No More Teachers' Dirty Looks: Effects of Teacher Nonverbal Behavior on Student Outcomes

No More Teachers' Dirty Looks: Effects of Teacher Nonverbal Behavior on Student Outcomes

Teachers change lives. As the popular bumper sticker attests, the power of teachers to influence the course of a child's life is enormous, rivaling in some cases even that of the child's parents. This influence can be positive or, regrettably, negative. Common lore and our own memories tell us that our images of good and bad teachers are heavily influenced by their nonverbal behaviors. We can all remember our fa­vorite teacher from grade school whose warm smile and kind voice made us feel special. If we were unlucky enough, we might also remem ber a hated teacher whose sarcasm and obvious hostility undermined our liking for school.

Anecdotally, then, few people would argue against the idea that non verbal behavior is a critical aspect of teaching effectiveness. Much edu cational literature makes a similar assertion (Doyle, 1977; Galloway, 1971a, 1971b, 1984; Grant & Hennings, 1971; Grubaugh, 1989; Philippot, Feldman, & McGhee, 1992; Woolfolk & Brooks, 1985; Woolfolk & Galloway, 1984). However, the issue of the impact of teach ers' nonverbal behavior on student outcomes is ultimately an empiri­cal one, and thus a logical question is what is the actual empirical evidence regarding the effects of teacher nonverbal communication? The purpose of the current chapter is to begin to address this question

by reviewing the available empirical literature on the relations between teacher nonverbal behavior and student outcomes.

APPROACH AND SCOPE OF THE CURRENT CHAPTER

In order to identify as comprehensively as possible the available liter­ature on teachers' nonverbal behavior, we conducted a series of liter­ature searches of the PsycINFO and ERIC databases. We began with more global searches that used the keywords 'nonverbal behavior' and 'nonverbal communication' crossed by 'teacher*.' After the global searches had identified the major categories or themes of the relevant research, we refined our database searches by using more specific keywords, for example, 'teacher nonverbal immediacy.' Fi­nally, we adopted the ancestry approach of examining the articles we obtained for references that might have been omitted in our comput­erized searches.

Our literature searches yielded a pool of over 150 articles tap-ping into some aspect of teachers' nonverbal behavior. A sizable portion of these articles were nonempirical in nature, that is, mostly theoretical pieces in which the author(s) assert that teach­ers' nonverbal behavior is important and then speculate on why that might be so, without offering data to support their claims. Those articles will not be reviewed in the current chapter. The re­mainder of the articles could be classified into one of five major categories, the first four of which will comprise the scope of this chapter: (a) studies examining the relation between perceptions of teachers' global nonverbal immediacy and student outcomes; (b) studies that related one or two specific nonverbal cues, or combi­nations of cues, to student outcomes; (c) studies on the mediation of teacher expectancy effects; (d) studies that investigated differen­tial teacher behavior displayed toward students varying in race, gender, or ethnicity; and (e) miscellaneous studies that did not fall into any of these four categories, such as qualitative, in-depth anal­yses of individual teachers' use of nonverbal behavior (e.g., Allen, 2000; Galloway, 1971b; Hendrix, 1997).

For each of the four substantive areas of research on teacher non-verbal behavior, we will offer a succinct review and critical evaluation of the available literature. Our primary goals are twofold: to provide researchers in nonverbal communication with a call to action by de-scribing both what is known in this area and directions for future re-search, and to offer teachers and other education professionals an empirically-grounded understanding of the role of teacher nonverbal behavior as well as potentially practical guidelines for incorporating nonverbal cues in actual practice. We conclude the chapter with an overall assessment of the strengths and limitations of the literature on teacher nonverbal behavior.

MAKING A CASE FOR THE SPECIAL RELEVANCE
OF NONVERBAL BEHAVIOR IN TEACHING

Before embarking on our review of the literature, one could reasonably ask why researchers in nonverbal communication need to treat educa tion as a special case. Why can we not simply apply the already very large literature on nonverbal behavior in general to the classroom? Are there any reasons to suspect that basic principles of nonverbal com­munication will differ in teacher-student interactions?

We will are that there are indeed several aspects of the classroom context that render it a unique setting where the traditional rules gov erning communication, both verbal and nonverbal, do not always ap­ply and thus demands special research scrutiny. Take, as only one example, the well-established body of research on turn-taking mecha nisms, that is, the cues that govern the back and forth nature of conver sation. We know a great deal about how dyads negotiate talking turns in everyday interaction, with researchers identifying four basic catego­ries of turn-taking cues: turn-requesting, turn-denying, turn-main­taining, and turn-yielding (Duncan, 1972; Wiemann & Knapp, 1975).

Yet it is not at all obvious that the same processes and cues would hold in the classroom. First, and obviously, the classroom typically does not involve dyadic conversations but rather groups of varying size, interacting across greater distances than are the norm in dyadic interactions. Thus, some of the more subtle turn-taking cues that are highly effective in dyadic conversations (e.g., the barely audible intake of breath that precedes a speaking turn) would generally not be notice-able in the classroom. A second difference is that, unlike everyday group conversations where each participant has a more-or-less equal chance at speaking up, in the classroom the teacher calls the conversa tional shots. Students speak primarily at the mercy of the teacher and indeed in many classrooms must explicitly request permission to speak by raising their hands, a turn-taking mechanism that does not appear on the standard lists of cues (Wiemann & Knapp, 1975).

Turn-taking thus highlights one of the structural features of the classroom context that has important ramifications for researchers in nonverbal communication, namely that it generally consists of a highly unequal power structure. The typical classroom has one teacher, usu ally older, who possesses considerable power over a group of students. As Susan Fiske has documented, people in positions of lower power are especially attentive to the behavior of high-power individuals (Fiske, 1993). Thus, because the teacher tends to do the greater share of the talking, and because students wi11 be especially motivated to at tend to the teacher, it is likely that students will notice their teacher's nonverbal behavior to a greater extent and such behavior may be more influential than in ordinary conversation.'

In short, the stereotypical classroom has a higher-power teacher standing (and therefore highly visible) in front of an audience of lower-

power and generally attentive students. The teacher controls the con­versational flow and must use nonverbal behavior effectively to do so, for example, by directing his or her gaze and nodding at one particular child in a sea of waving arms to designate who is supposed to answer a question, or by flashing a warning look at a disruptive student without interrupting the delivery of lesson material. Even when teachers are not intentionally controlling their nonverbal behavior to convey a given message, the situational demands of the classroom create hypersensi­tivity among students to all behaviors, verbal and nonverbal, given off by teachers. To the student, a teacher's smile in response to a sug­gested answer could be a validation of his or her sense of intellectual achievement and thus affect the student's self-esteem much more strongly than perhaps the teacher could ever suspect. Regrettably, the converse is also true, and a cold or hostile glance (whether intended or not, or even caused by any action of the student or not) can evoke in the student a sense of shame or despair.

Given the great prominence that teachers' nonverbal behavior can have in an academic context, and given the unique features of the class-room that make it difficult to apply directly traditional theories of non-verbal communication (Doyle, 1977), the need for nonverbal research that takes place in the classroom is great. In the sections that follow, we review what is known about the possible effects of teachers' nonverbal behavior in the four areas identified earlier.

Teacher Nonverbal Immediacy and Student Outcomes

The nonverbal research question attracting perhaps the greatest atten­tion in the education literature has been the relation between teachers' nonverbal immediacy and various student outcomes. Nonverbal im­mediacy is a global construct, originally introduced to the nonverbal literature by Mehrabian (1966) and defined as the degree of perceived physical or psychological closeness between people. FYom the begin­ning, immediacy was conceptualized as being inextricably linked with positive affect and liking toward another person, with Mehrabian (1971) declaring that immediacy and liking are 'two sides of the same coin. That is, liking encourages greater immediacy, and immediacy produces more liking' (p. 77).

Meta-Analysis of the Teacher Nonverbal Immediacy Literature

The nonverbal immediacy construct was first applied in the educa­tional literature in a highly influential dissertation and subsequent journal publication by Janis Andersen (1978, 1979). This study sparked a rush of other studies investigating the relation between teacher nonverbal immediacy and student outcomes. Because these studies share a similar hypothesis (as immediacy increases, so does the positivity of student outcomes), and because they demonstrate a

remarkable homogeneity of methodological and analytic approaches, even to the extent of often using identical measures, this body of litera­ture is ideally suited for a meta-analytic review. In this section, then, we present the results of a meta-analysis of the teacher nonverbal imme­diacy literature .2

Our literature search identified a total of 37 independent studies looking at the relation between teacher nonverbal immediacy and stu­dent outcomes. From each study, we coded: (a) the size and nature of the sample (undergraduate, secondary, or primary school students); (b) whether nonverbal immediacy was experimentally manipulated or measured as a naturally occurring variable; (c) the type of student out-come measured; and (d) the effect size, as indexed by the Pearson r, for the immediacy-outcome relation for each of the dependent measures reported. The vast majority of teacher immediacy studies adopt an af­fective, behavioral, and cognitive learning distinction, and our meta-analysis thus preserves this distinction by computing and reporting results separately by category. Affective learning' refers to students' evaluative reactions toward either the course or the teacher. 'Behav­ioral learning' is somewhat of a misnomer; although it can refer in principle to the mastery of performance tasks (cf., Comstock, Rowell, & Bowers, 1995), in an overwhelming number of the studies used in the current meta-analysis, behavioral learning instead refers to behav­ioral intention variables, for example, asking students how likely they would be to take another class with the same teacher or on the same topic. To avoid confusion and represent the spirit of those results more accurately, we wi11 refer instead to this category as 'behavioral intentions.^'

'Cognitive learning' refers to more traditional conceptions of stu­dent academic performance. In the teacher immediacy literature, cog­nitive learning is generally measured in one of two ways: scores on actual performance measures, such as a recall test of material pre­sented in a lecture or course grades, or students' self-reports of how much they had learned in class. Most of the studies using the latter type of cognitive learning variable relied on Richmond, Gorham, and McCroskey's (1987) learning Ioss measure, which asks students, first, how much they learned in the class with the current instructor and, second, how much they think they could have learned in the class if they had an ideal instructor. The learning loss score is created by subtracting responses to the first question from responses to the sec­ond question. Because we suspected that students' self-reports might yield different results than actual performance measures, we analyzed effect sizes for the two subcategories of cognitive learning measures separately.

