The Sourcebook of Nonverbal Measures

Going Beyond Words

Valerie Lynn Manusov

• Publisher: Psychology Press
• Year: 2014
• ISBN:
  9781135704216
• Edition: 1
• Topic: Langues et linguistique
• Subtopic: Études sur la communication

The Sourcebook of Nonverbal Measures provides a comprehensive discussion of research choices for investigating nonverbal phenomena. The volume presents many of the primary means by which researchers assess nonverbal cues. Editor Valerie Manusov has collected both well-established and new measures used in researching nonverbal behaviors, illustrating the broad spectrum of measures appropriate for use in research, and providing a critical resource for future studies.With chapters written by the creators of the research measures, this volume represents work across disciplines, and provides first-hand experience and thoughtful guidance on the use of nonverbal measures. It also offers research strategies researchers can use to answer their research questions; discussions of larger research paradigms into which a measure may be placed; and analysis tools to help researchers think through the research choices available to them.With its thorough and pragmatic approach, this Sourcebook will be an invaluable resource for studying nonverbal behavior. Researchers in interpersonal communication, psychology, personal relationships, and related areas will find it to be an essential research tool.

I. INTRODUCTION TO NONVERBAL COMMUNICATION RESEARCH

Researcher Choices and Practices in the Study of Nonverbal Communication

Cindy H. White

University of Colorado

Sargent Jack

Kean University

INTRODUCTION

There is sometimes a temptation to believe that a book like this provides researchers with ready-made approaches for examining interaction and answering research questions. So, we begin this chapter with a caution that fulfilling such a belief is unlikely because ready-made research solutions are rare. A book like this one can, however, be a great asset to researchers, because such volumes allow researchers to see clearly what others are doing and to make thoughtful decisions about how to examine nonverbal behavior. Decisions about measurement, whether the measures involve self- (or other-) reports, coding, rating, or physiological measures are theoretical acts (Bakeman & Gottman, 1997). They are the researchers’ assessments of what is important in the interaction and what will have relevance to the theory being tested or the context being examined. It makes sense, then, to begin this book with a discussion of some of the decisions that researchers face as they begin to examine nonverbal cues.

Once a researcher makes that first decision to study nonverbal behavior, he or she is put on a path that requires many other decisions. These decisions include who is going to be studied, what type of interaction will be studied, where the study will take place, how the study will be designed, which behavior will be examined, and how long the behavior will be observed (Scherer & Ekman, 1982, see Cappella, this volume and Dunbar & Burgoon, this volume). Some of these decisions are easy because they are indicated clearly by the behavior or situation being studied (for instance, if a researcher wants to study affection displays during airport departures, he or she has to go to the airport as Heslin & Boss, 1980, did). But many of these choices are decisions the researcher must make along the way as he or she considers how to best capture the elements of interaction that are important and meaningful (e.g., should he or she study touch during departure or facial expressions or both, across what time frame, and among which passengers?). Moreover, each of these choices has implications for the others, so that the researcher is faced not with discrete choices but with a set of interrelated judgments.

In the sections that follow, we examine the advantages and disadvantages of some selected research choices related to the study of nonverbal communication. We first discuss the potential uses and limitations of self-reports for the study of nonverbal behavior. Then we explore in depth the decisions related to categorization of observations, because oftentimes the nature and the practical enactment of coding or rating schemes are somewhat undetectable once a study comes to publication. Specifically, we examine choices researchers make about the type of observational system to employ (type of observation scheme, coding versus rating), decisions regarding where to collect data (field or laboratory settings), and issues related to the nature of the interaction being examined (coding individuals versus dyads/groups, examining interaction events or coding slices of time). We then discuss briefly when a researcher may choose physiological measures. Our goal is to provide readers with an introduction to some of the issues nonverbal researchers grapple with and to highlight chapters in this volume where readers can see that choice in action.

Self-Reports of Nonverbal Behavior

One choice researchers have is whether to use self-report measures in their studies. Self-report tools rely on participants’ own perspectives (or, occasionally, the perspectives of others who are connected to the one observed), rather than the views of independent observers, to not only determine what specific behaviors were enacted but also delineate their meanings and significance. These measures are often used to gather information about microbehaviors, naturally occurring social interactions, meanings assigned by observers to behaviors, or to assess numerous communication-related skills and abilities across a variety of contexts (see Riggio, this volume; Riggio & Riggio, this volume). For instance, in a number of studies, Stanley Jones has used the self-report Touch Log Record to record the touching behavior of individuals. In his chapter in this book, he describes work by Jones and Yarbrough (1985), who were interested in the meanings associated with specific touching behaviors. Their study identified 12 mutually exclusive meanings associated with specific touching behaviors, and the self-report coding system generated in their work is one available framework. In another example, Palmer and Simmons (1995) used a self-report method that asked research confederates to record in an open-ended questionnaire the specific behaviors they used to convey liking to their interaction partner. Later these responses were content analyzed. This allowed the authors to compare participants’ understanding of the behaviors they had used with observer assessments of that behavior.

