Language and Persuasion

Lawrence A Hosman. The Persuasion Handbook: Developments in Theory and Practice. Editor: James Price Dillard & Michael Pfau. Sage Publications, 2002.

Persuasive messages contain various elements, but one of the most critical is language. This chapter reviews research that has examined the Persuasive impact of various components of language. Because other chapters also focus on issues relating to language and Persuasion, such as metaphor and message framing, they are not reviewed in this chapter.

Nature of Language

Although many characterizations of language exist (e.g., Bradac, 1999), most agree that language has two general components: a structural component and a use component (Crystal, 1995). The structural component focuses on the hierarchical organization of language and consists of several parts, four of which are important for this chapter. These are phonology, syntax, lexicon, and texts or narratives.

Phonology deals with a language’s sound system and how sounds are combined into meaningful units. Syntax addresses the rules underlying the construction of sentences. The lexicon originally referred to the vocabulary of a language. More recently, study of the lexicon has diversified (Crystal, 1995) and includes a language’s words and meanings, idioms, abbreviations, euphemisms, and other meaningful units. Texts or narratives are “self-contained units of discourse” (p. 2), usually with some form of internal organization. Often, text is a frame of reference for the interpretation of a language’s phonological, grammatical, and lexical elements.

The use component focuses on how speakers use language in communicative contexts. It includes several areas of study including regional variations in language use, ethnic and social variation in language use, and pragmatic variations in language use.

These general components and individual parts are interrelated, and the boundaries between them are fuzzy. For example, speakers can use lexical elements of a language in different ways and for different purposes. Not only are narratives an organized collection of sentences, but speakers can use them for particular purposes. Nonetheless, these various language components will serve as the organizational basis for the research reviewed later in this chapter.

Assumptions and Questions Underlying Research on Language and Persuasion

The central question that scholars of language and Persuasion address is deceptively simple: What effects do variations in the phonological, syntactical, lexical, textual, and use elements of a message have on Persuasion? Two aspects of the question are critical. First, what language variations are important? As Bradac, Bowers, and Courtright (1979) pointed out, variations in nearly all of the levels of language can be important. Later sections of this chapter review language variations that scholars have studied.

The second critical element is what aspects of the Persuasion process these language variations affect. Most research assumes that language variations affect one of three elements of the Persuasion process: judgments of the speaker, message comprehension or recall, or attitude toward the message. Numerous studies have focused on judgments of the speaker. The assumption is that language variation affects the impression formation process, and in a Persuasion context an important impression affected is that of the speaker. Language variations may affect listeners’ judgments of a speaker’s source credibility, attractiveness, likability, and/or similarity. Other research has examined the impact of language variation on listeners’ comprehension, recall, and/or understanding of a message. Finally, some research has investigated the effect of language variations on attitude toward the message. Research focusing on judgments of the speaker and message comprehension or recall implicitly assumes that effects in these two areas will ultimately affect attitude toward the message and Persuasion. That is, research assumes that if a particular language variation has a positive impact on speaker credibility, it will also have a positive impact on attitude toward the message. These assumed links among various elements are intuitively plausible but do not always exist. As discussed later in this chapter, researchers need to investigate these assumptions more explicitly.

Language and Persuasion

Subsequent sections review research investigating the effects of language variations on the Persuasion process. The research reviewed is limited to (a) research with a quantitative or empirical methodology and (b) research that focused on outcomes relevant to the Persuasion process such as judgments of the speaker, message recall, and attitude toward the message.

Phonological Level

The phonological level deals with the sound system of a language. Although the study of phonology includes the formation of sounds or the combination of sounds, the research that is relevant to Persuasion focuses on the Perceptual outcomes of different sound combinations. Certain sound combinations may have different outcomes for the Persuasion process than do others.

Little research has looked at the Persuasive impact of this level of language, but two studies suggest its potential importance. Barry and Harper (1995) found that men’s and women’s first names could be distinguished by their phonetic attributes. Most important for the Persuasion process is their claim that “phonetic attributes might contribute to the Perception of a name as attractive or powerful” (p. 817). At least at the level of impression formation, a speaker’s name might have Persuasive implications.

Smith (1998) conducted a study in which he looked at the Persuasive impact of political names. Using a category scheme that assigned weights to various phonetic features in a politician’s name, such as the number of syllables or pattern of emphasis, he found that the model could predict 83% of the winners of presidential elections; 73% of the 1995 local elections in Spokane County, Washington; nearly 65% of the U.S. Senate elections in 1996; and 59% of the most competitive House elections in 1996. Although other factors undoubtedly influenced these election results, they clearly show that the phonetic properties of a politician’s name may influence the electorate.

In short, the sound system of a language may have important consequences for Persuasion. Because the research is not very extensive in this area, the nature and extent of these implications are not well-known.

