The effect of social presence on the facial and verbal expression of emotion and the...
Transcript of The effect of social presence on the facial and verbal expression of emotion and the...
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
1/15
http://www.gwu.edu/~ed220ri/reading/LeeSP.htm
The effect of social presence on the facial and verbal expression of emotion and
the interrelationships among emotion componentsJournal of Nonverbal Behavior; New York; Spring 2002; Victoria Lee;Hugh Wagner;
Volume: 26Issue: 1
Start Page: 3-25
ISSN: 01915886
Subject
Terms:
Emotions
Women
Nonverbal communication
Interpersonal communication
Abstract:In a study, women were videotaped while they spoke about a positive and a negative
experience either in the presence of an experimenter or alone. They gave self-reports of
their emotional experience, and the videotapes were rated for facial and verbal expression.
Full Text:Copyright Human Sciences Press, Inc. Spring2002
[Headnote]ABSTRACT. Women were videotaped while they spoke about a positive and a negative experience either in the presence ofan experimenter or alone. They gave selfreports of their emotional experience, and the videotapes were rated for facialand verbal expression of emotion. Participants spoke less about their emotions when the experimenter (E) was present.When E was present, during positive disclosures they smiled more, but in negative disclosures they showed less negativeand more positive expression. Facial behavior was only related to experienced emotion during positive disclosure whenalone. Verbal behavior was related to experienced emotion for positive and negative disclosures when alone. These resultsshow that verbal and nonverbal behaviors, and their relationship with emotional experience, depend on the type ofemotion, the nature of the emotional event, and the social context.
KEY WORDS: nonverbal behavior; emotion; social presence; facial expression; verbal expression.
Understanding the interaction among different components of emotional states is necessary for a
comprehensive account of emotion (Nichols & Zax, 1977). In this paper, we examine the
interrelations of three components of emotion: facial, verbal, and experiential. We also examine
how these interrelationships depend on the social situation in which the emotion is experienced,
and on the valence of the experienced emotion.
Interrelations Among the Components of Emotion
It has widely been assumed, tacitly or explicitly, by those researching expressive behavior that
people vary on a continuum of general expressiveness; that is, expression in one channel isaccompanied by expression in other channels. Among those who have made this explicit,
Manstead (1991) wrote, "Some persons are highly expressive, making their feelings apparent
through facial expressions, gestures, tone of voice, and what they say, whereas others are much
more reserved" (p. 285). Similarly, Halberstadt, Crisp, and Eaton (1999) wrote, "Expressiveness
is a persistent pattern or style of exhibiting facial, body, vocal, and verbal expressions . . . Our
judgments about a person's style of expressiveness are based on aggregates of that individual's
facial, body, vocal, and verbal expressions over time and across situations" (p. 110). The
Affective Communication Test (ACT; Friedman, Prince, Riggio, & DiMatteo, 1980) assesses
general expressiveness using a number of self-report items relating to various modes of
expression and various situations. In a related field, the postulated relationship between
expression (or, conversely, inhibition) and health has been investigated operational izingexpressiveness in different ways, including facial (e.g., Malatesta, Jonas, & Izard, 1987) and
verbal (e.g., Pennebaker & O'Heeron, 1984). Thus, expression should covary in different
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
2/15
expressive channels, degree of expressiveness should be persistent and consistent, and the effects
of inhibiting expression should not depend on the channel of expression.
Considering how widespread this assumption is, there is little direct evidence that expression in
different channels is correlated. Buck, Miller, and Caul (1974) found a nonsignificant correlation
of .32 between the frequency of "personal" verbal responses to affective slides and their measure
of facial communication. However, this correlation diminishes to .15 when the effect ofcovariation with skin conductance responses is partialed out. Indirect evidence of a lack of
association comes from a study by DePaulo, Blank, Swaim, and Hairfield (1992). They used the
ACT to identify people who were "dispositionally expressive and unexpressive." Participants'
behavior while they attempted to appear natural, or more or less expressive than usual was
videotaped, and judged for expression by viewers who saw the full audiovisual record, visual
only, audio only, or read a verbal transcript. All effects were smaller for the last two judgments
than for judgments involving nonverbal channels, giving reason to suppose that there is some
lack of generality of expression. One aim of the present research was to investigate the
relationship between verbal and facial expression during the oral description of positive and
negative emotional experiences.
