Broadcasters' nonverbal communication in 9-11
More than words alone:
Broadcasters' nonverbal communication
In the first 24 hours of the Sept. 11 terrorist attacks
Renita Coleman, Ph.D.
Manship School of Mass Communication
Louisiana State University
Denis Wu, Ph.D.
Manship School of Mass Communication
Louisiana State University
[log in to unmask]
220 Johnston Hall
Manship School of Mass Communication
Louisiana State University
Baton Rouge, LA 70803-7202
Submitted to the Visual Communication Division of AEJMC
for the 2003 Annual Conference, Kansas City, MO.
More than words alone:
Broadcasters' nonverbal communication
In the first 24 hours of the Sept. 11 terrorist attacks
A content analysis the first 24 hours of 9-11 shows that TV new journalists
were not objective in their nonverbal behavior. They showed significantly
more positive and negative expressions than neutral expressions despite a
journalistic commitment to objectivity. The time of day mattered, with
broadcasters showing more negative expressions during the second 8-hours of
coverage, as did length of shot with broadcasters who were on camera for
longer consecutive periods showing more positive and negative expressions
through nonverbal channels. There was no significant difference between
male and female journalists, despite studies showing women show more
expressions of emotion than men.
More than words alone:
Broadcasters' Nonverbal Communication
In the First 24 Hours of the September 11 Terrorist Attacks
Ethical norms in broadcasting require journalists to present the news in an
unbiased manner regardless of the journalists' personal beliefs, attitudes,
moods, or emotions (Elliott, 1986; Lewis, 1984; Cohen, 1987). Whether
covering politicians on the campaign trail, routine city council meetings,
dangerous breaking news or emotional events, broadcast journalists are
expected to be calm, detached, and unemotional.
The study of objectivity or bias has had a long and prolific history ever
since Spiro Agnew accused the press of preferential treatment toward his
opponent. Political coverage has been a natural focus of such research,
which also has tended to target almost exclusively the verbal channels of
communication, even when television, a primarily visual medium, is examined.
This study aims to extend the inquiry of objectivity or bias in two ways;
first, it examines coverage of non-political news, specifically the
breaking news of September 11, 2001. Second, it adds to the growing field
of research in nonverbal communication rather than the more traditional
examination of bias only in words.
This study analyzes the nonverbal behavior of broadcasters from four news
networks covering the events of 9-11. Our central research question is
whether broadcasters exhibited significant positive and negative nonverbal
behavior instead of neutral behavior. How successful were professionally
trained broadcasters in controlling their nonverbal communication during
what was the most traumatic and emotional event in recent memory? If
broadcasters are found to exhibit differential expressive displays in their
nonverbal behavior, then we are interested in discovering what factors best
predict a broadcaster's display of biased nonverbal behaviors. The
nonverbal communication of network broadcasters is particularly important
because of the potential for these journalists to affect a large audience
and the effects that nonverbal communication can have on viewers.
This study does not intend to criticize the journalists who covered this
unfolding disaster, often at great personal risk. We acknowledge they did
an exceptional job under unprecedented pressures. But we also realize that
journalists are human and nonverbal displays can be difficult to control,
even for trained professionals. Even though journalists may not intend to
communicate nonverbal messages, they can still have consequences for
viewers. A look at the communication of implicit messages in coverage of
9-11 can be instructive in understanding audiences' reactions.
The study of nonverbal behavior, which is grounded in theories of
perception, has grown in importance since the 1970s (Burgoon, 1980; Babad,
1999). It has been shown that much subtle and implicit information is
conveyed through nonverbal channels in a few seconds or even a fraction of
a second (Rosenthal et al., 1979), and people are quite accurate in
decoding these brief instances of nonverbal communication (Ekman, Sorenson
& Friesen, 1969; Izard, 1977; Burns & Beier, 1973). It is well established
that certain nonverbal behaviors accompany certain feelings and that
nonverbal cues are especially adept at communicating information about
emotion and mood (Edman, Friesen & Ellsworth, 1972).
Additionally, there is a growing awareness of the need to include affect,
or emotion, in studies alongside cognition (Garramone, 1992).
Agenda-setting theory, for example, has come to include an affective
dimension to people's motives for watching television (McCombs, Shaw &
Weaver, 1997). The concept of affective framing expands the theory to
consider production issues such as image size, placement, editing and
most importantly for this study emotional tone. Similar to the framing
concept, the second-level agenda setting effect seeks to capture the impact
of various attributes that accompany the campaign agendas such as
appearance and personal style (McCombs et al., 1997).
Definitions of objectivity or bias are not well differentiated and the two
terms are sometimes used interchangeably (Lichtenberg, 2000). Objectivity
is the overarching concept that informs this study as it is equated with
neutrality and the absence of value judgments (Lichtenberg, 2000). More
specifically, bias has been defined as any systematic slant (Waldman &
Devitt, 1998), and differential amounts of negative, positive, and neutral
content (Moriarty & Popovich, 1991). Scholars have operationally defined
this concept in a number of different ways, including the relative amount
of positive and negative quotes or comments (Robinson & Sheehan, 1983;
Patterson, 1993; Lowry & Shidler, 1995), and the positivity and negativity
of facial expressions (Friedman, DiMatteo & Mertz, 1980; Friedman, Mertz &
DiMatteo, 1980; Mullen et al., 1986; Coleman & Granberg, 2002). We adapt
these definitions for this study to define bias as differential amounts of
negative, positive, and neutral nonverbal behaviors of broadcasters.
Nonverbal behavior is defined as including facial expressions, posture, or
other behavioral cues such as gestures (Englis, 1994).
Effects of nonverbal communication
Research suggests that the nonverbal component of communication is at least
as important as the verbal content (Graber, 1990; Mehrabian, 1968; Argyle,
Alkema & Gilmour, 1971). For example, when verbal and nonverbal messages
contradict, receivers typically believe the nonverbal message (Richmond,
McCroskey & Payne, 1991). Furthermore, of all the nonverbal cues, facial
expressions carry the most information (Mehrabian, 1968). Facial close-ups
are rich sources of direct and inferred information (Graber, 1990) because
they readily reveal mental states (Ekman, 1983). Nonverbal cues contain the
"covert" information that people send out to others (Burns & Beier, 1973,
p. 118). Decades of research have determined valid and reliable ways to
measure nonverbal behavior (Mehrabian, 1972; Clore, Wiggins & Itkin, 1975;
Rosenthal, Hall, DiMatteo, Rogers & Archer, 1979).