Syntactic Level

The syntactic level of language deals with the rules governing the construction of sentences from the component parts of a language. More recently, scholars have studied the rules that govern the construction of larger chunks of discourse such as narratives or stories. The study of syntax shows that sentences can vary in their complexity. Some have a relatively simple structure such as “The cat chased the mouse.” Other sentences are more complex, usually because a grammatical transformation has been applied. For example, if the passive sentence transformation were applied to the preceding sentence, it would become “The mouse was chased by the cat.” Other sentences are more complex such as “The cat, focusing on its prey, chased the mouse.” Sentences with more complex grammatical structures would be expected to be more difficult to understand or comprehend. This comprehension difficulty could affect the Persuasion process negatively, presumably because comprehension of a message is an antecedent to Persuasion or attitude change. This assumption is consistent with information processing models of Persuasion (McGuire, 1969).

Jacoby, Nelson, and Hoyer (1982) looked at the effects of various syntactic constructions on the comprehension of corrective advertising. They found that positively worded statements (e.g., “Research has proven X”) were more easily comprehended than negative ones (e.g., “Research has not proven X”). This result is consistent with other psycholinguistic research that has found that negative grammatical transformations are more complex than positive grammatical transformations because they either require a longer time to process (see Fodor, Bever, & Garrett, 1974) or tax the cognitive system more during processing (see Ratner & Gleason, 1993).

Motes, Hilton, and Fielden (1992) examined the effects of active and passive sentence structure on the Perceived believability, clarity, appealingness, and attractiveness of print advertisements. They found that readers more favorably evaluated advertisements with active rather than passive sentence structure.

More recently, Lowrey (1998) conducted three studies examining the effect of syntactic complexity on advertising comprehension and attitudes toward the product. In one study, she found that simple syntax produced better recall than did complex syntax, but she also found that syntactic complexity was unrelated to attitude toward the product. A second study found that argument strength moderated the effects of syntactic complexity. With a complex syntactic structure, attitudes did not differ as a function of argument strength; however, with a simple syntactic structure, strong arguments were more Persuasive than weak arguments. Cognitive response data also reflected this pattern. A third study showed that receiver involvement affects the motivation to process complex syntax. Only highly involved participants were willing to assess strong and weak claims when the syntax was complex.

Although the research literature is sparse, it suggests that the complexity with which Persuasive materials are written affects their outcomes. Research has also begun to look at units larger than a sentence. Thorson and Snyder (1984) looked at the structure of television commercial scripts and their impact on the recall of these commercials. They used an “advertising language model” based on Kintsch and van Dijk’s (1978) macropropositional model of discourse. This model provides several structural measures of advertising content. They found that several of these measures predicted commercial recall.

Adaval and Wyer (1998) studied the effect of narratives on the Perceived attractiveness of vacation promotion literature. Two travel brochures described a vacation. One brochure described the vacation in a narrative form, while the other brochure described it in a list form. The authors also looked at the effect of undesirable information being contained in the two conditions. The results showed that participants evaluated vacations presented in a narrative form more positively than when the vacations were presented in a list form. This effect was enhanced when the brochure included undesirable information about the vacation site. That is, participants attended to negative information more when presented in a list form than when presented in a narrative form. The effects of a narrative information presentation were also enhanced when pictures accompanied the text.

The nature of a sentence’s grammatical construction or of a narrative’s construction has important Persuasive consequences. Grammatically complex materials are more difficult to recall than grammatically simple materials. This research has yet to address whether these differences have consequences for other aspects of the Persuasion process such as speaker judgments and attitude change.

Lexical Level

Persuaders’ choices about the words to use and the meaning of words in a Persuasive message are critical. This section reviews research that has looked at the effect of lexical variation and semantic variation on the Persuasion process.

Lexical Diversity. Lexical diversity refers to the vocabulary richness or vocabulary range that speakers exhibit and is assessed via a type-token ratio (TTR)—the number of different words in a message (types) divided by the total number of words (tokens). A low TTR means that a speaker’s vocabulary is relatively redundant, while a high TTR means that it is relatively diverse. Lexical diversity affects listeners’ judgments of speakers through a principle of “preference for complexity” (Bradac, Desmond, & Murdock, 1977). Simply stated, listeners prefer complexity because it is interesting, and lexical diversity should be preferred because it represents more complex lexical choice.

In a series of studies, Bradac and his associates (e.g., Bradac, Courtright, Schmidt, & Davies, 1976; Bradac, Davies, Courtright, Desmond, & Murdock, 1977; Bradac, Desmond, & Murdock, 1977) supported this principle, finding that lexical diversity is directly related to judgments of a speaker’s competence and socioeconomic status and to Perceptions of message effectiveness. Another study (Burroughs, 1991) found that these types of evaluations occurred when adults evaluated child speakers.

Subsequent studies (e.g., Bradac et al., 1976; Bradac & Wisegarver, 1984) found that ascribed speaker status interacted with diversity to affect a number of speaker judgments. A high-status speaker exhibiting high lexical diversity was Perceived positively, while a high-status speaker exhibiting low lexical diversity was Perceived negatively. In addition, some studies (Carpenter, 1990; Dulaney, 1982) have found that those who lie or are duplicitous exhibit higher lexical diversity than do those who do not lie. The explanation for this latter finding is that the process of lying requires speakers to plan their utterances more carefully, thus increasing the use of new words.

In sum, these studies show that the richness of a speaker’s vocabulary is related to listeners’ judgments about a speaker’s credibility or status. No research has explored the relationship between lexical diversity and attitude change. The preference for complexity principle would suggest that high lexical diversity would have a positive effect on the Persuasion process.

