MEMORY, 2003, 11 (1), 101109

The power of the spoken word: Sociolinguistic cues influence the misinformation effect
Lana A. Vornik, Stefanie J. Sharman, and Maryanne Garry Victoria University of Wellington, New Zealand
We investigated whether the sociolinguistic information delivered by spoken, accented postevent narratives would influence the misinformation effect. New Zealand subjects listened to misleading postevent information spoken in either a New Zealand (NZ) or North American (NA) accent. Consistent with earlier research, we found that NA accents were seen as more powerful and more socially attractive. We found that accents per se had no influence on the misinformation effect but sociolinguistic factors did: both power and social attractiveness affected subjects' susceptibility to misleading postevent suggestions. When subjects rated the speaker highly on power, social attractiveness did not matter; they were equally misled. However, when subjects rated the speaker low on power, social attractiveness did matter: subjects who rated the speaker high on social attractiveness were more misled than subjects who rated it lower. There were similar effects for confidence. These results have implications for our understanding of social influences on the misinformation effect.

INTRODUCTION
More than 25 years of research has created a substantial body of evidence showing that misleading postevent information (PEI) can negatively affect memory reports of events (Belli, 1989; Dodd & Bradshaw, 1980; Greene, Flynn, & Loftus, 1982; Loftus, 1975, 1991; Loftus, Miller, & Burns, 1978). The misinformation effect is the influence of erroneous PEI on memory-test responses (Belli, 1989; Loftus & Hoffman, 1989). In the standard misinformation experiment, subjects are first shown an event, in the form of a slide sequence or a video depicting something like a shoplifting episode or a car accident. After a delay, they are presented with PEI that is misleading for some aspects of the event, but not others. Subjects are typically more accurate about control items (aspects not targeted by misleading information) than about misled items.

Factors that influence the misinformation effect

Research shows that certain conditions affect the likelihood that subjects adopt inaccurate PEI, such as the delay between the original event and the PEI, the reliability of the person who delivers the PEI, and the time given to subjects to view the original event. For instance, the longer the delay between the original event and the PEI, the more likely it is that subjects incorporate the misinformation into their final reports, presumably because the passing of time weakens their memories for the event (Loftus et al., 1978). We also know that some social factors influence whether people capitulate to the misleading suggestions (see, for example, Assefi & Garry, in press). For instance, the source of the PEIwhom we might call the ``misinformation messenger''matters (Ceci, Ross, & Toglia, 1987; Dodd & Bradshaw, 1980; Lampinen & Smith, 1995). In their experi-

Requests for reprints should be sent to Maryanne Garry, School of Psychology, Victoria University of Wellington, Box 600, Wellington, New Zealand. Email: maryanne.garry@vuw.ac.nz We thank Scott Brown, Peter Smith, Seema Assefi, Sonia Cunningham, Ann Weatherall, Rachel Sutherland, Beth Loftus, the Cognitive Workshop, and especially Louise Frost for recording the postevent information. We also thank Don Read for suggesting the covariate analysis.

# 2003 Psychology Press Ltd http://www.tandf.co.uk/journals/pp/09658211.html

DOI:10.1080/09658210244000054

102

VORNIK, SHARMAN, GARRY

ment, Dodd and Bradshaw asked subjects to view slides of a car accident, and then answer some questions that contained misleading information. Half the subjects were told that the questions were asked by a lawyer representing the driver who caused the accidentan unreliable sourceand the remaining subjects were not given any information about the source. Dodd and Bradshaw found that misleading questions led the ``no source'' subjects to adopt the misleading information, but the ``defence lawyer'' subjects to reject it. Therefore, subjects do not simply accept information from any source: Certain conditions such as unreliability of the speaker can dampen the effect of misleading PEI on memory. However, most researchers investigating whether social factors influence the misinformation effect have explicitly telegraphed that information to subjects. For example, Dodd and Bradshaw (1980) manipulated the misinformation messenger's profession: Subjects were told the PEI was provided by a defence lawyer or given no information about source. Ceci et al. (1987) manipulated the messenger's age, and Lampinen and Smith (1995) manipulated both age and credibility to convey information about the source of the PEI. Would these social factors still influence the misinformation effect if they were not telegraphed as explicitly; that is, if the information about the messenger were conveyed in a more subtle manner? One way in which information about the messenger may be conveyed more subtly is by delivering spoken PEI in various accents. Sociolinguistic research has shown that accent conveys much information about the speaker. The question we ask here is whether the sociolinguistic information conveyed by different accents has an influence on the misinformation effect. In this experiment, we examined whether the information conveyed by accent can make people more or less susceptible to misinformation effects. New Zealand subjects watched a slide sequence, and then listened to misleading PEI in either a New Zealand (NZ) or North American (NA) accent. They later provided us with information about the social attractiveness and power conveyed by the accent. Finally, subjects were tested for their memory of the original event.

