Stroop manuscript_29

Transcription

Stroop manuscript_29
Dean & Klein!Font Effect
page 1 of 28
Dean & Klein!Font Effect
Dean, C. T. & Klein, R. M. (2010, February) In search of the font effect: The effect of conflict
upon reading and naming typefaces serially, Paper presented at the 4th International Conference
on Design Principles and Practices (Chicago, USA) and manuscript submitted for publication in
Design Principles and Practices: An International Journal.
In search of the Font Effect: the effect of conflict
upon reading and naming typefaces serially.
Christopher T. Dean, NSCAD University
[email protected]
Raymond M. Klein, Dalhousie University
[email protected]
page 1 of 28
Dean & Klein!Font Effect
Abstract
Stroop (1935) demonstrated that conflicting words interfered
with ink colour naming while conflicting colours did not
interfere with word reading. In our study — a collaboration
between a typographer and a psychologist — we re-created
Stroop's study by replacing colour names with typeface names,
ink colours with typefaces, and control colour squares with a
pseudoword. Relatively expert participants who could recognize
the five typefaces (Helvetica, Times, Papyrus, Garamond and
Ondine) were asked to read and name the typefaces from a 10 × 10
matrix, modeled after Stroop’s original materials, with and
without conflict. (Typographer:) “Despite the specialized skill
for typeface recognition in our participants” we replicated
Stroop’s finding of asymmetrical interference, (psychologist:)
“confirming the dominance of reading.”
page 2 of 28
Dean & Klein!Font Effect
In 1935 John Ridley Stroop published an article titled “Studies
of verbal interference in serial verbal reactions.” This article
has been cited 4,435 times1 and continues to be one of the most
famous experiments in the history of cognitive psychology. As
reported by MacLeod (1991), at least 700 Stroop related articles
have been written on this phenomenon between its original
publication and 1991. Figure 1 illustrates the number of Stroop
citations by decade since its original publication and 2004.
!
Stroop’s original study consisted of three separate
experiments. The results of the first two, which our study
combined into a single experiment, can be summarized as follows:
Stroop’s (1935) Experiment 1 was a reading task. It required
participants to read colour names printed in conflicting as well
as neutral ink colours. It showed no significant difference in
Number of citations
2,000
1,500
1,000
500
0
1935-1944
1945-1954
1955-1964
1965-1974
1975-1984
1985-1994
1995-2004
Decade since publication
Figure 1. Number of citations of Stroop (1935) by decade since publication.
1. ISI Web of Science database as of 10 August, 2008.
page 3 of 28
Dean & Klein!Font Effect
faratime between the two. Stroop’s (1935) Experiment 2 was a
naming task. Participants were required to name ink colours in
which conflicting colour names were spelled (for example when
presented with the colour name red printed in green ink the
correct response would be to say “green”) as well as coloured
squares. It showed that it took significantly longer to name the
ink colours in which conflicting colour names were spelled than
it did to name the ink colours of the coloured squares. See
Figures 2–42 for photographs of some of Stroop’s original
materials.
!
2. Courtesy of Colin M. MacLeod, Professor of Psychology, University of Waterloo. Digital image colour correction
courtesy of Marilyn S. Klein, Dentistry Instructional Resources, Dalhousie University.
page 4 of 28
Dean & Klein!Font Effect
Figure 2: Stroop’s (1935) original materials — conflict reading/naming
page 5 of 28
Dean & Klein!Font Effect
Figure 3: Stroop’s (1935) original materials — control naming
page 6 of 28
Dean & Klein!Font Effect
Figure 4: Stroop’s (1935) original materials from a display case at the
Peabody College of Vanderbilt University (Nashville, Tennessee).
!
These patterns of results, which has been replicated many
times (see Table 1 for Stroop’s results and a close
replication), indicate that the presence of conflict does not
interfere with reading, but does interfere with naming and that
consequently reading is a dominant response to colour naming.
The interference effect, as seen specifically in Stroop’s (1935)
Experiment 2 (bottom two rows of Table 1),
page 7 of 28
Dean & Klein!Font Effect
has since been referred to as “Stroop interference” or
“the Stroop effect.”
!