The prototypical study in this literature consisted of recruiting large classes of undergraduates to rate their teachers on nonverbal immedi­acy, and those ratings were then correlated with student self-reports of affective, behavioral, and cognitive learning. The nonverbal immediacy

Table 8.2 shows the meta analytic results for the four categories of dependent measures. When studies reported more than one result for each category (as would be the case, for example, if a study reported correlations for both attitudes toward the teacher and attitudes to-ward the course), we first computed the mean effect size across mea­sures for each category of dependent measures, thus preserving the independence of effect sizes within category. Because correlation coef­ficients are not normally distributed, all Pearson rs were transformed

Immediacy Behavior Items (Richmond et al., 1987)

2.   Gestures when talking to the class.

3.   Uses monotone/dull voice when talking to the class. (R)

4.   Looks at the class when talking.

5.   Smiles at the class as a whole, not just individual students.

6.   Has a very tense by position when talking to the class. (R)

7.   Touches students in the class.

8.   Moves around the classroom when teaching.

9.   Sits on a desk or in a chair when teaching. (R)

10.        Looks at board or notes when talking to the class. (R)

11.        Stands behind podium or desk when talking to the class. (R)

12.        Has a very relaxed body position when talking to the class.

13.        Smiles at individual students in the class.

14.        Uses a variety of vocal expressions when talking to the class.

Meta-Analysis of the Teacher Nonverbal Immediacy Literature

Note: Positive effect sizes mean greater immediacy is related to more positive outcomes. 'Cognitive Learning' refers to self-eport measures of learning, and 'Cognitive Performance' refers to measures of actual performance.

via the Fisher r-to-z transformation (Rosenthal, 1991) prior to any computations. For ease of interpretation, however, mean effect sizes were transformed back and are reported in terms of Pearson rs in pre­senting the results.

Looking first at the results for the affective learning measures, which was far and above the most common type of dependent measure reported in this literature, we identified 33 independent effect sizes. The mean effect size for the 33 samples was an r of .43, which using Cohen's (1969) criteria could be considered a medium-to-large effect. Thus, as teachers' nonverbal immediacy increased, so did students' positive evaluations of the course and teacher. Put another way, using the Binomial Effect Size Display (BESD; Rosenthal & Rubin, 1982), this effect size can be interpreted as meaning that having a teacher high in nonverbal immediacy is associated with an increase in students' positive evaluations from 28.5% to 71.5%.

With respect to behavioral intentions, the mean effect size for the 19 independent samples investigating this category of dependent variable was an r of .32. Students of teachers high in nonverbal immediacy were more likely to report that they would be interested in taking an-other course from this teacher or on a similar topic. In terms of the BESD, this effect can be interpreted as meaning that having a high-im­mediate teacher is associated with an increase in favorable behavioral intentions from 34 to 66%.

As noted earlier, we analyzed results for the cognitive learning de-pendent variables separately according to whether the measure was a self-report of perceived learning or a measure of actual cognitive per­formance on a task. Far more studies incorporated the former type of measure than the latter. With respect to self-reports of cognitive learn­ing, 21 independent effect sizes were located. The mean effect size for these samples was comparable to that found for the behavioral mten-

Lions variables, corresponding to an r of .36, indicating that teachers' nonverbal immediacy was positively related to students' self-reported learning. In BESD terms, this means that having a teacher high in non-verbal immediacy was associated with an increase in self-reported learning from 32% to 68%.

As we will discuss in more detail later, the use of self-reported cogni­tive learning measures raises obvious validity concerns. Thus, an im­portant question for this literature is whether the same relation between teacher nonverbal immediacy and learning would hold for measures of actual cognitive performance. as, very few of the studies in this area collected data that would enable us to answer this ques­tion. We were able to locate only six independent effect sizes based on cognitive performance measures such as recall tests or course grades. The mean effect size for these six samples was an r of only .14, sub­stantially lower than that obtained for self-reported learning. In BESD terms, this means that having a teacher high in nonverbal immediacy was associated with an increase in test performance from 43% to 57%.

In short, the results of the meta-analysis reveal that teacher nonver­bal immediacy is strongly related to many positive student outcomes: liking for the course and teacher, willingness to take more classes with the teacher, and students' perceptions that they have learned a lot in the class. at is not yet clear is the degree to which these positive out-comes may be translated into gains in actual student achievement. Al-though this meta-analysis found a small effect for the cognitive performance studies, the estimate was based on a very small number of studies and thus should be interpreted only cautiously.

As researchers, we must always be wary of committing the cor­relational fallacy. While it may be tempting to conclude that the causal pathway of the teacher immediacy-student outcome finding flows from the teacher to the student, the reality is that a meta-analytic result based on both correlational and experimental studies does not permit such a conclusion. The causal pathway could potentially run in the op­posite direction. Certainly, any teacher can tell you that a friendly, re­ceptive audience helps to elicit expressive teaching behavior-and that an unfriendly audience can dim the enthusiasm of even the best teach­ers. In order to draw a firm causal conclusion that teacher immediacy leads to positive student outcomes, one needs to review studies that experimentally manipulated teacher nonverbal immediacy and ran­domly assigned participants to condition. Only five studies in the cur-rent data set met those criteria by including such an experimental manipulation, which was usually in the form of preparing two video-taped versions of the same lecture-one delivered by the instructor in a highly immediate style (vocal variety, gaze at camera/students, ex­pressive gestures and movement, etc.) and one delivered in a non-im­mediate manner (speaking in a monotone, looking steadily at notes, etc.). The mean effect size for the five experimental studies was an r of .34, not appreciably different from the mean effect size of .39 for the

correlational studies. Moreover, a one-sample t test of the five effect sizes yielded a t(4) = 2.94,p = .02, one-tailed, with 95% confidence in­tervals of .02 to .59. Thus, even though there were only five studies, we can generalize both to other studies of this ilk (experimental manipula­tions involving one lecturer providing both treatments) as well as to the four dependent variable domains (affective learning, behavioral inten­tions, cognitive learning, and cognitive performance), as the mean effect sizes obtained for those domains fall within the confidence interval for the experimental studies.

Implications for Research on Teacher Nonverbal Immediacy: Where Do We Go From Here?

Our meta-analysis demonstrates convincingly that there is a strong, positive relation between perceptions of teachers' nonverbal immedi­acy and students' evaluations of the teacher, class, and self-reported learning. Although only a small number of these studies involved ex­perimental manipulations, the effect sizes obtained from the experi­mental studies were consistent enough with the overall effect sizes to allow us to conclude that teacher nonverbal immediacy is very likely to be a causal factor in affecting student outcomes. The consistency and overall magnitude of the mean effect sizes obtained in this literature thus provide empirical support for McCroskey and Richmond's (1992) claim that 'teaching immediacy may be one of the most critical variables in determining teaching effectiveness' (p. 119). We explore now the implications of these results and offer directions for future re-search on teacher nonverbal immediacy.

First, it seems obvious that there is little left to be learned from stud­ies asking undergraduates to rate their teachers' nonverbal immediacy and provide self-reports of outcome variables. We had hoped to report comparisons between results from studies using undergraduates and those using elementary or secondary school students, because one could offer a fairly compelling argument that variations in nonverbal immediacy should have a greater effect for younger students given the more affectively-tinged environment of elementary classrooms and in-creased one-on-one interaction between teachers and students in the lower grades. Unfortunately, we were unable to address that question in the current meta-analysis, as only one of the 37 studies we located used anything other than an undergraduate sample (and it was a study of secondary school students). Clearly, one of the directions for future research in this area should be to probe the nature of the teacher immediacy-student outcome relation with younger students.

Equally important, more research needs to be done using behav­ioral outcomes (i.e., actual cognitive performance) rather than relying on self-reports of learning. The cognitive learning self-reports used in this literature have been defended heavily by the researchers employ­ing them, who argue that students have accurate insight regarding how

much they have learned and that course grades are confounded by variables such as 'attendance, writing skills, participation, student preparation, and perceived motivation and may reflect student com­pliance as much as learning' (Gorham, 1988, p. 44). Because most of the studies assessing cognitive learning relied solely on self-report, data on the validity of the self-reports are scarce. Chesebro and McCroskey (2000) reported a correlation coefficient of r = -.50 be­tween the standard learning loss measure and scores on a quiz cover ing the material presented in the lecture in question (with a negative correlation indicating greater validity, as a high score on the learning loss variable indicates less learning), but in a similar study, Witt and Wheeless (2001) obtained an r between the two variables of only -.21. Thus, the magnitude of the validity coefficients for the self-reported learning loss variable is not appreciably large. Also telling is that a dif­ferent pattern of results is obtained with the two types of dependent measures, with moderate-to-large effect sizes obtained for self-reports but much smaller effect sizes obtained for actual cognitive perfor­mance measures. This suggests one of two things: (a) either shared method variance is contributing to the large effect sizes obtained in the self-report studies, or (b) the two types of measures tap into overlap-ping but distinct constructs. In either event, it may not be safe to generalize from the self-report studies that immediacy has the same positive effects on actual learning.

For researchers of nonverbal behavior, perhaps the most pressing question stemming from this literature is an analogous concern re­garding the validity of the most commonly used paper and pencil mea sures of immediacy and the specification of what, exactly, comprises nonverbal immediacy. Andersen (1979) originally defined nonverbal immediacy as the nonverbal manifestation of high affect or the behav iors that indicate physical or psychological closeness (p. 545), a defini tion that is more functional than theoretical in origin. The immediacy scale items shown in Table 8.1 (Richmond et al., 1987) are certainly broad in scope. However, what we do not yet know is the relative contri­bution of these individual behaviors to producing immediacy. Some researchers have reported results separately for each of the items of the immediacy scale; these studies indicate stronger relations for some items (e.g., vocal variety, eye contact, smiling, and relaxed body position) than for others (e.g., McCroskey, Fayer, Richmond, Sallinen, & Barraclough, 1996; Richmond et al., 1987). However, there are probably halo effects occurring when participants rate an instructor on all 14 items at one time, making it difficult to parse the contribu­tions of individual behaviors.

Moreover, a troubling aspect of this literature is its over-reliance on student report of teacher nonverbal behavior. As was the case with the cognitive learning variables, data on the validity of the paper and pencil measures of nonverbal immediacy are sparse. The first Andersen (1979) study on this topic was actually one of the few studies to collect

independent ratings of nonverbal immediacy in an effort to validate the student report measures. This study found that the objective judges' ratings of immediacy correlated highly with the student reports, r = .80 across 13 class sections. Other studies where nonverbal immedi­acy was experimentally manipulated have similarly shown that stu­dents' perceptions of immediacy are significantly and substantially affected by the immediacy manipulation (Booth-Butterfield, Mosher, & Mollish, 1992; Comstock et al., 1995).