Language Imagery and Vividness. Another aspect of the lexicon studied by language and Persuasion scholars is verbal imagery or the ability of words to elicit images in listeners. Some researchers call this the vividness effect. Some words or expressions seem to elicit more imagery than others. Typically, concrete words, use of detail, and/or emotional language should elicit more images or be more vivid than should abstract or unemotional language. Similarly, one would expect verbal imagery or vividness to have more of a positive impact on Persuasion than would pallid language. Vivid language should be more memorable and accessible and should more favorably influence attitude change than should pallid language (Nisbett & Ross, 1980). These predictions are consistent with theories that focus on attitude accessibility (Fazio, 1989), theories such as information processing (McGuire, 1969) that include attention to the message, and theories such as the Heuristic-Systematic Model (Eagly & Chaiken, 1993) that incorporate the availability of heuristics as part of the Persuasion process.

Despite these expectations, research on the Persuasive impact of language imagery is contradictory. Some early studies found that verbal imagery had a positive impact on Persuasion. For example, Rossiter and Percy (1978) found that concrete words produced nearly twice as many favorable attitudes toward a product than did abstract words. An important literature review (Taylor & Thompson, 1982) concluded, however, that no conclusive evidence existed demonstrating that vividly presented information was more Persuasive than nonvividly presented information.

Since Taylor and Thompson’s (1982) review, the literature has been mixed with respect to the vividness effect. Some studies have continued to find a vividness effect (e.g., Burns, Biswas, & Babin, 1993; Rooks, 1986), while other studies have not (e.g., Collins, Taylor, Wood, & Thompson, 1988; Rooks, 1987).

Much recent research has attempted to account for these contradictory results. Some have argued that vividness effects will occur only under conditions of differential listener attention (Taylor & Thompson, 1982). That is, when listeners’ attention is constrained, listeners attend to vivid information more than to pallid information, and this vivid information is more Persuasive. When listeners’ attention is not constrained, listeners attend to vivid and pallid information equally. Frey and Eagly (1993), however, did not support this account.

Others have argued for a resource-matching Perspective (Keller & Block, 1997). This Perspective contends that vivid or pallid information’s impact depends on a match between the cognitive resources demanded by the information and the cognitive resources allocated by a listener. Other scholars have argued that the effect of vivid information depends on other receiver characteristics. Block and Keller (1997) found, for example, that vivid information in health communications was more Persuasive when the receivers were high in self-efficacy.

Two problems surround this area of research. First, many different conceptualizations and operationalizations of vividness exist. Some scholars (e.g., Taylor & Thompson, 1982) consider vivid information to include concrete and specific language, pictures and videotapes, direct experience, and case histories. Other studies use only concrete language items (e.g., Keller & Block, 1997). Some investigators consider vivid information to be that communicated by face-to-face interaction as opposed to print (Herr, Kardes, & Kim, 1991). Other investigators include grammatical structures such as active voice and present tense (Burns et al., 1993) in their operationalizations of vividness. These various operationalizations make it difficult to compare results across studies.

Second, the concept of vividness overlaps with other ideas discussed in this chapter and others. For example, emotional language and concrete words could be related to the work on language intensity and equivocation. Abstract language may be equivocal and less effective Persuasively than are specific or concrete words. Case histories are conceptually linked to narratives. Much of this research on vividness has continued independent of work in other related areas.

Language Intensity. Hamilton and Hunter (1998) noted that two major approaches exist to the definition of language intensity. The first views language intensity as a stylistic feature of messages. Intense language could include emotion-laden words, such as horrible and excellent, or specific graphic language, such as astronomical and completely. The second approach views intensity as reflecting the extremity of a source’s position on an issue (e.g., Bowers, 1963). A speaker describing a government policy as horrible is using more intense language than a speaker who describes the policy as disconcerting, and this shows greater deviation from attitudinal neutrality on this issue.

Although Hamilton and Hunter (1998) noted that some conceptual overlap exists between the two approaches, they argued that the approaches should be considered discrete. That is, someone can use intense language and not express an attitudinally extreme position. Conversely, a speaker can express an attitudinally extreme position without using particularly emotional or specific language. This conceptual distinction is important because it has consequences for how language intensity affects the Persuasion process.

Hamilton and Hunter (1998) summarized the language intensity research using meta-analytic techniques. Based on information processing theory (Hamilton, 1997; McGuire, 1969), they tested a causal model of language intensity’s Persuasive effects. The results supported two causal paths between intensity and attitude toward the source. The first path showed that language intensity (via Perceived intensity) increased Perceived speaker dynamism. In turn, speaker dynamism increased Perceived message clarity. Speaker dynamism apparently increased receivers’ interest in the message, causing receivers to focus more on the message and increasing its clarity.

Message clarity was positively related to Perceived source competence, which in turn was positively related to Perceived source trustworthiness. Finally, Perceived source trustworthiness was positively related to attitude toward the source.

A second causal path existed between language intensity and Perceived source competence. Specifically, language intensity was positively related to the Perceived extremity of the source’s position, which in turn was negatively related to Perceived source competence. As in the other causal model, Perceived source competence was linked to attitude toward the source through Perceived source trustworthiness. Importantly, when language intensity suggested an extreme source position, it had negative effects on Perceived source competence and, ultimately, on attitude toward the source.

