The authors have contributed equally to the text and the research behind it. 2 For stylistic reasons the pronouns of choice are he and his in the remainder of this paper. 0378-2166/$ – see front matter ß 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.pragma.2010.02.013

2532

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

Watts and his coworkers (Watts et al., 1992; Watts, 2003) take the criticism on Brown and Levinson one step further by presenting an alternative goal for politeness theory. Watts suggests that the theory should be able to explain why certain types of (linguistic) behavior used by conversationalists in a particular society are perceived as deferential, rude, conventional, or embarrassing by other competent members of that society. The experiments that we present in section 4 and following are thus relevant to Watts’ plea for an ‘emic’ (as opposed to Brown and Levinson’s ‘etic’) approach of politeness phenomena (cf. Haugh, 2007). As for the criticism of politeness theory that are based on empirical considerations, most scholars have focused on Brown and Levinson PDR-model: Wx = D(S,H) + P(H,S) + Rx. In this model they relate the weight of the face-threatening act (Wx) to three social factors: the distance between speaker and hearer, the power of the hearer over the speaker, and the inherent weight of the imposition of the face-threatening act. The simplicity of this model made empirical testing easy. Most of the work in this direction has been carried out by studying language production in situations that have been characterized by varying the independent variables ‘distance’, ‘power’ and ‘relative weight’ and measuring the dependent variables ‘type’ and ‘amount’ of politeness strategies. The technique of the discourse completion test (for example Blum-Kulka et al., 1989; Hendriks, 2002; Le Pair, 2005) has been popular in this field, especially in cross-cultural communication. Other techniques include the vignette technique (Holtgraves and Yang, 1992), the analysis of actual conversations (Rees-Miller, 2000), business correspondence (Yeung, 1997), or dialogues of subjects playing a role in a scenario presented by the analyst (for example van der Wijst, 1996; Fe´lix-Brasdefer, 2006). Baxter (1984), on the other hand, presents her participants with various compliance-gaining situations and a set of descriptions of 32 politeness strategies, and then requests them to select the strategy that is most polite in the given situation. Relating the results of these production methods to the evaluation of the effectiveness of politeness strategies may be possible. The strategies employed in situations with a great distance between speaker and hearer, a great power of the hearer over the speaker, and a weighty imposition are considered the most effective politeness strategies. However, it is evident that the experiments mentioned earlier only provide indirect evidence. In order to gain more insight into the effectiveness of politeness strategies, it is still necessary to conduct experiments in which language users directly evaluate the use of politeness in specific situations. In this article we will present an example of such an experiment. In this experiment we will also test another assumption underlying the quantification of strategies: the assumption that ‘more is better’, that the more politeness strategies participants use in a discourse completion experiment, the more face threatening the ‘bare’ speech act is. Considering the theoretical importance of the ‘emic’ assessment of the effectivity of politeness strategies, it strikes us that empirical investigations of the effect of the use of specific politeness strategies on the receivers’ evaluation of the sender and the message are relatively scarce. We do find several conversational analytic studies in which the relational meaning of some conversational turn is reconstructed by means of the analysts’ interpretation of the reaction of the hearer (Watts, 2003; Mills, 2003 give numerous examples). However useful this method may be to obtain an understanding of what is occurring in discourse and society, the problem is that we cannot evaluate the exigency of the interpretations and their bearing on the theory as it is unknown how frequent alternative reactions would have been at the first conversational turn. Jary (1998), for instance, points out that ‘Brown and Levinson’s view is counterintuitive as it predicts that whenever so-called polite forms/ strategies are applied then an additional layer of meaning is necessarily communicated, while our experience as conversationalists tells us that polite forms often go unnoticed by the participants’ (p. 2). Jary does not provide much empirical evidence for his claim; we would certainly like to see a little more substantiation than his ‘experience as a conversationalist’. In our opinion, the assessment of employing politeness strategies calls for an experimental design: a message with a facethreatening act with or without politeness strategies as independent variable, and the evaluation of the message as dependent variable. An example of such experiment design is found in Clark and Schunk (1980) (see also Kemper and Thissen, 1981) who discuss a series of six experiments in which subjects had to evaluate the perceived politeness of (indirect) requests in terms of acceptance or refusal. Although matters are complicated by differences in conventionality, Clark and Schunk conclude that their subjects scale the politeness of the utterances according to the extent that they convey that the speaker takes the face needs of the hearer into consideration, as predicted by Brown and Levinson. Hill et al. (1986) used the same method to study the differences between Japanese and American politeness cultures, and their work was the inspiration for Huls (1991) to conduct the same kind of experiments in European cultures. Although these approaches have proved to be fruitful, there are also disadvantages. First, while it is common knowledge that the evaluation is heavily influenced by aspects of context and situation, subjects in these experiments have to evaluate utterances or fragments of dialogues that are presented in isolation. Second, subjects are requested to evaluate the utterance with special reference to its ‘politeness’. This undermines the validity of these experiments in two ways. The researcher may evoke a so-called ‘framing effect’: when confronted with the request to assess the ‘politeness’ a participant will activate his own ‘folk-theory’ of what constitutes polite behavior in this situation, which puts the subject in the position of the researcher. Furthermore, the focus on politeness is too narrow. In real interactions the evaluation of a message is the result of a balancing act between immediate success (S understands and accepts the message) and long-term success (S continues having a good relation with H). In other words, we think that it is not the perceived politeness of the message per se that is interesting but the general evaluation of the message. Therefore, we are inclined to prefer the approach of Goldsmith and MacGeorge (2000) who manipulated the advice given to a distressed person by adding different politeness strategies and asked their subjects to evaluate the quality of the advice as a whole.