While these studies provide reassuring evidence for the validity of the student rating measures, the fact that so few studies of teacher immedi­acy incorporate direct observational, behavioral measures of teachers' nonverbal behavior remains a limitation of this area. An important thrust of the research agenda for teacher nonverbal immediacy thus should be a more detailed analysis of the precise nature of the nonverbal behaviors that communicate immediacy through detailed coding of vid­eotapes of high- and low-immediacy teachers. Follow-up studies should then systematically manipulate the identified behaviors to determine which are most critical in affecting student outcomes.

OTHER RESEARCH ON THE RELATIONS BETWEEN TEACHER
NONVERBAL BEHAVIORS AND STUDENT OUTCOMES

In this section, we review the findings of studies that systematically explored the relations between nonverbal behaviors and student out-comes. Whereas the nonverbal immediacy literature was more molar in perspective, looking at the contribution of general nonverbal be­havior, globally defined, studies in this section often took a more mo­lecular approach: How do nonverbal cues relate to teaching effectiveness on an individual basis? Because these studies involve such a diverse range of independent variables, we do not attempt a meta-analytic review but instead conduct a more traditional narra­tive review of this literature.

We discuss first the subset of studies that address 'positive vs. nega­tive' teacher nonverbal behavior or similar constructs such as 'high or low affect' toward students or 'still vs. active' teaching styles. We did not include these articles in the immediacy section because the au­thors did not identify their studies as falling in that domain, and their methodology was often fairly different, although the degree of concep­tual overlap between immediacy and such things as 'positive teacher behavior' is obviously high.

Not surprisingly given the conceptual overlap, the results of these studies are generally consistent with the teacher nonverbal immediacy literature. Several studies looked at the effects of 'vivid' or 'active' nonverbal behavior on the part of teachers, with three of the studies in­volving experimental manipulations of teacher behavior (Schiefer, 1986; Seals & Kaufman, 1975; Sims, 1986) and one involving self-re­ported teacher nonverbal expressiveness (Hamann, Lineburgh, &

Paul, 1998). These studies found consistent positive relations between a more expressive teaching style and students' evaluation of the teacher or class, but no consistent relation with student performance outcomes. For example, undergraduates watching a lecture presented in a vivid manner did not have better recall on a test over the material (Schiefer, 1986); similarly, preschoolers listening to musical selec­tions while the teacher displayed enthusiastic facial expressions did not show greater recognition of the selections than when the teacher exhibited a bored expression (Sims, 1986).³

Other studies manipulated the valence of nonverbal cues more so than their intensity. These studies, on the whole, yielded a similar pattern whereby positive teacher nonverbal behavior had positive ef­fects on students' subjective reactions but either nonsignificant or even negative effects on students' cognitive performance. For exam ple, Anita Woolfolk and her colleagues have conducted a program of research in which teachers' verbal and nonverbal behavior are ma­nipulated in a factorial design. In one study that was reported in a se­ries of articles, 128 sixth grade students were given a short vocabulary lesson by one of four undergraduate 'teachers,' who were trained to administer the lesson accompanied by either positive or negative verbal and nonverbal cues, with positive nonverbal behavior operationalized by head nods, positive voice tone, and smiles. Woolfolk and her colleagues found that the students had more posi tive reactions toward the teachers when they used positive nonverbal behaviors, especially the female teachers (Woolfolk, Woolfolk, & Garlinsky, 1977), and they were equally willing to self-disclose to nonverbally positive and negative teachers (Woolfolk, Garlinsky, & Nicolich, 1977). However, students receiving positive nonverbal be­haviors from the teacher actually produced fewer sentences using the vocabulary words and, for female students, performed worse on the subsequent spelling test (Woolfolk, 1978). In a separate experi­ment, Woolfolk and Woolfolk (1975) found that positive teacher non-verbal behavior was associated with an increased willingness to self-disclose. Thus, these studies reveal once again that positive, friendly nonverbal behavior-while perceived favorably by stu­dents-does not necessarily result in improved learning, and indeed they raise the possibility suggested by Woolfolk (1978) that 'teacher negative nonverbal behavior may be more motivating than teacher positive nonverbal behavior' (p. 93).

A similar study was conducted by Goldberg and colleagues, who had 120 2nd and 6th grade students watch a videotape of a teacher presenting material using either positive, neutral, or negative nonver bal behaviors. A similar pattern of results was obtained. When using positive nonverbal behavior, the teacher was evaluated more positively by the students (Goldberg & Mayerberg, 1973). However, in a second article reporting the student performance data, there was no consis tent effect of nonverbal behavior. For one task, students in the neutral

condition performed worse, with no difference between students in the negative and positive nonverbal conditions, whereas in a second task, students in the negative nonverbal condition performed worse than students in the neutral and positive conditions, who did not differ (Goldberg & Mayerberg, 1975).

In contrast to the experimental studies just described, Harris, Rosenthal, and Snodgrass (1986) reported an observational study linking teachers' warmth to student cognitive performance. Ten teachers were videotaped while administering a ten-minute lesson, consisting of sentence-completion problems and arithmetic word problems, to students from kindergarten to second grade. Video-tapes were coded for a variety of verbal and nonverbal behaviors. Re­gression analyses indicated that judges' global ratings of teacher warmth significantly predicted students' performance on the lesson, r = .32. However, because teacher warmth was not experimentally manipulated, the possibility exists that good student performance elicited more positive teacher behavior rather than vice versa.

Other studies show similarly positive effects of positive teacher nonverbal behavior on non-cognitive student outcomes such as eval­uations of the instructor, liking for the class, or attentive behavior (Bettencourt, Gillet, Gall, & Hull, 1983; Chaikin, Gillen, Derlega, Heinen, & Wilson, 1978; Guerrero & Miller, 1998; Harris et al., 1986; Kazdin & Klock, 1973; Keith, Tornatzky, & Pettigrew, 1974; Kleinfeld, 1974; Neill, 1986, 1989a, 1989b; van Tartwijk, Brekelmans, & Wubbels, 1998; Wass, 1973). Clearly, an important riddle for educa tional and nonverbal researchers to solve is why, exactly, teacher non verbal behavior appears to affect students' evaluative reactions more so than their cognitive performance. There are several possible ex­planations. The studies reviewed here that assessed cognitive perfor mance almost always did so in the context of an artificial, short-term teaching encounter, e.g., watching a 10-minute videotape of an unfa miliar teacher and then being tested immediately over the content of that videotape. Perhaps teachers' nonverbal behavior has a greater impact on cognitive performance in the context of a long-term teach­ing relationship, when there is more opportunity for it to affect im­portant mediators of academic performance such as student motivation. This could explain the seemingly counterintuitive find­ings of Woolfolk (1978) and other researchers that negative nonver­bal behavior resulted in better task performance. Perhaps negative teacher behavior can be effective in the short term, but in the long run it undermines students' liking for school and motivation and thus would be damaging. A second possible explanation for the failure to find strong effects of teacher nonverbal behavior on cognitive perfor mance is that perhaps performance on these tasks is determined so strongly by other factors (student ability, content of the material, na ture of the assessment, etc.) that teachers' nonverbal style simply is less relevant.

HARRIS AND ROSENTHAL Studies Looking at Discrete Nonverbal Behaviors.

Most of the studies we identified looking at teachers' nonverbal behav­ior did so in a holistic manner, operationalizing teacher behavior in global terms such as 'immediate,' 'positive,' or 'active,' or measuring or manipulating several discrete nonverbal cues simultaneously. This strategy has both positive and negative consequences. The positive fea­tures of operationalizing nonverbal behavior holistically is that it re­flects the reality that nonverbal behaviors do not occur in isolation; teachers emit a broad constellation of behaviors, verbal and nonver­bal, and all behaviors take place and are interpreted in the larger con-text of the other behaviors that are being emitted.

There are drawbacks to operationalizing teacher nonverbal behav­ior globally, however. The primary limitation, as discussed earlier, is that it becomes impossible to determine which discrete nonverbal cues play a causal role in affecting student outcomes. Yet such a deter­mination is an important consideration when teacher training pro-grams are contemplated. Because we know that immediate, positive teacher nonverbal behavior is perceived positively by students, teacher education programs may want to offer explicit training in be-having in a more nonverbally immediate and positive manner. But if research were to show, for example, that teacher smiles were the pri­mary causal factor in producing positive student outcomes, training could be more efficient, simpler, and ultimately more successful if it fo­cused on increasing the frequency of teacher smiles rather than trying to increase simultaneously all 14 behaviors on the nonverbal immediacy scale shown in Table 8.1.

Unfortunately, the teacher nonverbal behavior literature is not at the stage where such detailed considerations can be made. We located few studies that manipulated or measured discrete nonverbal behaviors and analyzed them in isolation. Some of the studies reviewed earlier come close to doing this, by manipulating or measuring two or three nonverbal cues simultaneously. For example, in Kleinfeld (1974), non-verbal warmth was operationalized by having the guidance counselor sit closer, smile often, and touch the student twice briefly; Kazdin and Klock (1973) similarly manipulated nonverbal approval through smil­ing and touch, and Sims (1986) manipulated gaze and facial expres­sions. Manipulating a smaller number of variables as in these studies narrows down the possible causal factors when significant results are obtained, but it still remains impossible to determine which one (or all) of the two or three cues manipulated was the most important.

Guerrero and Miller (1998) attempted to answer that question by asking students to rate instructors on four different nonverbal style variables separately (nonverbal warmth, speaking style, eye contact, and articulation) and then correlated those variables with students' im­pressions of the teacher. They found that nonverbal warmth was most strongly related to student evaluations, although the other three vari-

ables were nonetheless significantly related. However, as in the immedi­acy literature, because the nonverbal variables were coded subjectively, the possibility remains that they are contaminated by halo effects.

Wass (1973) was one of the few researchers to manipulate discrete behaviors in a factorial manner. Students (307 3rd-6th graders) watched videotaped clips of a teacher providing feedback to a hypotheti­cal student via verbal, voice tone, and facial expression channels, with the valence of the channels manipulated factorially; for example, the verbal statement was either praising or critical, and the facial expres­sion was either smiling or frowning. Students then rated whether the feedback given was good, bad, or neither good nor bad. Results showed that the valence of the verbal channel overwhelmingly influenced stu­dents' judgments; for example, over 800 of the students rated the mes­sage good if the verbal statement was positive, regardless of the valence of the nonverbal channels. Of course, given that the content of the verbal statement included an explicit evaluation of the hypothetical student's performance, it is not surprising that the verbal content predominated in students' judgments. It would be interesting in future research to de­termine the relative impact of nonverbal channels using this design with more standard academic verbal content.