Experts in the fields of typography, type design and
graphic design, are capable of rapidly identifying many
typefaces, often when presented with only a few characters. This
capacity for typeface recognition can become almost as easy to
them as naming colours is to the lay-person. The typeface
aficionado, like the first author of this paper, may even have a
hyper-sensitivity and attentional bias with regard to typefaces.
It is virtually impossible for such a skilled typographer to
determine what is on the menu and how much it costs without
first inspecting the typeface.
!
The purpose of this study, which resulted from a
collaboration between a typographer and a cognitive psychologist
each learning about the other’s field of expertise, is to extend
the Stroop effect with color naming to the naming of typefaces.
Table 1: Mean times in seconds for reading and naming in conflict and
no-conflict conditions for Stroop (1935) and MacLeod (1986) as cited in
MacLeod (1991).
Stroop (1935)
MacLeod (1986)
Reading with conflict
43.3
41.58
Reading without conflict
41.0
41.16
Naming with conflict
110.3
102.27
Naming without conflict
63.3
59.76
page 8 of 28
Dean & Klein!Font Effect
Modeled on Stroop (1935), in this study we will replace the
names of colours with the names of typefaces, ink colours with
typefaces and control colour squares with a control pseudoword3 .
Instead of reading colour names and naming ink colours,
participants in this study were asked to read typeface names and
name typefaces in conflicting and non-conflicting conditions.
Table 2 illustrates various combinations of tasks, conditions
and correct responses used in Stroop’s (1935) and in this study.
!
Because reading is a dominant response we both expected
that we would replicate Stroop’s finding from the naming
condition: Conflicting typeface words should interfere with the
task of naming the typeface in which these words were printed.
With regard to the reading task, however, the two authors
generated different hypotheses;
Table 2. Stimuli and correct responses in the different conditions of
Stroop's classic experiment and the present one.
Stroop (1935)
Dean and Klein (2008)
Stimuli
Stimuli
Correct response
Correct response
Reading without conflict
Green
“Green”
Helvetica
“Helvetica”
Reading with conflict
Green
“Green”
Helvetica
“Helvetica”
“Green”
Egrubma
“Helvetica”
“Red”
Helvetica
“Garamond”
Naming without conflict
Naming with conflict
Green
3. The pseudoword “Egrubma” — a portion of the pseudoword “Hamburgerfons” spelled backwards — was used in this
study. “Hamburgerfons” is a pseudoword commonly used by typographers, type designers and graphic designers in which
they set a typeface in order to ascertain its unique characteristics. The pseudoword “Hamburgerfons” was shortened
because it was longer than all of the typeface names used in this study. The order of the letters were reversed to
generate a novel yet pronounceable pseudoword.
page 9 of 28
Dean & Klein!Font Effect
!
The psychologist hypothesized that typeface names, even for
experts, would have the same status as do colour names for
typical Stroop participants (they are well-known but speeded
naming of them would not be a dominant response), and
consequently that there should be no interference when reading
typeface names from the conflicting typeface in which the words
were printed. The typographer, however, hypothesized that when
reading typeface names, the participants who were skilled in
typeface recognition would experience interference from the
conflicting typefaces in which these words were printed. If
observed, he would have called this “The Font Effect.”
page 10 of 28
Dean & Klein!Font Effect
Method
Participants
!
Eleven participants (six males and five females whose ages
ranged from 22–58 years, mean = 37.523, SD = 12.846) were
recruited from faculty, students and alumni of NSCAD University
and Dalhousie University during various presentations and
teaching sessions on typography. Their level of education ranged
from third year undergraduate to third year graduate.
Professional experience ranged from none to 20 years. All had
previous training in the arts, graphic design and/or typography.
All participants reported normal or corrected to normal vision
and spoke English as a first language. One participant was
bilingual, fluent in English and French. Participants received
$8.00 Canadian as compensation. All participants were treated in
accordance with the Tri-Council Policy Statement’s of Ethical
Conduct for Research Involving Humans.
page 11 of 28
Dean & Klein!Font Effect
Apparatus and stimuli
!