The second aim of this research was to investigate the relationship between the facial and
experiential components of emotion. Most theorists have held that there is generally a positive
relationship between the two, although this relationship can be modified by situationally
determined display rules (e.g., Buck, 1984; Ekman, 1972; Izard, 1991). However, the evidence
for a positive relationship is equivocal. In Ekman, Friesen, and Ancoli's (1980) study,
participants were shown positive and negative films, and the results suggest a positive
relationship between emotional feelings as reported by the participants and their facial activity as
measured by the Facial Action Coding System (FACS; Ekman & Friesen, 1978). However, it
should be noted that Ekman et al. (1980) failed to report data for a film that produced equally
strong positive affect but little smiling. Wagner, MacDonald, and Manstead (1986) found some
support for this relationship by showing that observers could identify from videotaped facial
expressions the emotions reported by people viewing emotional stimuli. However, this coherence
of emotion terms applied to anger, disgust, surprise, and happiness, but not to fear or sadness.
Furthermore, they found no relationship between intensity of reported emotion and rated
intensity of facial expression. Rosenberg and Ekman (1994) used films intended to elicit disgust
and fear, and cued participants to report momentary emotions a number of times during each
film. Their results showed coherence between a wide range of self-reported emotions and FACS
codings of the videotaped facial behavior at those times. However, only half of the emotional
facial expressions were accompanied by subjective feelings. Jakobs, Manstead, and Fischer
(1999) found that people who smiled while viewing humorous films reported more positive
emotions than those who did not smile.
In contrast, Fridlund (1991) and Fridlund, Kenworthy, and Jaffey (1992) reported that the
intensity of facial expression during viewing of humorous or sad films was not related to
reported intensity of emotional experience. Fernandez-Dols, Sanchez, Carrera, and Ruiz-Belda
(1997) compared emotion self-reports and FACS codings of facial behavior during film viewing,
and found that for only two of 35 participants did the FACS codings match those predicted for
the intense emotions reported. Jakobs, Manstead, and Fischer (1996) found no significant
relationships between experienced emotions and the expected facial displays in people viewing
sad films. Reviewing this literature, Manstead, Fischer, and Jakobs (1999) concluded that
whether a relationship between self-report and facial display exists and how strong it is depends
upon the intensity of the emotional stimulus, the type of emotion, and the social context.
The Effect of Social Presence on Emotional Expression
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
3/15
The majority of research on facial behavior has examined individuals in isolation from other
people. Often, this has been with the intention of investigating the nature of "spontaneous" facial
expressions, unmodified by social influences (e.g., Buck, Savin, Miller, & Caul, 1972), or
because the focus of interest has been to isolate the expressions of "basic" emotions (e.g.,
Ekman, 1989). In recent years, however, greater attention has been paid to facial expression, and
nonverbal behavior in general, in social situations (e.g., Lee & Beattie, 1998). Wagner and Lee
(1999) have argued that, in order to understand nonverbal behavior in social situations we muststudy it together with the verbal behavior that accompanies it. Thus, the social situation is crucial
in understanding not only facial behavior, but also its relationship with verbal behavior.
Wagner and Lee (1999) reviewed research comparing emotional facial expression when people
are alone or in social situations. These studies have produced conflicting results. On the one hand
is a series of results suggesting that social presence increases facial expression (e.g., Andrus,
1946; Bainum, Lounsbury, & Pollio, 1984; Chapman, 1973; 1975; Chapman & Wright, 1976;
Fridlund, 1991; Kraut & Johnston, 1979; Young & Fry, 1966). For example, Chapman and his
colleagues showed that 7- and 8-year-old children smiled and laughed more at funny stories
when another child from their own class was present as an audience than when alone. They
laughed even more when the other child was a fellow listener, but less when the companion wasolder or engaged in a different task. In contrast, a number of studies show that social presence
inhibits facially expressive behavior (Friesen & Ekman, see Ekman, 1972 and Friesen, 1972;
Kilbride & Yarczower, 1980; 1983; Kleck, Vaughan, Cartwright-Smith, Vaughan, Colby, &
Lanzetta, 1976; Kraut, 1982; Yarczower, Kilbride, & Hill, 1979). For example, the studies of
Kilbride and Yarczower (1980) showed that the presence of experimenters inhibited children's
facial expression compared to when they were alone. Wagner and Lee (1999) pointed out that
studies in these two groups are distinguished by two factors: the role of the other person, and the
relationship between the people involved. They review further studies showing that when the
other person is a friend facial expression is increased, but when the other is a stranger expression
tends to decrease (e.g., Buck, Losow, Murphy, & Costanzo, 1992; Wagner & Smith, 1991).