Other researchers have similarly studied differences across verbal and nonverbal channels. Schiefer (1986) manipulated verbal and non-verbal 'vividness' in a 2 x 2 factorial. Although no significant main ef­fect of nonverbal vividness was obtained, 'there was a positive effect of verbal vividness combined with nonverbal vividness' on students' ability to follow the lecture (p. 1106). Schmidt and McCutcheon (1994) had 180 undergraduates provide evaluations of seven lectures under two conditions: audio-only and video-only. Although no significant dif­ferences emerged on the composite teacher ratings between condi­tions, when items were analyzed separately, teachers received more negative ratings in the audio-only condition on hesitant speaking style, being well prepared, and communicating effectively.

We located two studies that entailed a fairly exhaustive coding of a wide range of discrete nonverbal cues. Keith et al. (1974) coded video-tapes of 43 preservice teachers (students in teacher training pro-grams) on a total of 38 separate behaviors in such categories as gestures; body movements and orientation; facial expression; head, physical, and visual orientation; and facial attractiveness. They also coded 19 verbal and nonverbal responses of the students (grades K-6). Using a similar methodology, Fox and Poppleton (1983) coded a total of 47 variables while observing physical education teachers inter-acting with students, of which 11 were nonverbal in nature, such as proximity, touch, general body orientation, body movement, and facial expression. Both of these studies thus collected the data that in princi­ple could allow a determination of the relations between individual nonverbal cues and student outcomes. Somewhat disappointingly, from our perspective at least, in both studies the analyses that were re-

ported were factor or cluster analyses of all variables (teacher and stu­dent verbal and nonverbal behaviors) combined together. These analyses provide interesting insight into the patterns of naturally co-occurring teacher and student behaviors; for example, Keith et al. (1974) identified three major clusters of behaviors that they termed (a) positive task-relevant interaction, (b) observation and group interac­tion, and (c) teacher disapproval and pupil misbehavior. But they do leave unanswered the question of which individual cues best predict student outcomes.

A handful of studies have examined the impact of a single nonverbal behavior on student outcomes. For example, in a study of elementary school children, Otteson and Otteson (1978) found that teachers who read a short story while engaging in eye contact with the students facili­tated their recall of the story. Gesture is the nonverbal cue that has re­ceived the most rigorous research attention, and studies on gesture-unlike studies of other aspects of teacher nonverbal behavior-have focused primarily on student learning as the outcome variable of inter-est. (See Roth, 2001, for a recent excellent review of the role of gesture in education.) Several empirical studies have documented that a teacher's gestures can predict student learning, especially when the material covered is complex and/or involves scientific or mathematical concepts (Flevares & Perry, 2001; Goldin-Meadow, Kim, & Singer, 1999; McNeil, Alibali, & Evans, 2000; Roth, 2001; Valenzeno, Alibali, & Klatzky, 2003).

A prototypical example of this research is a study by Goldin-Meadow et al. (1999). In this study, eight teachers were videotaped while present­ing a math lesson to 49 3rd and 4th grade students. The videotapes were coded for the problem-solving strategies conveyed both verbally and gesturally by the teachers (e.g., pointing to the addends to be summed), and students' comprehension of the strategies was assessed, as operationalized by their ability to repeat it back to the teacher. Analy­sis of the videotapes indicated that 60% of the teacher's speaking turns contained both speech and gesture; of these, there was a 2:1 ratio of speaking turns where the verbal and nonverbal strategies were matched vs. mismatched. Looking at how the teachers' gestures related to stu­dent comprehension, the researchers found that children were more likely to reiterate the strategy when it was accompanied by gesture than when no gesture accompanied it, and they were less likely to reiterate the strategy when the gesture did not match it than when no gesture ac­companied it. Goldin-Meadow et al. (1999) thus concluded, 'gesture aided comprehension of teacher speech when it matched that speech, and hurt child comprehension of teacher speech when it mismatched that speech' (p. 726).

Similar results were found by Valenzeno et al. (2003), who showed preschool children one of two videotapes explaining the concept of sym­metry. In one version, the teacher did not use any gestures, whereas in the other version, she produced pointing and tracing gestures as she ex-

plained symmetry. Children were then administered a posttest wherein they were asked to judge six objects as being either symmetrical or asymmetrical. Those who saw the verbal + gesture videotape scored significantly higher on the posttest than did children who saw the ver­bal-only videotape.

Summary and Conclusions

at can teachers conclude from this set of studies? Consistent with the immediacy literature, teachers who engage in warm, active nonver­bal behavior are evaluated more positively by their students. Also con­sistent with the immediacy literature, it is not clear, though, that this behavior predicts greater amounts of learning, and at least a couple of studies found better task performance associated with negative teacher nonverbal behavior. Does that mean that the old teachers' ad-age, 'Don't smile until Christmas,' is correct? Not necessarily. Again, too few of the studies included learning as a dependent variable to ar­rive at any firm conclusions. The research on gesture, moreover, indi­cates definite positive effects of teacher nonverbal behavior on learning. It would also be unwise, perhaps, to discount the strong mo­tivating role of positive student evaluations in facilitating academic performance. Students may initially work harder for a teacher who is strict and stern, but overly negative teacher behavior can lead to the students' intrinsic motivation for schoolwork being undermined. Fur­ther research is needed that tracks teachers' nonverbal behavior and its influence on students' evaluative reactions to the teacher and cogni­tive performance longitudinally.

For researchers in nonverbal communication, this set of studies represents an advance over the nonverbal immediacy literature, as most of these studies involved the objective coding or manipulation of nonverbal behaviors, rather than measuring them subjectively through Likert-type scales that ask for relative judgments of behav­iors. Moreover, most of the studies in this section involved experimen­tal manipulations of teachers' nonverbal behavior, thus permitting stronger causal inferences regarding the role of nonverbal behavior. However, there is much more that can be done with this area. As noted earlier, most of the studies manipulated or measured nonverbal cues in combination, making it impossible to untangle the contributions of a teacher's smiles from his or her tone of voice, for example. More studies need to be done that analyze the effects of individual cues, with the recent work on gestures being an excellent example of the kinds of careful, informative studies that can be done in this regard.

MEDIATION OF TEACHER EXPECTANCY EFFECTS

The teacher nonverbal behavior literature reviewed so far has con­cerned 'main effects,^' as it were, of teacher behavior, that is, the rela-

Lions between a teacher's behavior toward an entire class and student outcomes. In the remainder of this chapter, we look at the question of differential teacher behavior toward certain subgroups of students. In other words, within a single classroom, do teachers treat some stu­dents differently than others, and how does that differential behavior affect those students' outcomes? We begin with a review of the large lit­erature on the mediation of teacher expectancy effects and conclude with a review of the literature on differential teacher behavior toward students varying in race, ethnicity, and gender. Because the teacher ex­pectancy literature has been the topic of extensive reviews in the past (see Babad, 1992; Harris & Rosenthal, 1985), we do not describe indi­vidual studies in this section but rather summarize the major conclu­sions of this literature.

Beginning in the mid-1950s, there has been a growing literature on the phenomenon of 'interpersonal expectancy effects.' This term re­fers to the finding that one person's expectation for the behavior of an-other can come to serve as a self-fulfilling prophecy. Thus, experimenters collecting psychological data have been shown to ob­tain data consistent with the type of data they had been led to expect. For example, when experimenters were told that the research subjects they were assigned were 'success-perceivers,' those subjects rated the faces of other people as more successful. When experimenters were told that the research subjects they were assigned were 'failure-per­ceivers,' those subjects rated the faces of other people as less success­ful. In these and many other studies with human subjects, psycho-logical experimenters tended to obtain the data they had been led to expect (Rosenthal, 1963, 1969; Rosenthal & Rubin, 1978).

Subsequent research showed that even when animal subjects were employed, experimenters led to expect better learning of mazes by their rats obtained better learning of mazes than did experimenters led to expect less learning from their rats (Rosenthal & Fode, 1963). Similar findings were obtained when Skinner boxes were employed instead of mazes. Rats expected to learn sequences of responses more readily learned those sequences more readily (Rosenthal & Lawson, 1964).

If rats learned better when expected to, then it seemed not far-fetched to think that children might learn better when expected to by their teachers. Indeed, that turned out to be the case. When teachers in an elementary school were led to believe that certain children in their class would show unusual gains in intellectual performance over the course of a school year, those certain children (whose names had been selected by means of a table of random numbers), actually did gain more in intellectual performance than did the children of the control group (Rosenthal & Jacobson, 1992). Meta-analyses of the 19 studies of the effects of teachers' expectations showed not only that there was an overall non-negligible effect of experimentally induced effects of teacher expectations, but that the magnitude of the effect increased lin-

early with the plausibility of the manipulation (Raudenbush, 1984, 1994; Rosenthal, 2002).

Mediating Variables

As more was learned about the occurrence and generality of interper­sonal expectancy effects, more was also learned about the processes of communication that probably served to mediate the effect. If we label the experimentally induced expectancy for the behavior of another per son, E, the communicating or mediating variables, M, and the outcome responses of the person for whom the expectations were held, 0, we can diagram the mediation process by means of the following three arrows:

The E-O arrow or link describes the experimental effect of Eon O. The E-M arrow or link describes the experimental effect of E on M. The M-0 arrow or link describes the relationship between the mediating vazi­able and the outcome variable. We should note that the E-O and E-M arrows are causal links because we have experimentally induced E. The M-O link, however, is not usually a causal link because M has not been experimentally manipulated. We return to this important, but of-ten overlooked, point later.

Four Factors

On the basis of the first 30 or so published studies relevant to media­tion, a four-factor 'theory' of the mediation of teacher expectancy ef­fects was proposed (Harris & Rosenthal, 1985, 1986; Rosenthal, 1994). The 'theory' describes four major groupings of teacher behav­iors hypothesized to be involved in mediation. The first factor, cli­mate, refers to the warmer socioemotional climate that teachers tend to create for high expectancy students, a warmth that can be commu­nicated both verbally and nonverbally. The input factor refers to the tendency for teachers to teach more material to their 'special' stu­dents. The output factor refers to the tendency for teachers to give their 'special' students greater opportunities for responding. Finally, the feedback factor refers to the tendency for teachers to give more differentiated feedback to their 'special,' high expectancy students. By differentiated, we mean that the feedback will be contingent on the correctness or incorrectness of the student's response and that the content of the feedback will tend to be directly related to what the stu­dent has said.

Meta-analyses conducted by Harris and Rosenthal (1986) were de-signed to summarize the many studies examining either the E-M or M-O links (or both) and to come up with a quantitative estimate of the importance of each of the four factors in the mediation of interpersonal expectancy effects. Table 8.3 gives the average magnitude of the role of each factor separately for the E-M and M-O links. While all four factors received ample support in terms of significance testing, the magni­tudes of the effects for the climate and input factors were especially impressive. Teachers appear to teach more and to teach it more warmly to students for whom they have more favorable expectations.