The advantages of nonverbal self-reports are many. They are easy and inexpensive to administer (see Duke & Nowicki, this volume), the costs may be lower than for typical behavioral observation, and they provide access to a wide variety of interactions that may not be available any other way (see Keeley, this volume, and Floyd & Mikkelson, this volume). In addition, because participants are not required to report to a research site, they may be more willing to participate. Finally, many self-report measures have strong reliability (see Andersen & Andersen, this volume; Riggio & Riggio, 2001) and are correlated positively with independent observer rating tests.

Despite the benefits of using self-report measures to examine nonverbal communication, a number of limitations to their use also exist. These types of measures can be less precise than coders’ observations and more prone to response biases, such as social desirability. Additionally, participants may recall their own behaviors more accurately than they recall the behavior of the person with whom they are interacting (Metts, Sprecher, & Cupach, 1991; Riggio & Riggio, 2001). Some researchers even question if self-report measures record only “impressionistic” rather than the “actual” nonverbal behaviors (Metts et al., 1991). Furthermore, participants may be overwhelmed by the amount of recording that is required from them. Jones (this volume) points out that participants need to be “motivated to record events conscientiously” but that this can be remedied through training. Participants, also, need to understand what specific behaviors or proximal cues the researcher is looking for, as well as the differences in the various constructs that the participant is expected to measure.

Numerous self-report measures of nonverbal behavior exist—particularly measures of emotional expressiveness—a number of which are discussed in this book. For a more thorough review of a number of these measures, as well as their internal and construct validity, see the section on self reports in PART II.

Observer Assessments of Nonverbal Behavior

Although we can learn much about human interaction by asking communicators to reflect on their own behavior and the behavior of others, such reflections cannot always provide detailed information about the myriad behaviors that are actually enacted as people communicate. If a researcher wants to learn how the behavior of one person influences another or observe differences in how interaction in managed (see Cappella’s discussion of coding, this volume), for example, he or she may need to examine interaction as it occurs.

Observational studies of nonverbal behavior provide insight into a number of communication and relational processes. Researchers have demonstrated, among other things, the ways in which deceptive interactions are shaped jointly by both participants in a deceptive conversation (e.g., Burgoon, Buller, White, Afifi, & Buslig, 1999; see Burgoon, this volume), the way interaction between relational partners is impacted by their respective attachment styles (e.g., Guerrero, 1996), the differences in nonverbal behavior that can distinguish couples whose relationships are successful from those who are not (e.g., Gottman & Levenson, 1992; Manusov, 1995), and the importance of nonverbal behavior in adult-infant bonding and interaction (Bernieri & Rosenthal, 1991; Cappella, 1981). Observational studies focus the attention of researchers on the actual behavior of participants and provide an opportunity for researchers to explore how meanings and feelings are manifest in interaction. Although it is clear that observational studies can reveal many interesting aspects of interaction, the decision to observe behavior is not a simple one.

A researcher who decides to observe behavior is faced with a number of choices about what to observe and how to catalog observations in an observational system. Although it might seem that what to observe would be clear once a researcher has decided on a research question or identified a focus of the study, the decision about what to code actually entails many judgments about what constitutes communicative action and how social interaction is organized (see Bakeman, this volume).

Bakeman and Gottman (1997) noted that coding schemes could be thought to exist along a continuum, with one end anchored by physically based schemes that reflect the organism’s physiology, and the other end anchored by socially based schemes “that deal with behavior whose very classification depends far more on the mind of the investigator (and others) than on the mechanisms of the body” (p. 18). In the discussion that follows, we focus on what we consider to be socially based schemes; that is, observational systems that examine behaviors or messages that have more to do with social categories of interaction (such as smiling or involvement) than with physiological elements of behavior (such as amplitude; for more on physiological measures, however, see Kinney, this volume; Tusing, this volume).

Socially Based Measurement Schemes

One very important decision that researchers make when examining nonverbal communication via a socially based coding scheme is the level of measurement. By this term we mean that researchers make choices about the amount of behavior they will examine within an interaction and the extent to which the assessment involves more concrete indicators of the behavior’s occurrence or more abstract assessments of the social meaning of the behavior. Burgoon and Baesler (1991) discussed this choice as one of micro versus macro levels of measurement. Specifically, they argue that micro level measurement “involves single, concrete behaviors,” whereas macro level measurement involves “larger samples of a given behavior or collection of behaviors” (p. 59). They also noted that macro level measurement typically entails more abstract terms and larger time intervals than does micro level measurement (see Cappella, this volume, on time choices), but they distinguish the level of measurement from the abstractness of what is measured.

In nonverbal research, it is typical for coding of larger amounts of behavior to also involve perceptual judgments that entail interpretation by coders. Likewise, micro level coding tends to involve more concrete assessments of single behaviors. For instance, assessment of involvement typically involves coders’ assessments of the extent to which the relational message is displayed based on examination of a number of behaviors that are used to convey involvement (see Guerrero, 1996, this volume); this choice can be contrasted with coding that assesses the occurrence of a specific behavior such as smiling (Julien, this volume; Segrin, 1992). Thus, we discuss this distinction as one of rating versus coding.

