Presence
in Mediated Instruction:
Bandwidth,
Behavior, and Expectancy Violations
Department
of Computer Science
Joseph
B. Walther
Department
of Language, Literature, & Communication
Rensselaer
Polytechnic Institute
Karen
P. Swan
The
State University of New York at Albany
Presented to the annual
meeting on Asynchronous Learning Networks
November, 2001
Orlando
Abstract
This paper proposes an integrated model
of the development of social presence in mediated communication whereby the
level of intimacy in a mediated interaction is a function of the bandwidth and
immediacy behaviors. In such a model,
the affective reactions of users should be a joint function of the degree to
which the combined effects of bandwidth and immediacy violate equilibrium, with
respect to the reward valence of the communicator(s) involved. The paper reports on preliminary research
that supports an equilibrium model with respect to bandwidth and immediacy behaviors
and proposes research that would investigate equilibrium taking into
consideration reward valence as well.
Among the concerns facing distributed education using telecommunication media is the degree of presence or “personalness” that students at a distance experience. Whether through interaction with peers in constructivist group-based learning processes, through interactive web-based content delivery, or through video-mediated interaction, the degree to which students experience that almost intangible aspect of the traditional classroom—the presence of others—is often seen as a critical affective dimension that is requisite to effective learning.
Presence has long been a central topic in the evaluation of telecommunication systems. Short, Williams, and Christie’s (1976) seminal work on this topic defined the construct as pertaining to the involvement, warmth, and immediacy that communicators experience during interaction that may be limited by the bandwidth, or cue-carrying capacity, of various telecommunication systems. While Short et al. (1976) focused on audio- and video-teleconferencing, the social presence construct has been imported into theorizing Internet-related communication systems as well (e.g. Rice & Case, 1983). Early research contended that presence is diminished by text-based computer-mediated communication, since it is devoid of the visual and vocalic cues that higher-bandwidth and face-to-face systems provide. Recent research has challenged this contention (e.g. Fulk, Schmitz, & Steinfield, 1990) or established its validity in certain specifiable conditions related to the longevity of online relationships (e.g. Walther, Slovacek, & Tidwell, 2001). A new approach to the presence/bandwidth question has been tendered in the idea of a medium’s “affective channel capacity” (Picard, 1997). Affective channel capacity is defined as how much affective or emotional information a channel lets through as compared to the total amount of information is passed. It may or may not be synonymous with bandwidth. Pointing a camera at a wall uses the bandwidth, but does not transmit any affective information.
Among the challenges to the notion that bandwidth alone determines presence has been a growing recognition that the active behaviors of telecommunicators are quite amenable, and spontaneously adaptable, to the communication of such functions as may be referred to as presence. The Social Information Processing Theory of mediated communication (Walther, 1992) contends that even in strictly textual media such as email or computer conferencing, participants are able to acclimate to the restrictions of the media by adapting their language behavior to fulfill the functions of missing nonverbal cues in such ways as to build interpersonal impressions, build mental models of their colleagues, and develop relationships marked by affective exchange. While several mechanisms have been suggested to accomplish these functions, little research has concentrated on a theoretical specification of the interdependent roles of such variables as immediacy behaviors, bandwidth, equilibrium and communicator expectations and how these three dimensions interplay to predict students’ affective reactions. This research attempts to bring together principles from each of these domains, and develop a model of behavioral, technical, and cognitive/affective systems that interact in predictable ways in the production of student affect in distributed education settings.
Bandwidth, while technically
speaking refers to the speed and information-carrying capacity of an electronic
conduit, has come to represent the number of communication cue systems that
different media may carry.
Low-bandwidth systems are restricted to digitally-encoded text alone,
whereas higher bandwidth systems allow audio (voice). Even higher bandwidth may describe video or two-way video. While technically not correct, face-to-face
communication has the greatest “bandwidth” of all. According to social presence theory, the higher the bandwidth,
the more the salience there is of another person involved in a conversation,
and the greater the warmth, liking, and involvement of participants (Short et
al., 1976). Social presence theory
posits this relationship regardless of the content or other dynamics of communication
episodes.
Immediacy
Immediacy is a set of interchangeable behaviors which, when enacted, increases involvement and connotes liking (Mehrabian, 1971). These behaviors may be enacted nonverbally or verbally. Nonverbally, they include adopting a close interpersonal distance, forward lean, touch, facial expressiveness including nodding and smiling, and warm vocal tones (Burgoon et al., 1984). Verbally, they include linguistic constructions that indicate direct involvement in an action by the speaker, active participation, and unqualified attitudes; spatio-temporally indicative demonstratives, denotative specificity, selective emphasis, and “agent-action-object relationships” (Wiener & Mehrabian, 1968, Witt, Wheeless, Reyna, and Swigger, 2000). Similar constructs such as involvement include spending time with a partner. In an electronic mail environment, a fast reply to a task-oriented message also indicates immediacy Walther & Tidwell, 1995).