convey different types of information, including information about social attractiveness and power (Giles, 1971; Luhman, 1990). Social attractiveness. The research on language suggests that in a misinformation experiment, the impact of spoken misleading PEI is likely to depend on subjects' appraisal of social attractiveness. Social attractiveness is a wellknown construct in the sociolinguistic literature, and includes characteristics such as trustworthiness, dependability, and friendliness (Bayard, 1995; Podberesky, Deluty, & Feldstein, 1990). In general, English-speaking countries rate their own accent higher on social attractiveness, and locally accented speakers appear more trustworthy, friendly, and dependable (Giles, 1971; Luhman, 1990). However, New Zealanders buck this trend, and devalue NZ English on all characteristics when it is compared with other English language accents, such as NA. In what Bayard (1995) calls ``cultural cringe'', New Zealanders rate more valued forms of English such as NA and British Received Pronunciation (RP, otherwise known as the Queen's English) higher on social attractiveness than the local NZ accent (see also Hyugens & Vaughan, 1983). We also know that accents rated higher on social attractiveness are more persuasive. For example, Giles (1973) asked high-school students what they thought about capital punishment, then had them listen to arguments in either local English accents or RP. Then he asked the students for their views about capital punishment for a second time. In the end, the argument in the local accent was more likely to cause the students to change their thoughts on capital punishment. Therefore, to the extent that our subjects rate the NZ accent lower and the NA accent higher on social attractiveness, they should be more misled by PEI presented in a NA accent than PEI presented in a NZ accent. Power. Accents also carry information about the power of the speaker. In general, most research examining attitudes towards variants of English show that the greater the gap between standard speech (NA, RP) and the speaker, the less powerful listeners find that speaker (Anisfeld, Bogo, & Lambert, 1962; Callan, Gallois, & Forbes, 1983; Webster & Kramer, 1968). Both RP and NA are perceived as more powerful and prestigious than other English variants (Giles, 1971; Luhman, 1990; Stewart, Ryan, & Giles, 1985).

Information conveyed by accent

The social psychology literature has shown that accents are paralinguistic sources of information (Riches & Foddy, 1989; Seggie, 1983). Accents

SOCIOLINGUISTIC CUES AND MISINFORMATION

103

Therefore, to the extent that our subjects rate the NZ accent lower and the NA accent higher on power, they should be more misled by PEI presented in a NA accent than PEI presented in a NZ accent. Interestingly however, Giles (1973) has demonstrated that socially attractive speakers can be more persuasive than powerful speakers. Therefore there is reason to speculate that social attractiveness will have a different influence depending on the power of the speaker. Overall, if subjects are more misled by an accent they rate high on power and social attractiveness, and less misled by an accent they rate low on power and social attractiveness, it would be evidence that memory distortions are sensitive to more subtle types of information, such as sociolinguistic factors. However, if subjects were similarly misled regardless of power and social attractiveness ratings, it would be evidence that these sociolinguistic factors do not affect the misinformation effect and more explicit cues about the source of the misinformation are necessary.