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2533

In order to obtain a richer understanding of the impact of politeness strategies, we conducted two experiments in which subjects had to evaluate bad news letters (refusals) with and without politeness strategies. The strategies implemented in the letters were all examples of positive politeness. The first experiment explores the effect of one strategy, a combination of two strategies, and of a combination of three positive politeness strategies. The second experiment explores the effect of the layering of individual positive politeness strategies. In this article we will first discuss the relevant aspects of politeness theory (2). Then, we present a general introduction to the experiments (3), followed by an overview of the experiment in which we combined politeness strategies (4) and the experiment for the evaluation of distinctive strategies (5). The last chapter presents our conclusion and topics for discussion (6). 2. Theoretical framework 2.1. Brown and Levinson deductive approach of politeness As this study is based on Brown and Levinson’s work on politeness strategies, we will briefly discuss this very influential work (Brown and Levinson, 1987). Their first step was to assume a ‘Model Person’, a competent adult member of a society who has an inclination to keep up his face: a positive face (the model person presumes that his wants are desirable for at least some others) and a negative face (he hopes that his actions are unimpeded by others). Brown and Levinson’s second step is that they assume that both faces of this model person are inevitably threatened when he takes part in communication. When, for example, Model Person A requests Model Person B to lend him fifty dollars, A’s own positive face is threatened because he finds himself belonging to the unpleasant category of people who bother other people. Furthermore, A has to admit to himself that he is somewhat dependent on B, which threatens his negative face. As for B, his negative face is threatened because he knows that A expects him to comply; therefore, he has to do something that A asks, which limits his personal wants. If B refuses, he damages both his own positive face and that of requester A because he denies the legitimacy, or at least the normality of the request. The Model Person can decide to ignore the harm done by the face-threatening act, by doing the act right away, or ‘Going Bald on Record’. Besides ‘Bald on Record’, Brown and Levinson distinguish four other super categories, of which one category (‘Do not do the FTA’) is irrelevant for our purpose. In order to soften the harm done by the attack on the face, the model person can try to save his own positive face, or that of his communication partner, by using politeness strategies. Two of those, ‘Go off record’ (for example ‘Be vague’) and ‘Negative politeness’ are oriented to the restoration of their negative face and therefore they are not directly relevant for us here. The last super strategy is directed to the positive face. Those strategies aim at solidarity between the communication partners and the creation of a friendly atmosphere. Brown and Levinson have identified no less than fifteen positive politeness strategies: (1) Notice, attend to H, (2) Exaggerate attention to H (for example by giving compliments), (3) Intensify interest to H, (4) Use in-group identity markers, (5) Seek agreement, (6) Avoid disagreement, (7) Presuppose/raise/assert common ground, (8) Joke, (9) Assert or presuppose S’s knowledge of and concern for H’s wants, (10) Offer, promise, (11) Be optimistic, (12) Include both S and H in the activity, (13) Give or ask for reasons, (14) Assume or assert reciprocity, and (15) Give gifts to H. Brown and Levinson’s taxonomy is the result of a deductive approach: they deduce the strategies from the Model Person’s wants step by step. This makes it apparent that they devote less attention to the factual effects of politeness strategies.3 They make one exception however, when they claim that the super categories have a differential effect. They presume that these super strategies can be scaled on a politeness continuum: from bald-on-record as apt only in case of the mildest facethreatening acts, via positive politeness and negative politeness to off-record suitable for the severest face-threatening acts (Brown and Levinson, 1987). 2.2. Criticism of Brown and Levinson’s taxonomy Brown and Levinson’s claims about the differential effects of the super strategies have been challenged by several researchers. An early example is Baxter (1984) who questioned if ‘Brown and Levinson’s typology of politeness strategies [is] verified in people’s perceptions?’ Baxter’s experiment revealed that her participants did not categorize most of the perceived strategies in the same way as Brown and Levinson. The most interesting differences concern the intricate relationship between positive and negative politeness strategies. Baxter observed that positive politeness is a precondition for negative politeness. In the company of positive politeness strategies, negative politeness strategies are evaluated as polite, while the same negative politeness strategies are perceived as aggravating face threats when positive polite signals are absent. After Baxter, however, other researchers, using other methods (see section 1), obtained results that seem to confirm Brown and Levinson’s claim, with the exception of off-record strategies. Hints for instance are considered as rude in many cultures (Holtgraves, 2002). Another fundamental objection against the strategies has been discussed in section 1: instead of constructing a universal taxonomy of politeness strategies, the ultimate goal of politeness research should be to explain why members of a particular 3 Their deductive approach makes the taxonomy immune to several points of critique as well, for example that a particular turn in a conversation can be interpreted as an instance of more than one strategy, that some linguistic elements frequently occur in other contexts than face threatening acts, and that they do not offer a theory about the combination of strategies redressing one face threatening act.

2534

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

society are able to evaluate their own linguistic acts and the acts of others as deferential or rude (Watts, 2003; Watts et al., 1992). In other words, the role of a politeness theory should be to offer a scientific underpinning of the layman’s concepts of polite and impolite language behavior. In Watts’ own terms, we should start with Politeness-1, the way ‘normal’ people talk about those styles, use them to categorize others and to assess communication situations. Investigators of Politeness-1 will eventually design a scientific theory that explains the layman’s evaluation of styles, which is called Politeness-2. Watts (2003) and other researchers inspired by him approach the conceptualization of Politeness-2 with an inductive research strategy. In particular, they use the methods of conversational analysis to interpret the turns in the conversation as choices that speakers and hearers must have made among alternative ways of being polite. Although a great deal of valuable empirical work has been carried out within this framework, we do not think that this approach is the only research strategy to develop theories about Politeness-2. Actually, we propose a different strategy to contribute to Politeness-2. In our opinion, the politeness strategies identified by Brown and Levinson are to be considered as a reservoir or an inventory of resources that may be deployed by language users in specific communicative situations. Whether they will use one or more of the strategies depends on numerous conditions, one of the most important being the facilities offered by the grammar and lexicon of the language they use. Another important set of conditions is based on what the language user knows about the presumed effects of the respective strategies in his language. In other words, what he does is based upon what he knows about the communicative conventions in that culture. It is important to keep in mind that while the inventory of politeness strategies might be universal in theoretical respect, the interpretation of the incorporation of one or more concrete strategies is evidently dependent on language, culture, and situation. In reference to Watts’ distinction between Politeness-1 and Politeness-2, our intention is to design fragments of Politeness-2 by modeling the effects of some of the politeness strategies in a given language, and testing that model in a systematic way. In this article we will demonstrate this methodology by discussing two aspects of the evaluation of politeness strategies:
The selection problem: what is the differential effect to a face-threatening act of adding one particular strategy, compared to the addition of another politeness strategy? This question can be answered by contrasting the evaluation of a bald-onrecord text with texts with instances of politeness strategies.
The combination problem: what is the effect of adding a second (or third) politeness strategy to the evaluation of a text? 2.3. Practical relevance of the experimental approach for professional communication As demonstrated in the previous sections, our research is directly relevant for politeness research. Furthermore, we also think it is of practical relevance. One of the major challenges in business writing is communicating a clear message while keeping the client happy and preserving the image of the organization (Rothschild and Burnett, 1997; de Jong, 2002; Janssen, 2007). Although clear communication may be a relatively easy task when relaying positive messages—when writers’ and clients’ goals converge—it is often a challenge when conveying negative information. While reconciling the short-term goal of comprehension and acceptance of the message with the long term goal of customer relations, business writers have to comply with the demands of the communicative situation and preserve the client’s positive and negative face while also maintaining their own (Pilegaard, 1997; Hagge and Kostelnick, 1989; Janssen and Jaspers, 2004). Several studies have demonstrated that professional writers use politeness strategies to achieve their communicative goals (Hagge and Kostelnick, 1989; van der Mast and Janssen, 2001). In their analysis of accounting letters Hagge and Kostelnick (1989) observe several types of indirectness (negative politeness strategies) and vagueness (an off record strategy) in sections where the accountants point out the ‘problems’ they have encountered during an audit: Customer’s safekeeping records are not always updated for current charges. We noted some receipts in the customer safekeeping register for items that had been returned to the customers without record of the return (Hagge and Kostelnick, 1989:324) The same goes for the sections in which the auditor suggests solutions: We recommend that guidelines be established for the types of items and the dollar amounts that may be handled in this manner. van der Mast (1999) and van der Mast and Janssen (2001) came to the same conclusion after long-term studies of policy writers ‘at work’. Revised parts of policy documents contained more so-called ‘polyphonic text characteristics’ than the originals. Although van der Mast and Janssen and van der Mast do not use politeness theory as their point of departure, their examples can be easily analyzed in terms of politeness strategies. Policy writers consciously use shifts in perspectives, vagueness, impersonalizations, etc. to make their texts effective. Although research in business communication clearly indicates that politeness phenomena are omnipresent and policy writers are very aware of the interpersonal implications of their wordings, the aforementioned selection problem and the combination problem continue to be relevant for business communication. As for the selection problem, it is tempting to conclude that since politeness phenomena appear abundantly in business texts, the use of these strategies is advisable. Instead, we suggest evaluating the effectiveness of those politeness phenomena first to determine if the strategies are truly effective in strategic communication. The same concept applies to the combination problem. The strategies Brown and Levinson present as individual strategies for expository reasons are very