Although often unintentional, nonverbal behaviors can have powerful effects
on viewers (Englis 1994). Nonverbal behaviors that express emotion can
influence observers' evaluations and have great potential to generate
powerful similar emotions in viewers (Englis, 1994; McHugo, Lanzetta,
Sullivan, Masters & Englis, 1985).
Most of the research on nonverbal communication focuses on interpersonal
communication. Do these effects of nonverbal interpersonal communication
transfer when the nonverbal communication is mediated by television? The
research on nonverbal communication in media that has been done revolves
around political communication. Many studies show that television-mediated
visual imagery can have an impact on public opinion and voting intention
(Kepplinger, 1982; Kepplinger & Donsbach, 1987; Lanzetta, Sullivan, Masters
& McHugo, 1985; Moriarty & Garramone, 1986; Moriarty & Popovich, 1991;
Rosenberg & McCafferty, 1987; Tiemens, 1978; Tiemens, 1989; Rosenberg,
Bohan, McCafferty & Harris, 1986). Englis (1994) says that comprehensive
models of voting behavior should place importance on the effects of a
candidate's nonverbal behavior.
Other evidence provides ample support for the idea that nonverbal behaviors
of television communicators are potent elicitors of emotions in viewers
(Englis, 1994; Englis, Vaughan & Lanzetta, 1982; Friedman, DiMatteo &
Mertz, 1980; Haley, Richardson & Baldwin, 1984). These nonverbal
expressions evoke emotional reactions in observers, and the emotions
viewers experience generally parallel the emotion of the expresser (Apple &
Hecht, 1982). For example, fear or anxiety is experienced by viewers in
response to nonverbal displays of distress (Englis & Lanzetta, 1993), and
anger is experienced in response to displays of anger (McHugo, Lanzetta,
Sullivan, Masters & Englis, 1985).
Furthermore, even a widely used theory of information processing of
television, the Limited Capacity Model (Lang, 2000), suggests that
emotion-eliciting stimuli may lead to better processing. Emotional message
characteristics may increase the resources allocated. Because visual
information is automatically encoded, it is encoded regardless of the
capacity requirements of the stimulus. Research also shows that visual
information, including "talking heads" of journalists and others, results
in more information being encoded (Lang, 1995). Because nonverbal behavior
is one of the primary ways emotion is communicated, and since nonverbal
behavior is a form of visual information, it seems that emotional nonverbal
displays of newscasters could have the effect of increasing viewers'
processing of journalists' nonverbal messages. This is especially likely
considering that when journalists are on screen as "talking heads" there
are few novel visual images competing for viewers' attention. This study
does not examine the effect of broadcasters' nonverbal displays on viewers,
but theory and previous research suggest that such effects are likely.
Studies of news diffusion after major unanticipated events have rarely
examined what happens to people emotionally after they receive the news
(Kubey & Peluso, 1990), but there is evidence of increased depression,
post-traumatic stress disorder, sleeplessness, and other psychological
problems after 9-11, even in those who did not directly experience the
event but saw it on television (Pew Research Center, 2001).
One other consistent finding regarding nonverbal expressiveness is a gender
effect; it is well documented that women convey emotions better than men
(See Hall, 1984, for a review of the literature; also: Tucker & Riggio,
1988; Wagner, Buck, & Winterbotham, 1993). This is an important
consideration for our study given that at least 25% of broadcast
journalists are female (Weaver & Wilhoit, 1996).1
Such findings about nonverbal communication have important implications
for the study of nonverbal communication in broadcast news. They show that
impressions are more accurate than one would expect, and that people
communicate a great deal of information quickly and without intent.
Nonverbal communication has been studied mainly in the context of political
candidates on television; this study helps extend that to the nonverbal
communication of journalists on television.
There is evidence that people in certain occupations actors for example
are better at decoding and encoding nonverbal behaviors than others, but no
work has been done on journalists (Rosenthal, Hall, DiMatteo, Rogers &
Archer, 1979). Journalists are taught to eliminate bias from their written
and verbal reporting, but much less attention is paid to unintentional bias
in their nonverbal behavior. Broadcast journalists are also instructed to
conceal their personal feelings and attitudes in their facial expressions
and nonverbal behavior. Ethical norms require them to be neutral and
refrain from expressing personal views and emotions (Elliott, 1986; Lewis,
1984; Cohen, 1987). While expressions may be masked with facial management
techniques (Ekman & Friesen, 1975; See DePaulo, 1992, for a review), few
journalists are trained in these techniques. This "leakage" of affect can
be picked up readily in nonverbal behavior (Babad, 1992; Babad, Bernieri &
Rosenthal, 1989; Ekman, 1985).
The fact that broadcast journalists are instructed to be objective in their
display of emotion, attitudes, and bias makes this study even more
important. Most studies of nonverbal communication of emotion examined
expressive displays of school teachers, judges, political candidates, or
others not trained to be objective not journalists (Babad, Bernieri &
Rosenthal, 1991; Babad & Taylor, 1992; Ambady & Rosenthal, 1993; Blanck,
Rosenthal & Cordell, 1985). Does journalists' training and professional
practice enable them to better control unintentional displays of emotion?
Studies of broadcast journalists
Only a few studies could be found that investigated the nonverbal behavior
of TV journalists; all were in the context of political coverage. Babad
(1999) found that Israeli TV interviewers exhibited differential nonverbal
behavior toward politicians they were interviewing, and, in some cases,
they were blatantly preferential toward one candidate. Friedman and
colleagues (Friedman, DiMatteo & Mertz, 1980; Friedman, Mertz & DiMatteo,
1980) found significant differences in the facial expressions of four of
five U.S. network anchors when they referred to two presidential candidates
in the 1976 election. Mullen et al. (1986) replicated the Friedman study in
the 1984 presidential election and found one of the three network anchors
showed significantly more positive facial expressions when referring to one
candidate than the other. Coleman & Granberg (2002) replicated those
studies in the 1996 presidential election and again found biased emotional
displays in the facial expressions in two of five network anchors.
