Content uploaded by Frank J. Bernieri
Author content
All content in this area was uploaded by Frank J. Bernieri on Dec 31, 2013
Content may be subject to copyright.
THE
IMPORTANCE
OF
NONVERBAL CUES
IN
JUDGING
RAPPORT
Jon
E.
Grahe
and
Frank
J.
Bernieri
ABSTRACT:
This study examined
the
relative impact different channels
of
commu-
nication
had on
social
perception
based
on
exposure
to
thin
slices
of the
behavioral
stream.
Specifically,
we
tested
the
hypothesis that
dyadic
rapport
can be
perceived
quickly
through
visual channels. Perceivers
judged
the
rapport
in 50
target interac-
tions
in one of
five stimulus display conditions: transcript,
audio,
video,
video-
+
transcript,
or
video
+
audio.
The
data demonstrated that perceivers
with
access
to
nonverbal,
visual information
were
the
most accurate perceivers
of
dyadic
rapport.
Their judgments were
found
to
covary
with
the
visually
encoded
features that past
research
has
linked
with
rapport expression. This
suggests
the
presence
of a
nonver-
bally based
implicit
theory
of
rapport that more
or
less
matches
the
natural
ecology,
at
least
as it
occurs
within
brief samples
of the
behavioral stream.
Our
impressions
of
others
undoubtedly
are
formed primarily
during
our
initial
encounters
with
them.
First
impressions influence
a
perceiver's
desire
to
interact
with
another
(Zebrowitz,
1990). Confirmation
and
belief
perseverance
biases
then
can
maintain these initial impressions over time
(Fiske
&
Taylor, 1991).
In
order
to
understand fully
how
such things
as
relationship
formation,
satisfaction,
and
interactant rapport
develop,
we
must
begin
by
learning
the
processes
involved
in the formation of our
very
first
impression
of
another.
What
is
responsible
for
that
first
impression?
Is it
formed
initially
and
primarily
through
what
is
said
during
that
initial
en-
counter,
or is it a
result
of the
visual
and
nonverbal
features
we
perceive?
This
report examines
the
contribution
different channels
of
communication
make
in the
expression, transmission,
and
perception
of
rapport.
This
research
was
supported
by a NSF
Young Investigator Award (#DBS-9258686)
to the
second author.
The
authors
wish
to
thank
Michelle
Curtis,
Janet
Davis,
Keith Dempsey, Kath-
leen
Could,
Lisa Slotnik,
and
Jeranna Younker
for
their assistance
in the
data
collection
and
coding
for
this study.
Jon
E.
Grahe,
Department
of
Psychology,
Monmouth
College;
Frank
J.
Bernieri,
Depart-
ment
of
Psychology,
University
of
Toledo.
Address
correspondence
to Jon E.
Grahe,
Department
of
Psychology,
Monmouth
College,
Monmouth,
IL
61462;
e-mail:
jgrahe@monm.edu.
journal
of
Nonverbal Behavior
23(4),
Winter 1999
©
1999 Human
Sciences
Press,
Inc.
253
Researchers
examining
the
relative importance
of
verbal
versus
non-
verbal
information
in
person perception have offered inconsistent conclu-
sions
(Archer
&
Akert, 1977; Berry, Pennebaker, Mueler,
&
Hiller, 1997;
Gifford
&
Hine,
1994; Mehrabian
&
Ferris,
1967; Mehrabian
&
Weiner,
1967).
Mehrabian
and
colleagues (Mehrabian
&
Ferris,
1967; Mehrabian
&
Weiner, 1967) concluded that most interpersonal information
was
commu-
nicated nonverbally. However, Archer
and
Akert
(1977)
found that per-
ceivers
with
access
to
verbal
and
vocal information made more accurate
judgments.
More
recently,
Berry
et al.
(1997) reported that verbal content
conveyed
as
much information
as
nonverbal behavior.