Further systematic investigation of the effects of social presence on expressive behavior stems
from Fridlund's (1991) argument that sociality does not necessarily require the physical presence
of another person. He showed that people viewing humorous films believing that a friend was
viewing the same film in an adjacent room smiled just as much as those whose friend was
viewing with them. Fridlund et al. (1992) showed a similar effect for sad expressions during
viewing of sad films. Jakobs et al. (1999), in a modified and extended replication of Fridlund's
(1991) study, showed that both stimulus intensity and level of sociality had a distinct impact on
smiling. It would seem, however, that the impact of social context was limited to the physical
presence versus absence of others: whether the friend was engaged in the same or a different task
made no difference to the amount of smiling. Jakobs, Manstead, and Fischer (1996) conducted asubsequent study, which was procedurally similar to their first except that this time they used sad
film clips rather than happy ones. Those participants who viewed the sadness evoking film clips
alone showed less smiling and more negative emotional displays than did those viewing the
same clips in a more social situation.
The finding that the alone situation evoked more fear and sadness displays than did any of the
four social conditions could be seen to be consistent with the operation of a display rule that one
should conceal sadness or fear in situations where others are present. Such findings are similar to
those of the Japanese-American study of Friesen and Ekman (Ekman, 1972; Friesen, 1972).
Their study is, of course, usually cited as an example of the operation of a cultural display rule,
namely a Japanese display to mask negative affect resulting from viewing distressing films witha smile in the presence of another person. However, Wagner and Lee (1999) pointed out that in
both cultural groups studied, the presence of another person produced increased variation in
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
4/15
facial behavior. Specifically, in both groups the presence of an interviewer resulted in a decrease
in facial expressions of a type we might call "appropriate" to the stimulus material, and an
increase in expressions that are "inappropriate" to the stimuli, predominantly smiles.
Overview of the Present Study
In the present study, female participants recounted positive and negative emotional events eitherin the presence or absence of a female experimenter. The experimenter was known to all
participants, and would not be seen as an authoritative or threatening a figure. All talks were
videotaped, and judgments were made of facial and verbal expressiveness. The following
hypotheses were investigated:
Hypothesis 1.
The facial and verbal components of emotional expression will be positively related. As we saw
above, there is little direct evidence for this when people are alone, so the prediction in that
condition is based on the usual assumption of a trait of general expressiveness. When
participants are talking to the experimenter, we expect a positive relationship because verbal andnonverbal channels should work together to aid communication. However, it is not clear that we
should expect this for both positive and negative expression in both positive and negative talks.
Hypothesis 2.
Those who recount positive emotional events in the presence of the experimenter will show more
positive facial expression (e.g., smiling) than those recounting similar events in the absence of
the experimenter. This hypothesis is derived from the findings of Jakobs et al. (1999).
Hypothesis 3.
Those who recount negative events in the presence of the experimenter will evidence more
positive and less negative facial expression of emotion than those recounting similar events
alone. This prediction follows from the results of jakobs et al. (1996), and Wagner and Lee's
(1999) interpretation of Ekman (1972) and Friesen (1972).
Hypothesis 4.
During positive talks there will be a significant correlation between positive facial
expressiveness and selfreported positive emotion for participants in the alone condition.
Method
Participants
Fifty-four female' undergraduate students from the University of Manchester, UK participated in
this experiment, and received course credit for doing so. Their ages ranged from 18 years to 48
years, with a mean age of 20.5 years (SD = 4.8). Twenty-eight participants were randomly
assigned to talk about their positive and negative emotional events in the presence of the female
experimenter, while 26 recounted similar events alone. All participants would have recognized
the experimenter as a teaching assistant.
Procedure
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
5/15
Female undergraduates were approached individually by the experimenter and asked if they
would be willing to participate in a study on emotional disclosure, in which they would have to
talk about both the most positive and negative emotional events they had experienced in the past
year. Those who agreed to do so were recruited to come along to the laboratory at a designated
time within that same week.