This project extends these studies to examine broadcasters' nonverbal
behavior in a different context coverage of breaking news of a disaster
rather than political campaigns. Political campaigns may be unique in
evoking nonverbal displays in broadcast journalists due to the highly
partisan nature of politics. Journalists' facial displays have not been
studied in other situations, which may be expected to produce different
results. Regular nightly newscasts, for example, may be so routine that
broadcasters have little affect to convey. On the other hand, live coverage
of breaking news may be expected to produce the opposite effect. Despite
their professional training, journalists are human and may well be unable
to voluntarily control the nonverbal communication of strong autonomic
responses such as fear, anger, and anxiety in extraordinarily unusual
situations. In the interest of maximizing comparisons and finding effects
if they do exist, this study examines the nonverbal communication of
broadcasters under the most emotional event in recent memory the
disasters of September 11, 2001.
It is crucial to examine both the explicit and implicit information
conveyed to audiences during this event in order to fully understand the
American public's reaction in the aftermath. Potter and Lang (1999) say
that how the networks handle imagery is critical to reducing stress,
anxiety, and panic in the public, while Utley (1997) says that, in times of
crisis, TV news has the capacity to bring solace to viewers. Studies of
9-11 coverage have examined the verbal content but not the nonverbal; no
content analysis has yet examined whether broadcasters communicated solace
or panic through nonverbal means.
Studies of 9-11
Results of scientific studies of 9-11 coverage are only beginning to
emerge. What has been found is that for millions of viewers, television was
their main source of information concerning the events of September 11,
2001 (Robertson, 2001). One study that is particularly instructive for this
research found that viewers of 9-11 coverage perceived news anchors as
emotionally involved with the event (Litterst, 2002). Participants in two
surveys and two focus groups said the anchors projected sadness and, to a
lesser degree, anger. Viewers were also able to differentiate between
anchors' expressions of fear, anxiety, and tiredness (Litterst, 2002). This
study polled viewers on their perceptions of broadcasters' nonverbal
behavior and did not use trained, objective coders to examine the actual
content of broadcasters' expressions.
Hypothesis and Research Question
Based on studies showing people's limited ability to control their
nonverbal expression of attitudes and affect, it is hypothesized that
broadcast journalists, despite their professional training to be neutral
and objective, would show significant emotion in their nonverbal behavior.
To test this, we examine eight research questions. If any of these show
that broadcasters were likely to show significant non-neutral expressions,
then we want to develop a model that best predicts broadcasters' display of
nonverbal expressions. The first part of the research questions regarding
"Total Nonverbal Expressions" refer to displays that showed either positive
or negative expressions versus displays that did not so were neutral or
non-expressive. The comparison is between expressive displays (positive and
negative combined) and neutral displays. The second part of each research
question tests Valence; these compare expressions three ways positive vs.
negative vs. neutral. The research questions we test are:
RQ1a: Did broadcasters communicate more Total Nonverbal Expressions
(positive and negative combined) than Neutral Expressions during the first
24 hours of coverage of Sept. 11?
RQ1b: Did broadcasters show more positive, negative, or neutral Valenced
nonverbal expressions during the first 24 hours of coverage of Sept. 11?
RQ2a: Was there a significant difference between broadcasters' Total Nonverbal
Expressions and Neutral Expressions according to the time of day? That is,
broadcasters more likely to show emotion early or later in the coverage of
RQ2b: Was there a significant difference among the Valence of broadcasters'
nonverbal expressions according to time of day?
RQ3a: Were broadcasters' Total Nonverbal Expressions significantly related
to length of the shot?
RQ3b: Was the Valence of broadcasters' nonverbal expressions significantly
related to length of the shot?
RQ4a: Was there a significant difference between male and female
broadcasters in their Total Nonverbal Expressions?
RQ4b: Was there a significant difference between male and female
broadcasters in the Valence of their nonverbal expressions?
RQ5a: Was there a significant difference between broadcasters live on the
scene and broadcasters in a studio or set in Total Nonverbal Expressions?
RQ5b: Was there a significant difference between broadcasters live on the
scene and broadcasters in a studio or set in the Valence of their nonverbal
RQ6a: Was there a significant difference among networks in broadcasters' Total
RQ6b: Was there a significant difference among networks in the Valence of
broadcasters' nonverbal expressions?
RQ7a: Was there a significant difference in Total Nonverbal Expressions
between broadcasters in New York and broadcasters in Washington, D.C.?
RQ7b: Was there a significant difference in Valence of nonverbal
expressions between broadcasters in New York and broadcasters in
RQ8: What variables best predict broadcasters' Total Nonverbal Expression
during coverage of the first 24 hours of Sept. 11?
The methodology for this study was a quantitative content analysis
following the methods of Riffe et al. (1998), adapted for visual analysis
following the recommendations of Van Leeuwen and Jewitt (2001). Tapes of
the first 24 hours of coverage of the disasters of September 11, 2001 were
obtained from the Vanderbilt Television News Archives for four networks --
ABC, CBS, NBC, and CNN.
A random sample was generated from the universe of 24 hours. First, the 24
hours were divided into 15-minute increments. Twenty-five percent of the
15-minute units were randomly sampled, yielding 24 units consisting of 15
minutes each. The sample produced a total of 2,069 shots; 663 with
journalists. Because this was a probability sample, results can be
generalized to the entire 24 hours of coverage on the four networks.
The unit of analysis was the shot; the most basic unit of audiovisual
message construction (Gianetti, 1982). A shot is defined as a fragment of
visual material that has no break in continuity of action, that is, video
that does not contain editing cuts (Gianetti, 1982). In a shot, the camera
movement is unedited; if the camera's position changes it may be due to
zooms or pans, but not cuts (Van Leeuwen & Jewitt, 2001). We coded only
shots that lasted at least 4 seconds in order to increase coder accuracy;
typically, instruments that measure nonverbal behavior using segments of 2
seconds have about 84% accuracy; scenes of 5 seconds have a nearly perfect
accuracy rate (Rosenthal, Hall, DiMatteo, Rogers & Archer, 1979).