This
inconsistency
may
reflect differences
in the
situational context within which
the
behavior
was
observed (Ambady
&
Rosenthal, 1992; Bernieri, Gillis, Davis,
&
Grahe, 1996)
and the
information content being judged (Ambady
&
Rosen-
thai, 1992; DePaulo, 1992; Noller, 1985). Noller
(1985)
argued,
for ex-
ample, that constructs containing affect
are
communicated more quickly
through nonverbal channels.
Other construct domains including personality
and
communication
motives
(e.g., deception)
may
manifest themselves more strongly
within
the
verbal
channel. Gifford (1994)
found
that
job
candidates encoded their
personality
weakly
in
their nonverbal behaviors during their interview.
Berry
et al.
(1997)
found that
key
words
in
dyadic conversation (e.g., emo-
tion
words,
the use of
present tense, self-referents, etc.) communicated
im-
portant information about
the
competency, dominance,
and
warmth
of the
target.
Such
studies
indicate that
the
verbal channel
can
communicate
much
to the
impression formed
of
another's
traits
and
dispositions.
In
contrast
to
personality constructs, certain social
and
affective-based
constructs
may be
more readily communicated
via
nonverbal behaviors
(De
Paulo, 1992).
For
example, Babad, Bernieri,
and
Rosenthal (1989;
1991)
found that teachers'
biases
and
expectations towards their pupils
were communicated immediately
and
primarily through nonverbal chan-
nels.
Furthermore, Ambady
and
Rosenthal (1992) reported
in a
meta-anal-
ysis
that social outcomes were predicted from very brief samples
of
behav-
ior.
Incredibly,
they found that
the
predictive validity
of a
thin slice
of an
interaction
did not
increase
when
the
length
of the
slice increased from
20
s
to 5
min.
The
sheer
amount
of
verbal content that
is
lost
as one
moves
from
a 5 min
long segment
of a
conversation
to 20 s is not
trivial. There-
fore, since validity does
not
decrease
with
exposure length,
one may
con-
clude that
the
critical information present
in
these
thin
slices
of
behavior
could
be
chronic, expressive,
and
very likely nonverbal
in
nature.
As De-
Paulo
(1992)
noted,
nonverbal behaviors
can
occur very quickly
and ob-
server
attributions
in
response
are
irrepressible.
254
JOURNAL
OF
NONVERBAL BEHAVIOR
JON E.
GRAHE,
FRANK
J.
BERNIERI
Rapport
and Its
Perception
The
construct
of
rapport differs slightly from those previously
discussed
because
it is
defined
at the
dyad
or
group level. Rapport
is
unlike
affect
or
a
disposition
in
that
it
refers
to a
quality
of an
interpersonal relationship
and
does
not
exist
within
any
individual. Despite
this
difference
in the
unit
of
analysis that
is
being
considered,
the
perception
of
rapport appears
to
occur very much like other person specific constructs.
Bernieri
and
colleagues have reported
a
number
of
studies
document-
ing
that
passive
observers
can
assess
the
rapport between target interac-
tants
from very brief video clips
(30-50
s)
extracted from
a
much longer
interaction (Bernieri
&
Gillis, 1995; Bernieri
et
al.,
1996;Gillis,
Bernieri,
&
Wooten,
1995).
More
recently, Bernieri
and
Grahe
(1999)
demonstrated
that
a
significant degree
of
target-observer agreement
in
rapport
assessment
was
attainable
with
exposures
limited
to a
mere
10 s. One
conclusion
made throughout these
studies
was
that interactant rapport must have been
expressed
and
communicated nonverbally rather than verbally
in
order
for
observers
to
detect
it in
such brief exposures (Bernieri
et
al., 1996).
Tickle-Degnen
and
Rosenthal (1990) formulated
a
purely nonverbal
construct
of
rapport
after
a
meta-analysis
of the
existing literature. They
identified
three distinct components
of
rapport that when taken together
best
represent what psychologists
and lay
people
are
referring
to
when
using
the
term rapport.
