Social Psychological Models of Interpersonal Communication

states or predispositions. Any particular behavioral episode can be characterized in a variety of ways at different levels of abstraction: "A Models of Interpersonal Communication page 32 punches B," or "A hurts B," or "A dislikes B." The most abstract way to characterize a behavior would be as evidence of a predisposition: "A is aggressive." Maass et al. found that negatively valent behaviors of outgroup members tend to be characterized at relatively high levels of abstraction, and those of in-group members are characterized more concretely, but for positively valent behaviors the pattern is reversed. Positively valent behaviors of out-group members are characterized as specific episodes, while those of in-group members are characterized abstractly. Maass et al. call this the "linguistic intergroup bias" (see also Hamilton, Gibbons, Stroessner, & Sherman, 1992; Maass & Arcuri, 1992). One consequence of the linguistic intergroup bias is to help make stereotypes resistant to disconfirmation, since behaviors that are congruent with the negative out-group stereotype will tend to be characterized as general properties ("Smith is lazy"), while behaviors that are inconsistent with the stereotype will tend to be characterized in quite specific terms ("Smith painted his house"). Although examining the causal implications of language has yielded fascinating results, there are reasons to be cautious about generalizing these findings to language use. Edwards and Potter (1993) have pointed out that simple, out of context subject-verb-object sentences of the kind typically used in studies of implicit causality are rarely encountered in discourse. Consequently, the judgments subjects make from them may have little to do with the way language normally is processed in communication. Seen in isolation, "Alan desires Jane" may be understood as consequence of Jane's desirability, but in the context of a narrative that depicts Alan as a compulsive womanizer, his desire for Jane may be attributed less as to her desirability than it is to his proclivity. Models of Interpersonal Communication page 33 Is implicit causality really a matter of encoding and decoding? Or, to put it another way, is an interpersonal verb's causal implications part of its linguistic meaning, or is it an inference an addressee will draw in a particular context of usage about what the speaker intended? Semin and Marsman (in press) argue that interpersonal verbs invite inferences about a variety of properties (e.g., the perceived temporal duration of the action or state, how enduring a quality they imply, affective consistency, etc.), causal agency being only one of them. Researchers have assumed that interpersonal verbs automatically trigger inferences about causal agency, but Semin and Marsman suggest that such inferences are themselves a consequence of contextual factors (e.g., the question the subject is asked). Much of the work on implicit causality has approached the phenomenon in linguistic terms, but it may be more readily understood as part of the addressee's attempt to infer an intended meaning. The general question of how addressee's extract intended meanings from messages is discussed in Section 3. 2.3 Issues and Limitations Two features of the Encoder/Decoder model should be highlighted. One is implicit in the very notion of a code, and is illustrated in the early color codability studies. It is that the meaning of a message is fully specified by its elements—i.e., that meaning is encoded, and that decoding the message is equivalent to specifying its meaning. The other feature is that communication consists of two autonomous processes—encoding and decoding. We have tried to illustrate the Encoder/Decoder schematically in Figure 1. Despite the fact that language can in certain respects be regarded as a code, and the fact that both encoding and encoding processes are involved in communication, encoding and decoding do not adequately Models of Interpersonal Communication page 34 describe what occurs in communication, as will be discussed in the next three sections. Here we will just briefly point to some areas where the approach falls short. In the first place, it is often the case that the same message can (correctly) be understood to mean different things in different circumstances. For example, some messages are understood to mean something other than their literal meaning. While there is not universal agreement on the value of the literal vs. nonliteral distinction (Dascal, 1989; Gibbs, 1982, 1984; Katz, 1981; Keysar, 1989; Searle, 1978), it is abundantly clear that the most commonplace utterance (e.g., "You're leaving") can be understood differently in different contexts (e.g., as an observation of a state of affairs, as a prediction of a future state of affairs, etc.). Without making the relevant context part of the code, a model that conceptualizes communication as simply encoding and decoding will have difficulty explaining how the same message can be understood to mean different things at different times. Moreover, even when context is held constant, the same message can mean different things to different addressees. And there is considerable evidence to indicate that speakers design messages with their eventual destinations in mind (Bell, 1980; Clark & Murphy, 1982; Fussell & Krauss, 1989a; Graumann, 1989; Krauss & Fussell, 1991). Similarly, there is growing evidence that nonverbal behaviors are not simply signs that encode internal state in a straightforward way. A facial expression may be related to a person's internal state, but comprehending its significance can require considerably more than simply identifying the expression as a smile, a frown, an expression of disgust, etc. For example, smiles are understood to encode a affectively positive internal state, but they hardly do this in a reflexive fashion. In a series of ingenious field Models of Interpersonal Communication page 35 experiments, Kraut (1979) found smiling to be far more dependent on whether or not the individual was interacting with another person than it was on the affective quality of the precipitating event, and Fridlund (1991) has shown that even for people who were alone, the belief that another person was engaged in the same task (albeit in another room) was sufficient to potentiate smiling. In dyadic conversations, the facial expressions of the listener (i.e., the person not holding the conversational floor at a given moment) may change rapidly. Some of these changes (e.g., smiles) may represent back-channel signals (Brunner, 1979; Chen, 1990), while others (e.g., wincing at the other's pain) may serve to signal the listener's concern (Bavelas, Black, Chovil, Lemery, & Mullet, 1988; Bavelas, Black, Lemery, & Mullet, 1986). Even aspects of voice quality cannot be straightforwardly interpreted. For example, a speaker's vocal pitch range is a consequence of the architecture of the vocal tract. However, social factors can influence how a given speaker places his or her voice within that range. Men seem to place their voices in the lower part of their vocal range, and women do not, which, incidentally helps explain why a man's size can more accurately be predicted from his voice than a woman's (Gradol & Swann, 1983). In addition, a speaker's pitch and amplitude will be influenced by the pitch and amplitude of the conversational partner (Gregory, 1986, 1990; Lieberman, 1967; Natale, 1975). In a similar fashion, a speaker's internal state can induce changes in voice quality, but the relationship is hardly one-to-one. For example, stress profoundly affects voice fundamental frequency, but in any specific instance the effect can vary considerably depending on the conversational partner (Streeter et al., 1983). So, while encoding and decoding may characterize the role of nonverbal behavior is Models of Interpersonal Communication page 36 some communication situations, the applicability of the model is far from universal. 3. INTENTIONALIST MODELS