Following previous studies, nonverbal behaviors were defined as facial
expressions, posture, and gestures (Englis, 1994). In this study, we
measured six nonverbal dimensions that had been tested in previous studies
eyebrows, mouth and lips, head, overall face, overall body, and overall
gesturing (Mullen, 1999; Sullivan & Masters, 1988; Moriarty & Popovich,
1991; Moriarty & Garramone, 1986). Following tests of nonverbal behavior
from psychology, the coders did not judge the emotion shown in they face;
rather, particular movements within a facial area were distinguished
(Ekman, 1972). The movements were coded along a three-point dimension
positive, negative, and neutral. These procedures have been shown to
increase intercoder reliability and measurement validity.
Eyebrows were coded as negative if they were lowered or furrowed; positive
if they were raised or not furrowed; and neutral if they were normal or
expressionless. Mouth and lips were coded negative if the corners were
retracted, tight or frowning; negative if they were raised, retraced and
raised, smiling or laughing; neutral if they were normal or nonexpressive.
Head was coded negative if it was turned from vertical or down; positive if
it was up; neutral if it was normally positioned. Overall face was coded
negative if it was serious, tense, unhappy, or worried; positive if it was
happy, light-hearted, calm, or peaceful; neutral if it was normal or
expressionless. Overall body was coded negative if it was stiff or tense;
positive if it was relaxed; neutral if it was normal or expressionless.
Overall gesturing was coded negative if the journalist engaged in much
gesturing, handwaving, etc.; positive if there was some gesturing, and
neutral if there was none (Mullen, 1999; Sullivan & Masters, 1988; Moriarty
& Popovich, 1991; Moriarty & Garramone, 1986).
This resulted in two interval level dependent variables: The first (Total
Nonverbal Expressions) was an index of the total amount of nonverbal
expressions across the six measures. This DV was a sum of the number of
non-neutral expressions without regard to valence; in other words, both
negative and positive expressions were coded as nonverbal expressions of
emotion, and were summed across the six measures. The second (Valence) was
an index of the valence for the six variables; for each measure, positive
expressions were given a value of 1, negative expressions a value of 1,
and neutral expressions a value of 0. The values of all six measures were
We also coded length of time in seconds that each shot lasted; time of day
when each shot occurred; the city (New York, Washington, D.C.,
Pennsylvania, Foreign, Other) and demographics such as gender, network
(ABC, CBS, NBC, CNN), role (anchor/reporter), and location of journalist
(in the studio or live on scene).2
The only variable suggested by previous studies is gender; however, we
considered it important to examine variables unique to journalists' norms
and routines that have been shown to have significant effects in studies of
other dependent variables (Shoemaker & Reese, 1996). Since journalists have
not been studied often in terms of their nonverbal behavior, this is
exploratory research. It seems intuitive that journalists' nonverbal
behavior would be affected by these variables; for example, whether they
are in the studio watching events unfold from afar or live on the scene of
the tragedy could reasonably be expected to produce different nonverbal
behaviors. We examine these variables in the interest of beginning to build
a body of knowledge that helps us understand and explain the conditions
under which trained, objective journalists can be expected to display
nonverbal expressions that are not neutral.
Two independent coders were trained and then coded 10% of the shots.
Reliabilities ranged from .85 to 1.0 using Scott's Pi. Individual Scott's
Pi's were as follows: Eyebrows = .85; Mouth and Lips = .94; Head = 1.0;
Face = .87; Body = .93; Gesture = 1.0; Gender = 1.0; Journalists' Location
= .93; City Location = .97.
The volume was turned off so that spoken language was not audible to the
coders. This was designed to eliminate possible emotional bias based on
verbal cues (Burns & Beier, 1973). Coders also were able to pause the video
to examine the shots for emotional expression for a longer time than would
occur in normal television viewing. This allowed the coders to consider
longer the subtleties of the facial expression, as well as to avoid any
confusion occurring by a facial expression (Swenson & Casmir, 1998).
This study of the first 24-hour newscast about September 11 generated a
total of 2,067 shots, of which, CNN contributed about 30% of the total
shots, ABC 26%, CBS 22%, and NBC 22%. In addition, a total of 654
journalists from the four TV networks during the time frame and their
nonverbal expressions were coded. CNN had slightly more shots of
anchors/reporters (217) than the three networks (ABC - 164, CBS - 136, and
NBC - 134). These shots and reporters are the basis of our analysis. What
is reported below follows the order of our research questions.
Our first finding indicates that 600 journalists -- out of 654 total --
made at least one nonverbal expression during the first 24 hours of the
September 11 coverage. Seventy-six percent of the reporters expressed
nonverbally about negative emotion; 18% showed no, or neutral, nonverbal
expressions; and 6% had positive nonverbal expressions. This finding may
not be too surprising given the nature of the event; yet the journalistic
tenet of impartiality and objectivity in the news coverage was not upheld
at least on the nonverbal part.
We also were interested in knowing whether broadcasters differed
nonverbally across various times of the day. As reported in our method
section, we divided the 24 hours evenly into three periods and discovered
that during the period of September 11 from 5 p.m. to September 12 at 1
a.m. the average of nonverbal expressions reaches the peak; followed by the
next period 9 a.m. to 5 p.m. on September 11. September 12 from 1 am to 9
pm had the fewest number of nonverbal expressions. In terms of valence of
the nonverbal expressions (see Table 1b), we found that reporters made the
most negative nonverbal expressions during the first period (46%),
immediately followed by the second period (44%).
Table 1 about here.
In the third research question, we asked whether Total Nonverbal
Expressions and Valence differ due to the length of the shot. Pearson
correlation test was conducted and the results are mixed. Even though both
relationships between Total Nonverbal Expressions and Valence and the
length of shot are positive (.156 and .015, respectively), only the former
relationship is statistically significant (p< .001). In other words, the
longer the shot, the more nonverbal expressions were made. Gender was the
key variable in our fourth research question. Our statistical test does not
show any significant difference between male and female broadcasters
regarding the number of nonverbal expressions and the overall valence.
Regarding the association between nonverbal expression in studios or on the
scene, we used ANOVA to examine the potential difference. Our finding
indicates journalists in studios are more likely than live journalists to
make non-neutral expressions, with F(1,1)=123.35, p<.001. However, the
difference of valence between them is not statistically significant, which,
perhaps, is affected by the overall negative expressions.