Each
of the
three components—mutual attention,
coordination,
and
positivity—was
found
to
have
established
nonverbal
correlates
for
their expression (Tickle-Degnen
and
Rosenthal, 1987; 1990).
Ultimately,
one
must acknowledge that
the
rapport between
two
individ-
uals
is
influenced
by: (a) the
entire behavioral stream, verbal
and
nonver-
bal,
(b)
prior expectations
and
cognitive schemas,
and (c) the
attribution
of
relevant
actions
and
events. Tickle-Degnen
and
Rosenthal
did not
argue
that rapport should
be
defined exclusively
as a
nonverbal behavior,
but
rather
that rapport
was
encoded strongly
within,
and
communicated
throughout,
the
visual nonverbal channels. Thus,
the
best
place
to
look
for
and
assess
rapport
would
be
within
the
visual
features
of the
behavioral
stream.
Bernieri
et al.
(1996) identified behaviors predictive
of
self-reported
rapport
by
examining
the
behavioral
cues
present
within
two
social con-
texts.
Gesturing, interactional synchrony (i.e., coordination),
and
proximity
were particularly potent indicators.
The
variance
in
these
cues occurring
within
the
time frame sampled,
which
was
less
than
60 s,
predicted about
half
of the
criterion variance
of
target self-reports that were collected
after
the
completion
of the
entire 5-30
min
long
dyadic interaction. Clearly,
255
within
the
contexts examined interactant rapport
was
encoded strongly
within
the
behavioral stream.
In
that study,
the
authors also reported that outside
observers
made
use
of
this encoding. Naive judgments made after watching
30-50
s
video
clips were correlated
with
the
target criterion
of
self-reported rapport (Ber-
nieri
et
al., 1996). Observer judgments
of
rapport were remarkably consis-
tent
and
unchanging
across
situations, suggesting
a
relatively stable
im-
plicit
nonverbal social perception theory
of
rapport that
may be
applied
universally
across
different social context. This resulted
in
observers being
more accurate
in
assessing
cooperative,
as
opposed
to
adversarial, interac-
tions
because target expressivity,
a cue on
which
they relied
so
heavily,
was
statistically more predictive
of
self-reported rapport
in
that context.
Although
these
findings
suggest
that rapport
is
transmitted nonverbally,
they
do not
preclude
the
possibility that rapport
is
also transmitted verbally.
No
direct
test
has
been made comparing perceivers
with
access
to
verbal
versus
nonverbal channels
of
communication. Aside from general talk-
ativeness,
the
cues
measured
by
Bernieri
and
colleagues were nonverbal
in
nature.
Unlike
other related
studies
(e.g., Gifford
&
Hine,
1994) verbal
content
was
never
assessed
explicitly
in any of
their reported
studies
(Ber-
nieri
&
Gillis, 1995; Bernieri
et
al., 1996;
Gillis
et
al.,
1995).
In
other
words, these studies were biased methodologically towards finding nonver-
bal
primacy
in the
encoding
and
perception
of
rapport
due to
their defi-
cient
analysis
of
information contained
within
the
verbal channel
of
com-
munication.
Overview
and
Hypotheses
The
objective
of the
current investigation
was to
determine
the
relative
importance
of
verbal
versus
nonverbal information
in the
perception
of
rapport
employing
thin
slices
of the
behavioral stream.
In the
current
ex-
periment,
perceivers were
limited
to
information
provided
in a
given chan-
nel
or
selected composite channel
of
communication (transcript,
audio,
video,
video-(-transcript, video
+
audio). They were
asked
to
judge
the
rap-
port
that existed between
50,
unacquainted, mixed-sex dyads. Perceiver
judgment
policies were also examined.
It was
predicted generally that perceivers
would
rely primarily
on the
nonverbal information present
within
thin
slices
and
would
be
relatively
insensitive
to
variations
in
verbal content
across
target stimuli given this
methodology
(i.e.,
brief exposure times).