As this second causal path suggests, the positive correlation between language intensity and attitude change may depend on the message’s position—whether the Persuasive message is attitudinally congruent with or discrepant from receivers’ attitudes. The meta-analysis supported this. When a message was attitudinally congruent, language intensity had little Persuasive impact. However, when a message was attitudinally discrepant, language intensity’s effect was dependent on a receiver’s ego involvement. With a receiver high in ego involvement, language intensity had a negative relationship with attitude toward the source. When the receiver was low in ego involvement, language intensity positively affected attitude toward the source. A field study of language intensity’s effects in skin cancer prevention messages supported this meta-analysis (Buller, Borland, & Burgoon, 1998). This study found that high-intensity messages produced less attitude change in listeners not intending to increase their skin protection than in those intending to increase their skin protection, particularly when the message drew explicit conclusions for the listeners.

The relationship among language intensity, message discrepancy, and ego involvement was also dependent on the source’s credibility. Language intensity had a negative impact on attitude change when a high-credibility speaker delivered a discrepant message to a receiver with high ego involvement. Language intensity positively affected attitude change when a high-credibility speaker delivered a discrepant message to a receiver with low ego involvement.

Another meta-analysis by Hamilton (1998) further explored the relationship between language intensity and source credibility. He found that argument quality enhanced the positive effects of language intensity on source competence, while opinionated language (i.e., receivers’ positions on the issue are evaluated) enhanced the negative effect of language intensity on Perceived source competence.

Equivocal Language. One choice communicators have to make is how clear or how vague to be in a Persuasion context. Should politicians, for example, state their position on abortion clearly and unequivocally, or should they be vague and equivocal? Although some view equivocation negatively, Eisenberg (1984) discussed the valuable role that strategic ambiguity plays in organizations. Ambiguity, for example, helps to build consensus on abstract goals, such as a mission statement, while simultaneously allowing for individual interpretations of these goals.

Williams and his colleagues conducted some of the earliest empirical research in this area. In a series of three studies (Goss & Williams, 1973; Williams, 1980; Williams & Goss, 1975), he examined equivocation’s impact on Perceptions of source credibility, message recall, and agreement with the message. He defined equivocation in this research as vagueness. For example, a speaker would equivocate if the speaker said that he or she favored a change in abortion policies rather than specifically stating that he or she favored a ban on abortions. Williams found that equivocal, attitudinally incongruent messages led to higher ratings of speaker character, greater message acceptance, and greater recall of argument content than did unequivocal, attitudinally incongruent messages. These results suggested that receivers could easily reject clear, attitudinally incongruent messages but that receivers could not as easily reject vague, attitudinally incongruent messages. In this latter case, Williams argued that the vagueness allowed receivers to Perceive the messages as congruent with their attitudes.

Bavelas, Black, Chovil, and Mullett (1990) developed a more extensive theory of equivocation. They argued that equivocal messages avoid one of four elements in a communicative situation: sender, content, receiver, or context. An equivocal message may avoid showing whether the message is a speaker’s own opinion. A speaker using the expression “noted authorities say” is not clearly saying whether it is his or her opinion, and the message is therefore equivocal. A message may be equivocal because it does not have clear content. This is consistent with the operational definition used by Williams (1980). An equivocal message does not address a particular receiver in the setting. It may address someone as a category, as in “Conservatives really bother me,” or not address any particular Person. Finally, equivocation occurs if someone wants to avoid the immediate context. This occurs, for example, if someone does not offer a direct answer to a question. If a teenage girl asks a parent whether she can go 60 miles to a concert with a boyfriend, the parent might respond equivocally by saying “Is it hot in here?”

Fundamental to this theory of equivocation is the contention that equivocation is the result of avoidance-avoidance circumstances. Speakers equivocate because they must make some response, but each response has a negative consequence. For example, equivocation is likely to occur when a speaker must choose between telling a hurtful truth and telling a harmless lie (Bavelas et al., 1990). This may help to explain why politicians equivocate to such a large extent. Bavelas and her colleagues (1990) presented data supporting this contention.

Most of Bavelas’s research has focused on the production of equivocal messages rather than on their Persuasive consequences. Hamilton, however, has pursued the Persuasive consequences of equivocal messages. In one study (Hamilton & Mineo, 1998), he found that equivocation, defined in the study as a lack of linguistic specificity, decreased Perceived message clarity. This, in turn, made it more difficult for receivers to identify a speaker’s position. This study also found that longer messages led to decreased linguistic specificity or greater equivocation. A second study (Hamilton, 1998) found that unequivocal or specific language enhanced source credibility, but only with high-quality arguments. Specific language apparently enhanced the Perceived quality of the arguments and consequently enhanced the Perceived credibility of the source. With low-quality arguments, unequivocal language accentuated the poor quality and negatively affected Perceived source credibility.