Of the four channels we examined (CNN, ABC, NBC, and CBS), CNN had the
fewest nonverbal expressions, while NBC yielded the highest average. The
ANOVA test resulted in F(3,1)=20.82, p<.001, showing the four channels are
significantly different. The follow-up Tukey's HSD test (see Table 2) shows
that CNN is statistically different from the three networks -- CNN
broadcasters expressed far less nonverbally. Regarding valence, NBC's
reporters were the most negative (mean score= -2.31), followed by CBS
(1.94), CNN (1.46), and ABC (1.29). ANOVA resulted in significant
differences of valence among the four channels F(3, 1)=13.86, p<.001. The
follow-up analysis using Tukey's HSD indicates that ABC and CNN are
statistically different from CBS and NBC.
Table 2 about here.
As Washington and New York were both venues of the September 11 attacks, we
were interested in finding whether the visual presentations of the two
sites were any different. Our ANOVA analyses indicate that Washington
generated significantly more nonverbal expressions than New York, F
(1,1)=20.04, p< .001; whereas the valences of the two venues' reportage are
not significantly different, F (1,1)=1,249, p= .265, even though New York's
coverage conveyed slightly more negative nonverbal messages than that of
In the last part of our data analysis, we set out to examine the strength
and magnitude of the found predictors of volume and valence of nonverbal
expressions. Our suspect predictors include time category, shot length,
location, gender of reporter, studio setting, all of which were from our
research questions. The variable of broadcasters CNN, ABC, NBC, and CBS
was not included in the test primarily because it would be tautological,
and also the difference among the broadcasters resides on the extent of
nonverbal communication made rather than on the direction.
The five predictors were regressed into the models of amount and valence
of the nonverbal expression with a stepwise method. The only independent
variable that stands out of the regression model for valence is time
category, F (1, 652)=22.96, p <.001, R2=.034, indicating that as time went
on TV reporters were less likely to convey negative nonverbal expressions.
As to the prediction model of the amount of nonverbal expression, the
regression test results in two variables — time category and shot length
(see Table 3). In other words, the earlier the time and the longer the shot
length, the more nonverbal expressions would be made by the TV journalists.
Table 3 about here.
Despite widespread industry commitment to the goal of objectivity and
elimination of bias, broadcast journalists are not always successful in
conveying neutrality in their nonverbal behavior. This study found that
broadcasters overwhelmingly exhibited nonverbal behavior that was negative
or positive rather than neutral during the first 24 hours of coverage of
the September 11 terrorist attacks. It is maybe not surprising given the
enormity of this tragedy that their underlying attitudes and emotions were
revealed nonverbally, or that the nonverbal information they conveyed was
overwhelmingly negative. We purposely chose to study coverage of 9-11
because this event is unlike any other since Pearl Harbor in 1941; if
seasoned journalists who are trained to be neutral were likely to convey
nonverbal behaviors that were not neutral, it surely would be during an
event as horrific as this. But the nature of the event we studied should
not serve to dismiss these findings; consider that numerous studies of
election coverage also show that broadcast journalists convey positive and
negative nonverbal expressions (Babad, 1999; Friedman, DiMatteo & Mertz,
1980; Friedman, Mertz & DiMatteo, 1980; Mullen et al., 1986; Coleman &
Granberg, 2002). Clearly, it is not only emotional events the magnitude of
9-11 that produce such nonverbal displays in journalists. While political
campaigns may no doubt stir partisan attitudes in journalists, the
magnitude of these attitudes should be much less than the magnitude of
emotions evoked by a massive terrorist attack resulting in 3,000 innocent
deaths. This study aimed to extend the findings on broadcasters' nonverbal
communication from one arena political coverage to another, that of
breaking news. That broadcasters' do exhibit valenced nonverbal behavior in
at least two different settings makes it less likely that the content of
the story is the only explanation. That seasoned, network anchors and
reporters were able to convey objectivity through neutral nonverbal
expressions only about 18% of the time is worth noting. We would expect
veteran professionals to be able to control their nonverbal expressions at
least to the level of chance, which is almost twice the rate that they did
These displays of nonverbal behaviors are better understood in the context
of the time of day of the coverage, the length of time broadcasters were on
air in a shot, and their location in a studio or live on the scene. During
the second eight hours of coverage, broadcasters were least able to control
their nonverbal expressions than in the first and third eight-hour periods.
This can be understood through Graber's (1980) three stages of crisis
framework. During the first stage, journalists are focused on describing
what happened, Graber says (1980). Our first time period corresponds to
this stage in Graber's scheme. When the first plane struck the World Trade
Center, journalists covering the news were focused on getting accurate
information and getting it out. It wasn't until the second stage, where
journalists turn toward making sense out of the situation (Graber, 1980),
that they had time to consider the enormity of the situation. Only after
the event unfolded and no more breaking news occurred did journalists begin
to consider the long-term implications of such a massive, planned
undertaking. When the shock wore off in this second period, they began to
feel and express emotions and attitudes.
The time of day when nonverbal expressions began to show most is also
enlightening it was not until the second eight hours, from 5 p.m. on
Sept. 11 until 1 a.m. on Sept. 12, that broadcasters began to show
significantly greater negativity in their nonverbal behavior. Not only does
it make sense that the enormity of the situation had begun to sink in, but
many of these journalists had been covering the story since it broke just
before 9 a.m. This means that they would have put in a full day's work by
now, even on an ordinary, rather uneventful news day. They were simply
tired. It makes sense that tired journalists are unable to exert the effort
necessary to conceal their underlying attitudes and emotions.
Furthermore, the longer broadcasters were on camera in a single shot, the
poorer their control over their nonverbal expressions. This finding can be
seen as a corollary to the fatigue factor explained above; the longer
broadcasters work, the less control they have over their nonverbal
behavior. This makes sense whether they have been working 8 to 16
emotionally exhausting hours, or whether they have been on camera 20 to 60
seconds when less than 5 seconds is the norm.
Some of the findings that may seem counterintuitive include journalists in
the studio showing more nonverbal expressions than those live on the scene,
and that journalists in Washington exhibited more nonverbal expressions
than those in New York. Without theory or findings from previous studies
that speak to these results we can only speculate on the possible reasons.