Upon arrival at the laboratory, the participant was seated in a comfortable armchair facing asecond chair approximately 2 m away. The experimenter chatted with the participants to help
them feel at ease. To one side of the second chair was a table on which a miniature video camera
was hidden among a number of files, tape-cassette and videocassette boxes. The camera gave an
almost full frontal, head and torso view of the participant. Hidden wires passed the signal to a
VCR in an adjacent room. There was also a tape recorder on the table, along with a number of
blank tapes, which were loaded and switched on and off before and after each talk. The
participant was fitted with a tie clip microphone which ostensibly led to the audio recorder, but
which in fact also led to the VCR. Before proceeding, the experimenter left the room, seemingly
to collect response sheets, and started the VCR.
For those in the presence condition, the experimenter sat in the chair facing the participant andinformed her that she should talk to the experimenter for 2 min about both a positive and a
negative emotional event experienced within the past year. The order in which the positive and
negative events were recounted was counterbalanced. It was explained that if she were to finish
before being told that the time was up, she should think about the event and decide if there was
anything more that she would like to add. In the absence condition the participant was informed
that she should talk about a positive and negative emotional event that she had experienced
within the past year. The experimenter would leave the room while she did so, and would return
after 2 min. She was told to start talking as soon as the experimenter had left the room, and that
if she finished before the time was up she should think about the event and decide if there was
anything more that she would like to add. The order in which the positive and negative events
were recounted was again counterbalanced.
In both experimental conditions, after the end of each talk period, participants completed a self-
report questionnaire assessing the extent to which they had experienced specific emotions while
talking about the events, that is their emotional response in the laboratory. Ratings were made for
five emotions (happy, sad, angry, afraid, disgusted), each on a five-point Likert scale (0 = not at
all to 4 = very much). To preserve the spontaneity of participants' behavior, they were not
informed about the covert videotaping until the experiment had been completed in its entirety,
when they were extensively debriefed. Permission to use the videotaped recordings was
solicited, and all participants gave consent. Anonymity and confidentiality regarding collected
data were assured.
Ethical Considerations
The procedures used in this study raise two ethical issues. The study was designed following the
Ethical Principles of the British Psychological Society, dating from 1992 (see BPS, 2000), which
are broadly similar to the Ethical Standards of the American Psychological Association (APA,
1992). It was considered and approved by the Research Ethics Committee of the Psychology
Department of the University of Manchester, itself operating within the BPS Principles.
Coercion of participants. The same investigator personally recruited participants, conducted the
laboratory study, and debriefed them. For most of the participants, she served as an assistant instatistics classes, helping them individually with their learning. She was not involved in any way
with the assessment of the students, and it was made clear to them that they were completely free
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
6/15
to opt out of participating in this study. They were aware that alternatives were available, in the
form of participation in studies for other researchers, or of completing an assignment on
experimental methodology. These steps protected participants in accordance with Standards
6.09, 6.11 (b), (c), and (d) of the APA Ethical Standards. Although all participants appeared to
be keen to help, in hindsight it would have been preferable to avoid any suggestion of coercion
by having recruitment conducted by a person with no teaching contact with the participants.
Covert videotaping. Standard 6.13 seems to expressly forbid filming or recording of participants
without their consent in this type of study. However, after considering the BPS Ethical
Principles, and after consultation with colleagues and members of the Research Ethics
Committee, it was considered that the scientific aims of the study would be compromised by
revealing the videotaping beforehand, and that this partly justified the withholding of this
information (in accordance with APA Standards 6.12 and 6.15(a) and (c)). In addition,
participants were informed that they were being audiotaped, so they were aware that they did not
have privacy during the recording sessions. Participants were allowed to view the video
recording after the completion of the study, and were told at that point that they could opt to have
the recording erased, without prejudice to the credits they had earned for participating. Finally,
Principle 4.1 of the BPS Ethical Principles states, "The withholding of information or themisleading of participants is unacceptable if the participants are typically likely to object or show
unease once debriefed." Over many years of experience with covert recording, requests to have
the recordings destroyed have been extremely rare, and a strong objection is almost unknown,
even when the researchers have not been in a position of authority. On the contrary, the typical
response has been interest and/or amusement. In view of these points, the Research Ethics
Committee agreed that the procedures adopted in this study were acceptable.