Thus, the diversity of words used by Persuaders, the images words create in listeners’ minds, the intensity of their language choices, and the vagueness of their language choices affect judgments of speaker credibility and attitude change. As noted in this section, the vividness research suffers from inconsistent operationalizations of the concept. Some operationalizations of vividness are similar to those used in the equivocal language research. Conceptual overlaps between the work in language intensity and equivocal language also exist. Equivocal language also hides the degree to which a speaker’s attitude deviates from neutrality. These are issues for future researchers to pursue.

Language Use

Language use refers to the use of language in social contexts. Variables that fit in this area include pragmatics, power of speech style, and standard and nonstandard language varieties.

Pragmatic Implication

When a receiver listens and tries to understand language or messages, part of the process involves making inferences about the speaker’s meaning. Harris and Monaco (1978) elaborated on the way in which this occurs and distinguished between logical implications and pragmatic implications. A logical implication occurs when a sentence necessarily implies some information. For example, the sentence “Bill hit Tom” logically implies that Tom was hit. Pragmatic implication is information that is neither directly stated nor logically implied. The statement “The hungry lion caught the gazelle” pragmatically implies that the lion killed the gazelle but does not logically imply it.

Several studies (e.g., Harris, 1977; Harris, Teske, & Ginns, 1975; Harris, Trusty, Bechtold, & Wasinger, 1989; Searleman & Carter, 1988) in both advertising and legal contexts have revealed that people are unable to distinguish between information that is directly asserted and information that is pragmatically implied. People were also more likely to remember as true pragmatically implied material than directly asserted information. These findings suggest that receivers add to the Persuasive message via pragmatic inferences. These inferences may be warranted or unwarranted, but they would be related to processes critical to receivers’ comprehension, understanding, and memory of Persuasive messages.

Power of Speech Style

Erickson, Lind, Johnson, and O’Barr (1978) first distinguished between powerful and powerless speech styles. Based on their examination of several trial transcripts, they observed that those high in status (judges, lawyers, and expert witnesses) spoke differently from those low in status (lay witnesses and defendants). Those low in status exhibited a relatively high frequency of language features such as hedges (e.g., “sort of,” “kind of”), hesitations (e.g., “um,” “er”), intensifiers (e.g., “certainly,” “surely”), and polite forms (e.g., “please,” “sir”), and Erickson et al. labeled this a powerless style. Those high in status spoke with relatively few of these language features, and this was called a powerful style. The authors subsequently conducted an experimental study comparing participants’ evaluations of these styles and found that participants Perceived a speaker exhibiting a powerful style as more credible, sociable, attractive, and certain than a speaker exhibiting a powerless style.

Since then, two lines of research have evolved. One approach continues to use the molar concepts of powerful and powerless speech styles. These studies (e.g., Bradac, Hemphill, & Tardy, 1981; Bradac & Mulac, 1984b; Gibbons, Busch, & Bradac, 1991; Grob, Meyers, & Schuh, 1997; Hahn & Clayton, 1996; Sparks, Areni, & Cox, 1998) have generally found that participants Perceive speakers exhibiting a powerful style as more credible, attractive, sociable, and dynamic than speakers exhibiting a powerless style.

The other line of research has focused on the individual components contained in the styles and their implications for the impression formation process. This chapter earlier discussed one of these components—language intensity. Other studies (e.g., Bradac & Mulac, 1984a; Haleta, 1996; Hosman, 1989; Hosman & Siltanen, 1994) examining individual components have found that participants Perceive speakers exhibiting hedges and hesitations as less credible, attractive, and dynamic than speakers not using them. Polite forms, however, constitute something of an anomaly. Bradac and Mulac (1984a) found that listeners Perceive polite forms to be as powerful as a powerful style. Other researchers (e.g., Lakoff, 1975) have contended that polite forms are a powerless form of speech.

Two links exist between power of speech style and the Persuasion process. The first is an indirect link among power of speech style, impression formation, and attitude change. Most of the research shows that a powerful speech style will enhance a speaker’s Perceived credibility, attractiveness, dynamism, and sociability, and to the extent these impressions will positively affect attitude change, a powerful style should be more Persuasive.

The more direct link focuses on the direct impact of powerful and powerless speech styles on attitude change. This is an area of substantial controversy. One study (Gibbons et al., 1991) found that a powerful speech style did not produce more attitude change than did a powerless style. Two studies (Erickson et al., 1978; Hahn & Clayton, 1996) found that a powerful speech style resulted in a more favorable verdict than did a powerless style. A meta-analysis of studies prior to 1991 found that powerful speech styles produced positive effects on attitude change (Burrell & Koper, 1998). One recent study (Sparks et al., 1998) suggested that differences in the ability to find direct effects of power of speech style may be due to the modality of message presentation. The authors found that a powerful speech style was more Persuasive than a powerless style when the message was presented via audiotape, but no significant differences between styles emerged when the messages were written.

Standard and Nonstandard Language Varieties

For years, scholars have studied the effects of regional and ethnic language variation on the impression formation process (for reviews, see Bradac, 1990; Giles & Coupland, 1991). For example, Hopper and Williams (1973) investigated the impact of Standard American, Black, Mexican American, and Southern White speech characteristics on employment decisions. They found that participants evaluated Standard American speech more positively than they did the other three speech types.