Adapting Susan Sontag's ideas (1977) about photographic images to these
live images may help us understand. It may be reasonable to assume that
reporters who are routinely in the field reporting on shootings, stabbings,
fatal accidents and such may be more immune to shocking images than
journalists in a studio who do not see such carnage live and up close
daily. Sontag (1977) says that images of atrocities lose the power to shock
when one is repeatedly exposed to them. It may be reasonable to extrapolate
this argument to reporters who witnessed repeated attacks and much greater
devastation and loss of life in New York than those in Washington. Sontag's
(1977) essential argument is that the greater the exposure to horror, the
less the reaction, which is in line with studio- and New York-based
journalists showing fewer nonverbal expressions than those on-scene and in
Finally, we turn to the findings regarding gender. While these findings are
counter to what much previous research shows, we take them as entirely
understandable given the nature of the population studied. Unlike teachers,
judges, couples, and others, the female broadcasters in this sample were
not more likely to reveal emotion in their nonverbal expressions than male
broadcasters. While broadcasters overall may not have been entirely
successful in concealing their nonverbal expressions, men and women had
about the same rate of success (or failure) in this regard. This finding
should help put to rest any lingering concerns over women being able to
report as well as men on certain stories. We are unable to say from this
correlational study whether women broadcasters learn to conceal their
emotions better than women in other occupations, or whether women who go
into broadcasting already have this trait (DePaulo, 1992). We take it as
encouraging that, for once, there is no gender difference when all other
indications say there should be.
There were significant differences between networks; this is congruent with
other research on bias in political coverage (Friedman, DiMatteo & Mertz,
1980; Friedman, Mertz & DiMatteo, 1980; Mullen et al., 1986; Coleman &
Granberg, 2002). The results of this study mirror the findings of the most
recent study of facial expressions of newscasters in an election (Coleman &
Granberg, 2002). In this and the election study, broadcasters at NBC and
CBS showed the most bias in their nonverbal behavior. Conclusions should be
drawn with caution since the two studies differed in many significant ways.
For example, in the election study, only the facial expressions of network
anchors were examined, not all nonverbal behaviors of numerous anchors and
reporters. Also, the valence was different; in the election study, NBC's
and CBS's anchors showed more positive expressions when mentioning one
candidate; in this study of 9-11, NBC's and CBS's journalists showed
significantly more negative nonverbal behaviors. The election study used
untrained viewers; the 9-11 study used trained coders.
The findings from this study give critics of journalism some ammunition
when they say that journalists are not objective at least not in the
nonverbal content of their messages. In this content analysis, the
nonverbal behaviors of broadcasters covering the first 24 hours of 9-11
more often conveyed positive or negative expressions than neutral
expressions. The goal of objectivity in news coverage is deeply entrenched
in journalistic culture, yet this study and others have found that
journalists routinely do not achieve that goal at least nonverbally.
While journalists are taught the value of objectivity, and broadcast
journalists are cautioned to keep their nonverbal behaviors neutral, few
courses or continuing education seminars actually train broadcasters in
facial management techniques that would help them achieve this goal of
objectivity in nonverbal behavior. One outcome of studies such as this
should be to make broadcasters aware of this tendency to betray objectivity
and point out the work in psychology on impression management (see DePaulo,
1992, for a review).
Beyond the implications for whether objectivity is an impossible goal for
journalism, the results of this study raise questions of what effects these
biased nonverbal expressions of broadcasters have on viewers. Such
conclusions cannot be determined by this study, but numerous other projects
have found that nonverbal behavior of journalists can influence viewers, no
matter how well hidden (Moriarty & Garramone, 1986), stir viewers'
emotions, influence the opinions they form (Graber, 1988), and even affect
their attitudes and behavior (Sullivan & Masters, 1988). Future studies
should link the broadcasters' nonverbal messages with their effects on
This project extends the study of broadcasters' nonverbal behavior beyond
the context of political coverage. Future studies should consider other
story contexts as well to discover what types of stories and at what level
of severity are broadcasters better able to conceal their nonverbal
expressions. For instance, it may be hypothesized that routine, nightly
newscasts may lead to more neutral nonverbal communication but that other
breaking news events reduce the likelihood of broadcasters' control of
their nonverbal expressions. Future studies could also compare network
newscasters to local TV journalists, and market size could also be a
factor. Comparative data are needed to generalize beyond a single event.
While future research should provide answers to some of these questions,
the study reported here contributes substantially to a better understanding
of the way broadcasters communicate using more than words alone. By
studying different story content, and also using trained coders and
objective measures instead of untrained viewers, these results help confirm
the proposition that nonverbal displays of broadcasters vary considerably,
and that journalists are not always objective, despite their best efforts
and commitment to journalistic values.
1. These data were collected in 1992; it is likely that women represented a
higher percentage of broadcast journalists in 2001. In fact, our sample
consisted on 31% female journalists.
2. Significant correlations between role and location of journalist
suggested multicollinearity. Being an anchor and being in the studio were
significantly and positively related as was being a reporter and being live
on the scene. We eliminated the anchor/reporter variable since it seemed
more plausible that a journalists' nonverbal expressions were related to
his or her physical location experiencing the event in person at the
scene or vicariously through video in the studio than to his or her job
description as an anchor or reporter. In fact, sometimes reporters
performed the duties of anchor during 9-11 coverage.
Ambady, N. & Rosenthal, R. (1993). Half a minute: Predicting teacher
evaluations from thin slices
of nonverbal behavior and physical attractiveness. Journal of Personality
Psychology, 64, 431-441.
Apple, W. & Hecht, K. (1982). Speaking emotionally: The relation between
verbal and vocal
communication of affect. Journal of Personality and Social Psychology, 42,
Argyle, M.; Alkema, F. & Gilmour, R. (1971). The communication of friendly
and hostile attitudes
by verbal and nonverbal signals, European Journal of Social Psychology, 1:
Babad, Elisha. (1999). Preferential treatment in television interviewing:
Evidence from nonverbal
behavior. Political Communication, 16:337-358.
Babad, Elisha. (1992). Teacher expectancies and nonverbal behavior. In R.
Applications of nonverbal behavioral theories and research (pp. 176-190).
Hillsdale, N.J.: Lawrence Erlbaum.
Babad, Elisha & Taylor, P. (1992). Transparency of teacher expectancies
cultural boundaries. Journal of Educational Research, 86, 120-125.