Procedure
As with previous misinformation studies (Belli, 1989; McCloskey & Zaragoza, 1985), subjects were initially told that the purpose of the study was to see if mode of information presentation (visual or verbal) affects the way that information is learned. Subjects watched a series of 61 slides depicting a young man shoplifting in a university bookstore (Loftus, 1991). The slides were the same for all subjects except for the eight critical slides displayed in one of two versions. We manipulated the viewing time of the slides because the misinformation effect tends to be medium to large, and may overshadow the relatively weaker effect of sociolinguistic factors. One way to weaken the misinformation effect is by varying slide exposure time. The longer the slides remain on the screen, the better the chance for subjects to encode the original information successfully, and the less likely they are to be misled by PEI. Thus, half of the subjects viewed each slide for 3 seconds and the other half viewed each slide for 6 seconds. Counterbalancing ensured that each version of the eight critical slides was shown equally often in all accent conditions. The slide sequence was followed by a 20 minute filler activity during which subjects completed a word puzzle. After the filler task, the subjects listened to a 3 minute, 500 word narrative describing the event. The wording of the narratives was changed slightly to accommodate NZ colloquialisms (for instance, ``elevator'' was changed to ``lift''). Four versions of the narrative were prepared and each audiotaped in two accents, NZ and NA English. The narratives were recorded in the same female voice by a volunteer actor whose native language was NZ English.1
1 The matched guise technique was used in the experiment. In the matched guise technique a narrative is tape-recorded by the same speaker in different guises (for instance, in two different accents). It is not revealed to the subjects that the narratives have been recorded by the same speaker (Edwards, 1982; Lambert, Hadgson, Gardner, & Fillenbaum, 1960). Potentially confounding paralinguistic variations found to affect the perception of speech, such as speech rate, and intonation and pitch (Brown, 1980; Giles, Henwood, Coupland, & Harriman, 1992; Gumperz, 1982) are therefore held constant and the judgements are believed to represent the stereotyped attitudes towards the language varieties under investigation. To ensure face validity, our NA accented version was played to two Americans, who were simply asked to guess where the speaker was from, but we made no mention of North America. Both of them guessed a region of the US.

METHOD Subjects
Subjects were 259 Victoria University of Wellington students who completed the experiment as a voluntary part of their lab programme. Of these, 118 subjects were assigned, on the basis of their intact laboratory classes, to the NA condition, and 141 were similarly assigned to the NZ condition. Data from subjects whose native language was not NZ English were excluded (approximately 16% in each condition), leaving 99 NA subjects and 118 NZ subjects. Although they were advised that the narrative had been recorded by a NZ or a US postgraduate student, a further 33 subjects (15%) were excluded because they did not correctly identify (or left blank a question about) the accent of the speaker. Although 94% of NA subjects correctly reported the speaker's accent, 77% of NZ subjects did so. However, another 10% of NZ subjects mistakenly identified the accent as Australian, a finding that fits with research showing New Zealanders and Australians often have difficulty distinguishing the two accents and react to them similarly (Weatherall, Gallois, & Pittam, 1998). Phase two proceeded with 93 subjects in the NA condition and 91 subjects in the NZ condition.

104

VORNIK, SHARMAN, GARRY

In each version of the narrative four of the critical items were misleading (for instance, the yellow candle was called white). The other four items were used as a control, and were referred to by the generic category name, such as a candle. Counterbalancing ensured that each version of a critical slide acted equally often as a control item and a misinformation item. After listening to the narrative, subjects completed scales to measure their attitude towards the PEI speaker. These scales contained 22 sixpoint unipolar rating scales adapted from Podberesky et al. (1990) and Bayard (1995). Each rating scale contained a personality trait or characteristic followed by ratings from 1 (not at all like the speaker) to 6 (very much like the speaker). See the Appendix for the list of characteristics that subjects rated. Immediately afterwards, they took part in another 5 minutes of filler activity. In the final stage, subjects completed a 19-item, two alternative forced choice standard recognition test on the slide sequence. They were explicitly told to base their answers only on what they saw in the slides and not on anything they might have heard afterwards. Eight items were critical, and the others were fillers. For each item subjects rated their confidence in the accuracy of their answers on a scale from 1 (not confident at all) to 5 (very confident). The last two questions on the test asked subjects to indicate their native language, and to identify the accent of the speaker as North American, New Zealand, British, Australian, or other. If they chose ``other'', subjects were asked to write down the speaker's accent. All subjects were then debriefed.