Equilibrium is a normative
level of interaction involvement, violation of which causes reciprocal or
compensatory reactions, more often the latter (Argyle & Dean, 1965).
Communicators adopt levels of gaze, physical proximity, and other behaviors
indicative of intimacy that are derived partly from cultural norms (Argyle
& Cook, 1976), as well as from a need for affiliation. When one
conversational partner disrupts equilibrium through one nonverbal channel
(e.g., proxemic distance reduction) the other attempts to restore it through
the same or an an alternative channel (e.g., reduced gaze) in order to maintain
desired levels of intimacy. For
example, when one partner adopts an interpersonal distance closer than normal,
the other partner may attempt to move back a similar distance. Likewise, when one partner moves away, the
other may move forward.
When considering mediated
communication, it is extremely important to consider whether immediacy and
equilibrium may be exchanged through complementary rather than through
identical cues. Walther (1992) argues
that it is likely that the relative lack of nonverbal cues in computer-mediated
communication may prompt accommodation via verbal mechanisms (Figure 1)—people
make more explicit in text their feelings and attitudes--but limited research
has found that equilibrium restoration takes place across channels. Donohue,
Diez, Stahle, and Burgoon’s (1983) research on verbal and nonverbal immediacy
suggest that face-to-face interactants compensate for reduced nonverbal
affiliativeness with verbal cues, in order to restore “normal” conversational
style. They found that when one
conversational partner reduced nonverbal immediacy (i.e., decreased proximity),
the other partner exhibited significantly greater (spatio-temporal) verbal
immediacy.
Short et al. (1976) were also
aware of equilibrium theory and research.
While they did not embrace equilibrium theory per se, they did speculate
that language may substitute or even “overcompensate” for missing nonverbal
information. Reviewing teleconferencing
research, they suggested that a participant, “aware of the reduced-cue
situation,...will modify his behaviour; thus head-nods indicating agreement may
be replaced by verbal phrases such as ‘I quite agree’....This constitutes a
clear case of interchangeability between non-verbal cues (in this case
head-nods and facial expressions) and verbal messages (in this case explicit
expressions of agreement or disagreement)” (Short et al., 1976, p. 64).
Figure 1
Burgoon’s theory of nonverbal
expectancy violations (NEV) refines the general equilibrium approach,
predicting affective outcomes rather than approach/avoidance behaviors, and
qualifying equilibrium to depend on receivers’ assessments of the partner’s
reward value. Like equilibrium theory,
NEV theory holds that people have normative expectations for interpersonal
distance and involvement. When one
person violates these expectations, the perceiver’s reaction is affected by the
reward characteristics of the violator (see Figure 2). Reward value has to do
with the person’s social attractiveness or status, based on social stereotypes
or relational history, and may be high or low.
A violation may be toward greater immediacy or less (in the case of a
proxemic violation, close or far, respectively).
When a high-reward or
low-reward communicator conforms to expectations (i.e. maintains equilibrium),
no particular interpretive dynamic is aroused; evaluations and reactions to
such persons are derived straightforwardly from the established level of regard
for that person (i.e. high- or low-reward corresponding to a positive or
negative evaluation, respectively).
Figure
2
However, when a person
violates expectations, several dynamics come into play. Theoretically, when a violator deviates from
the expected, a perceiver experiences cognitive and affective arousal and
attempts to make sense of it; drawing on the nature of the violation and the
apparent communicator reward characteristics, interpretations and affective
reactions are derived. When a
high-reward person violates expectations by enacting greater immediacy, this is
interpreted doubly favorably; affective evaluations are more positive, and reciprocation
may occur. When a high reward person
violates expectations by enacting lesser immediacy than normal, this also leads
to a very positive reaction, but may result in compensation (i.e. the perceiver
may try to close the physical/psychological distance). When a low-reward communicator violates
expectations, that person is denigrated, i.e. affective reactions are doubly
negative; if the violation was one of greater immediacy, compensation (i.e.
withdrawal) takes place, whereas if the violation was toward lesser immediacy,
reciprocation (a similar reduction in immediacy) takes place. The NEV theory advances our predictive
abilities beyond that afforded by equilibrium theory, by positing the influence
of communicator reward and the affective outcomes associated with deviations
from baseline equilibrium or expectations.