Rating. Rating entails having coders attend to a set of behaviors that comprise the meaning or message conveyed within an interaction. For instance, a number of deception researchers have examined the involvement displayed in interaction (Burgoon et al., 1999; White & Burgoon, 2001). Ratings of involvement reflect observers’ assessments of the degree of involvement displayed (i.e., the meaning), but such assessments are based on careful observation of a set of behaviors that are related to involvement (e.g., eye contact, forward lean, body orientation). Similarly, Knobloch and Solomon (this volume) trained coders to rate the level of conversational equality displayed in interaction based on verbal contributions and nonverbal elements of interaction that contribute to equality of the conversation (such as eye contact, volume/rate of speaking, and gestures).

One advantage of rating is that it tends to make effective use of raters’ time by asking them to make more comprehensive judgments of larger chunks of behavior. Another advantage of this type of coding is that it reflects what Burgoon and Baesler (1991) refer to as isomorphism between the “coder’s [or rater’s] task and the phenomenological experience of communicators” (pp. 60–61). In others words, these types of judgments are similar to the types of judgments made by communicators during interaction, and they take advantage of the ability of raters to understand how behaviors work together to convey social messages. Finally, rating is often sufficient for research projects that seek to understand how social meaning relates to aspects of the relationship or interaction task (see Koenig Kellas & Trees, this volume; Manusov, this volume) or how it influences interaction outcomes (see Buller, this volume; Roberts & Noller, this volume; Trees, this volume).

Rating does, however, limit researchers’ understanding of interaction in important ways. First, ratings require considerable inference on the part of raters; this means that the perceptual judgment being made by the raters must be elaborated clearly, the observers must be well trained, and the reliability of their perceptions must be checked and recalibrated throughout the entire rating process. Additionally, rating does not provide insight into the extent to which specific behaviors contribute to interaction processes (Burgoon & Baesler, 1991). Finally, when behaviors are changing frequently across an interaction, rating may be ineffective because raters cannot capture the dynamic nature of the interaction effectively.

Coding. Coding, on the other hand, provides a level of precision and accuracy that may be very useful when researchers wish to understand how micromomentary actions contribute to interaction dynamics (see Aune, this volume; Chovil, this volume). This type of assessment has been used successfully in research on marital dyads (Buehlman, Carrère, & Siler, this volume; Doohan, Carrère, & Taylor, this volume; Jones, Carrère, & Gottman, this volume; Noller, this volume), mother-infant interaction (Cappella, 1981), initial interaction (Palmer & Simons, 1995), and couples’ communication (see Afifi & Johnson, this volume; Dillard & Solomon, this volume).

One advantage of coding is that it often yields high levels of reliability because single behaviors are examined, and their presence or absence is relatively easy to identify. When a number of separate behaviors are coded, researchers also have the opportunity to combine behaviors in their analyses so that they can examine the influence of individual behaviors and sets of behavior as they co-occur as well as the individual variance contributed by each behavior vis-à-vis the other cues. Finally, when assessments of a behavior are made across time, coding is a good way to capture if and how behaviors change across an interaction.

The disadvantages of coding relate to the fact that typically it does not address the social meaning of a behavior (i.e., the focus is on the behavior’s occurrence). As a result, coded behavior may not predict interaction outcomes as well as rated behavior does. Additionally, coding of behavior is often more time and cost intensive than is rating (Burgoon & Baesler, 1991). Also, although coding specific behaviors allows researchers to combine behaviors for analysis, it is unclear how this should be accomplished, as it seems certain that the impact of different behaviors is not directly “additive.” As a result, the combinations of behaviors that researchers examine may not have relevance to the experience of communicators in interaction.

Choosing Rating or Coding. Cappella (1991) argued that coding and rating (as well as participant judgments and untrained observer assessments) provide “different frames of reference” (p. 111) from which to view what is happening in interaction. He asserted that although each approach yields data that are somewhat different, information from each approach can be reconciled if researchers take into account the differences in point of view that influence what is captured by each approach (see Bernieri, this volume; Dunbar & Burgoon, this volume). Such a view of rating and coding is useful because it reminds us that choosing to rate or code is really a decision about the point of view from which we will see things. Coding provides a tighter focus, fixing our view on specific behaviors, whereas rating shifts our gaze to a wider angle, encouraging us to see the social features of interaction. Researchers’ decisions about which form of observational assessment to use should reflect the aspects of interaction they wish to describe. Of course, the type of communication situations we are studying and the nature of the data we have collected can affect the point of view we select for our analyses.

Selecting and Training Raters/Coders. The issue of coder/rater selection and training has received little attention in writings about nonverbal research. Researchers involved in observational research, however, are aware that coder selection and training is an important aspect of the research process. Although there is not space here to provide a complet...

Table of contents

• Cover Page
• Half Title Page
• Title Page
• Copyright Page
• Contents
• Preface
• Part I Introduction to Nonverbal Communication Research
• Part II Nonverbal Measures
• Part III Paradigms and Practices
• Author Index
• Subject Index