RESULTS
Our main research question was whether the sociolinguistic information conveyed by different accents had an influence on the misinformation effect.

First, we evaluated the attitude scale that we used to assess subjects' views towards the variously accented PEI. The attitude scale showed a high Cronbach's alpha (a = .91), indicating strong internal consistency. Having demonstrated good internal consistency, we then performed a factor analysis to verify the a priori grouping of the attitude scale. Principal components factor analysis with Varimax rotation extracted three distinctive factors. The cut-off point for inclusion of a variable in the interpretation of a factor was set to .40. As the Appendix shows, the items presumed to be under each factor did indeed load highly on that factor. Two items did not load highly on any factor; one item loaded highly on two factors, and these items were excluded from any further analyses. Communality values were moderately high for all items, suggesting that a large proportion of the variance in the variables was accounted for by the factors, and that the variables were well defined by the factor solution (Tabachnick & Fidell, 1996). The first factor was the social attractiveness factor, and the second factor the power factor. There was a third factor, which was similar to a factor identified in the sociolinguistic literature as charisma (Bayard, 1995; Brown & Gilman, 1960; Giles & Coupland, 1991). Although we generated three factors from our analysis, in the analyses we report here charisma had no effect on the pattern of results. Therefore, we include only the effects of social attractiveness and power. We generated factor scores for each subject by taking the mean of the variables associated with each factor. For instance, the power factor consisted of eight individual items: intelligent, ambitious, selfconfident, a leader, powerful, strong, educated, and prestigious. These factor scores were used in subsequent analyses. Analysis of variance of the rating scales. Did subjects treat NA and NZ accented speakers differently? To answer this question, we classified the two attitude factors by accent, as shown in Figure 1. Each pair of bars in Figure 1 represents subjects' mean ratings on each factor for either the NA or NZ accent. Both power and social attractiveness ratings were higher for the NA accent than for the NZ accent. In fact, subjects did indeed respond to the NZ and NA accents differently, as shown by the main effect for accent in a 2 (accent) 6 2 (attitude factors) mixed ANOVA, F(1, 75) = 13.72, p < .01. For both accents, speakers were seen as more socially attractive than powerful,

Attitude scales
Before we examined the relationship among social attractiveness, power, and the misinformation effect, we needed to determine whether our subjects rated the NZ and NA accents in line with the sociolinguistic literature. Recall that NA speakers should be rated higher than NZ speakers on both power and social attractiveness.

SOCIOLINGUISTIC CUES AND MISINFORMATION

105

Figure 1. The mean rating for each factor for PEI spoken in a NA or NZ accent.

F(1, 75) = 32.05, p < .01. In short, subjects had different attitudes towards NA and NZ accents on both power and social attractiveness factors.

Influence of accent on the misinformation effect

Were subjects more misled by the NA accent than the NZ accent? The top panel of Figure 2 shows subjects' performance (number of correct responses out of a total of four) when they received misinformation and when they did not. Note that subjects were correct less often on misled items than on control items. However, Figure 2 also shows that accent did not matter. The nearly parallel lines show that accent did not influence the misinformation effect, F < 1. In both accent conditions, subjects were less accurate when they were misled than when they were not, F(1, 182) = 39.02, p < .01, a medium-sized effect, f = 0.33 (Cohen, 1988).

Figure 2. The effect of accented PEI on the misinformation effect (top panel) and mean misled and control confidence estimations (bottom panel).

Confidence estimations
Confidence overall. The bottom panel of Figure 2 shows subjects' confidence on misled and control items. They were more confident that their responses to misled items were correct than they were about their responses to control items, F(1, 182) = 45.61, p < .01, f = 0.50. Once again, however, the parallel lines show that accent did not matter: there was no effect of accent on confidence, F < 1. Confidence when wrong. Another way of looking at subjects' confidence is to examine their

ratings when their answers were incorrect; we can look specifically at the confidence of subjects who have ``bought into'' the misleading suggestion. When they were wrong, misled subjects were more confident about their answers than control subjects, F(1, 182) = 35.17, p < .01, f = 0.53. There was no effect of accent on these confidence estimations, F < 1. In performing these analyses, we have assumed that accents are relatively pure carriers of social attractiveness and power information. In other words, we have assumed that accents can proxy for these factors. However, accents may also carry other information about the speaker, information that we did not or could not measure. Therefore, even though we found no influence of accent per se on the misinformation effect, it may be more useful to examine the influence of social attractiveness and power specifically. Next, we examine the specific influence of these two sociolinguistic variables on the misinformation effect.