Over the years, this research has shifted its characterization of these speech styles from one of regional and ethnic variation to one of standard and nonstandard language varieties. A standard variety is one linked with high socioeconomic status and power, while a non-standard variety is one linked with low socioeconomic status and power. Nonstandard language varieties are usually associated with regional or ethnic minority dialects or accents.

In general, the research shows that listeners positively evaluate standard language varieties across several evaluative dimensions such as intelligence, competence, and social attractiveness. For example, de la Zerda and Hopper (1979) studied interviewers’ reactions to Mexican American speech, finding that the degree of accentedness predicted evaluations of an interviewee’s ambition, intelligence, and cooperativeness and predicted interviewers’ likelihood of hiring the Person. Furthermore, the greater the status of the position being interviewed for, the greater the importance language attitudes played in these evaluations. Atkins (1993) more recently confirmed this finding comparing Black English and Appalachian English speakers in an employment context. This pattern can be affected by the extent to which listeners identify with a language variety. If listeners Perceive the language variety to be similar to their own, then they may be less likely to downgrade a nonstandard variety (Giles & Coupland, 1991).

Although this research suggests links between standard and nonstandard language varieties and Persuasion, little research has directly addressed the issue. If listeners’ Perceptions of a speaker’s competence, status, or attractiveness are related to the Persuasion process, then one would expect standard language varieties to be more Persuasive than nonstandard varieties. Two studies provide some evidence that standard and nonstandard language varieties may have effects that extend beyond speaker judgments. Koslow, Shamdasani, and Touchstone (1994) examined Hispanic consumers’ reactions to print advertisements containing varying degrees of Spanish- and English-language use. They found that Spanish-language use enhanced Hispanic consumers’ Perceptions of an advertiser’s sensitivity toward their culture and their attitude toward the advertisement. However, sole reliance on Spanish-language use decreased their attitude toward the advertisement, apparently reflecting Hispanic insecurities about language use. Rubin, Healy, Gardiner, Zath, and Moore (1997) examined reactions to a Standard American or South Asian accent in an AIDS prevention clinic. Participants judged a physician using a Standard American accent to be more interpersonally attractive and as possessing more general ability than a physician using a South Asian accent. Rubin et al. observed no significant differences between accents for recall of the AIDS message or intention to comply.

The use of language in Persuasive contexts has important implications for the Persuasion process. Most of this research, however, has focused on how stylistic or dialectical variations affect the impression formation process. Much more research needs to focus on their role in the attitude change process.

Directions and Challenges for Future Research

As the preceding literature review shows, research on language and Persuasion has been conducted in many areas, with researchers working relatively independent of one another. Consequently, they have often not integrated findings in one area with relevant findings in other areas. Thus, theories or models that put research in a common framework or point to potentially fruitful areas of study need to guide the research. Two directions might help to accomplish this.

First, future research on the relationships between language variation and the impression formation process needs to be integrated within a more general model of the process. Most research has proceeded by merely investigating the impact of language variations on listener impressions related to the Persuasion process such as Perceived competence, attractiveness, and trustworthiness. A viable framework is a process model of language attitudes (Cargile, Giles, Ryan, & Bradac, 1994). Process models of language attitudes would encourage investigators to think more extensively about the link between language variation and impression formation. The Cargile et al. (1994) model emphasizes five features of the process by which listeners form attitudes about language variation: listener dynamics, interpersonal history, outcomes, the social situation, and Perceived cultural factors.

Listener dynamics include several listener characteristics that may affect the process such as listeners’ social group membership. Listeners’ goals are relevant. The goals that listeners have for attending to a speaker or a Persuasive message may influence the language features to which they attend. Listeners’ moods may also be significant. A listener in a negative mood may be more receptive to messages that are syntactically easy to process than to messages that are syntactically difficult to process. A listener’s expertise on an issue is potentially relevant as well. As Cargile et al. (1994) pointed out, if a listener has expertise on an issue, then he or she can invoke scripts or schemata to process information. This ability to invoke scripts or schemata frees cognitive resources, which allows the listener to process schema-inconsistent information more efficiently.

Researchers have neglected the area of listener dynamics. For example, they have not examined listener goals extensively. Fiske’s work (Fiske, Morling, & Stevens, 1996) on power, social control, and anxiety suggests that listeners high in trait anxiety are motivated to regain control via impression formation processes when powerful people threaten their needs. Under certain conditions, they disregard negative information about a speaker and form an overly positive evaluation. They may also react differently to language variations in a message. For example, Hosman (1997) found that listeners with an external locus of control reacted more positively to a powerful speech style than did those with an internal locus of control. Similarly, little work has explored the impact of a listener’s emotional state on his or her processing of various language forms. For instance, if a listener is in a negative mood, will this influence his or her processing of syntactically complex Persuasive messages?

Listener dynamics also point toward a more sophisticated conceptualization of listener attitudes. Not only may listeners’ attitudes toward language variations have a cognitive component, they also may have an emotional component and a behavioral component. Most current work on language and Persuasion has focused on the cognitive component. That is, language variation causes listeners to believe that the speaker is trustworthy, competent, and/or suitable for employment. Little work has focused on the emotional and behavioral components.