Eight black and white pages of 36 × 24 in. (91.44 × 60.96
cm) were used as stimuli in this experiment:
Stimulus page 1 (CN1) — Typeface names printed in
conflicting typefaces:
On this page five typeface names (Courier, Garamond, Helvetica,
Ondine and Papyrus) were printed in one of five different
typefaces (ITC Courier Medium, ITC Garamond Regular, Helvetica
Neue 55 Roman, Adobe Ondine Regular and Adobe Papyrus Regular)4 .
Each page had the words laid out so that there were 10 words per
line and a total of 10 lines. The five typefaces used in this
experiment were selected based upon their lack of confusability
with one another and ease of pronunciation (see Appendix A: The
card sorting task). Once an arbitrary mapping rule was chosen,
the ordering of the words and typefaces corresponded exactly
with the ordering of words and colours used in Stroop’s original
materials (see figures 2-4), but not exactly with his
description of how he arrived at this ordering5.
4. For a complete description of the methodology used to select these five typefaces see Appendix A.
5. Quoting Stroop (1935) “The words were also arranged so that the name of each color would appear twice in each
line. No word was printed in the color it named but an equal number of times in each of the other four colors.”
However, in figure 2 you can clearly see that the word brown appears three times in the first column. Additionally,
you can see three brown control patches in the last column of figure 3..
page 12 of 28
Dean & Klein!Font Effect
!
The size of each word was adjusted so that they shared a
common x-height of 20 postscript points (pts, 7.056 mm). Words
were horizontally separated from one another with five en6 spaces
of Helvetica Neue 55 Roman at 40.63 pts (14.333 mm). Inter-liner
spacing was set to 117.61 pts (49.410 mm), tracking was set to
zero and kerning was set to automatic. An additional practice
row of 10 words appeared at the top of each page separated
vertically from the rest of the words by 235.22 pts (82.980 mm).
Top, left/right and bottom margins were 2.3 in. (58.420 mm), 1.8
in. (45.720 mm) and 3.2 in. (81.280 mm) respectively. The task
for this page was to name the typefaces. Figure 5 illustrates
this stimulus page.
Stimulus page 2 (CN2) — Typeface names printed in
conflicting typefaces:
This page was similar to CN1 except the orders of the words and
typefaces were reversed both horizontally and vertically. The
task for this page was to name the typefaces.
6. A unit of measurement equal to half of the typesize in use.
page 13 of 28
Dean & Klein!Font Effect
Figure 5: A sample stimulus page illustrating a condition of conflict
(scaled to 18%).
Stimulus page 3 (CR1) — Typeface names printed in
conflicting typefaces:
This page was physically identical to CN1. The difference was
that the task for this page was to read the words.
Stimulus page 4 (CR2) — Typeface names printed in
conflicting typefaces:
This page was physically identical to CN2. The difference was
that the task for this page was to read the words.
page 14 of 28
Dean & Klein!Font Effect
Stimulus page 5 (N1) — Control pseudoword printed in a nonconflicting typeface:
On this page, instead of the name of a typeface, the pseudoword,
Egrubma, appeared laid out following the same specifications of
CN1. The task for this page was to name the font.
Stimulus page 6 (N2) — Control pseudowords printed in nonconflicting typefaces:
This page was similar to N1 except the order of the typefaces
was reversed both horizontally and vertically. The task for this
page was to name the typefaces.
Stimulus page 7 (R1) — Typeface names printed in nonconflicting typefaces:
This page was similar to page CN1 except the font names were set
in Linotype Univers 530. The task for this page was to read the
words.
Stimulus page 8 (R2) — Typeface names printed in nonconflicting typefaces:
This page was similar to R1 except the order of the words were
reversed both horizontally and vertically. The task for this
page was to read the words.
page 15 of 28
Dean & Klein!Font Effect
!
A 36 × 24 in. (91.44 × 60.96 cm) sheet of 3∕16 in. (4.763
mm) Bainbridge All Black Foamboard was used as a shield to cover
the pages preventing participants from seeing them until it was
appropriate for them to do so.
!