Data Reduction
Verbatim transcripts were made of every talk session from the videotaped recordings. Two
female judges, psychology graduate students blind to the hypotheses of the study, independently
read the transcripts and, for each one, made two ratings: the extent to which the participant
verbally expressed positive emotion, and the extent to which she verbally expressed negative
emotion. The ratings were made on five-point Likert scales (0 = not at all, 4 = very much).
Global ratings of positive and negative emotion, rather than individual ratings of specific
emotions, were given in order to coincide with the facial expressiveness data (see below). One
judge scored the transcripts in their recorded order, the second scored in reverse order. A few
weeks after reading and rating the transcripts the same two judges independently viewed the
videotaped recordings of the talk sessions. judges watched each recording twice, without sound,
and made one rating after each, again on five-point Likert scales (0 = not at all, 4 = very much):
first, the extent to which the participant showed positive facial expression, and second the extentto which she showed negative facial expression. One judge rated the videotapes in their recorded
order, the second in reverse order.2
Results
Content of Disclosures
Among the positive talks, 20 related to academic achievement, and 20 to being with friends or
family. Of the negative experiences recounted, 23 concerned relationship problems and
relationship breakdown. Others included deaths of family or friends, being assaulted, and leaving
home for the first time.
Data Analysis
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
7/15
Effective reliabilities of the ratings of the two judges were computed for each measure for each
event type using the Spearman-Brown formula (see Rosenthal, 1982). They ranged from R = .61
for ratings of facially expressed negative emotion during the negative talk, to R = .84 for facially
expressed negative emotion during the positive event. For each measure the mean of the judges'
ratings was computed and used in all analyses. At the end of each talk period, participants rated
the extent to which they had experienced specific emotions (happy, sad, angry, afraid, disgust)
while recounting the emotional experience. "Happy" ratings were used as a measure ofexperienced positive emotion, and the remaining four emotion categories were averaged to give
a measure of experienced negative emotion.
Manipulation Checks
It was necessary to establish that emotions appropriate to the specific talk conditions were
recounted and that, comparatively, other emotions were not talked about to a significant degree.
Mean verbally expressed emotion in all of the conditions of the study is shown in Table 1.
Planned comparisons were performed with an alpha of .05, and showed that verbal expression of
negative emotion was greater in negative than in positive talks, t (53) = 11.52, p < .0001, and
was greater than positive expression in negative talks, t (53) = 16.80, p < .0001; verbalexpression of positive emotion was greater in positive than in negative talks, t (53) = 12.05, p < .
0001, and was greater than negative expression in positive talks, t (53) = 9.73, p < .0001. Thus, it
was established that participants verbally expressed the appropriate emotions, and did so more
than they expressed inappropriate emotions.
TABLE 1
TABLE 2
It was also necessary to establish that the presence and absence conditions did not directly affect
the participants' emotional experience during the talks they gave. A 2 (E present vs. E absent) x 2
(positive talk vs. negative talk) X 2 (positive emotion vs. negative emotion) ANOVA, with
repeated measures on the last two factors, was conducted on self-reported emotion. Table 2
shows the mean experienced emotion for each condition of the study. No effects involving E
presence were significant, all Fs < 1.00. Thus, for both the positive and negative disclosure
sessions, there were no differences between the two experimental conditions in terms of the
extent to which participants reported feeling positive and negative emotion while recounting the
emotional experiences.
Finally, global ratings of expression might be subject to distortion if participants in different
conditions spoke for different lengths of time. For example, if people spoke for less time about
negative than positive topics, their expression might make less of a global impact on the raters.
To find if this was a problem in this study, the mean talk durations were examined. Mean
durations were: positive/E present-112.9 s; positive/E absent115.9 s; negative/E present-114.8 s;
negative/E absent 111.6 s. There are no significant differences among these means, and we are
justified in concluding that there is unlikely to be an effect of talk duration on ratings.
Effects of Social Presence on Behavioral Responses
Mean ratings of facially expressed emotion for the conditions of the experiment are shown inTable 3. Ratings were submitted to a 2 (E present vs. E absent) x 2 (positive talk vs. negative
talk) X 2 (positive emotion vs. negative emotion) ANOVA, with repeated measures on the last
-
8/7/2019 The effect of social presence on the facial and verbal expression of emotion and the interrelationships among emo
8/15
two factors. This showed a significant main effect of emotion, multivariate F (1, 52) = 10.42, p