Babad, Elisha; Bernieri, F. & Rosenthal, R. (1989). Nonverbal communication
leakage in the
behavior of biased and unbiased teachers. Journal of Personality and Social
Psychology, 56, 89-94.
Babad, Elisha; Bernieri, F. & Rosenthal, R. (1991). Students as judges of
teachers' verbal and
nonverbal behavior. American Educational Research Journal, 28, 211-234.
Berry, Diane. (1991). Accuracy in social perception: Contributions of
facial and vocal information.
Journal of Personality and Social Psychology, 61(2):298-307.
Blanck, P.; Rosenthal, R. & Cordell, L. (1985). The appearance of justice:
Judges' verbal and
nonverbal behavior in criminal jury trials. Stanford Law Review, 38, 89-164.
Burgoon, Judee K. (1980). Nonverbal communication research in the 1970s: An
overview. In Dan
Nimmo, (Ed.) Communication Yearbook IV. New Brunswich, NJ: Transaction
Books, p 179-197.
Burns, Kenton L. & Beier, Ernst G. (1973). Significance of vocal and visual
channels in the
decoding of emotional meaning. Journal of Communication, 23, March, p. 118-130.
Clore, G.L.; Wiggins, N.H. & Itkin, S. (1975). Gain and loss in attraction:
nonverbal behavior, Journal of Personality and Social Psychology, 31:706-712.
Cohen, A. (1987). The television news interview. Newbury Park, CA: Sage.
Coleman. Renita & Granberg, Donald. (2002). Visual bias in broadcasters'
facial expressions and
other factors affecting voting behavior of TV news viewers in a
Presented to the Radio-Television Journalism Division, AEJMC annual
August, Miami Beach, FL.
DePaulo, Bella M. (1992). Nonverbal behavior and self-presentation.
Ekman, P. (Ed.) (1983). Emotion in the human face (2nd ed.). New York:
Ekman, P. (1985). Telling lies. New York: Norton.
Ekman, P. & Friesen, W.V. (1975). Unmasking the face. Englewood Cliffs,
Ekman, Paul; Sorenson, E.R. & Friesen, W.V. (1969). Pan-cultural elements
in facial displays of
emotion. Science, 164, 86-88.
Ekman, Paul; Friesen, Wallace V. & Ellsworth, Phoebe. (1972). Emotion in
the human face. New
Elliott, D. (1986). Responsible journalism. Beverly Hills, CA: Sage.
Englis, Basil G. (1994). The role of affect in political advertising: Voter
emotional responses to the
nonverbal behavior of politicians. Chapter 13, pp 223-247 in Eddie M.
Clark; Timothy C.
Brock & David W. Stewart (Eds.) Attention, attitude and affect in response
Hillsdale, N.J.: Lawrence Erlbaum Associates, Publisher.
Englis, Basil G. & Lanzetta, J.T. (1993). Effects of the perceived group
membership of a coactor
on observers' vicarious emotional responses. Unpublished manuscript. As
Englis, B. (1994). The role of affect in political advertising: Voter
emotional responses to
the nonverbal behavior of politicians. Chapter 13, pp. 223-247 in Eddie M.
C. Brock & David W. Steward (Eds.) Attention, attitude and affect in
advertising. Hillsdale, N.J.: Lawrence Erlbaum Associates, Publisher.
Englis, Basil G.; Vaughan, K.B. & Lanzetta, J.T. (1982). The conditioning
emotional responses. Journal of Experimental Social Psychology, 18, 375-391.
Friedman, H.S.; DiMatteo, M.R. & Mertz, T.I. (1980). Nonverbal
television news: The facial expressions of broadcasters during coverage of
election campaign, Personality and Social Psychology Bulletin, 6 (3): 427-435.
Friedman, H.S.; Mertz, T.I. & DiMatteo, M.R. (1980). Perceived bias in the
expressions of television news broadcasters, Journal of Communication, 30:
Garramone, Gina. (1992). A broader and warmer approach to schema theory.
Yearbook 15, pp. 146-154.
Gianetti, L.D. (1982). Understanding movies. Englewood Cliffs: Prentice Hall.
Graber, Doris. (1980). Mass media and American politics. Washington, D.C.:
Graber, Doris (1988). Processing the news: How people tame the information
tide, 2nd ed. White
Plains, N.Y.: Longman.
Graber,Doris. (1990) Seeing in remembering: How visuals contribute to
learning from television
news. Journal of Communication, v40, n3, p134-155. Summer.
Haley, R.L.; Richardson, J. & Baldwin, B.M. (1984). The effects of
nonverbal communication in
television advertising. Journal of Advertising Research, 24, 11-18.
Hall, J.A. (1984). Nonverbal sex differences: Communication accuracy and
Baltimore/London: Johns Hopkins University Press.
Izard, C.E. (1977). Human emotions. New York: Plenum Press.
Kepplinger, Hans M. (1982). Visual biases in television campaign
Kepplinger, H.M. & Donsbach, W. (1987). The influence of camera
perspectives on perception of
a politician by supporters, opponents, and neutral observers, in Political
Research, D.L. Paletz (Ed.). Norwood, N.J.: Ablex. Pp. 62-72.
Kubey, Robert W. & Peluso, Thea. (1990). Emotional response as a cause of
diffusion: The case of the space shuttle tragedy. Journal of Broadcasting &
Media, 34(1):69-76. Winter.
Lang, Annie. (1995). Defining audio/video redundancy from a
limited-capacity information processing
perspective. Communication Research, 22(1):86-115. February.
Lang, Annie. (2000). The limited capacity model of mediated message
processing. Journal of
Communication, 60(1):46-70. Winter.
Lanzetta, J. T.; Sullivan, D.G.; Masters, R.D. and McHugo, G.J. (1985).
Viewers' emotional and
cognitive responses to televised images of political leaders, in Sidney
Kraus and Richard M. Perloff (eds.), Mass media and political thought: An
information processing approach. Beverly Hills: Sage.
Lewis, C. (1984). Reporting for television. New York: Columbia University
Lichtenberg, Judith. (2000). In defence of objectivity revisited, pp.
238-254, in James Curran &
Michael Gurevitch, (Eds). Mass Media and Society, 3rd. Ed. London: Arnold.