An Integrated
Model
By importing these theories into the realm of
computer-mediated interaction an integrated model is derived. The level of intimacy in a mediated
interaction may be a function of the bandwidth and the immediacy behaviors of
the communicators, and the affective reactions of users should be a joint
function of the degree to which the combined effects of bandwidth and immediacy
violate equilibrium in respect to the reward valence of the communicator(s)
involved. The following assumptions are
specified based on the forgoing:
·
Assumption 1: There is
an equilibrium value for intimacy in mediated interaction. Violations of equilibrium prompt restoration
or evaluation of the violator.
·
Assumption 2: Behavioral
immediacy and bandwidth are complementary systems in the achievement of
intimacy in mediated interaction.
Derived from these assumptions are the following
propositions:
·
Proposition 1:
Communicators maintain equilibrium through the adjustment of immediacy to complement
bandwidth in mediated communication contexts.
·
Proposition 2:
Unexpectedly high bandwidth/immediacy combinations or unexpectedly low
bandwidth/immediacy combinations result in violations of expectations,
prompting evaluations and affective reactions toward the violator.
We posit that in mediated communication, an
equilibrium level of bandwidth and immediacy is maintained such that as
bandwidth is lower, immediacy must become higher in order for a desirable level
of intimacy to be communicated. Reciprocally,
as bandwidth is higher, enacted immediacy may be lower, to achieve comparable
equilibrium levels of overall intimacy.
If the combined levels of
bandwidth and behavioral immediacy exceed equilibrium, this constitutes a
violation of expectations, and evaluations will accrue as determined by
participant reward level in accordance with the NEV model. Likewise, if combined bandwidth/immediacy
are below equilibrium, a violation of expectations takes place, and will be
processed accordingly with respect to communicator reward. A further implication of this model suggests
that a high reward communicator has great liberty to use lower bandwidth and
immediacy behavior, and yet effect positive reactions—if that individual is
truly rewarding.
To illustrate these propositions, and example may be helpful. In a situation where distance education takes place using text-based computer conferencing only (a low bandwidth system), over time participants will in general develop greater immediacy in their language, and greater familiarity and intimacy in the content, style, structure, typography, and timing of their messages. Analysis should be able to detect a change toward greater intimacy and affection over time. However, in a system in which videoconferencing is used entirely—a high bandwidth system—one should find little change in the intimacy of language and other cues over time. While certainly participants get to know one another and regard each other with somewhat greater familiarity over time, their communication would not be expected to reach the same level of intimacy over time as who use text-based communication alone, since the nonverbal cues automatically conveyed by video will function to create the level of presence that would not exist without being enacted in the text-based system. Both of these predictions illustrate equilibrium models.
Violations of expectations might characterize these
situations. In one, a communicator
(teacher or learner) uses cues connoting a high degree of immediacy communicated
through videoconferencing, exceeding expected intimacy. In such a case, the receivers’ reactions
depend on the reward value of the sender.
If the sender is the teacher (who holds some reward value by virtue of
grading resources) who is in other ways liked (highly qualified, instructive,
enjoyable, and and/or physically attractive), the violation should be evaluated
positively. If the teacher is a low
reward source in general (punitive, dull, questionable qualified, physically
unattractive, etc.) then the reaction should be negative. Similarly, a peer student may be rewarding
(attractive, enjoyable) or unrewarding (unpleasant, unattractive), and that
factor would affect whether another student likes or dislikes a combination of
high bandwidth intimacy (such as video) plus higher immediacy behavior (in
language, duration, frequency, etc.).
The NEV model also includes predictions about under-immediate violations of expectations. An example of such a violation might constitute a teacher who has videoconferencing available and uses a text-based communication system instead, while maintaining the level of behavioral intimacy expected in higher-bandwidth systems.
Results of several previous
studies suggest that the proposed dynamics operate in computer-mediated
discussion, substantiating Proposition 1 (that students maintain equilibrium
through the adjustment of immediacy to complement bandwidth in mediated
communication contexts. One study shows
that students do indeed exhibit immediacy behaviors in mediated, text-based
class interaction; the other attests to students’ achievement of perceived
intimacy levels online comparable to those of face-to-face groups. While neither study directly tests these
reactions in direct response to variations in others’ reward levels—thus the
need for further research—they help establish the foundation for such research
by demonstrating how students compensate for restricted bandwidth through
active and effective behavioral changes.
Swan, Polhemus, Shih and Rogers (2001) developed a coding scheme based on the categories developed by Rourke, et. al. (2001) to explore the equilibrium model in asynchronous course discussions.