The stimulus pages were produced on a MacBook Pro, model
identifier: MacBookPro2,2 with a 15 in. (38.1 cm) display at a
native resolution of 1440 × 900 pixels set to millions of
colours, a 2.33 GHz Intel Core 2 Duo processor, 2 GB of RAM and
a 210 GB hard drive running Mac OS X version 10.4.11. Software
used to produce the pages was Adobe Illustrator CS version
11.0.0. Files were saved as Adobe Acrobat PDF’s with no downsampling or compression saved with version 4.0 compatibility.
Pages were printed by an Océ TDS 450 printer at 600 dots per
inch (DPI) on white 24 lb Océ bond paper and laminated by a GBC
Discovery 80 laminator with a 3 mil matte lamination. The
luminance of the stimulus pages varied from approximately 20–25
candelas per meter squared (cd/m2).
!
For each of the previously described pages, there was a
corresponding 8.5 × 11 in. (21.59 × 27.94 cm) score sheet for
use by the experimenter to identify any errors the participants
might make while naming or reading. On each score sheet were
printed the words Courier, Garamond, Helvetica, Ondine and
page 16 of 28
Dean & Klein!Font Effect
Papyrus in a 10 × 10 matrix corresponding with the correct
responses for each of the afore mentioned stimulus pages.
The specimen sheets
This six page document measured 8.5 × 11 in. (21.59 × 27.94 cm)
consisted of five different specimen sheets, one for each of the
five typefaces used in the experiment, followed by a small
naming exercise. Each specimen sheet contained a complete
alphabet of both uppercase and lowercase letters, numerals and
selected glyphs such as a dagger, ligature and ampersand.
Typefaces for each specimen sheet were set at 42 pts (14.817 mm)
with an inter-linear spacing 52 pts (18.344 mm). Due to
differences in the proportional spacing of each typeface,
tracking was adjusted so that each specimen sheet shared the
same number of letters per line. Each page also contained two
small passages of lorem ipsum7. The first passage was set at 10
pts (3.528 mm) with an inter-linear spacing of 12 pts (4.233 mm)
and a line length of 264 pts (93.133 mm). The second passage was
set at 12 pts (4.233 mm) with an interlinear spacing of 14 pts
(4.939 mm) and a line length of 264 pts (93.133 mm). Figure 6
illustrates a page from this document.
7 Lorem ipsum is placeholder text often used by typographers to illustrate how actual text will appear on a page. It
is used because it resembles normal letter distribution in the English language as opposed to random letter strings.
page 17 of 28
Dean & Klein!Font Effect
Ondine
abcdefghijklmnop
qrstuvwxyz
ABCDEFGHIJKLMM
NOPQRSTUVWXYZ
1234567890
!@#$%^&*
{ [ ( fi fl † ‡ § ¶ ‘ “ , … ) ] }
10 /12
Lorem ipsum dolor sit amet, consetetur sadipscing elitr, sed
diam nonumy eirmod tempor invidunt ut labore et dolore
magna aliquyam erat, sed diam voluptua. At vero eos et
accusam et justo duo dolores et ea rebum. Stet clita kasd
gubergren, no sea takimata sanctus est Lorem ipsum dolor sit
amet. Lorem ipsum dolor sit amet, consetetur sadipscing elitr,
sed diam nonumy eirmod tempor invidunt ut labore et dolore
magna aliquyam erat, sed diam voluptua. Duis autem vel eum
iriure dolor in hendrerit in vulputate velit esse molestie
consequat, vel illum dolore eu feugiat nulla facilisis at vero eros
et accumsan et iusto odio dignissim qui blandit praesent
12/14
Lorem ipsum dolor sit amet, consetetur sadipscing
elitr, sed diam nonumy eirmod tempor invidunt ut
labore et dolore magna aliquyam erat, sed diam
voluptua. At vero eos et accusam et justo duo dolores
et ea rebum. Stet clita kasd gubergren, no sea
takimata sanctus est Lorem ipsum dolor sit amet.
Lorem ipsum dolor sit amet, consetetur sadipscing
elitr, sed diam nonumy eirmod tempor invidunt ut
labore et dolore magna aliquyam erat, sed diam
voluptua autem. Duis autem vel eum iriure dolor in
Figure 6: A page from the specimen sheets booklet.
page 18 of 28
Dean & Klein!Font Effect
!