Litterst, Judith. (2002). Affective framing of crisis: Broadcaster
nonverbal communication during
the twenty-four hours following the "Attack on America." Paper presented to
the National Communication Association annual conference, New Orleans,
Lowry, Dennis & Shidler, Jon. (1995). The sound bites, the biters, and the
bitten: An analysis of
network TV new bias in campaign '92. Journalism & Mass Communication
McCombs, Maxwell; Shaw, D. L. & Weaver, D. (Eds.) (1997). Communication and
Exploring the intellectual frontiers in agenda-setting theory. Mahway,
McCombs, Maxwell; Llamas, Juan Pablo; Lopez-Escobar, Esteban; Rey, Federico
(1997). Candidate images in Spanish elections: second-level agenda-setting
effects. Journalism & Mass Communication Quarterly, 73(4): 703-717.
Mehrabian, A. (1968). Inference of attitudes from the posture, orientation,
distance of a communicator, Journal of Consulting and Clinical Psychology,
Mehrabian, A. (1972). Nonverbal communication. Chicago: Aldine/Atherton.
McHugo, Gregory J.; Lanzetta, John T.; Sullivan, Denis G.; Masters, Roger
D. & Englis, Basil G. (1985). Emotional reactions to a political leader's
expressive displays. Journal of Personality and Social Psychology,
Moriarty, Sandra E. & Garramone, Gina M. (1986). A study of newsmagazine
photographs of the
1984 Presidential campaign, Journalism Quarterly, 63;728-734.
Moriarty, Sandra E. & Popovich, Mark N. (1991). Newsmagazine visuals and
presidential election, Journalism Quarterly, 68(3):371-380.
Mullen, Brian; Futrell, David; Stairs, Debbie; Tice, Dianne M.; Baumeister,
Roy F.; Dawson,
Kathryn E.; Riordan, Catherine A; Radloff, Christine E.; Goethals, George
John G. & Rosenfeld, Paul. (1986). Newscasters' facial expressions and
of viewers: Can a smile elect a president? Journal of Personality and
Mullen, Lawrence. (1998). Close-ups of the president: Photojournalistic
distance from 1945 to
1974. Visual Communication Quarterly, Spring, 4-6
Noggle, Gary A. & Kowalik, Timothy. (2002). Hard news/soft news
image/issue: A comparative content analysis of network newscasts of
2001. Paper presented at the National Communication Association annual
conference, New Orleans, LA, November.
Patterson, Thomas. (1993). Out of order. New York: Knopf.
Pew Research Center (2002, Feb. 23). American psyche reeling from terror
Potter, David & Lang, Annie. (1999). Making TV news memorable: The seven
habits of highly effective
storytellers. RTNDA Communicator magazine.
Richmond, V.P.; McCroskey, J.C.; & Payne, S.K. (1991). Nonverbal behavior
interactions, 2nd ed. Englewood Cliffs: Prentice-Hall.
Riffe, Daniel; Lacy, Stephen & Fico, Frederick G. (1998). Analyzing media
quantitative content analysis in research. Mahwah, N.J.: Lawrence Erlbaum
Robertson, L. (2001). Anchoring the nation. American Journalism Review, 23,
Robinson, Michael & Sheehan, Margaret. (1983). Over the wire and on TV.
(New York: Russell
Rosenberg, Shawn W. & McCafferty, Patrick. (1987). The image and the vote:
voters' preferences. Public Opinion Quarterly, 51:31-47.
Rosenberg, Shawn W.; Bohan, Lisa; McCafferty, Patrick & Harris, Kevin.
(1986). The image and the vote:
The effect of candidate presentation on voter preference. American Journal
of Political Science,
Rosenthal, R.; Hall, J.; DiMatteo, M.R.; Rogers, P. & Archer, D. (1979).
non-verbal communication: The PONS test. Baltimore, M.D.: Johns Hopkins
Shoemaker, Pamela & Reese, Stephen. (1996). Mediating the message: Theories
on mass media content. White Plains, N.Y.: Longman.
Sontag, Susan (1977). On photography. New York: Anchor Books.
Sullivan, Denis G. & Masters, Roger D. (1988). "Happy warriors": Leaders'
viewers' emotions, and political support. American Journal of Political
Science, 32(2):345-368. May.
Tiemens, R.K. (1978). Television's portrayal of the 1976 presidential
debates: An analysis of
visual content, Communication Monographs, 45(4):362-370.
Tiemens, R.K. (1989). The visual context of argument: An analysis of the
September 25, 1988
presidential debate, in, B.E. Gronbeck (Ed.), Spheres of Argument.
Annandale, VA: SCA. Pp. 140-146.
Tucker, Joan S. & Friedman, Howard S. (1993). Sex differences in nonverbal
Emotional expression, personality, and impressions. Journal of Nonverbal
Behavior, 17(2):103-120, Summer.
Tucker, J.S. & Riggio, R.E. (1988). The role of social skills in encoding
posed and spontaneous
facial expressions. Journal of Nonverbal Behavior, 12: 87-97.
Utley, G. (1997). Web vs. television: It comes down to emotion. Cable News
Van Leeuwen, Theo & Jewitt, Carey. (2001). Handbook of visual analysis.
Wagner, H.L.; Buck, Ross, & Winterbotham, Meg (1993). Communication of
Gender differences in sending accuracy and communication measures. Journal of
Nonverbal Behavior, 17(1):29-54. Spring.
Waldman, Paul & Devitt, James (1998). Newspaper photographs and the 1996
election: The question of bias. Journalism & Mass Communication Quarterly,
75(2): 302-311. Summer.
Weaver, David H. & Wilhoit, G. Cleveland. (1996). The American journalist
in the 1990s: U.S.
news people at the end of an era. Mahwah: N.J.: Lawrence Erlbaum Associates.
Total Nonverbal Expression (TNE) by time category
5pm 9/11-1am 9/12
1am 9 pm 9/12
Chi-square=7.083 df=2 p=.029 n=654
Valence by time category
5pm 9/11-1am 9/12
1am 9 pm 9/12
Chi-square=15.009 df=4 p=.005 n=654
Total Nonverbal Expression (TNE) and Valence by time category
Note. Means in the same column with different superscripts are different at
p< .05 in the Tukey HSD comparisons. Cell entries are mean scores.
Table 3. Predictors of Total Nonverbal Expression (TNE)
R square= .054 F (2, 651)=18.569 p< .001