A language attitude model also incorporates interpersonal history or how familiar a speaker and listener are with each other. Cargile et al. (1994) pointed out that when speakers and listeners are unfamiliar with each other, uncertainty reduction processes are more likely to occur, resulting in greater impact of language variations. Similarly, expectancy violation processes are more likely to play a role when speakers and listeners are familiar with one another. For example, if a listener knows a speaker’s status, then expectancy violations are more likely to affect the language attitude process.

The immediate social situation is also a factor important in the language attitude process. Many studies (e.g., Brown, Giles, & Thackerar, 1985; Johnson & Buttny, 1982; Street, Brady, & Lee, 1984) have found that listeners’ attitudes toward a particular language variation vary from one context to another. For example, listeners may positively evaluate a fast rate of speech in informal contexts, but they may negatively evaluate it if technical or complex material is presented. The nature of the context may make some variations more salient than others. A courtroom context might make language variables that show uncertainty more salient than other language variables because judgments of witness uncertainty play a more pivotal role in the courtroom.

Finally, the cultural context also plays an important role in language attitudes. A particular aspect of the cultural context that is important is social norms. These social norms establish what the preferred or expected language behaviors are within a community. Holtgraves and Dulin (1994) found, for example, that African American and European American listeners evaluated bragging differently because the two cultures’ norms about bragging differ.

This model begins to suggest additional interesting areas of research for language and Persuasion. A potentially interesting research area is at the intersection of listener dynamics and cultural context. For example, do listeners from diverse cultures differ in their knowledge or expectations about what language variations are important in particular Persuasion contexts? Friestad and Wright (1994) developed a Persuasion knowledge model that emphasizes “the culturally supplied folk wisdom on Persuasion” (p. 1) that people bring to Persuasion contexts. We know very little about this knowledge and how it may vary between individuals from different cultures.

A second direction for researchers in language and Persuasion is to conduct more theoretically grounded research (see Burgoon & Dillard, 1995). Much of the research on language and Persuasion has been atheoretical in two senses. First, research has examined language variables without a well-developed explanation of why these variables should have particular effects. For example, although a powerful speech style clearly has positive effects across a variety of impression formation dimensions, it is unclear why it has such effects. At least two candidate explanations have been offered. One is that a powerful speech style indicates speaker self-control (e.g., behaving appropriately in a situation, exhibiting self-confidence). Research shows that receivers positively evaluate those who exhibit self-control (Stern & Manifold, 1977). The second explanation is that a powerful style suggests control over others. This control over others explanation has both a positive and a negative element. Control over others may be positive if it indicates effective behavior, but it may be negative if it suggests threatening or domineering behavior. Bradac and Street (1989–1990) argued that it is the positive Perception that leads to the positive evaluation of a powerful style. Although a few studies have examined these explanations and found support for them (e.g., Hosman, 1997; Hosman & Siltanen, 1994), additional investigation is needed.

Second, and more important, most of the research has not integrated language variables into a coherent Persuasion theory such as the Elaborated Likelihood Model of Persuasion or the Heuristic-Systematic Model of Persuasion. These process models highlight the different ways in which listeners process Persuasive elements. Recent explications of dual-process models (e.g., Petty & Wegener, 1998) tend to ignore the role that language variables play in the process. However, such variables might be easily incorporated and studied within such models. In particular, investigators could examine these variables to see whether they operate as central, Peripheral, or biasing cues. Most likely, they operate as Peripheral or biasing cues, interacting with other elements of a message (e.g., argument quality) or other elements of the Persuasion process (e.g., speaker credibility). For example, Hamilton’s work on language intensity suggests that it interacts with argument quality. Other variables, such as syntactical complexity and narrative structure, could also affect the way messages high or low in argument quality are processed. Some variables may operate as biasing cues that affect the processing of some message elements more than others. For example, power of speech style may influence the processing of a low argument quality message more than that of a high argument quality message.

Examining language variables within these more comprehensive theoretical frameworks accomplishes two important goals. First, it moves language and Persuasion research away from a focus on how language affects message comprehension or speaker judgments and toward an increased focus on how language affects attitude change. As mentioned earlier, too many researchers assume that messages positively affecting speaker credibility, for example, will also positively affect attitude change when in fact they need to investigate these linkages.

Second, because dual-process models emphasize cognitive responses to messages, they would encourage exploration of how listeners respond to language variables contained in Persuasive messages. As Giles, Henwood, Coupland, Harriman, and Coupland (1992) pointed out, few studies have examined cognitive responses to language variables. For instance, we know little about how listeners cognitively respond to messages containing high-intensity language.

Studying the cognitive responses to these language variables will require more sophisticated content analytic schemes than are typically used in dual-processing research. Most of this research employs simple positive-negative coding schemes, sometimes dividing the categories into smaller units such as thoughts about the speaker. More sophisticated schemes, such as that used by Giles et al. (1992), allow for a more fine-grained understanding of cognitive responses to Persuasive messages and the role they play as mediators in the Persuasion process.

Other theories may also be valuable. For example, expectancy violation theory might enhance the study of language use variables (Burgoon, 1990). Put simply, this theory contends that listeners develop expectations about the language Persuaders should use. When speakers violate these expectations by using language that is unexpected, listeners will evaluate them negatively. If, for example, listeners expect high-status speakers to speak with high lexical diversity and they do not, then listeners may Perceive the speakers as having low source credibility. This theory has been successfully used to examine the impact of language intensity on attitude change and might be fruitfully used to explore other language variables.