Participants were timed with a Nexxtech digital stopwatch,
model number 6315022, accurate to one hundredth of a second.
Data was collected, analyzed and presented using Numbers ‘08
version 1.0.2, pro Fit version 6.1.8 and Adobe Illustrator CS3
version 13.0.2. Luminance was measured8 with a UDT Technology
photometer model number 161 calibrated by the National Science
Laboratory in Washington DC.
8. Courtesy of Dr. Donald E. Mitchell, Professor Emeritus, Department of Psychology, Dalhousie University.
page 19 of 28
Dean & Klein!Font Effect
Design
Each participant was exposed to four of the eight pages. The
assignment of pages to conditions was accomplished by a
factorial combination of the two tasks (naming and reading), the
two page types (conflict and no-conflict) and the two orders
(the same pages previously mentioned with the orders of the rows
and columns reversed). In order to avoid switching between tasks
(and hence to minimize the difficulty of the tests) the two
conditions (conflict and no-conflict) of each task (reading and
naming) appeared in successive pairs. The eight participant
sample condition and the order of tasks were fully
counterbalanced across the eight blocks to remove any untoward
effects of order.
Procedure
In order to ensure that participants were capable of naming the
typefaces, which would allow us to measure the effects of
interference, they were provided with the specimen sheet
document for review least 24 hours prior to the experiment.
Before the experiment began participants were informed about the
tasks they were about to perform and gave their written consent.
Participants entered a 3 × 3 × 2.5 yard (2.743 × 2.743 × 2.286
m) room and were seated in a chair 18 in. (45.720 cm) high at a
page 20 of 28
Dean & Klein!Font Effect
30 × 36 × 22 in. (76.200 × 91.440 × 55.880 cm) table at an angle
of 180°. The room was lit by three individually mounted 100 watt
tungsten light bulbs covered with translucent shades. Each
participant entered the experimental room at least three minutes
prior to beginning the experiment in order to allow their eyes
to adjust to the lighting conditions. Figure 7 shows the
experiment room with the experimenter on the left and a pilot
participant on the right.
!
Figure 7: The experimenter (typographer) running a pilot participant
(psychologist) in the experimental room.
page 21 of 28
Dean & Klein!Font Effect
Participants were assigned to one of the eight different orders
of the eight conditions as seen in table 3.
!
Each session began with a stack of four pages placed face
up on the table. Each page was held in place by a small piece of
packaging tape and the stack was covered by the foamcore shield.
Before the shield was removed, participants were informed of the
requirements of the upcoming test. They were instructed to
either read the words or name the typefaces. They were
instructed to go as quickly as possible without making mistakes
and, following Stroop’s original instructions, to correct any
mistakes that they noticed they had made. They were encouraged
to follow along with their finger to avoid losing their place.
Each condition began with a single 10 word practice line
followed by the remainder of the page which consisted of 10
lines of 10 items of 100 items.
Table 3: Condition orders for the
eight participants used in this study.
Ss1:
CR1
R2
CN2
N1
Ss2:
CN1
N2
CR2
R1
Ss3:
R1
CR2
N1
CN2
Ss4:
N1
CN2
R1
CR2
Ss5:
R2
R1
CN1
N2
Ss6:
N2
N1
CR1
R2
page 22 of 28
Dean & Klein!Font Effect
The practice
On the signal “Ready! Go!” — the same signal used by Stroop in
1935 — the experimenter started timing and the participant moved
the shield approximately 5 in. (12.7 cm) downwards exposing the
10 item practice row at the top of the page. Participants then
performed the practice task while the experimenter followed
along the corresponding check sheet and kept track of any errors
the participant made. Upon completion of the practice
participants were informed of any mistakes they had made.
The test
On the signal “Ready! Go!” the experimenter started timing and
the participant removed the shield exposing the rest of the
page. Participants then performed the test while the
experimenter followed along the corresponding score sheet to
keep track of any errors. Upon completion of each condition,
participants were informed of any mistakes they had made.
!
This process was repeated for each of the four conditions.