Data were collected from the discussions that took place in a graduate level course in Educational Computing given entirely online in the Spring, 2001 semester. The course consisted of four modules that ran sequentially across the semester. In each module, there were three discussions initiated by instructor questions and roughly corresponding to the three weeks students were directed to spend working in each module. Students were required to submit one response to the instructor prompt and two responses to their classmates in each discussion. They could, of course, submit as many responses as they liked, and many participated a good deal more than required.
Three categories of immediacy indicators were identified:
· Affective indicators: personal expressions of emotion, feelings, beliefs, and values. The affective indicators we coded for included the use of paralanguage, expressions of emotion, statements of values, humor, and self-disclosure
· Cohesive indicators: verbal immediacy behaviors that build and sustain a sense of group commitment or group presence. Cohesive indicators coded for included greetings and salutations, the use of vocatives, group reference, social sharing, and course reference.
· Interactive indicators: evidence that the other is attending. Indicators we coded for included acknowledgement, agreement, approval, invitation, and personal advice.
Swan et al. found a great many
immediacy/social presence indicators in the online discussions we reviewed, a
total of 1,336 (663 affective, 468, interactive, and 235 cohesive) in 235
postings, or an average of almost 6 indicators per posting. Figure 3 shows the raw numbers of responses
and indicators across modules. Only one
posting had no immediacy indicators in it and that message was not responded
to. We believe these findings provide
evidence that participants in the online discussions we studied made up for the
lack of affective communication channels by employing more immediacy behaviors
in those channels that are available to them.
Finally, research has also demonstrated that student
groups using text-based electronic communication systems achieve intimacy
levels comparable to those of face-to-face student groups. An experiment by Walther and Burgoon
(1992) involved student groups randomly assigned to use an asynchronous
text-based computer conferencing system or to meet face-to-face over a period
of six weeks, to discuss three decision-making tasks. Participants completed measures assessing their perceived
relational communication after each task was completed. Repeated measures analyses of immediacy and
other social dimensions (composure, receptivity, and social orientation) were
expected to show that face-to-face groups developed these attributes more
quickly than mediated groups, whereas mediated groups were predicted to “catch
up” to face-to-face levels over time.
These predictions were partially supported: While all groups on average
experienced greater immediacy over time, there were no significant differences
between mediated and unmediated groups even at the inception of their
discussions. Subsequent research
identified that the anticipation of long-term relationships was one causal
force leading to initial immediacy in both conditions (Walther, 1994), although
anticipated longevity seems to have prompted more immediacy behavior in online
groups than in offline groups.
Altogether, these results suggest that there is an equilibrium-restoring
dynamic taking place using text-based computer-mediated communication: Faced
with the loss of bandwidth, students exhibit immediacy in their communication,
and report normal levels of intimacy, in online interaction.
Figure 3
In order to test whether extending the NEV framework to see if it effectively accommodates bandwidth as an intimacy cue, and in order to see whether these dynamics pertain to distance education, a series of experiments is proposed. Such research is best to be conducted in a real distance education setting. However, the potential effects on students’ affective reactions and hence, the value of the lessons and the course may be affected. Therefore, simulations within actual courses will be conducted. In each of the following cases, a guest lecturer—the same actual person--will be employed within an ongoing course, with variations in that person’s bandwidth and behavior in accordance with the above framework.
LOW REWARD: Guest lecturer version 1 will be introduced to the students as an undergraduate student who took the same course previously, who is filling in for the professor who must be absent. In the video condition, the lecturer will appear unkempt and in informal, wrinkled clothes. The lecturer will be said to be filling in, with no grading authority.
HIGH REWARD: This guest lecturer will be introduced as a young professor who has done exemplary and original research and development on the topic and who won student-nominated teaching awards last year. The lecturer will appear smartly dressed and well groomed. The lecturer will be presented as if s/he will be evaluating the students on their attentiveness and remarks, which s/he will feed back to the course professor.
HIGH IMMEDIACY BEHAVIOR: The lecturer will greet the students using high immediate language and personal references to both his/herself and the students. S/he will present a fifteen minute lecture from notes, and follow that with a twenty minute question and answer period during which s/he will maintain immediate and personal/informal language. The lecturer will conclude by offering the students means by which s/he can be contacted in the future.
LOW/NORMAL IMMEDIACY: The lecturer will greet the students using low-immediacy language and impersonal content, and present a fifteen minute presentation from notes. The presentation will be followed by a twenty minute question and answer period, in which the lecturer will continue to use impersonal/formal and non-immediate language. In the video condition, s/he will maintain a conventional distance from the camera. The lecturer will conclude by announcing s/he is finished.