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2535

often used not in isolation but in combination in everyday communication. Letters, e-mails, and conversations usually contain more than one politeness strategy. The question remains: are politeness strategies in combination more effective than single strategies? 3. General introduction to the experiments The idea behind our study is simple: we asked our participants to evaluate refusal letters with or without combinations of politeness strategies by using 7-point Likert scales. The implementation of the study, however, is more complex; therefore, in this section we will address the design of the experiments, our choices of medium, genre, positive politeness strategies and the speech act ‘refusal’. Because the validity of an experiment is threatened when participants know what the experiment is about, we opted for a between-subject design in which our participants evaluated only one text in one condition. Furthermore, we made sure that the dependent variable (the scales used for evaluating the text) gave no clues about our intentions either. By mixing the Likert scales measuring ‘politeness effects’ with scales measuring other aspects of text quality, the validity of the experiment was protected. While most work on politeness has been done with conversational data, we opted for written language for two main reasons. First, texts lack the interactive dynamics of conversations, which is advantageous as it makes it possible to present our subjects stimulus material that can be held constant. Second, evaluating a text is more familiar for most subjects rather than evaluating a conversation (but see: Goldsmith and MacGeorge, 2000). We used business letters because, as teachers and researchers in a communication department, we need to advise our students on efficient and effective ways to communicate with and within organizations. One of the central topics in business correspondence is customer orientation. Many textbooks suggest ample use of politeness in order to maintain the client’s face (although terms such as linguistic politeness and ‘face’ are seldom used). In our research we feel obliged to validate this advice (Jansen, 1998). We have chosen refusal letters because they are ubiquitous and everybody has at least some experience with reading them. We presume that the participants’ recollection of these experiences will make empathizing with the receiver of the refusal letter easier. Furthermore, this type of letter is usually brief, which makes it appropriate for stimulus material. The amount of time required to read the letter and complete the questionnaire was minimal. Within the genre of business letters we chose to focus on refusal letters because they have three characteristics that make them well suited for politeness research. First, the central speech act ‘refusal’ makes them easily recognizable as a genre. Second, refusals are a very common genre that makes it relatively simple to find suitable participants for our experiments. Third there are several subgenres based on the content of what is refused (a job, a claim, a request, etc.). For us this differentiation is a major advantage because it makes studying varieties of a genre possible and, technically speaking, replicating the experiment and maintaining external validity becomes feasible. Refusals are also inherently interesting for politeness research because they belong to the so-called ‘dispreferred responses’ (Antaki, 1994:80–81, Hayashi, 1996). In most contexts in which a speaker requests something from a hearer, speaker and hearer both know that the speaker expects the hearer to comply. When the hearer refuses, he knows that he thwarts the speaker’s expectation by doing so. Using conversational analysis, Antaki (1994) and Hayashi (1996) demonstrate that the refuser feels the need to give an account for his or her choice of a dispreferred act, but this action is by no means the only option language user has. Liao and Bresnahan (1996) for instance, offer a list of 24 refusal strategies in Chinese and American English. Although some of them are rather special (for example ‘Divert the addressee’, ‘Switch to another language’) most of them are easily categorizable in Brown and Levinson’s typology. The mere existence of an extensive list of refusal strategies may be explained by the fact that they may redress the positive and the negative face of the addressee (Manno, 1999). The addressee’s negative face is threatened because he intended, or at least hoped, to obtain compliance with his request, and because of the refusal he could not act as he planned before. The addressee’s positive face is also threatened since the refuter dissociates himself from the requester by thwarting his expectations. This dichotomous situation enables us to investigate the effects on both types of face of the hearer. As stated earlier, the scope of our research is limited to positive politeness strategies for the following reasons. Positive politeness strategies are relatively underexposed in politeness research. The majority of politeness studies focus on negative politeness. This focus is skewed even more so in written communication. More importantly, positive strategies are less vulnerable to cross-cultural points of view, since the universal validity of the face concept has been contested for negative face only (Holtgraves, 2002; Watts, 2003) and effects of cultural differences found in empirical studies are also limited to offrecord strategies (Holtgraves, 2002) and negative politeness strategies (Fukada and Asato, 2004). Additionally, the diversity of positive politeness strategies makes them worthy of comparison. A perusal of Brown and Levinson’s strategies reveals that most of them are very different in scope. Considering their variations, exploring the effects of their use on language users is interesting and necessary. A final consideration is that the addition of positive politeness strategies is not dependent on the way the face-threatening act itself is formulated. Holtgraves and Yang (1992) found that the addition of positive strategies did not vary whether the face-threatening act was formulated bald on record, positive polite, negative polite or off record, while the amount of negative polite strategies varied greatly. Positive politeness strategies are stable in another way as well: Holtgraves and Yang (1992) found no relation between positive politeness strategies and power or weight of exposition, which make it somewhat easier to generalize from our results to other communicative situations. However, the same

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2536

Fig. 1. The additive effect of presenting 0, 1, 2 and 3 politeness strategies on the evaluation of a letter.

researchers stipulate that the occurrence of positive politeness strategies is positively related to a great distance between the communication partners (Holtgraves and Yang, 1992; Pilegaard, 1997). As the distance between sender and addressee in our refusal letters (insurance company–client) is also large, we expect the positive politeness strategies to be particularly apt. Our experience as consultants has taught us that the implementation of positive politeness strategies in bad news letters is not self-evident: letters presented to us by different insurance companies very often lacked signals of interest in the reader or signals of a common concern (see also Janssen and Jaspers, 2004). Therefore we believe it may be fruitful to explore the receivers’ reactions to combinations of positive politeness strategies by way of an experiment. 4. The combination experiment 4.1. Introduction: combinations of strategies As mentioned in section 2, we are interested in the question of how the combination of two or more positive politeness strategies influences the reader’s evaluation of written discourse. If we try to model the possible effects, we can construct at least three theoretical models: the additive model, the saturation model and the optimum model.4 Because, as we mentioned in the previous section, Brown and Levinson claim only a differential effect between the super categories we assume that all strategies within the super category of positive politeness have the same effect, (however see section 5). In the additive model (Fig. 1), the addition of each successive strategy has a distinct positive effect on the reader’s evaluation of the text. This model presupposes that strategies contribute to the perceived text quality independently. This is the simplest of models, which makes this model the most straightforward in view of the relation between Politeness-1 and Politeness-2. The relation between the dosage and effect is direct and proportional. While there is a lower limit (bald on record), there is no upper limit. In the saturation model, the same positive relation exists between addition and evaluation as in the additive model, but this time each successive extra strategy has a smaller effect on the evaluation than its predecessor (see Fig. 2). In this model the contribution of strategies to evaluation is practically bound to an upper limit that is reached asymptotically. The saturation model is relevant for cases in which every extra effort yields diminishing returns. The optimum model is different from the first two models in terms of its ultimate effects. While the additive and saturation models result in positive effects, in the optimum model the addition of more than the ‘‘optimum’’ number of politeness strategies will result in a negative evaluation (see Fig. 3). This model is relevant in cases where the absence of, or shortage of, politeness influences the reader’s perception of the writer as ‘rude’ and, at the same time, too many strategies lead to the suspicion that the writer is verbose or insincere or that the writer understands his act to be severely face-threatening. In other words, the optimum model predicts that there is an optimal dosage of politeness strategies for each and every message in a given culture. 4.1.1. Empirical studies of positive politeness strategies in bad news letters Combination has been the subject of a few studies and these studies suggest there is a positive evaluative effect as a result of combining positive politeness strategies. The corpus study of American refusal letters in an application procedure by Jablin and Krone (1984:397) shows that the positive politeness strategy ‘Praise the applicant’ occurs in 42% of the letters in their corpus, ‘Give reasons for the refusal’ in 82% and the expression of appreciation for the application, a form of paying attention to the needs of the applicant in 94%. The results of this study indicate that ‘praise strategy’ had a limited but positive effect on the subjects’ evaluation of the letters. The effects of the other strategies were less evident, partly because the design of this component makes it hard to estimate them. The corpus of business request letters of Bargiela-Chiappini and Harris (1996) 4 In theory, we can distinguish a fourth model: the retrogression model. This model would predict that using politeness strategies in the context of a refusal letter is less effective than presenting a message Bald on Record. We consider this model in conflict with the most fundamental findings of Brown and Levinson (1987) and for that reason disregard this option.

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2537

Fig. 2. The effect of adding one, two or three politeness strategies on the evaluation of texts in the saturation model.