Mediating Variables and Causal Inference

Teachers' expectations for their pupils' intellectual functioning have been shown to serve as self-fulfilling prophecies. The performance ex­pected came to pass because the teacher expected it. We can draw that causal conclusion about the E-O link and the E-M link because of the randomized experiments that have been conducted, reported, and summarized quantitatively. But there is a causal conclusion we can not draw, and it's a big one. Although the relationships between mediating variables and outcome variables are well-established, almost nothing can be said about the effects of the mediating variables on the outcome variables because the mediators have rarely been manipulated experi­mentally. We may find that experimentally manipulated teacher expec­tations cause both greater teacher warmth and better student intellectual performance, and that greater teacher warmth predicts

Meta-Analytically Derived Average Correlations Indexing the Effect Sizes of the Four Fac‑
tor Theory (After Harris & Rosenthal, 1986)

Note: All correlations are significantly greater than zero at p < .002. The correlation between the mag­nitudes of the average E-M and M-0 links is .88.

better student intellectual performance. From this it is tempting to conclude that the expectancy-caused warmth caused the improved outcome. Tempting, but quite unjustified. If we want to conclude that warmth is the mediator variable caused by raised teacher expectancy and, in turn, causing improved student performance, we must manip­ulate teacher warmth directly. Valid causal inferences are not available 'on the cheap' no matter how fancy the 'causal inferential' software. When randomization is truly impossible (e.g., for ethical reasons) we should use those statistical procedures such as subclassification on propensity scores (Rubin, 1998) ranking highest not only in sophisti­cation about validity, but also in transparency.

Applications and Remaining Questions

We have learned a good deal about some processes of communication in classrooms. But for all we have learned there is more that we do not yet know. For example, now that we know that teachers' behaving more warmly toward their students is associated with better performance on the part of their students, can't we just apply this knowledge and train teachers to treat students more warmly in order to improve stu­dent performance? This simple question is really two questions, nei­ther of which we can answer very well. First, we don t know the degree to which teacher warmth can be trained. Although various training programs have been developed for teachers, experimental studies of their effectiveness are few and far between. In a later section of this chapter we describe a few of those programs. Second, we don't know whether teachers trained to be warmer would in fact elicit better per­formance from their students. It would take specifically designed ran­domized studies to learn the degree to which we can train teachers to treat their students more warmly, and if these studies showed that we could, it would take additional randomized studies to show that this increased teacher warmth brought about improved student perfor­mance. These unanswered questions are not cause for pessimism; they just indicate that our task is not yet completed.

DIFFERENTIAL TEACHER NONVERBAL BEHAVIOR
IN OTHER DOMAINS

The research summarized above on the mediation of teacher expec­tancy effects focused on the question of whether teachers treat high-ex­pectancy students differently than they treat low-expectancy students. In this section, we focus instead on whether teachers display differen­tial nonverbal behavior to students varying on dimensions besides ac­ademic expectations. In other words, do teachers act differently toward students varying in race, gender, ethnic group, or other vari­ables? These studies obviously overlap conceptually with the teacher expectancy literature, as teachers' academic expectancies stem in part

from their beliefs about ability differences across racial, gender, and other subgroupings of students (Dusek & Joseph, 1985). A primary difference is that these studies typically do not report data to show that teachers' differential behavior culminates in differences in student outcomes.

Do Teachers Demonstrate Different Nonverbal Behaviors Toward Girls and Boys?

One of the most controversial issues in education in the past decade has been the assertion that teachers pay disproportionately greater at­tention to boys, on both subtle and overt levels, creating a 'chilly cli­mate' in the classroom for girls (American Association of University Women, 1992; Lance, 1985; Sandler, Silverberg, & Ha11, 1996; Sadker & Sadker, 1994). This notion has been argued persuasively and passionately by academic feminists and disseminated so widely in the popular media that many-if not most-educators would agree un­blinkingly that there is a chilly climate in today's schools for girls. However, this assertion has not gone unchallenged by critics, who note that chilly climate proponents do not offer convincing empirical evi­dence for their claims, and they point to a large array of data suggesting instead that boys are at greater risk for poor educational outcomes than are girls (HIeinfeld, 1998; Sommers, 2000).

Most of the chilly climate articles have focused on differential verbal behaviors, such as calling on boys more than girls or discouraging girls from taking math or science courses. For our purposes, we are in terested in the narrower question of whether teachers exhibit different nonverbal behavior toward girls and boys. Given the 'hotness' of gen der issues in psychology today, the paucity of empirical studies on this question (and, as we will see later, the question of differential behavior according to student race) is disheartening.

at have the few empirical studies on this question found? The re sults of two studies suggest that the sex of the teacher interacts in im portant ways with the sex of student to determine teachers' nonverbal behavior. Perdue and Connor (1978) analyzed touching patterns be­tween both male and female teachers and their male and female pre-school students. The primary result is that there were greater amounts of teacher touch in same-sex teacher-student dyads than in oppo­site-sex dyads. Instances of touch were also coded as to whether they were friendly, helpful, attentional, or incidental, and there were differ ences due to student sex with respect to the frequencies of these cate­gories. Most of the touches received by girls were of the helpful category (40%), with only 18% of the touches being deemed 'friendly.' Boys, on the other hand, received significantly more friendly touches than girls (29%) and fewer helpful touches (23%).

In a study with an undergraduate student population, Hechtman and Rosenthal (1991) videotaped 60 preservice students while teach-

ing two short lessons-one verbal, one mechanical/quantitative-to undergraduates, with the students tested on the content before and af­ter the lesson. Analyses revealed that teachers appeared more nonverbally hostile when teaching sex role counter-stereotypic les­sons, that is, when teaching the verbal lesson to the men and the me chanical lesson to the women. There was also a three-way interaction between lesson, student sex, and teacher sex such that men showed this biased teaching behavior more than did the women. This differen tial behavior was evidently picked up by the students, who in turn were less satisfied with the lesson when male teachers taught counter-ste­reotypic material.

In an observational study examining both race and gender, Simpson and Erickson (1983) found that boys received more nonverbal criti­cism than did girls. More specifically, White teachers gave the most nonverbal criticism to Black boys relative to Black girls, White boys, and White girls. In another study examining race and teacher liking for students, Lyon (1977) found that boys who were evaluated negatively by the teacher also received more negative nonverbal behavior from them, in the form of frowns, head shakes, glares, and restraining touches.

Given the extremely limited number of studies directly addressing this issue, conclusions regarding sex bias-or the lack thereof-in teachers' nonverbal behavior would be premature, as the evidence in these studies is mixed with respect to which gender is 'favored' nonverbally. Clearly, more descriptive research is needed to document patterns of teachers' nonverbal behavior toward male and female stu­dents. Equally important, future studies should collect data that would allow linking differential teacher behavior to student outcomes. Merely establishing that teachers treat boys and girls differently would not be sufficient grounds for alarm; one would want to show that these differences affect one gender unfairly. One would also want to rule out student behavior as a causal eliciting factor prior to making claims of teacher bias. For example, if boys do receive more negative nonverbal criticism as suggested by Simpson and Ericson (1983), it would be im­portant to discover whether this held true even after controlling for, say, student misbehavior or off-task behavior.

Do Teachers Show Different Nonverbal Behavior to Students of Different Races?

As conscious or unconscious racism in the classroom is often sug­gested as a root cause of the achievement gap between Blacks and Whites, the effects of student race on teachers' nonverbal behavior is a research question of obvious importance. A leading researcher on this topic is Robert Feldman, whose group has conducted several studies examining differential teacher nonverbal behavior as a function of stu dent race (see Feldman, 1985; Feldman & Saletsky, 1986, for reviews

of this literature). In the first study in this research program, Feldman and Donohoe (1978) videotaped 36 undergraduates delivering an analogy lesson to Black or White confederates. These videotapes were rated for the extent to which the teacher appeared pleased with the stu­dent. Analyses indicated a very large effect of student race, with teach­ers rated as appearing more pleased when interacting with a White confederate than a Black confederate. Student race also interacted sig­nificantly with teacher prejudice level, such that the magnitude of the race difference was greater for high-prejudice teachers than low-preju­dice teachers. A second experiment reported in the article replicated the main effect of student race found in the first study. This study also included a sample of Black undergraduates role-playing the teacher, and there were no differences due to student race for Black teacher nonverbal behavior, at least when ratings were made by White raters. When Black raters were used, however, the teachers appeared more pleased with Black students, raising the possibility of same-race favor­itism that can only be detected by members of that race.

Feldman and Orchowsky (1979) used a similar methodology and once again obtained a strong effect of student race, with undergraduate role-playing teachers appearing more pleased with White confederates than Black confederates. This particular study also manipulated the task performance of the confederates, and an even larger effect was found for this variable, with teachers appearing more pleased when a student did well than when he or she did poorly.

An observational study of the behavior exhibited by one teacher to-ward 12 educationally handicapped students, however, failed to dem­onstrate differences in teacher behavior as a function of student race (Lyon, 1977). Correlations between student race and teacher gaze, smiles, head movements, touch, and proximity were all reported to be nonsignificant, although given the very small sample size, it is not possi­ble to interpret the null results as reflecting a zero population effect size.

In addition to the studies reviewed above, which do not yield a strong consensus in their findings, several studies from the teacher ex­pectancy domain also included race of student as an independent vari­able and thus are relevant here. Taylor (1979) looked at the joint effects of student race and teacher expectancy, using a sample of pre-service teachers asked to administer a lesson to a 'student,' described as varying in ability and race, who was allegedly on the other side of a one-way mirror. Although no significant main effects of student race were obtained for teachers' nonverbal behavior, a Race x Expectancy interaction was obtained such that teachers displayed the greatest amount of nonverbal warmth to high-ability Blacks and low-ability Whites. Chaikin and Derlega (1978) had undergraduate role-playing teachers administer a lesson to two Black and two White 10-year-old confederates. The White confederates received more smiles and gaze than did the Black confederates. In another role-playing study where Black undergraduates were asked to teach a fire safety lesson to con-

federates of differing race, Derlega, McAnulty, Strout, and Reavis (1980) found that the teachers maintained greater physical distance with White confederates than they did with Black confederates. Rubovfts and Maehr (1973) found that Black junior high schools stu dents were treated more negatively on a range of variables, primarily verbal, but also including ignoring by the teacher.

Taken together, these studies suggest that, when differential behav­ior occurs, it is more likely to favor White students than Black stu­dents, but that there also appears to be a tendency for teachers to behave more positively toward students of the same race as the teacher. Again, the paucity of empirical research on student race ef­fects is discouraging given the obvious applied importance and great social interest in these issues.