106

VORNIK, SHARMAN, GARRY

speaker, and this information was strong enough to influence the misinformation effect.

Attitudes and confidence

Confidence overall. To determine whether power and social attractiveness ratings influenced subjects' confidence overall, we performed an ANOVA. Confidence was not related to the power and social attractiveness ratings of the speaker: subjects were equally confident in their answers whether they listened to speakers with high or low ratings, F < 1. Confidence when wrong. Subjects' confidence when they were misled and wrong was related to both power and social attractiveness ratings. Figure 4 shows that the pattern of results is similar to that of the misinformation effect: when the speaker was rated high on power, confidence was similar for speakers who were rated high on social attractiveness and low on social attractiveness. However, when the speaker was rated low on power, confidence was higher for speakers who were rated high on social attractiveness than low on social attractiveness. This interaction between power and social attractiveness was significant, F(1, 68) = 4.37, p = .04. In short, when subjects ``bought'' the misleading suggestion, power and social attractiveness affected their confidence in the same way that it affected their accuracy.

Figure 3. The interaction between power and social attractiveness for subjects' performance on misled items.

Attitudes and misinformation

We looked at whether subjects' performance on misled items was different for high or low power and social attractiveness ratings. First, we used median splits to classify subjects into high and low groups for each type of rating (power Md = 3.75, social attractiveness Md = 4.14).2 Next we ran a 2 6 2 analysis of covariance (ANCOVA). The independent variables were power group (high or low), social attractiveness group (high or low), and the power 6 social attractiveness interaction. The covariate was subjects' performance on control items (number correct out of a total of four). Subjects' control performance was a significant covariate, F(1, 68) = 6.43, p < .02. The significant interaction between power group and social attractiveness group is displayed in Figure 3, F(1, 68) = 5.38, p = .02. Higher lines represent subjects who were more misled. Figure 3 shows when subjects rated the speaker highly on power, social attractiveness ratings did not matter: subjects were equally misled whether they rated the speaker high or low on social attractiveness, F < 1; the effect size was a negligible 0.01. However, when subjects rated the speaker low on power, social attractiveness ratings did matter: subjects who also rated the speaker high on social attractiveness were more misled than subjects who gave a lower rating, F(1, 68) = 5.14, p < .03, a small to medium-sized effect of 0.31. In short, accent was a vehicle for information about the power and social attractiveness of the
2 Eight subjects gave ratings on the median so their data were removed from this analysis.

DISCUSSION
Our main research question was whether the subtle sociolinguistic information conveyed by accent had an influence on the misinformation

Figure 4. The interaction between power and social attractiveness for subjects' confidence when misled and incorrect.