Fig. 3. The effect of adding one, two or three politeness strategies on the evaluation of texts in the optimum model.

contains predominantly negative polite strategies, a result that may be attributed to the fact that they only analyzed the request and its direct context in the letters. Analyzing a larger corpus of several types of business letters, Pilegaard (1997) found that positive politeness strategies (alone or in combination) make up 41.2% of the strategies employed in the letters in which business partners negotiate an order, the type of letter in his corpus where we expect refusals to occur most frequently. Manno (1999) finds ‘Give Reasons’ in 84% of his corpus letters and the expression of appreciation of the candidates, which we consider as a form of ‘Exaggerate sympathy with the addressee’ in 33% of the letters. Furthermore, Manno gives examples of combinations of strategies in his corpus of French-Swiss refusals of job-applicants. Analyzing a corpus of 31 Dutch refusal letters of applications, Kok (1993) found combinations of two or more positive strategies in all of the letters. In all but one text ‘Give Reasons’ was present. Pilegaard, Kok nor Manno, however studied the effect of these combinations. Using an experimental design, Locker (1999) studied the effect of several types of so-called buffers (an initial clause or paragraph before the presentation of bad news). Locker found that a buffer consisting of a combination of strategies (mainly compliments and reasons) was appreciated more than a neutral buffer, consisting of a simple introduction without politeness. van Waes and van Wijk (2000) studied the effects of the implementation of a combination of several positive politeness strategies (and a combination of negative strategies as another independent variable) on the evaluation of product recalls. In one of their experiments, they found a positive effect of positive politeness strategies on the evaluation of the acceptance of the bad news, of the text and of the issuing company. Wang et al. (2008) studied the effect of a pedagogical style using bald on record or politeness strategies of ‘virtual agents’ on the effectiveness and efficiency of e-learning. They report a considerable positive effect of the use of positive politeness strategies, especially for e-learners without much computer experience. All in all it is evident that positive politeness strategies are used frequently in bad news letters and that they also occur in different combinations. Furthermore, layering seems to be effective in general. What we do not know, however, is how layering affects the readers’ evaluation. 4.2. Method 4.2.1. Material To ensure an externally valid design the texts used as stimuli for the experiment consisted of four different refusal letters of insurance claims. Had we limited the experiment to one letter, any effects found could be attributed to specific characteristics to this specific text. We constructed our four sample letters on the basis of real examples from Kok (1993) and a corpus presented to us by a major insurance company in the Netherlands. The following characteristics of the letters were held constant for all letters:

2538

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

- Authentic lay out: the conventional business letter style, with logo and address information - Format: ca 150 words The letters differed in the name of the company sending the letter (all names were made up in order to avoid an existing company’s image affecting the readers’ evaluation) and in the content of the insurance claim (a stolen watch and a broken bike). We utilized a one-way ANOVA to determine if these four text versions (2 insurance companies with 2 different claims) could be considered as replications. As it turned out that there were no significant differences (p > .05), text version was discarded as a factor in this experiment. In line with Brown and Levinson’s PDR-model, we held the difference in power and social distance between writer and reader constant. All letters came from insurance companies, and the subjects had to picture themselves as clients who had made a claim. The weight of the imposition expressed by the face-threatening act was the refusal of the moderate claim ( p < .7). In other words: men and women responded in the same way to our manipulations. The participants were between 22 and 64 years of age, had at least a 4-year college degree, some type of insurance, and a job. The average age of our participant was 39.6 (SD 11.2). We did not have any hypotheses about (nor any interests in) age. However, we checked our data for age effects nevertheless. We divided our participant into two groups: 39 and ran a two-way ANOVA with condition as second factor (besides age group). Again, we found no interactions (.06 > p < .47). So both age groups reacted consistently to the manipulations. Therefore our conclusions are assumed to be valid for both sexes and age groups. Finally, all participants were native Dutch speaking and had a Dutch cultural background. 4.2.5. Design and procedures All participants evaluated one letter in a between-subject design. They read the instructions that placed them in the position of a person who had just claimed a lost or broken item. Subsequently, they read the refusal letter completely unaware of the specific content of the questionnaire. Reading the letter took five minutes on average. Once they finished reading, the participants filled out the questionnaire. If needed they could reread the letter before they made any assessments on the scales. On average, they required eight to twelve minutes to complete the questionnaire. The experiment was pretested on 32 subjects. 4.3. Results We evaluated the homogeneity of the scales with a reliability analysis. The reliability of all the clusters was between high and very high (all Cronbach’s α > .75). The results of the experiment are presented in Table 1, and Fig. 4. Table 1 and Fig. 4 demonstrate that addition of one strategy to a Bald on Record-text has a positive effect on the evaluation. One-way ANOVA’s show that for almost all dependent variables the difference between Bald on Record and all other Table 1 Means (SD) of the evaluation of refusal letters without politeness strategies (Bald on Record) and with one, two and three politeness strategies. Means (standard deviations) in the 4 conditions (1 = low evaluation, 7 = high evaluation). Dependent variables

Strategies BoR

1 strategy

2 strategies

3 strategies

Writers kindness Image of W0 company Importance of relation with R Empathy W0 position vs. R Quality of motivation Compliance with refusal Tone of voice of the text Text structure

3.7 4.0 4.1 3.2 3.9 3.2 2.8 4.3 5.8

5.2 5.5 5.5 4.6 5.5 5.0 4.7 5.7 5.9

5.2 5.1 5.5 4.7 5.3 4.7 4.1 5.4 5.6

5.2 4.8 5.3 4.7 5.1 4.6 4.2 5.1 4.9

(1.4) (1.5) (1.6) (1.4) (1.4) (1.3) (1.3) (1.7) (0.9)

(1.7) (1.3) (1.7) (1.4) (2.2) (1.3) (1.4) (1.3) (1.0)

(1.2) (1.2) (1.1) (1.4) (1.2) (1.2) (1.3) (1.3) (1.2)

(1.2) (1.4) (1.2) (1.4) (1.3) (1.5) (1.6) (1.5) (1.4)

2540

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

Fig. 4. Effects of combining politeness strategies.

conditions is statistically significant (F > 25, df 3, p < .001). However, adding a second and a third strategy has no additional effect for most dependent variables as a post hoc Scheffe´ test reveals (for all contrasts p > .05). Letters with two or three politeness strategies are generally evaluated the same as letters with one strategy. The only exception is the evaluation of the letter’s structure. For this variable we found no effect of the addition of the first strategy, but we found a negative effect of adding a third strategy (F = 22.9, df 3, p < .001). We suppose that the greater length of the text, caused by the extra strategy, has made a less organized impression on the readers. 4.4. Preliminary conclusions What has this experiment revealed about the evaluation of politeness strategies? Addition of a politeness strategy to a refusal letter has a positive effect on the evaluation of the letter. The effects are not only strong but also general as they pertain to very different aspects of the letter, the writer and the organization. Furthermore, the evaluation of text and writer has an effect on the acceptance of the refusal. Adding a second strategy, and then a third strategy has no effect on the evaluation of most variables—and even a negative effect on the appreciation of text structure. Put quite simply, if we had to choose between the three models presented in section 4.1, our choice would be the saturation model because just one politeness strategy suffices to achieve the maximally positive evaluation after just one positive politeness strategy. However, there may still be an alternative explanation for this result. In the experiment, the strategies were added in the same order: 1. 2. 3. 4.

‘Bald on Record’ (condition 0), ‘Give Reasons’ (condition 1), ‘Give Reasons’ and ‘Give Compliments’ (condition 2), ‘Give Reasons’, ‘Give Compliments’ and ‘Assert Knowledge of S Wants’ (condition 4).