Differential Teacher Behavior as a Function of Other Variables

Although race and gender are the individual difference variables at­tracting the most concern among educators, a handful of studies exists looking at differential teacher behavior as a function of other variables. Some have looked at differences in teacher behavior directed toward other ethnic and cultural groups and differences between socioeco­nomic groups. For example, Greenbaum (1985) showed that teachers at Indian reservation schools paused longer than did teachers in largely-White schools, although this difference is difficult to interpret given the small number of teachers involved (four) and the confound ing of teachers with schools. Davis, Dobson, and Shelton (1973) coded 'encouraging' and 'restricting' nonverbal behaviors of 20 first grade teachers, 11 of whom taught in schools with primarily low-SES stu­dents and 9 of whom taught in schools with largely middle-class stu­dents. This study found no differences between SES categories with respect to either total quantity of nonverbal behaviors or encouraging behaviors, and only a small, nonsignificant trend for low-SES classes to receive fewer restrictive behaviors (r = .22).

Elisha Babad and his colleagues have also conducted an impressive program of research documenting differential teacher nonverbal be­havior (see Babad, 1993, for a review of this literature). In addition to studies showing that teachers' expectations for their students are leaked through nonverbal channels (Babad, Bernieri, & Rosenthal, 1989a, 1989b; Babad & Taylor, 1992), these studies have also found that teachers behave differently according to the type of class being taught (preschool vs. elementary), and whether or not the student is a teacher's pet. For example, Babad, Bernieri, and Rosenthal (1987) had raters judge short (10-sec) clips of preschool, remedial, and elemen­tary school teachers' behavior. Preschool teachers showed the least amount of negative nonverbal affect and teacher dogmatism, whereas elementary school teachers showed the most, with remedial teachers falling between the two groups. Interestingly, there were no significant

differences on the more positive composite of 'active' teaching behav­ior. Examination across nonverbal channels indicated that the differ­ences in negative affect were found only for those channels showing the face, whereas the differences in dogmatism were generally obtained across all nonverbal channels.

Because most people can remember either being a teacher's pet themselves or being irritated by somebody else who was a teacher's pet, one of the more intriguing questions for researchers of teachers' nonverbal behavior is whether teachers display favoritism nonverbally and how that affects those students and others in the classroom. We lo­cated several studies that investigated differences between teachers as a function of their liking for the students. In a study that involved an ex­perimental manipulation of liking for particular students, Feldman (1976) found that undergraduates role-playing teachers appeared more nonverbally pleased with students they had been led to like. This result was confirmed by an observational study by Lyon (1977), who found teachers displaying more negative nonverbal behaviors (e.g., glares and frowns) to disliked students, but there was no difference in frequencies of positive nonverbal behaviors. With respect to proximity, Brooks and Wilson (1978) found that teachers stood further away from disliked students.

In a survey study of 80 elementary classrooms, Babad (1995) dis­covered that in many classrooms, students arrived at good consensus regarding which student(s) were teacher's pets, and this consensus was related to perceived differences in teachers' behavior toward stu­dents, especially affect-related behaviors. Moreover, perceived differ­ential affect was in turn related negatively to students' morale and satisfaction. Thus, to the extent that students can pick up on favorit­ism in teachers, such favoritism may have adverse effects on the class-room climate. Given the universal awareness of 'teacher's pets,' and the negative consequences that can accrue both for those students who are and those who are not favored by the teacher (Tal & Babad, 1990), more research on this critical dimension is clearly needed. For exam­ple, observational studies are needed to identify the cues that students use to determine who is regarded as a teacher's pet. A reasonable hy­pothesis is that these cues are primarily nonverbal in nature, as most teachers are motivated to avoid showing favoritism verbally.

IMPLICATIONS AND SUGGESTIONS FOR RESEARCHERS AND TEACHERS

at Have We Learned So Far And Where Do We Go From Here?

Taking the four domains of literature on teachers' nonverbal behavior together, one is left with the unmistakable conclusion that progress on these research questions is uneven. We have a very good grasp on the effects of teacher nonverbal immediacy and warmth on students' affec-

tive reactions to their educational experience. We also have a good grasp on the general question of how teachers' expectations are com­municated nonverbally. In other domains, however, the literature is disquietingly sparse. Moreover, the literature is also disquietingly dated, with most of the studies reviewed here published during the 1970s and 1980s, and much fewer dating from the 1990s or later.

Interest in specific topics waxes and wanes in any research do-main, and there are probably several reasons for the recent neglect of teacher's nonverbal behavior as a research endeavor. As any nonver­bal researcher can attest, doing nonverbal research is a difficult, la­bor-intensive, time-consuming process. Coding even a fairly small data set (say, 30 minutes of videotaped classroom interactions from a sample of 30 teachers) can take months. And as any educational re-searcher can attest, the prevailing ethical requirements for studying schoolchildren have become increasingly stringent and conducting research in the classroom correspondingly more difficult. For exam­ple, most IRBs today would require researchers interested in video-taping classroom interactions to obtain the active consent of the families of every student in the class before they could appear on the videotape, a difficult feat. School administrators are also increas­ingly reluctant to allow researchers in the classroom to administer surveys or other dependent measures, given increasing demands on instructional time. Thus, researchers interested in teachers' nonver­bal behavior are faced with a 'double whammy' that makes conduct­ing this research extremely difficult.

Given the logistical difficulties in doing such research, it is not sur­prising that so many researchers turn to role-play analogues of the teaching relationship, using undergraduates, and/or rely on self-re­port measures of nonverbal behavior. Although such studies can serve to provide encouraging preliminary data and to generate hypotheses to be tested in an actual school context, we feel that such reliance on un­dergraduate analogues and self-reports limits the external validity of our research. Undergraduates may act entirely differently than teach­ers, who have acquired hard-earned professional expertise through years in the classroom trenches. Self-reports, while expedient, also raise considerable validity concerns, especially in the domain of non-verbal behavior, given the extent to which nonverbal expression takes place outside the conscious awareness and control of an individual.

As noted earlier, more research is needed that looks at the relation between individual nonverbal cues and student outcomes, especially performance outcomes such as student grades or recall. The recent work on gestures described earlier (Goldin-Meadow et al., 1999; Valenzeno et al., 2003) serves as an excellent model for future research in which nonverbal behaviors are carefully defined, coded, and ana­lyzed within the context of a meaningful theoretical framework. More research is also needed on the intriguing possibility that teachers' non verbal behaviors may be more important in some contexts than in oth-

ers. The Wass (1973) study described earlier found that nonverbal behavior had a negligible impact when accompanying verbal feedback; other researchers have similarly found that teachers' words often carry more weight than their nonverbal behaviors (Woolfolk & Woolfolk, 1974). However, a considerable body of literature exists pur porting that teachers nonverbal cues, gestures in particular, may be particularly helpful in the teaching of foreign language, a domain where the verbal channel is not necessarily straightforward (Allen, 2000; Antes, 1996; Barnett, 1983; Beattie, 1977; von Raffler-Engel, 1980; Ward & von Raffler-Engel, 1980).

Training in Nonverbal Behaviors

Within the domain of nonverbal behavior in the classroom, perhaps the topic of greatest interest to educators is the extent to which it is possible to train teachers to use nonverbal behaviors more effec­tively. Identifying the behaviors that are associated positively with student motivation, affect, and learning is all well and good, but one could make a case that such knowledge is not really helpful unless it can be translated into training programs to help teachers become more effective. Although many teacher education programs address nonverbal behavior in their curricula, few empirically validated non verbal training programs exist. For example, French (1971) designed an inservice training program that included two 'assignments' in-tended to enhance teachers' use and awareness of nonverbal commu­nication: (a) teaming up with a trusted colleague, observing each other's classes, and exchanging feedback regarding each teacher's use of nonverbal behavior, and (b) devoting two five-minute intervals in the classroom to presenting lesson material exclusively non-verbally as a way of increasing awareness of nonverbal cues and their impact on students. Unfortunately, the article describing this train­ing model did not provide any evaluation data, so its efficacy is un­known. Love and Roderick (1971) developed an awareness unit targeting ten categories of teacher nonverbal behavior. The unit con­sisted of having teachers (a) read about nonverbal behavior; (b) ob­serve a videotape of another teacher and attend to that teacher's nonverbal behavior in a general way; (c) learn to recognize the ten cat­egories of nonverbal behavior more specifically; (d) practice the be­haviors in small role-playing groups; and then (e) practice the behaviors in a real setting. Love and Roderick (1971) state that they pilot tested this awareness unit; unfortunately they do not report de-tails about the pilot study (such as number of teachers sampled or any detailed results), stating merely that teachers used more catego­ries of nonverbal behavior on the posttest compared to the pretest.

Fetter (1983) developed another training program whereby teachers were asked first to fill out a self-report inventory of their own nonver bal behavior and then were asked to observe in private a videotape of

themselves teaching while referring to the nonverbal behaviors identi fied in the self-report instrument. A pilot study comparing 14 teachers who undertook this training to 14 control teachers showed that teach ers in the training condition increased we and movement toward stu­dents and decreased frowns and exasperated looks, although with these modest sample sizes, the differences were not significant.

Finally, Richmond, McCroskey, Plax, and Kearney (1986) found that teachers who had taken a course in nonverbal communication stress­ing the nonverbal behaviors associated with immediacy were per­ceived as being more nonverbally immediate by their students than did students of teachers who had not received such training. These find­ings must be interpreted with caution, although, given the posttest-only nature of the design and the failure to assign teachers randomly to training conditions.

In sum, although several nonverbal training programs have been developed, none of them has been subjected to the type of rigorous ran­domized, controlled trials and empirical evaluation that most re-searchers would want to see prior to advocating their widespread use in teacher education programs. These training programs, moreover, are fairly dated and thus do not reflect the advances in research on teachers' nonverbal behavior made in the past couple of decades. A critical direction for future research, therefore, would be the develop ment and empirical evaluation of nonverbal training units that could be incorporated into teacher training programs. In designing these units, we echo the sentiment raised earlier that the selection of nonver bal behaviors to be targeted for training should be done on an empiri cal basis, based on experimental studies indicating a true causal effect of the targeted behavior. Given the relative lack of solid experimental data on individual teacher behaviors, it could well be that the first step must be to expand the body of basic, experimental research on teachers' behavior where nonverbal cues are experimentally manipu­lated and their effects on students measured.