SOCIOLINGUISTIC CUES AND MISINFORMATION

107

effect. The results support a qualified yes: accents per se did not have an influence on the misinformation effect, but sociolinguistic factors did. Although subjects rated NA accents higher on power and social attractiveness, accents alone were not good proxies for either kind of sociolinguistic information. Why were accents alone not influential enough to change susceptibility to misleading suggestions? One possible explanation is that accents carry additional information that in turn influences power and social attractiveness (Bayard, 1991, 1995; Gordon & Abell, 1990). In our study, it is possible that other factors associated with the speaker influenced ratings of her power and social attractiveness in such a way as to counteract them. However, we did find that power and social attractiveness work together to affect susceptibility to misleading suggestions. When subjects rated the speaker high on power, social attractiveness did not matter: they were equally misled by speakers rated high and low on social attractiveness. When subjects rated the speaker low on power, social attractiveness did matter: they were more misled by speakers rated high on social attractiveness. We found the same pattern of results for confidence estimations when subjects committed to the misleading suggestion. How should we make sense of these social influences on memory distortion and confidence judgements? Dodd and Bradshaw (1980) argued that social connotations are attached to the message and processed simultaneously with the message. Thus, subjects may evaluate postevent messages as a whole, and the influence of misinformation may depend not only on the content of the message itself, but on the context in which the message is delivered. Therefore, we might draw two broad conclusions. First, our findings are consistent with earlier work showing that both the semantic content of the message and its social content should be considered when assessing its potential influence on eyewitnesses (Ceci et al., 1987; Dodd & Bradshaw, 1980; Lampinen & Smith, 1995). Second, we have shown that social content does not necessarily have to be made explicit to influence susceptibility to misleading suggestions. The source-monitoring framework (SMF) (Johnson, Hashtroudi, & Lindsay, 1993) offers one of the best accounts of the misinformation effect (Belli, Lindsay, Gales, & McCarthy, 1994; Lindsay, 1990; Multhaup, de Leonardis, & Johnson, 1999). In this framework, subjects experience a sourcemonitoring failure and attribute misleading PEI to

the event. Other research investigating social influences on memory distortions have put forth SMF-based explanations. For instance, Sherman and Bessenoff (1999) showed that when their ability to engage in deliberate source monitoring was compromised, subjects mistakenly attributed information about people in line with stereotypes. In another study, Roediger, Meade, and Bergman (2001) found that subjects were more likely to report schema-consistent but incorrect information if it had been suggested by a confederate. Roediger et al. argue that subjects confused the source of this false information, believing they had seen it themselves, when in fact it had been given to them by another person. Overall, these studies show that the SMF can account for social influences on memory distortions. Although we did not include a direct test of source monitoring, the SMF may account for our results as well: subjects may be more likely to remember what the powerful and socially attractive speaker told them, and incorporate it into their reports with high confidence, but they are unlikely to remember that the information came from the speaker. Interestingly, one aspect of our results does not square with the SMF, or with the source misattribution explanation of misinformation effects proposed by Lindsay (1990). According to Lindsay, errors in recall may be due to sourcemonitoring confusions, and errors are more likely to decrease as PEI becomes more discriminable or distinctive. By Lindsay's definition, being presented with a postevent narrative in a NA accent in New Zealand almost certainly constitutes a high-discriminability condition compared to the local NZ accent condition. Therefore, a NA accent should make it easier for subjects to discriminate between original information and PEI, which should lead to a reduction in the misinformation effect. However, we found no reduction in the misinformation effect when subjects heard PEI in a NA accent. These results may suggest that the discriminability of a source alone does not predict whether or not subjects will be misled; rather, the social information that the source carries also plays a role. However, as we noted previously, we did not include a sourcemonitoring measure in our experiment, and so this conclusion should be seen as tentative. Finally, we do not know whether the misinformation effects we have found come as a result of a genuine memory-source confusions, or from some kind of demand-based misinformation acceptance (see Belli, 1989). We used the standard

108

VORNIK, SHARMAN, GARRY

``Loftus-type'' test rather than the modified ``McCloskey and Zaragoza'' test, and the standard test cannot isolate the relative contributions of these mechanisms. However, we did include an instruction to subjects that they base their answers only on what they saw in the slides, and not on what they heard, to discourage them from relying only on the PEI to answer the memory test. Of course, in many real-life situations it does not matter what kind of mechanism is driving a false report: if an eyewitness identifies the wrong person, it does not matter if that misidentification was caused by genuine memory impairment, or misinformation acceptance (see also Loftus & Hoffman, 1989). There are practical and theoretical implications to this research. First, the influence of sociolinguistic factors on susceptibility to misleading information demonstrates that information about the source of the PEI can be conveyed subtly yet still have an effect. Future research might examine whether subjects are more susceptible to explicit or implicit cues about the misinformation messenger. A second, and more practical, implication comes into play when eyewitnesses with varying sociolinguistic characteristics discuss an event. Our research suggests that sometimes, memories are no match for the powers of the spoken word.
Manuscript received 10 October 2001 Manuscript accepted 20 February 2002