This fixed order means that the validity of our interpretation must be qualified with the proviso that all three strategies have the same effect on the readers. Because we are unaware of empirical evidence for this assumption, we decided to test it in a second experiment. 5. The evaluation of positive politeness strategies 5.1. Introduction: modeling the effects of positive politeness At the end of the previous section we mentioned a possible alternative explanation for the fact that the addition of two or more positive politeness strategies has no effect on the evaluation of letters: positive politeness strategies have a differential effect on the evaluation; in particular, ‘Give Reasons’ is effective, and both ‘Give Compliments’ and ‘Assert Knowledge for Hearer’s Wants’ are not. The simplest way to model this non-differentiality that was implicated in experiment 1 is: Effect Give Reasons ¼ Effect Give Compliments ¼ Effect Assert Knowledge The alternative model is that individual strategies have differential effects on the reader. We have reasons to assume that there is an inherent difference between ‘Give Reasons’ on the one hand and ‘Give Compliments’ and ‘Assert Knowledge’ on

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2541

the other hand. The first reason is based on the fact that many politeness strategies are molded in formulaic expressions with a relational meaning that is ‘bleached’ to the point of meaninglessness. An example is the Dutch equivalent of ‘please’: alstublieft (its literal meaning is ‘If it pleases you’). The etymological meaning validates the assumption that using ‘alstublieft’ is a negative politeness strategy, expressing that the hearer has the power to decide for himself, thereby mitigating the facethreatening act. However, in present-day conversational requests ‘alstublieft’ is considered an intensifier, so the chances are that its addition to a request has a strengthening effect. We think that standard compliments in letters underwent the same ‘bleaching’ process. Aside from these general grounds for assuming there are different effects from different politeness strategies, we can also hypothesize why ‘Give Reasons’ in refusals differs inherently from the other strategies. Let us begin with rethinking ‘Assert Speakers’ knowledge of Hearer’s wants’. By showing empathy with the hearer, the speaker makes clear that he is interested in the hearer and wants a good relationship with him, so that this cooperative context and situation can compensate for the harm done by the FTA. In other words, this strategy is immediately directed at preserving or restoring the hearer’s positive face, more or less independently from the FTA itself. In the case of the positive politeness strategy ‘Give Compliments’ the contribution to the hearer’s face is not that direct because it works through appraising some aspect of the hearer. What this strategy has in common with the previous one, however, is that it is again a direct contribution to hearer’s face and that as such it is independent from the FTA itself. The utilization of ‘Assert Knowledge’ and ‘Give Compliments’ may be done with good intentions, but their deployment can be risky. The hearer (or reader) may deduce from the compliments he has received and from the recognition of his needs that the speaker shows solidarity. This perceived like-mindedness can easily lead to the expectation that the request (or the claim in our experiments) will be met. When the refusal is presented there is a chance that the hearer feels the unpleasant sensation of being manipulated by the strategies that are now believed to be perceived insincere. Furthermore, the hearer may experience asymmetry between the general (positive) characteristics attributed to him by the strategies and the negative circumstances of the refusal. Compared to the other positive politeness strategies, ‘Give Reasons’ works more indirectly. Compare the following two answers of a human resources manager on the request ‘Do you have a job for me?’ a

no

b

no, we just hired someone

The second choice (b) is an argumentation. The standpoint (‘no’) is accompanied by a proposition that is added for one reason only: to make the standpoint more acceptable. That the manager takes the trouble of adding this proposition is the first reason to call it a positive politeness strategy. In addition, the function of the added proposition is to present a fact from which the standpoint follows logically. This argumentative characteristic has a number of consequences for the positive face of the hearer: by disclosing a fact about the organization, the manager takes the applicant into confidence, which is good for the applicant’s positive face. Presenting grounds and letting the applicant deduce the logical conclusion himself is also comforting for his positive face because (at least) the Dutch like to think of themselves as logical and independent thinkers. Furthermore, the applicant’s positive face is comforted by the specific content of the argument. By presenting this argument, instead of an alternative ‘I do not like your moustache’, the applicant has less reason to assume that he was turned down for personal reasons or reasons of competence. However, following this last line of argumentation ‘Give Reasons’ may also be considered a negative politeness strategy: by presenting an objective circumstance of the refusal, the human resources manager detaches the applicant and himself from the refusal. In sum, ‘Give Reasons’ is a rather special positive politeness strategy because it restores the hearer’s face indirectly, through substantiation of the refusal. By consequence, ‘Give Reasons’ is more closely linked to the content of the face-threatening act than other positive politeness strategies and this may well explain the effects of the first experiment. There are a few extra arguments why ‘Give Reasons’ has a special status in the context of refusals. We have already mentioned that the refusal of a request is considered a dispreferred response (Antaki, 1994; Hayashi, 1996) and that a dispreferred response urges an account by the refuser. In the context of a request, giving the reason for the refusal is the more logical option. So examples of ‘Give reasons’ abound in the conversational analysts literature on refusals. All Antaki’s examples of refusal in dialogues are accompanied by reasons. ‘Give reasons’ is also omnipresent in the Mexican-Spanish refusals analyzed by Fe´lix-Brasdefer (2006) who calls them ‘grounds’. Chen (2001) mentions justification (=Give Reasons) as the first strategy in his typology of self-politeness strategies. Discussing experimental evidence, Holtgraves (1992) also mentions ‘Justification’ as one of three types of accounts (the other two are ‘Concession’ and ‘Excuse’). Holtgraves finds that this orientation towards the face of the sender makes the subjects of his experiment consider ‘Justification’ to be the least rewarding type of account. Finally, most of the proportions of the distinctive positive politeness strategies in business letters are in line with this prediction. Manno (1999) and Kok (1993) found more examples of ‘Give Reasons’ than of the other positive politeness strategies in their respective corpora of refusal letters, and in Jablin and Krone’s corpus, ‘Give Reasons’ was only outnumbered by ‘Appreciation’. How should ‘Give Reasons’ weigh up against the other two positive strategies: Assert Knowledge and Give Compliments? The Elaboration Likelihood Model (Petty and Caccioppo, 1984) may help us to find the answer. Essential in this model is the distinction between two routes of persuasion: the central and the peripheral. In the central route the writer presents

2542

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

arguments, giving the reader the opportunity to assess their correctness and relevance. The peripheral route is for readers who lack the time or cognitive energy to evaluate the argumentation. Readers can be persuaded by other text elements than arguments, such as allusions to people that bring about positive associations, flattery of the reader, clues for rules of thumb, etc. Although both routes to persuasion have their pros and cons, empirical evi demonstrates that the central route is superior to the peripheral route in that convincing arguments lead to a set of attitudes that is more stable than a set of attitudes brought about by flattery and the like. The elaboration likelihood model is a relevant framework, because it permits us to compare the presumed contribution of ‘Give Reasons’ (the central route) with the contributions of the other two strategies (peripheral route). It predicts that in refusals—when the personal interests of the reader are at stake—readers will be inclined to invest cognitive energy in evaluating the arguments presented in the text, rather than rely on compliments and others demonstrations of empathy: Effect Give Reasons > Effect Give Compliments and Assert Knowledge 5.2. Method 5.2.1. Material The construction of the material was in concordance with the first experiment. We composed two letters, in order to be able to generalize over letters that reject an insurance claim. One in which the claim referred to a missing watch and the other to a missing camera. The letters had the same layout: one text page, about 150 words and identical fonts. Furthermore, the letters had the same introductory and final paragraph in all conditions. The only variation was in the second paragraph where the face-threatening act, the refusal, was presented with or without redressing positive politeness strategies. 5.2.2. Manipulations: independent variables The independent variables were, of course, types of positive politeness strategies in different combinations. There were three basic conditions: 1. 0 strategies (Bald-on-record) 2. 1 strategy: ‘Give Compliments’ 3. 1 strategy: ‘Assert Knowledge for Hearer’s Wants’ In this way we could test if the effects of one strategy that we found in the first experiment should be attributed to ‘Give Reasons’ or to the use of merely one politeness strategy in general. In order to be able to test for combination effects, we incorporated the following conditions: 4. 2 strategies: ‘Give Reasons’ and ‘Give Compliments’ 5. 2 strategies: ‘Give Reasons’ and ‘Assert Knowledge of H’s wants’ 6. 2 strategies: ‘Give Compliments’ and ‘Assert Knowledge of H’s wants’. To be more precise, if the results of Experiment 1 should be attributed to the saturation model, we expect a significant increase in the evaluation from condition 1 to 2, and 1 to 3, and a smaller increase (or none at all) from condition 2/3 to 6. However, if the results of Experiment 1 should be ascribed to the special nature of ‘Give Reasons’ as claimed in section 3, we expect conditions 4 and 5 (with ‘Give Reasons’) to differ from 6. 5.2.3. Participants 124 men and 116 women participated in the experiment. They were all between 18 and 74 years of age. The average age was 36 years (SD = 13). All participants had at least a 4-year college degree, some type of insurance, and a full-time or parttime job. They were all native Dutch speakers and shared a Dutch cultural background. Once again, a two-way ANOVA showed no significant interaction effect of gender with our main factor, the politeness strategies (.25 < p > .97). In order to measure an effect of age, we divided our participants into two groups: 36 years of age and performed another twoway ANOVA. with age as one of the main factors. There was no significant interaction effect (.27 < p > 93). In other words, men and woman and people in both age groups reacted identically to our manipulation. Therefore our conclusions are assumed to be valid for both genders and age groups. 5.2.4. Design and procedures A between-subject design was once again chosen in order to avoid a contamination of evaluations. After they had read the letter at their own speed, participants filled out the same questionnaire of 37 7-point Likert scales as in Experiment 1. Positively and negatively stated propositions were mixed, and the poles of the negatives were reversed in order to make reliability analysis and subsequent clustering possible. In our results, a higher score always means a more positive evaluation. The procedures were identical to the ones in the first experiment. The experiment was pretested on 48 subjects.