Closing Thoughts

That teachers exert a strong effect on students' lives, either in a posi tive or negative way, is undisputed. The literature reviewed here shows that an important part of teachers' influence is nonverbal in nature. Our review also indicates that, with the few notable exceptions de-scribed above, we do not have firm answers to questions regarding the precise nonverbal mechanisms underlying teachers' influence. Charles Galloway's apt comment in 1984 that 'the field of nonverbal has demonstrated complexities and variant interpretations beyond anything the pioneers in the field could have imagined' thus holds just as true today (Galloway, 1984, p. 412). The 'more research is needed' conclusion is trite, overused, and lame, yet in this case it may be the most honest way to close our chapter. Studying the nonverbal behavior

of teachers is fraught with difficulty, but we hope we have convinced readers that it is a challenge worth the effort.

ACKNOWLEDGMENTS

We thank MaryLu Rosenthal for her assistance in searching the litera­ture and data entry. Correspondence concerning this chapter should be addressed to Monica Harris, Department of Psychology, University of Kentucky, Lexington, Kentucky 40506-0044. E-mail: har­ris@uky.edu

NOTES

1.   Although it is beyond the scope of the present chapter, this unequal power structure, combined with the distracting cognitive demands placed on a teacher of formulating and delivering educational material, may well result in students' nonverbal behavior being less noticed by teachers and less in­fluential than in ordinary interaction.

2.   Space limitations do not permit full description of the meta-analytic pro­cedures and results for this literature; see Harris and Rosenthal (2003) for more detail, including a bibliography of articles included in the meta-analysis.

3.   However, neither article provides statistics from which an effect size can be computed, so we are unable to conclude that there is truly no effect (i.e., that r is near 0), only thatp is > .05. Such a failure to provide precise statistics is especially common in older studies, and thus this caveat holds true for many of the subsequent articles we summarize where it is stated that re­sults are not significant.

REFERENCES

Allen, L. Q. (2000). Nonverbal accommodations in foreign language teacher talk. Applied Language Learning, 11, 155-176.

American Association of University Women. (1992). How schools shortchange girls: A study of major findings on girls and education. Washington, DC AAUW Educational Foundation.

Andersen, J. F. (1978). The relationship between teacher immediacy and teaching effectiveness. Unpublished doctoral dissertation, West Virginia University, Morgantown, W.

Andersen, J. F. (1979). Teacher immediacy as a predictor of teaching effective­ness. In D. Nimmo (Ed.), Communication Yearbook, 3 (pp. 543-559). New Brunswick, NJ: Transaction Books.

Antes, T. A. (1996). Kinesics: The value of gesture in language and in the Lan­guage classroom. Foreign Language Annals, 29, 439-448.

Babad, E. (1992). Teacher expectancies and nonverbal behavior. In R. S. Feldman (Ed.), Applications of nonverbal behavioral theories and re-search (pp. 167-190). Hillsdale, NJ: Lawrence Erlbaum Associates.

Babad, E. (1993). Teachers' differential behavior. Educational Psychology Re-view, 5, 347-376.

Babad, E. (1995). The 'teacher's pet' phenomenon, students' perceptions of teachers' differential behavior, and students' morale. Journal of Educa­tional Psychology, 87, 361-374.

Babad, E., Bernieri, F., & Rosenthal, R. (1987). Nonverbal and verbal behavior of preschool, remedial, and elementary school teachers. American Educa­tional Research Journal, 24, 405-415.

Babad, E., Bernieri, F., & Rosenthal, R. (1989a). Nonverbal communication and leakage in the behavior of biased and unbiased teachers. Journal of Personality and Social Psychology, 56, 89-94.

Babad, E., Bernieri, F., & Rosenthal, R. (1989b). When less information is more informative: Diagnosing teacher expectations from brief samples of behaviour. British Journal of Educational Psychology, 59, 281-295.

Babad, E., & Taylor, P J. (1992). Transparency of teacher expectancies across language, cultural boundaries. Journal of Educational Research, 86, 120-125.

Barnett, M. A. (1983). Replacing teacher talk with gestures: Nonverbal com­munication in the foreign language classroom. Foreign Language Annals. 16, 173-176.

Beattie, N. (1977). Nonverbal aspects of the teaching and learning of foreign languages. Audio-Visual Language Journal, 15, 175-181.

Bettencourt, E. M., Gillet, M., Gall, M., & Hull, R. E. (1983). Effects of teacher enthusiasm on student on-task behavior and achievement. American Edu­cational Research Journal, 20, 435-450.

Booth-Butterfield, S., Mosher, N., & Mollish, D. (1992). Teacher immediacy and student involvement: A dual process analysis. Communication Re-search Reports, 9, 13-21.

Brooks, D. M., & Wilson, B. J. (1978). Teacher verbal and nonverbal behav­ioral expression toward selected pupils. Journal of Educational Psychol­ogy, 70. 147-153.

Chaikin, A. L., & Derlega, V. J. (1978). Nonverbal mediators of expectancy ef­fects in Black and White children. Journal of Applied Social Psychology, 8, 117-125.

Chaikin, A. L., Gillen, B., Derlega, V., J., Heinen, J. R. K., & Wilson, M. (1978). Students' reactions to teachers' physical attractiveness and nonverbal be­havior: Two exploratory studies. Psychology in the Schools, 15, 588-595.

Chesebro, J. L., & McCroskey, J. C. (2000). The relationship between stu­dents' reports of learning and their actual recall of material: A validity test. Communication Education, 49, 297-301.

Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic Press.

Comstock, J., Rowell, E., & Bowers, J. W. (1995). Food for thought: Teacher nonverbal immediacy, student learning, and curvilinearity. Communica­tion Education, 44, 251-266.

Davis, G., Dobson, R., & Shelton, J. (1973). Nonverbal behavior of first grade teachers in different socioeconomic level elementary schools. Journal of the Student Personnel Association for Teacher Education, 12, 76-80.

Derlega, V. J., McAnulty, M., Strout, S., Reavis, C. A. (1980). Pygmalion effects among Blacks: When and how expectancies occur. Journal of Applied So­cial Psychology, 10, 260-271.

Doyle, W. (1977). The uses of nonverbal behavior: Toward an ecological model of classrooms. Merrill-Palmer Quarterly, 23, 179-192.

Duncan, S. (1972). Some signals and rules for taking speaking turns in con­versations. Journal of Personality and Social Psychology, 23, 283-292.

Dusek, J. B., & Joseph, G. (1985). The bases of teacher expectancies. In J. B. Dusek (Ed.), Teacher expectancies (pp. 229-250). Hillsdale, NJ: Lawrence Erlbaum Associates.

Feldman, R. S. (1976). Nonverbal disclosure of teacher deception and inter‑

personal affect. Journal of Educational Psychology, 68, 807-816. Feldman, R. S. (1985). Nonverbal behavior, race, and the classroom teacher.

Theory Into Practice, 24, 45-49.

Feldman, R. S., & Donohoe, L. F. (1978). Nonverbal communication of affect in interracial dyads. Journal of Educational Psychology, 70, 979-987.

Feldman, R. S., & Orchowsky, S. (1979). Race and performance of student as determinants of teacher nonverbal behavior. Contemporary Educational Psychology, 4, 324-333.

Feldman, R. S., & Saletsky, R. D. (1986). Nonverbal communication in interra­cial teacher-student interaction. In F S. Feldman (Ed.), The social psychol­ogy of education: Current research and theory (pp. 115-131).

Fetter, M. P (1983). Nonverbal teaching behavior and the health educator. JOSH, 53, 431-432.

Fiske, S. T. (1993). Controlling other people: The impact of power on stereo-typing. American Psychologist, 48, 621-628.

Flevares, L. M., & Perry, M. (2001). How many do you see? The use of non-spoken representations in first-grade mathematics lessons. Journal of Ed­ucational Psychology, 93, 330-345.

Fox, C., & Poppleton, P. (1983). Verbal and nonverbal communication in teach­ing: A study of trainee P E. teachers in the gymnasium. British Journal of Educational Psychology, 53, 107-120.

French, R. L. (1971). Analyzing and improving nonverbal communication: A

model for inservice education. Theory Into Practice, 10, 305-309. Galloway, C. M. (1971a). Analysis of theories and research in nonverbal com‑

munication Journal of the Association for the Study in Perception, 6, 1-21. Galloway, C. M. (1971b). Nonverbal: The language of sensitivity. Theory Into

Practice, 10, 227-230.

Galloway, C. M. (1984). Nonverbal and teacher-student relationships: An intercultural perspective. In A. Wolfgang (Ed.), Nonverbal behavior: Per­spectives, applications, and intercultural insights (pp. 411-430). Kirkland, WA: Hogrefe & Huber Publishers.

Goldberg, G., & Mayerberg, C. K. (1973). Emotional reactions of students to non-verbal teacher behavior. The Journal of Experimental Education, 42, 29-32.

Goldberg, G., & Mayerberg, C. K. (1975). Effects of three types of affective teacher behavior on student performance. Child Study Journal, 5, 99-105.

Goldin-Meadow, S., Kim, S., & Singer, M. (1999). at the teacher's hands tell the student's mind about math. Journal of Educational Psychology, 91, 720-730.

Gorham, J. (1988). The relationship between verbal teacher immediacy be‑

haviors and student learning. Communication Education, 37, 44-53. Grant, B. M., & Hennings, D. G. (1971). The teacher moves: An analysis of

nonverbal activity. New York: Teachers College Press.

Greenbaum, P E. (1985). Nonverbal differences in communication style be­tween American Indian and Anglo elementary classrooms. American Edu­cational Research Journal, 22, 101-115.

Grubaugh, S. (1989). Non-verbal language techniques for better classroom management and discipline. High School Journal, 73, 34-40.

Guerrero, L. K., & Miller, T. A. (1998). Associations between nonverbal behaviors and initial impressions of instructor competence and course content in video taped distance education courses. Communication Education, 47, 30-42.

Hamann, D. L., Lineburgh, N., & Paul, S. (1998). Teaching effectiveness and social skill development. Journal of Research in Music Education, 46, 87-101.

Harris, M. J., & Rosenthal, R. (1985). The mediation of interpersonal expec­tancy effects: 31 meta-analyses. Psychological Bulletin, 97, 363-386.

Harris, M. J., & Rosenthal, R. (1986). Four factors in the mediation of teacher expectancy effects. In R. S. Feldman (Ed.), The social psychology of educa­tion (pp. 91-114). NY: Cambridge University Press.

Harris, M. J., & Rosenthal, R. (2003). Teacher nonverbal immediacy and stu dent outcomes: A meta-analysis. Manuscript in preparation.

Harris, M. J., Rosenthal, R., & Snodgrass, S. E. (1986). The effects of teacher expectations, gender, and behavior on pupil academic performance and self-concept. Journal of Educational Research, 79, 173-179.