Research examining the impact of language on Persuasion also needs to focus on the relationships among language, cognitive responses, attention, comprehension, recall, and attitude change. As noted previously, research has investigated the effect of language variables on a limited number of outcomes. It seems reasonable to expect that some of these cognitive elements (e.g., cognitive responses) will mediate or moderate the effects of language variables on outcomes (e.g., attitude change). Examining such relationships will require the use of statistical techniques (e.g., path analysis, structural equation modeling) that allow for the exploration of interconnections among variables. Using such techniques, Hosman, Huebner, and Siltanen (1999) found that power of speech style did not have a direct effect on attitude change but had an indirect effect on attitude change via cognitive responses. Hamilton (1998) also fruitfully used such statistical techniques.

A challenge in this area is to increase the generalizability of research results. This is particularly important if practitioners are to be able to use results meaningfully. Usually, increasing the generalizability of research means extending the results to different groups of participants. Increasing a study’s sample size and employing a sampling procedure that includes more diverse types of participants are methods to accomplish this. When researchers use language variables, however, generalization entails additional considerations. Here, generalization means to extend a study’s results beyond the particular language sample used in the study. In a typical language and Persuasion study, researchers construct a “template” message into which language variables of interest are inserted. Investigators might use a speech on a particular topic as a template message. They would then insert low- and high-language intensity forms into the template to create low- and high-intensity messages.

What investigators often forget is that these messages differ in more ways than simply the inclusion of high- and low-intensity language. The high- and low-intensity forms selected by the investigator might differ from the high-and low-intensity forms selected by another investigator. Furthermore, the template messages contain other language variables, and a template message written by one investigator might include different language variables from a template written by another investigator. These differences might interact in unknown ways with the language variables of interest. Thus, differences between low- and high-language intensity messages are limited to the template message into which researchers have inserted them.

Jackson and her colleagues (e.g., Brashers & Jackson, 1999; Jackson & Brashers, 1994a, 1994b; Jackson & Jacobs, 1983) have shown the implications of using single-message designs. Language effects can be extremely variable from one message to another. Brashers and Jackson (1999) discussed one study that looked at the impact of sexual content in advertising recall using 13 different topics. This study found that the impact of sexual content varied from “a standardized mean difference of Σ.49 to a standardized mean difference of .37” (pp. 469–470). Put simply, some messages had a positive impact on recall, while others had a negative impact. Not only do these findings have theoretical implications, but they also have practical implications. A practitioner developing messages designed to Persuade an audience must be aware of how highly variable language effects can be and must consider this when constructing messages.

Jackson argued that not only should multiple message replications be used in studies, but the results should be analyzed appropriately. This means treating replications as random effects rather than fixed effects in the statistical analyses. Her work offers suggestions about how this can be done (e.g., Jackson & Brashers, 1994a, 1994b).

This is not to suggest that this challenge is without controversy (e.g., Burgoon, Hall, & Pfau, 1991; Hunter & Hamilton, 1998). Scholars must address issues such as the number of replications that should be used, the statistical power of such designs, the ease of writing multiple messages, the length of such replications and its effect on participant fatigue, and the impact on sample size. Nevertheless, researchers must be concerned with the continued excessive reliance on single messages in studies and the conclusions drawn from them.

Conclusion

More than 20 years ago, Miller and Burgoon (1978) lamented on the decline of research on language and Persuasion. Since then, research has slowly increased as scholars in fields such as advertising, marketing, psychology, and communication investigate a variety of language variables. Some variables have been investigated extensively. We know, for example, the impact of lexical choices or standard and nonstandard language varieties on judgments of speaker credibility and attractiveness. We also have a better understanding of the factors that moderate and mediate the relationship between language intensity and attitude change.

At the same time, we still lack substantial knowledge about some important aspects of the relationship between language and Persuasion. How do the various levels of language structure affect Persuasion, and how do the various levels of language structure relate to each other in the Persuasion process? For example, the syntactic complexity of a message may affect its recall or comprehension, but we are less able to draw conclusions about its impact on attitude toward the message. More generally, we have substantial knowledge about how some language variables affect attitude toward the speaker, but we have little (if any) information about how it affects the attitude toward the message. Alternatively, how do lexical diversity and syntactical complexity affect each other in a Persuasive message?

The future of this area of research seems bright both theoretically and practically, but to achieve its potential, scholars must meet certain challenges. They should apply more systematic frameworks to organize their study of language variables. This chapter has suggested a process model of language attitudes as one possibility. A general model such as this not only helps to integrate research but also points to variables relatively unexamined such as the effect of listeners’ moods on the processing of language variables.

Even using such frameworks, investigators must integrate more research into comprehensive theories of Persuasion such as dual-process models. These theories will help investigators to focus on how language affects attitude change and how listeners respond cognitively to language variables.

Finally, researchers must increasingly worry about the generalizability of their results—generalizability that extends beyond the particular messages used in a study. Increasing the generalizability of results presents its own challenges, but being concerned with the issue is critical for practitioners to find the research valuable and useful.