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2543

Table 2 Means (standard deviations) of the evaluation (1 = low evaluation, 7 = high evaluation) in the 4 conditions. Dependent variables

Strategies BOR

1 strat (GR)

2 strat (+GR)

2 strat (GR)

Writers kindness W position towards R Importance of relation with R Quality of motivation Empathy Tone of voice of the text Compliance with refusal Text structure

4.3 4.3 2.9 2.1 3.8 4.6 2.3 5.7

3.9 3.9 3.7 2.5 4.0 4.4 2.1 5.0

4.6 4.7 4.3 4.4 4.6 4.9 3.0 5.0

3.5 3.5 3.1 3.1 3.9 3.6 1.6 3.9

(1.3) (1.2) (1.2) (1.2) (1.1) (1.5) (1.2) (1.2)

(1.5) (1.5) (1.5) (1.6) (1.3) (1.5) (1.1) (1.7)

(1.2) (1.3) (1.4) (1.6) (1.2) (1.5) (1.5) (1.8)

(1.1) (1.5) (1.5) (1.0) (1.1) (1.7) (1.7) (2.1)

Fig. 5. Results experiment 2.

5.3. Results and conclusions from Experiment 2 What were the results of the second experiment? First, we assessed the reliability of the scales. Cronbach’s alpha proved to be >.75 on all scales, indicating that the homogeneity of the clusters was high to very high. The two text samples (‘watch’ and ‘bicycle’) were evaluated in the same way; we therefore merged the data and analyzed the two samples as one. Furthermore, analyses of variance showed no statistically significant effects in condition 2 versus 3 and condition 4 versus 5. We also merged the strategies ‘Give Compliments’ and ‘Assert Knowledge’. In other words, these two independent variables were considered as one, which gave us more statistical power for further analysis. In Table 2 and Fig. 5 we give an overview of the results of our experiment (mean scores and standard deviations on the clusters). Table 2 and Fig. 5 demonstrate that there is no general rising slope between ‘Bald on Record’ and one strategy, and also no smaller slope between one strategy and the two strategies ‘Give Compliments’ and ‘Assert Knowledge’ just like the Saturation Model would have predicted. What we do see is that evaluations peak as soon as the letter contains the strategy ‘Give Reasons’, as predicted by our ‘Inherently Effectiveness Hypothesis’ (F > 7.6, df 3, p < .01). In other words the implementation of the two other positive politeness strategies seems to have no effect at all. When we evaluated the effects of these strategies in more detail with post-hoc analyses (Scheffe´)., it turned out that the combination of the strategies ‘Give Compliments’ and ‘Attend to H’ leads to a significantly lower evaluation on the dependent variables ‘Agreement’, ‘Tone’, and ‘Structure’ than the Bald on Record version. We interpret this outcome as an indication of a negative combination effect in accordance with the optimum model. Refusals with a reason do not benefit from the addition of one or two extra positive politeness strategies. The extra strategies do not harm perception either. But if the reasons are omitted, the successive addition of other strategies will have a cumulative negative effect. Presumably, in case of Bald on Record, the reader has no strong feelings about the absence of reasons (‘The writer must have been in a hurry’). But as the amount of other strategies increases, this omission is more evident, and it may cause irritation and suspicion: ‘Why is the writer showing so much concern about my personal well-being and at the same time withholding the explanation for the refusal?’ This second experiment supports the ‘Inherent Difference Model’ (section 4.1): not all positive politeness strategies are equal. At least in this genre (refusals) and this culture (Dutch), ‘Give Reasons’ is effective and the other two are not. We also found a limiting effect for layering strategies. The results favor the optimum model: in the context of a refusal letter without argumentation of the re, the more other positive politeness strategies added, the more negative the evaluation, and thus the acceptance, of the text will be.

2544

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

6. General conclusion and discussion Considering the limitations discussed below in section 6.1, we may conclude that positive politeness strategies differ dramatically in effectiveness. The addition of ‘Give Reasons’ always makes a difference, whether as a single strategy or in combination. Addition of one of the positive politeness strategies ‘Give Compliments’ or ‘Assert Knowledge’ has no effect, whether it is added as a single strategy or in combination with ‘Give Reasons’. Addition of a combination of ‘Give Compliments’ and ‘Assert Knowledge’ in absence of ‘Give Reasons’ has a negative effect on several variables. In terms of the presented models, the results point in the direction of the Inherent Difference Model, ‘Give Reasons’ being significantly different from the other strategies. As far as the models for the combination problem are concerned, the results are less conclusive because of the ineffectiveness of ‘Give Compliments’ and ‘Assert Knowledge’ as individual strategies. The only combination effect we found, the negative effect of the combination of ‘Give Compliments’ and ‘Assert Knowledge’, points in the direction of the optimum model. 6.1. Limitations The validity of both experiments is limited as they were performed in one language and culture (Dutch, the Netherlands), in one mode and genre (written language, business letters), and in the context of one face-threatening act (refusals). Furthermore, the design of the experiments results in inevitable limitations. As mentioned in section 1, a consequence of our approach is that we abandon any claim on universality of Politeness-2 effects. Therefore no predictions of similar effects can be inferred when these experiments are replicated in other cultures as cultural factors may have a significant influence on the evaluation of politeness strategies (Ambady et al., 1996; Blum-Kulka et al., 1989; Chen, 2001; Fukada and Asato, 2004; Haugh, 2007; Holtgraves and Yang, 1992; Hill et al., 1986; Mao, 1994; Yeung, 1997; Watts, 2003). Although the majority of this type of research seems to be based on the comparison of the AngloAmerican culture with Asian cultures, research contrasting European cultures (see for example Hendriks, 2002; Huls, 1991; van der Wijst, 1996) seems to point in the same direction. As these studies employ discourse completion tests and other methodologies that inherently underestimate chance as an explanation, we strongly encourage other researchers to replicate our methodology for studies in other languages and different cultures. As far as the language mode (written language, business communication) is concerned, it is conceivable that the predominance of ‘Give Reasons’ in the experiments, can be attributed to the fact that the conditions were presented in the written mode. This consideration is especially relevant for the two less successful strategies ‘Give Compliments’ and ‘Attend to H’s wants’ that capitalize on a direct redress of the face of H. This need for direct redress seems to be characteristic of high involvement communication, while written language is less suitable for high involvement communication because the writer and reader are absent (Biber, 1988; Chafe, 1982; Crystal, 2000:27). It can be argued that this aversion of positive face redress is even less relevant for written business communication since it strives for a certain independency between the communicators, especially in financial matters. However, this objection would be irreconcilable with the results of the corpus research discussed in section 4, which shows that these strategies occur with some frequency in business letters. Furthermore, the topic in our experiments was the refusal of a remittance claim. It remains to be seen whether other refusal letters will produce the same results. Finally, the evaluations of participants in an experiment may be different from readers who have just received a rejection of their remittance claim. The idea behind the effectiveness of politeness strategies is that they are balms for a threatened face. But whose face is really threatened in the experiments? It is the face of a fictitious addressee who submitted the claim, not the face of the participant. Although all the participants had experience with refusal of remittance, and trying to ‘get something back from your annual premium’ is a national sport in the Netherlands, the fact that they had not themselves submitted a claim, may have influenced their evaluations. As the experimental design demands that the stimulus material is identical (except for the independent variable), there is no way to circumvent this problem. It is inherent to experimentation. 6.2. Relevance of our findings We think that our results are directly relevant for researchers who, until now, have assessed the overall ‘politeness’ of a text by counting the number of politeness strategies, as we see in corpus studies by Pilegaard (1997) and Rees-Miller (2000) and in most analyses of discourse completion tests. If our findings are valid, this approach is questionable because one instance of one effective strategy proved to be more relevant for the assessment of overall politeness than several others. This result makes it probable that perceived politeness has little to do with aggregative attestations of politeness strategies. The deductive approach by Brown and Levinson made them characterize Justifications as positive politeness strategies. Brown and Levinson are right as long as we maintain the deductive approach. However, when we work ‘the other way around’ and interpret linguistic utterances in terms of politeness strategies, we must not forget that those utterances, just like every meaningful verbal expression, are more than mere instances of a politeness strategy from a semantic and a pragmatic viewpoint. Utterances with an argumentative function are subject to the evaluation of the content as well. And this effect emerged in our experiments. Our participants were totally unaware of our research goals. They evaluated a letter