Hechtman, S. B., & Rosenthal, R. (1991). Teacher gender and nonverbal be­havior in the teaching of gender-stereotyped materials. Journal of Applied Social Psychology, 21, 446-459.

Hendrix, K. G. (1997). Student perceptions of verbal and nonverbal cues lead ing to images of Black and White professor credibility. The Howard Journal of Communications, 8, 251-273.

Kazdin, A. E., & Klock, J. (1973). The effect of nonverbal teacher approval on stu­dent attentive behavior. Journal of Applied Behavior Analysis, 6, 643-654.

Keith, L. T., Tornatzky, L. G., & Pettigrew, L. E. (1974). An analysis of verbal and nonverbal classroom teaching behaviors. The Journal of Experimen­tal Education, 42, 30-38.

Kleinfeld, J. S. (1974). Effects of nonverbal warmth on the learning of Eskimo and White students. Journal of Social Psychology, 92, 3-9.

Kleinfeld, J. (1998). The myth that schools shortchange girls: Social science in

the service of deception. Washington, DC: The Women's Freedom Network. LaFrance, M. (1985). The school of hard knocks: Nonverbal sexism in the

classroom Theory into Practice, 24, 40-44.

Love, A. M., & Roderick, J. A. (1971). Teacher nonverbal communication: The development and field testing of an awareness unit. Theory Into Practice, 10, 295-299.

Lyon, S. (1977). Teacher nonverbal behavior related to perceived pupil so­cial-personal attributes. Journal of Learning Disabilities, 10, 52-56.

McCroskey, J. C., Fayer, J. M., Richmond, V P., Sallinen, A., & Barraclough, R. A. (1996). A multi-cultural examination of the relationships between non-verbal immediacy and affective learning. Communication Quarterly, 44, 297-307.

McCroskey, J. C., & Richmond, V. P. (1992). Increasing teacher influence through immediacy. In V. P. Richmond & J. C. McCroskey (Eds.), Power in the classroom: Communication, control, and concern (pp.101-119). Hillsdale, NJ: Lawrence Erlbaum Associates.

McNeil, N. M., Alibali, M. W, & Evans, J. L. (2000). The role of gesture in chil­dren's comprehension of spoken language: Now they need it, now they don't. Journal of Nonverbal Behavior, 24, 131-150.

Mehrabian, A. (1966). Immediacy: An indicator of attitudes in linguistic com­munication. Journal of Personality, 34, 26-34.

Mehrabian, A. (1971). Silent messages. Belmont, CA: Wadsworth.

Neill, S. R. St J. (1986). Children's reported responses to teachers' nonverbal

signals: A pilot study. Journal of Education for Teaching, 12, 53-61. Neill, S. R. St. J. (1989a). Children's reported responses to teachers' and

non-teachers' nonverbal communication. Educational Research, 31, 71-74. Neill, S. R. St. J. (1989b). The effects of facial expression and posture on chil‑

dren's reported responses to teacher nonverbal communication. British

Educational Research Journal, 15, 195-204.

Otteson, J. P. & Otteson, C. R. (1980). Effects of teacher gaze on children's story recall. Perceptual and Motor Skills, 50, 35-42.

Perdue, V. P., & Connor, J. M. (1978). Patterns of touching between preschool children and male and female teachers. Child Development, 49, 1258-1262.

Philippot, B, Feldman, R. S., & McGee, G. (1992). Nonverbal behavioral skills in an educational context: Typical and atypical populations. In R. S. Feldman (Ed.), Applications of nonverbal behavioral theories and research (pp. 191-213). Hillsdale, NJ: Lawrence Erlbaum Associates.

Raffler-Engel, W. von. (1980). Kinesics and paralinguistics: A neglected factor in second-language research and teaching. Modern Language Review, 36, 225-237.

Raudenbush, S. W. (1984). Magnitude of teacher expectancy effects on pupil IQ as a function of the credibility of expectancy induction: A synthesis of find­ings from 18 experiments. Journal of Educational Psychology, 76, 85-97.

Raudenbush, S. W. (1994). Random effects models. In H. Cooper & L. V. Hedges (Eds.), The handbook of research synthesis (pp. 301-321). New York: Russell Sage Foundation.

Richmond, V. B, Gorham, J. S., & McCroskey, J. C. (1987). The relationship between selected immediacy behaviors and cognitive learning. Communi­cation Yearbook, 10, 574-590.

Richmond, V. B, McCroskey, J. C., Plax, T. G., & Kearney, P. (1986). Teacher nonver bal immediacy training and student affect. World Communication, 15, 181-194.

Rosenthal, R. (1963). On the social psychology of the psychological experi­ment: The experimenter's hypothesis as unintended determinant of experi­mental results. American Scientist, 51, 268-283.

Rosenthal, R. (1969). Interpersonal expectations. In R. Rosenthal and R. L. Rosnow (Eds.), Artifact in behavioral research (pp. 181-277). New York: Academic Press.

Rosenthal, R. (1991). Meta-analytic procedures for social research (rev. ed.). Newbury Park: Sage.

Rosenthal, R. (1994). Interpersonal expectancy effects: A 30-year perspective. Current Directions in Psychological Science, 3, 176-179.

Rosenthal, R. (2002). Covert communication in classrooms, clinics, court-rooms, and cubicles. American Psychologist, 57, 839-849.

Rosenthal, R., & Fode, K. L. (1963). The effect of experimenter bias on the per­formance of the albino rat. Behavioral Science, 8, 183-189.

Rosenthal, R., & Jacobson, L. (1992). Pygmalion in the classroom (expanded edition). New York: Irvington.

Rosenthal R., & Lawson, R. (1964). A longitudinal study of the effects of exper­imenter bias on the operant learning of laboratory rats. Journal of Psychi­atric Research, 2, 61-72

Rosenthal, R., & Rubin, D. B. (1978). Interpersonal expectancy effects: The first 345 studies. The Behavioral and Brain Sciences, 3, 377-386.

Rosenthal, R., & Rubin, D. B. (1982). A simple, general purpose display of magnitude of experimental effect. Journal of Educational Psychology, 74, 166-169.

Roth, W. (2001). Gestures: Their role in teaching and learning. Review of Edu­cational Research, 71, 365-392.

Rubin, D. B. (1998). Estimation from nonrandomized treatment comparisons using subclassification on propensity scores. Annals of Internal Medicine, 127(8), 757-763.

Rubovits, P C., & Maehr, M. L. (1973). Pygmalion black and white. Journal of Personality and Social Psychology, 25, 210-218.

Sadker, M., & Sadker, D. (1994). Failing at fairness: How America's schools cheat girls. New York: MacMillan Publishing Co.

Sandler, B. R., Silverberg, L. A., & Hall, R. M. (1996). The chilly classroom cli­mate: A guide to improve the education of women. Washington, DC: Na­tional Association for Women in Education.

Schiefer, H. J. (1986). Effect of verbal and nonverbal vividness on students' in­formation-processing. Perceptual and Motor Skills, 63, 1106.

Schmidt, C. P., & McCutcheon, J. W. (1994). Verbal versus nonverbal cues in evaluations of teaching. Journal of Research and Development in Educa­tion, 27, 118-125.

Seals, J. M., & Kaufman, P. A. (1975). Effects of nonverbal behavior on student
attitudes in the college classroom. The Humanist Educator, 14, 51-55.

Simpson, A. W., & Erickson, M. T. (1983). Teachers' verbal and nonverbal communication patterns as a function of teacher race, student gender, and student race. American Educational Research Journal, 20, 183-198.

Sims, W. L. (1986). The effect of high versus low teacher affect and passive ver sus active student activity during music listening on preschool children's at­tention, piece preference, time spent listening, and piece recognition. Journal of Research in Music Education, 34, 173-191.

Sommers, C. H. (2000). The war against boys.Atlantic Monthly, 285, 59-74. Tal, Z., & Babad, E. (1990). The teacher's pet phenomenon: Rate of occur‑

rence, correlates, and psychological costs. Journal of Educational Psychol‑

ogy, 82, 637-645.

Taylor, M. C. (1979). Race, sex, and the expression of self-fulfilling prophecies in a laboratory teaching situation. Journal of Personality and Social Psy­chology, 37, 897-912.

Valenzeno, L., Alibali, M. W, & Klatzky, R. (2003). Teachers' gestures facilitate students' learning: A lesson in symmetry. Contemporary Educational Psy­chology, 28, 187-204.

Van Tartwijk, J., Brekelmans, M., & Wubbels, T. (1998). Students' perceptions of teacher interpersonal style: The front of the classroom as the teacher's stage. Teaching and Teacher Education, 14, 607-617.

Ward, L., & Rafller-Engel, W. von. (1980). The impact of nonverbal behavior on foreign language teaching. In W. von Raffler-Engel (Ed.), Aspects of nonver­bal behavior (pp. 287-311). Lisse: Swets & Zeitlinger.

Wass, H. (1973). Pupil evaluations of teacher messages in three channels of

communication Florida Journal of Educational Research, 15, 46-52.
Wiemann, J. M., & Knapp, M. L. (1975). Turn-taking in conversations. Journal

of Communication, 25, 75-92.

Witt, P. L., & Wheeless, L. R. (2001). An experimental study of teachers' verbal and nonverbal immediacy and students' affective and cognitive learning. Communication Education, 50, 327-342.

Woolfolk, A. E. (1978). Student learning and performance under varying con­ditions of teacher verbal and nonverbal evaluative communication. Journal of Educational Psychology, 70, 87-94.

Woolfolk, A. E., & Brooks, D. M. (1985). The influence of teachers' nonverbal behaviors on students' perceptions and performance. The Elementary School Journal, 85, 513-528.

Woolfolk, A. E., & Galloway, C. M. (1985). Nonverbal communication and the study of teaching. Theory Into Practice, 24, 77-84.

Woolfolk, A. E., Garlinsky, K. S., & Nicolich, M. J. (1977). The impact of teacher behavior, teacher sex, and student sex upon student self-disclo­sure. Contemporary Educational Psychology, 2, 124-132.

Woolfolk, A. E., & Woolfolk, R. L. (1975). Student self-disclosure in response to teacher verbal and nonverbal behavior. Journal of Experimental Educa­tion, 44, 36-40.

Woolfolk, R. L., & Woolfolk, A. E. (1974). Effects of teacher verbal and nonver­bal behavior on student perceptions and attitudes. American Educational Research Journal, I I, 297-303.

Woolfolk, R. L., Woolfolk, A. E., & Garlinsky, K. S. (1977). Nonverbal behavior of teachers: Some empirical findings. Environmental Psychology and Non-verbal Behavior, 2, 45-61.