Within
each presentation channel
condition,
mean correlations were computed between judgments
and
tar-
JOURNAL
OF
NONVERBAL BEHAVIOR
256
get
criterion. Planned
contrasts
performed
on
these effect
sizes
found
within
each channel condition allowed
us to
evaluate what contribution
verbal content made
to
observer judgments
of
rapport (see
Rosenthal
&
Rosnow,
1985).
For
example,
we
expected perceivers
who
were exposed
to the
visual channel
to
judge rapport more accurately—employing target
self-reports
as the
criterion
for
accuracy—than perceivers
who
were pre-
sented
with
only
the
audio channel.
In
addition,
it was
expected that perceiver judgments
would
correlate
more strongly with
cues
coded
within
the
channel they were experiencing.
Because
most
cues
coded
in
this study were nonverbal
in
nature,
it was
expected
that perceivers' judgments
would
covary more
with
our
coded
visual
cues when they
had
access
to the
visual
channel compared
to
per-
ceivers
whose judgments were made from audio
or
transcripted verbal
in-
formation.
The
purpose
of
these
predictions
was to
begin
to
document
em-
pirically
and
concretely
the
nature
of the
interpersonal perception
process
that
occurs
with
thin-slice exposure. This
is,
after
all,
the
very
basis
of
initial
impression formation
in
face-to-face interaction.
Method
Participants
and
Design
A
total
of 115
participants
(43
male,
70
female) received
extra
credit
for
their participation.
Two did not
report their gender. These participants
were excluded from
analyses
involving gender. Participants were
assigned
randomly
to one of
five conditions
in a 2
(sex
of
participant)
X 5
(tran-
script,
audio,
video, video
+
transcript,
and
video
+
audio) between-sub-
jects
design.
Judgment
Stimuli
1
Videotaped interactants were
120
high school students
and
under-
graduates
(60
male
and 60
females) participating
in a
study
of
face-to-face
social
interaction (Bernieri, Davis, Rosenthal,
&
Knee, 1994)
who
were
formed randomly
into
60
unacquainted mixed-sex dyads.
Target
interac-
tants
were seated
at a
table
with
a
world
map and
20,000
dollars
of
play
money.
The
experimenter instructed them
to
plan
a
trip
around
the
world.
Conditions
of the
interaction were that they needed
to
agree
on the
travel
locations,
and to
continue their discussion until they spent their trip
money.
The
interactants determined discussion
length,
which
varied from
4
to 15
minutes. Following
the
discussion, interactants
filled
out a
29-item
JON E.
GRAHE,
FRANK
J.
BERNIERI
257
self-report measure describing themselves
and the
interaction. Items were
derived
to
measure
the
three domains
of
rapport
identified
by
Tickle-
Degnen
and
Rosenthal (1990). They included:
(a)
positivity,
assessed
with
items
such
as
"friendliness"
and
"positivity";
(b)
attention,
assessed
with
items
such
as
"focused"
and
"interesting";
and (c)
coordination,
assessed
with
items such
as
"smooth"
and
"harmonious."
All
items
on the
question-
naire
were highly correlated.
The
29-item rapport scale yielded
a
Cron-
bach's
alpha
of
.94.
The
mean
of the
self-reports generated
by the two
targets
served
as our
target criterion
for
dyad rapport.
2
Of the 60
original unacquainted dyads,
50
were randomly selected
for
the
stimulus tape.
A 30 s
video
clip
was
taken from
the
second
to
last
minute
of the
interaction.
The end of the
interaction
was
sampled
so
that
perceivers'
judgments were made
at a
time very close
to
when
the
interac-
tants
made their self-reports.
The
interaction clips were assembled ran-
domly
onto
the
stimulus tape.
Twenty-five
potential
feature
cues
to
rapport were coded
within
the
thin-slice being analyzed. Appendix
A
contains
a
complete
list
of the
cues
coded
for
this study.
These
cues were coded
by a set of
independent
raters
not
involved
in the
video taping
or
rapport ratings
of the