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2545

but did not know that in fact they evaluated the politeness of the text. This may explain why we did not find any significant effects of politeness (just like Goldsmith and MacGeorge, 2000): the potential effects of politeness strategies disappeared amidst other stimuli. As stated earlier, we did not find any effects of gender. So, within the general limitations of our study, we may conclude that men were as unsusceptible to positive politeness (other than ‘Give Reasons’) as women. The absence of gender effects isn’t surprising. Holtgraves (2001 and 2002:70–72) and Mills (2003) noticed that some studies find an effect of sex while others do not. However, the picture of gender effects is not very clear. Most studies that have found differences in gender reaction were studies in how politeness is used by men and women and mainly focused on negative politeness (cf. Lakoff, 1977; Crosby and Nyquist, 1977; Tannen, 1990). In our experiments, however, our subject judged letters that were written by men and did not differ in directness. Furthermore, there were no relevant differences in power or distance; all subjects were placed in the same experimental situation of a claim holder whose claim was refused by the insurance company. If we try to interpret our results within a cognitive framework such as the ELM model, as mentioned in section 5.1, it is plausible that positive politeness strategies in general work as peripheral cues that are only taken into consideration when ‘Give Reason’ is absent. The presence of ‘Give Reasons’ in most of our experimental conditions then functions as a cognitive ‘black hole’, dragging our participants into the central route of persuasion, ignoring all peripheral stimuli. This explanation, however, is challenged by the fact that in the second experiment, no effects were found in all conditions without ‘Give Reasons’. Therefore, we may wonder whether or not positive politeness strategies have some form of cognitive representation. Penelope Brown (1995) states that ‘particular linguistic realizations are not ever intrinsically positively or negatively polite, regardless of context. Politeness inheres not in forms, but in the attribution of polite intentions, and linguistic forms are only part of the evidence used to assess utterances and infer politeness’ (p. 169, our italics). If this is the case, politeness research may profit from psychological experimentation that explains how people attribute those intentions and how politeness is inferred (cf. Holtgraves, 2002:4). Finally, there may even be a more ‘fundamental’ explanation for our results. It is not inconceivable that ‘being treated as a rational individual’ is one of the basic wants in Dutch society and that fulfilling this need by ‘giving reasons’ is sine qua non in interaction. Of course, it is still too early to challenge Brown and Levinson’s presumptions about people’s basic needs on the basis of our experiments. But nevertheless, it is an interesting and plausible idea that basic wants differ among cultures. It is understood that methods other than experimenting, such as discourse completion tests, reveal more differential effects of individual and layered politeness strategies than we have found in our experiment. Does mean that the jury is still out? We do not think so. In section 1 we have expressed our concerns about the validity of discourse completion tests. The participants in completion tests are able to reconstruct the goal of the test while they are doing it because of the systematic repetitions and variations. Therefore there is a reasonable chance that they are focused on the politeness phenomena and that the researchers will thus overestimate the relevance of politeness strategies on the basis of the data collected. In our experimental approach, we have a precise idea about what we know and what we do not know about the interface between the ‘etic’ conception of the strategies and the ‘emic’ conception of their effects. We know that ‘Give Reasons’ has an impact on the evaluation of refusal letters and that other strategies have none. In this respect, we are now in a more comfortable position than in the case of the conversation analyst who has to make strict interpretations within the dynamics of a specific interaction. Acknowledgements We thank our students for their collaboration in designing and implementing the experiments, the attendants of the presentation of this paper on the 2007 ABC-conference in Istanbul for their interesting questions and our colleagues Luuk van Waes (University of Antwerp) and Geert Jacobs (University of Gent) for their helpful comments on an earlier draft. We also thank two anonymous reviewers for their helpful and distinct suggestions. Derek Webb and Alaina Denney and Mrs Denney thoroughly edited an earlier version and the final version of this article.

2546

Appendix A

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

2547

References Ambady, Nalini, Koo, Jasook, Lee, Fiona, Rosenthal, Robert, 1996. More than words: linguistic and nonlinguistic politeness in two cultures. Journal of Personality and Social Psychology 70 (5), 996–1011. Antaki, Charles, 1994. Explaining and Arguing. The Social Organization of Accounts. Sage, London. Bargiela-Chiappini, Francesca, 2003. Face and politeness: new (insights) for old (concepts). Journal of Pragmatics 35, 1453–1469. Bargiela-Chiappini, Francesca, Harris, Sandra J., 1996. Requests and status in business correspondence. Journal of Pragmatics 28, 635–662. Baxter, Leslie A., 1984. An investigation of compliance-gaining as politeness. Human Communication Research 10, 427–456. Biber, David, 1988. Variation Across Speech and Writing. Cambridge University Press, Cambridge. Blum-Kulka, Shoshanna, House, Juliane, 1989. Cross-cultural and situational variations in requesting behavior. In: Blum-Kulka, S., House, J., Kasper, G. (Eds.), Cross-Cultural Pragmatics: Requests and Apologies. Ablex, Norwood (NJ), pp. 123–155. Blum-Kulka, Shoshanna, House, Juliane, Kasper, Gabriele (Eds.), 1989. Cross-Cultural Pragmatics: Requests and Apologies. Ablex, Norwood (NJ). Brown, Penelope, 1995. Politeness strategies and the attribution of intention: the case of Tzeltal irony. In: Goody, E.N. (Ed.), Social Intelligence and Interaction: Expressions and Implications of the Social Bias in Human Intelligence. Cambridge University Press, Cambridge, pp. 153–174. Brown, Penelope, Levinson, Stephen C., 1987. Politeness: Some Universals in Language Use. Cambridge University Press, Cambridge.