REFERENCES
Anisfeld, M., Bogo, N., & Lambert, W.E. (1962). Evaluational reactions to accented English speech. Journal of Abnormal and Social Psychology, 65, 223 231. Assefi, S.L., & Garry, M. (in press). Absolute memory distortions: Alcohol expectancies affect the misinformation effect. Psychological Science. Bayard, D. (1991). A taste of Kiwi: Attitudes to accent, speaker gender, and perceived ethnicity across the Tasman. Australian Journal of Linguistics, 11, 138. Bayard, D. (1995). Kiwitalk: Sociolinguistics and New Zealand society. Palmerston North, NZ: The Dunmore Printing Company Ltd. Belli, R.F. (1989). Influences of misleading postevent information: Misinformation interference and acceptance. Journal of Experimental Psychology: General, 118, 7285. Belli, R.F., Lindsay, D.S., Gales, M.S., & McCarthy, T.T. (1994). Memory impairment and source misattribution in postevent misinformation experiments with short retention intervals. Memory & Cognition, 22, 4054. Brown, B.L. (1980). Effects of speech rate on personality attributions and competency ratings. In H. Giles, W.P. Robinson, & P.M. Smith (Eds.), Lan-

guage: Social psychological perspectives (pp. 293 300). Oxford: Pergamon. Brown, R., & Gilman, A. (1960). The pronouns of power and solidarity. In T. Sebeok (Ed.), Style in language (pp. 253276). Cambridge, MA: MIT Press. Callan, V.J., Gallois, C., & Forbes, P.A. (1983). Evaluative reactions to accented English. Journal of Cross-Cultural Psychology, 14, 407426. Ceci, S.J., Ross, D.F., & Toglia, M.P. (1987). Suggestibility of children's memory: Psycholegal implications. Journal of Experimental Psychology: General, 116, 3849. Cohen, J. (1988). Statistical power analysis for the behavioural sciences (2nd Edn.). Hillsdale, NJ: Lawrence Erlbaum Associates Inc. Dodd, D.H., & Bradshaw, J.M. (1980). Leading questions and memory: Pragmatic constraints. Journal of Verbal Learning and Verbal Behavior, 19, 695704. Edwards, J.R. (1982). Language attitudes and their implications. In E.B. Ryan & H. Giles (Eds.), Attitudes toward language variation: Social and applied contexts (pp. 2033). London: Edward Arnold. Giles, H. (1971). Patterns of evaluation in reactions to R.P., South Welsh and Somerset accented speech. British Journal of Social and Clinical Psychology, 10, 280281. Giles, H. (1973). Communicative effectiveness as a function of accented speech. Speech Monographs, 40, 330331. Giles, H., & Coupland, N. (1991). Language: Contexts and consequences. Pacific Grove, CA: Brooks/Cole. Giles, H., Henwood, K., Coupland, N., & Harriman, J. (1992). Language attitudes and cognitive mediation. Human Communication Research, 18, 500527. Greene, E., Flynn, M.S., & Loftus, E.F. (1982). Inducing resistance to misleading information. Journal of Verbal Learning and Verbal Behavior, 21, 207219. Gordon, E., & Abell, M. (1990). ``This objectionable colonial dialect'': Historical and contemporary attitudes to New Zealand speech. In A. Bell & J. Holmes (Eds.), New Zealand ways of speaking English (pp. 2148). Wellington, NZ: Victoria University Press. Gumperz, J.J. (1982). Fact and inference in courtroom testimony. In J.J. Gumperz (Ed.), Language and social identity (pp. 163195). Cambridge, UK: Cambridge University Press. Hyugens, I., & Vaughan, G.M. (1983). Language attitudes, ethnicity and social class in New Zealand. Journal of Multilingual and Multicultural Development, 4, 207223. Johnson, M.K., Hashtroudi, S., & Lindsay, D.S. (1993). Source monitoring. Psychological Bulletin, 114, 328. Lambert, W.E., Hadgson, R., Gardner, R.C., & Fillenbaum, S. (1960). Evaluation reactions to spoken languages. Journal of Abnormal and Social Psychology, 60, 4451. Lampinen, J.M., & Smith, V.L. (1995). The incredible (and sometimes incredulous) child witness: Child eyewitnesses' sensitivity to source credibility cues. Journal of Applied Psychology, 80, 621627. Lindsay, D.S. (1990). Misleading suggestions can impair eyewitnesses' ability to remember event details. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 10771083.