After the experiment was complete participants were told about
the nature or the original Stroop experiment and the aim of this
study. After this debriefing, any questions were answered and
participants were compensated for their participation.
page 23 of 28
Dean & Klein!Font Effect
Results
!
Mean time to read pages in the no-conflict condition was
57.010 seconds (SD = 8.009). On average, participants took 1.566
seconds longer to read the typeface names in a conflict
condition than they did in a no-conflict condition (SE = 1.994).
A paired t-test indicated that this difference (representing, on
average, about 16 ms per item) was not significant (t(10) =
0.749, p = 0.471).
!
Mean time to name typefaces in a no-conflict condition was
123.473 seconds (SD = 20.630). On average, participants took
30.220 seconds longer to name typefaces in a conflict condition
than they did in a no-conflict condition (SE = 7.350). A paired
t-test indicated that this difference (representing an
interference effect from the conflicting typeface names upon
typeface identification times of about 300 ms per item) was
significant (t(10) = 3.920, p = 0.003).
page 24 of 28
Dean & Klein!Font Effect
Discussion
The results of this experiment showed the following:
a) No significant difference in reading time between conditions
of conflict and no-conflict.
b) Participants took significantly longer to name typefaces in
conditions of conflict than they did in conditions of nonconflict.
!
These findings are consistent with Stroop (1935) and
MacLeod (1991). They extend the finding of the Stroop effect
from naming colours to naming fonts and show that even for
individuals who are skilled at recognizing typefaces9 , reading
still presents itself as a dominant response to naming as first
observed by John Ridley Stroop in 1935. So, is the Font Effect a
figment of the typographer's imagination or does it simply
remain to be discovered? In the future, the typographer would
like to test type designers specifically, as he hypothesizes
that they would have an even greater sensitivity towards
typeface recognition thereby increasing the chances for typeface
interference.
9 Including the primary author of this paper who has spent over 20 years investigating various aspects of typography
and typographic specimens, and freely admits that he spends very little time reading compared to the layperson.
page 25 of 28
Dean & Klein!Font Effect
References
MacLeod, C. M. (1991). Half a century of research on the Stroop
effect: An integrative review. Psychological Bulletin. 109(2)
163 – 202.
Stroop, J. R. (1935). Studies of verbal interference in serial
verbal reactions. Journal of experimental psychology. 18(6)
643 – 662.
page 26 of 28
Dean & Klein!Font Effect
Appendix A: The card sorting task
The purpose of this task was to ascertain which typefaces
participants could correctly identify, to discover commonalities
between correct and incorrect identifications and to assist us
in the selection of the five typefaces used in this experiment.
Sixteen typefaces (Arial, Bodoni, Comic Sans, Cooper Black,
Copperplate Gothic, Courier, Fette Fraktur, Futura, Garamond,
Helvetica, Keunstler Script, Ondine, Optima, Papyrus, Univers
and Zaph Chancery) were selected based upon their ease of
identification, frequency of exposure and visual distinction
from one another. Participants were given two stacks of 16 cards
each. On each card in the first stack was printed each of the 16
typeface names set in Franklin gothic (the same font Stroop used
in his original materials—see figure 2). On each card in the
second stack the pseudoword “Hamburgerfons” was printed set in
each of the 16 previously mentioned typefaces.
!
Twenty six participants were selected from students, staff,
faculty and friends from both NSCAD University and Dalhousie
University. This group included all 11 of the participants who
were subsequently tested in the main experiment. The task was to
match each typeface name with the pseudoword set in the
corresponding typeface. The results of this task are presented
in the form of a confusion matrix (Figure 8).
page 27 of 28
Dean & Klein!Font Effect
Figure 8: A confusion matrix of the stimuli and responses from the card
sorting task (Appendix A). The right hand column illustrates specimens of the
typefaces used in this portion of the study.
The average performance on the five selected typefaces was about
the same as the average for the 16 typefaces (58% correct).
Considering only the 5 selected typefaces and responses with
these five typeface names, the 26 participants made 75 correct
responses and only five miscategorizations (for 94% correct).
The average score for the entire population of 26 subjects used
in the card sorting task was 58.4% and the average score for the
population of 11 participants used in The Font Effect experiment
was 77.4%.
page 28 of 28