2548

F. Jansen, D. Janssen / Journal of Pragmatics 42 (2010) 2531–2548

Chafe, Wallace L., 1982. Integration and involvement in speaking writing and oral literature. In: Deborah Tannen (Eds.), Spoken and Written Language. Ablex, Norwood, 53-53. Chen, Rong, 2001. Self-politeness: a proposal. Journal of Pragmatics 33, 87–106. Clark, Herbert H., Schunk, Dale H., 1980. Polite responses to polite requests. Cognition 8, 111–143. Crosby, F., Nyquist, L., 1977. The female register: an empirical study of Lakoff’s hypothesis. Language in Society 6, 313–322. Crystal, David, 2000. Language and the Internet. Cambridge University Press, Cambridge. de Jong, Jaap (Ed.), 2002. Handboek bedrijfscorrespondentie. (Handbook Business Correspondendence).Martinus Nijhof, Groningen. Eelen, Gino, 2001. A Critique of Politeness Theories. St. Jerome Publishing, Manchester/Northampton. Fe´lix-Brasdefer, J. Ce´sar, 2006. Linguistic politeness in Mexico: refusal strategies among male speakers of Mexican Spanish. Journal of Pragmatics 38, 2158– 2187. Fukada, Atsushi, Asato, Noriko, 2004. Universal politeness theory: application to the use of Japanese honorifics. Journal of Pragmatics 36, 1991–2002. Goffman, Ervin, 1967. Interaction Ritual; Essays on Face to Face Behavior. Garden City, New York. Goldsmith, Daena J., MacGeorge, Erina L., 2000. The impact of politeness and relationship on perceived quality of advice about a problem. Human Communication Research 26 (2), 234–264. Hagge, John, Kostelnick, Charles, 1989. Linguistic politeness in professional prose: a discourse analysis of auditors’ suggestion letters, with implications for business communication pedagogy. Written Communication 6, 312–339. Haugh, Michael, 2007. Emic conceptualisations of (im)politeness and face in Japanese: implications for the discursive negotiation of second language learner identities. Journal of Pragmatics 39, 657–680. Hayashi, Takuo, 1996. Politeness in conflict management: a conversation analysis of dispreferred message from a cognitive perspective. Journal of Pragmatics 25 (2), 227–255. Hendriks, Berna, 2002. More on Dutch-English . . .Please. A study of request performance by Dutch native speakers, English native speakers and Dutch learners of english. Doctorate Dissertation, Radboud University, Nijmegen. Hill, B., Ide, S., Ikuta, S., Kawasaki, A., Ogino, T., 1986. Universals of linguistic politeness: quantitative evidence from Japanese and American English. Journal of Pragmatics 10, 347–371. Holtgraves, Thomas, 1992. The linguistic realization of face management: implications for language production and comprehension, person perception and cross-cultural communication. Social Psychology Quarterly 55, 141–159. Holtgraves, Thomas, 2001. Interpreting questions and replies: effects of face-threat, question form, and gender. Social Psychology Quarterly 54 (1), 15–24. Holtgraves, Thomas, 2002. Language as Social Action. Lawrence Erlbaum Associates, Publishers, London/New Jersey. Holtgraves, Thomas, Yang, Joong-Nam, 1992. Interpersonal underpinnings of request strategies: general principles and differences due to culture and gender. Journal of Personality and Social Psychology 62 (2), 246–256. Huls, Erica, 1991. Een Landenvergelijkend Onderzoek naar de Formulering van een Verzoek.(An study in the wording of requests in different countries). In: van Hout, R., Huls, E. (Eds.), Artikelen van de Eerste Sociolinguistische Conferentie. Eburon, Delft, pp. 205–219. Jablin, Frederic M., Krone, Kathleen, 1984. Characteristics of rejection letters and their effects on job applicants. Written Communication 1 (4), 387–406. Jansen, Frank, 1998. In search of valid advice for the optimalization of functional texts: introduction to communication studies in the Netherlands. South African Journal of Linguistics 35, 15–27. Janssen, Daniel (Ed.), 2007. Zakelijke communicatie 1. (Business Communication 1).Wolters Noordhoff, Groningen. Janssen, Daniel, Jaspers, Joep, 2004. Client-oriented correspondence: how do you actually do it? Document Design 12 (3), 187–189. Jary, Mark, 1998. Relevance theory and the communication of politeness. Journal of Pragmatics 30, 1–19. Kemper, Susan, Thissen, David, 1981. How polite? A reply to Clark & Schunk. Cognition 9, 305–309. Kok, Anne-Mieke, 1993. ‘Zo’n Onaangename Standaardbrief’ Een Onderzoek naar Afwijzingsbrieven aan Sollicitanten (‘‘Such an unpleasant form letter’’. A study of rejection letters to job applicants). Doctoral Thesis ‘Algemene Letteren’. Utrecht University. Lakoff, Robin, 1977. Women’s language. Language and Style 10, 222–248. Le Pair, Rob, 2005. Politeness in the Netherlands: indirect requests. In: Hickey, L., Stewart, M. (Eds.), Politeness in Europe. Multilingual Matters Ltd., Clevedon, pp. 66–81. Liao, Chao-chih, Bresnahan, Mary I., 1996. A contrastive pragmatic study on American English and Mandarin refusal strategies. Language Sciences 18, 703– 727. Locker, Kitty O., 1999. Factors in reading responses to negative letters; experimental evidence for what we teach. Journal of Business and Technical Communication 13 (1), 5–48. Manno, Giuseppe, 1999. Savoir Refuser a` l’Ecrit: Analyse d’un Enchaıˆnement Non Pre´fe`re de Macro-actess de Discours. Journal of French Language Studies (9) 1, 39-68. Mao, LuMing R., 1994. Beyond politeness theory: ‘face’ revisited and renewed. Journal of Pragmatics 21, 451–486. Matsumoto, Yoshiko, 1988. Reexamination of the universality of face: politeness phenomena in Japanese. Journal of Pragmatics 12, 403–426. Mills, Sara, 2003. Gender and Politeness. Cambridge University Press, Cambridge. Petty, Richard E., Caccioppo, John T., 1984. The effects of involvement on responses to argument quality and quantity. Journal of Personality and Social Psychology 46, 69–81. Pilegaard, Morten, 1997. Politeness in written business discourse: text linguistic perspectives on requests. Journal of Pragmatics 28, 223–244. Rees-Miller, Janie, 2000. Power, severity and context in disagreement. Journal of Pragmatics 32, 1087–1111. Rothschild, Ewald H., Burnett, Rebecca, 1997. Business Communication. Prentice-Hall, New Jersey. Tannen, D., 1990. You Just Don’t Understand. Morrow, New York. van der Mast, Niels, 1999. Woordenwisseling; een onderzoek naar de manier waarop schrijvers consensus over beleidsteksten bewerkstelligen ((Ex) changes of Words; An Inquiry into the Way Writers Create Consensus on Policy Documents). Thesis, Amsterdam. van der Mast, Niels, Janssen, Daniel, 2001. Strategic revision of public documents: how policy writers revise documents in order to reach consensus. In: Janssen, D., Neuteling, R. (Eds.), Reading and Writing Public Documents. John Benjamins, Amsterdam/Philadelphia, pp. 171–210. van der Wijst, Per, 1996. Politeness and negotiations. Doctorate Dissertation. Katholieke Universiteit Brabant. van Waes, Luuk, van Wijk, Carel, 2000. The influence of politeness on the perception of product recall notices. Document Design 2, 272–279. Wang, Ning W., Johnson, Lewis, Mayer, Richard E., Rizzo, Paola, Shaw, Erin, Collins, Heather, 2008. The politeness effect: pedagogical agents and learning outcomes. International Journal of Human–Computer Studies 66, 98–112. Watts, Richard J., 2003. Politeness. Cambridge University Press, Cambridge. Watts, Richard, Ide, Sachiko, Ehlich, Konrad, 1992. Introduction. In: Watts, R., Ide, S., Ehlich, K. (Eds.), Politeness in Language: Studies in History, Theory and Practice. Mouton De Gruyter, Berlin, pp. 1–17. Yeung, Lorrita N.T., 1997. Polite requests in English and Chinese business correspondence in Hong Kong. Journal of Pragmatics 27, 505–522.