SOCIOLINGUISTIC CUES AND MISINFORMATION

109

Loftus, E.F. (1975). Leading questions and the eyewitness report. Cognitive Psychology, 7, 560572. Loftus, E.F. (1991). Made in memory: Distortions in recollection after misleading information. The Psychology of Learning and Motivation, 27, 187215. Loftus, E.F., & Hoffman, H.G. (1989). Misinformation and memory: The creation of new memories. Journal of Experimental Psychology: General, 118, 100104. Loftus, E.F., Miller, D.G., & Burns, H.J. (1978). Semantic integration of verbal information into a visual memory. Journal of Experimental Psychology: Human Learning and Memory, 4, 1931. Luhman, R. (1990). Appalachian English stereotypes: Language attitudes in Kentucky. Language in Society, 19, 331348. McCloskey, M., & Zaragoza, M. (1985). Misleading postevent information and memory for events: Arguments and evidence against memory impairment hypothesis. Journal of Experimental Psychology: General, 114, 116. Multhaup, K.S., de Leonardis, D.M., & Johnson, M.K. (1999). Source memory and eyewitness suggestibility in older adults. Journal of General Psychology, 126, 7484. Podberesky, R., Deluty, R.H., & Feldstein, S. (1990). Evaluations of Spanish- and Oriental-accented

English speakers. Social Behavior and Personality, 18, 5363. Riches, P., & Foddy, M. (1989). Ethnic accent as a status cue. Social Psychology Quarterly, 52, 197206. Roediger, H.L., Meade, M.L., & Bergman, E.T. (2001). Social contagion of memory. Psychonomic Bulletin & Review, 8, 365371. Seggi, I. (1983). Attribution of guilt as a function of ethnic accent and type of crime. Journal of Multilingual and Multicultural Development, 4, 197206. Sherman, J.W., & Bessenoff, G.R. (1999). Stereotypes as source-monitoring cues: On the interaction between episodic and semantic memory. Psychological Science, 10, 106110. Stewart, M.A., Ryan, E.B., & Giles, H. (1985). Accent and social class effects on status and solidarity evaluations. Personality and Social Psychology Bulletin, 11, 98105. Tabachnick, B.G., & Fidell, L.S. (1996). Using multivariate statistics (3rd ed.). New York: Harper Collins. Weatherall, A., Gallois, C., & Pittam, J. (1998). Australasians identifying Australasian accents. Te Reo, 41, 153162. Webster, W.G., & Kramer, E. (1968). Attitudes and evaluational reactions to accented English speech. Journal of Social Psychology, 75, 231240.

APPENDIX
Factor loadings and communalities (h2) for principal factors extracted with varimax rotation Item Intelligent Ambitious Self-confident A leader Attractive Powerful Strong Educated Prestigious Dependable Honest Sincere Conscientious Kind Sociable Of good character Good-looking Likeable Affectionate Pleasant Courageous Entertaining
a

F1a .59 .51 .61 .70 .60 .84 .77 .68 .69 .17 .14 .10 .23 .05 .17 .13 .36 .31 .08 .47 .39 .50

F2 .47 .11 .20 .15 .06 .05 .09 .23 .17 .75 .76 .69 .65 .32 .08 .51 .03 .17 .22 .06 .25 .16

F3 .20 .13 .31 .20 .52 .04 .01 .20 .07 .06 .14 .28 .19 .72 .65 .64 .57 .75 .67 .53 .07 .49

h2 .61 .29 .51 .55 .63 .71 .61 .56 .51 .60 .62 .57 .52 .62 .45 .68 .47 .69 .51 .50 .22 .52

Factor labels: F1 Power F2 Charisma F3 Social Attractiveness