BANDWIDTH: In the high bandwidth condition,
videoconferencing will be used allowing all participants to be audible and
visible. In the low bandwidth
condition, text-based chat will be used.
Dependent measures will include student assessments of the instructor, and students’ evaluations of other students’ responses, using Burgoon and Hale’s (1987) measures of communication intimacy (including perceived intimacy/immediacy), and Short et al.’s (1976) social presence scales. Additionally, a post-test will be administered that will assess students’ learning of the material that was covered in the lecture and discussion. We predict that immediacy and presence ratings will be greater in cases where high reward instructors violate expectations by using relatively excessive immediacy behavior/bandwidth combinations—either high or low in immediacy and bandwidth—relative to high reward instructors who use immediacy/bandwidth combinations that offset the extremity of either aspect. Moreover, we predict that that intimacy and presence will be lowest for low reward instructors who use high bandwidth/high immediacy or low bandwidth/low immediacy, compared to low reward instructors who use complementary high/low bandwidth/immediacy combinations. As communication and presence are expected to facilitate learning, we expect that post-discussion learning measures will also reflect the predicted directions for intimacy and presence assessments.
In conclusion, this research joins a growing recognition that bandwidth alone is insufficient to predict that affective and educational outcomes accompanying new techniques in educational communication. Moreover, it argues that social presence is only partly but not completely a function of channel capacity. Presence relies to a great extent on the active communication behavior of instructors and students. From that assumption, the question arises as to what kinds of social stimuli can encourage or retard the active adaptation by students to the constraints that technology has the potential to impose. This proposal also parts from traditional communication research through its recognition that bandwidth may function as a potent cue system. As such, bandwidth is a factor that holds the potential for interactions with other stimuli, cues, and behaviors, and the consideration of these factors simultaneously offers a complex but potentially comprehensive understanding of how technology and behavior combine in the prediction of social and educational outcomes using contemporary communication systems.
Argyle, M.,
& Cook, M. (1976). Gaze and mutual
gaze. Cambridge: Cambridge University Press.
Argyle, M.,
& Dean, J. (1965). Eye contact, distance and affiliation. Sociometry,
28, 289-304.
Burgoon, J. K.
(1978). A communication model of personal space violations: Explication and an
initial test. Human Communication
Research, 4, 129-142.
Burgoon, J. K., & Hale, J. L. (1987). Validation and measurement of the fundamental themes of relational communication. Communication Monographs, 54, 19-41.
Donohue, W. A.,
Diez, M. E., Stahle, R., & Burgoon, J. K. (1983, May). The effects of distance violations on verbal immediacy: An exploration.
Paper presented at the annual meeting of the International Communication
Association, Dallas.
Fulk, J., Schmitz, J., & Steinfield, C. W. (1990). A social influence model of technology use. In J. Fulk & C. W. Steinfield (Eds.), Organizations and communication technology (pp. 117-140). Newbury Park, CA: Sage.
Mehrabian, A. (1971). Silent messages. Belmont, CA: Wadsworth.
Picard, R. W.
(1997). Affective computing. Cambridge, MA: MIT Press.
Rice, R. E.,
& Case, D. (1983, Winter). Electronic message systems in the University: A
description of use and utility. Journal
of Communication, pp. 131-154.
Short, J., Williams, E., and Christie, B.
(1976). The social psychology of telecommunications. London: John Wiley.
Swan, K., Polhemus, L., Shih, L-F. & Rogers, D. (2001) Building knowledge building communities through asynchronous online course discussion. Paper presented at the Annual Meeting of the American Educational Research Association, Seattle.
Wiener, M.,
& Mehrabian, A. (1968). Language within language: Immediacy, a channel
in verbal communication. New York: Appleton-Century-Crofts.
Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational perspective. Communication Research, 19, 52-90.
Walther, J. B. (1994). Anticipated ongoing interaction versus channel effects on relational communication in computer-mediated interaction. Human Communication Research, 20, 473-501.
Walther, J. B., & Burgoon, J. K. (1992). Relational communication in computer-mediated interaction. Human Communication Research, 19, 50-88.
Walther, J. B., & Slovacek, C., & Tidwell, L. C. (2001). Is a picture worth a thousand words? Photographic images in long term and short term virtual teams. Communication Research, 28, 105-134.
Walther, J. B., & Tidwell, L. C. (1995). Nonverbal cues in computer-mediated communication, and the effect
of chronemics on relational communication.
Journal of Organizational
Computing, 5, 355-378.
