The Architectural Colour Design Process:
An Evaluation of Sequential Media via
Semantic Ratings
Sibel Ertez Ural,*Semiha Yilmazer
Department of Interior Architecture and Environmental Design, Faculty of Art, Design and Architecture, Bilkent University, Ankara 06800, Turkey
Received 11 May 2008; revised 21 April 2009; accepted 10 May 2009
Abstract: In recent studies, contextual situations of
applied colours are compared to colours presented as
samples or chips. Findings of such studies point out dif-
ferent results in terms of similarities or differences
between the evaluations of isolated/abstract colours and
contextualized situations. Architectural and spatial con-
texts have their own characteristics regarding colouring
criteria, so it is of great importance to examine the archi-
tectural/spatial colouring process from this point of view.
This study explores this process by investigating the con-
sistency of semantic ratings of four sequential stages of
the architectural colour design process, namely, colour
chips/samples, abstract compositions, perspective draw-
ings and 3D models. The architectural context for the
study was a simple interior space. Fifteen different colour
schemes were applied on the four media representing the
stages. Subjects rated the 15 sets against seven bipolar,
five-step semantic differential scales. The scales consisted
of harmonious-discord, pleasant-unpleasant, comfortable-
uncomfortable, spacious-confined, static-dynamic, excit-
ing-calming and extroverted-introverted. Findings indi-
cated that there are significant associations between the
evaluations of the abstract compositions, the perspective
drawings and the 3D models; however, the evaluations of
colour chips are significantly different than the others.
The medium effect observed mostly between abstract and
contextualized media. Additionally, factor analysis showed
that pleasantness, harmony, spaciousness and comfort are
connected in the evaluations of contextual situations,
while pleasantness and harmony differ from spaciousness
and comfort in the evaluations of colour chips and
abstract compositions. The factor of activity (arousal)
(dynamism, excitement, and extroversion) stays the same
for all four media. It is also found that different colour
characteristics are determinative over different media.
Ó2010 Wiley Periodicals, Inc. Col Res Appl, 35, 343 – 351, 2010;
Published online 22 January 2010 in Wiley InterScience (www.
interscience.wiley.com). DOI 10.1002/col.20583
Key words: architectural colour design process; spatial
colour; semantic ratings; architectural/spatial context;
colour emotions
INTRODUCTION
In the field of architecture, colour is a design element. It
stands as a composition of all its attributes and it exists at
a place where all related disciplines overlap. As Green-
Armytage says, ‘‘successful colour design depends on the
designer’s knowledge comes from several disciplines and
from design itself.’’
1
Theoretical knowledge comes from
several disciplines and arises from research mostly where
colours are abstracted and simplified into colour chips
and samples. The need for reconsideration of the theoreti-
cal knowledge within the architectural context and real-
life situations is inevitable.
In design itself, simplifications and abstractions are
widely used both in design education and by professio-
nals. Choosing isolated colour samples from various cata-
logues, developing abstract colour graphics for initial col-
our decisions and then elaborating on these decisions
through various drawings and models seem to be the most
common approach to the colour design process.
2–4
Each
of these visualization techniques is progressively used by
the participants in the design as a communication medium
throughout the process. Hence, corroboration between the
different visualization techniques is essential to a reliable
colour design method. The presented study deals with dif-
ferent visualization techniques, commonly used during the
architectural colour design process, by investigating the
consistency of their evaluations via semantic ratings.
*Correspondence to: Sibel Ertez Ural (e-mail: ural@bilkent.edu.tr).
V
V
C2010 Wiley Periodicals, Inc.
Volume 35, Number 5, October 2010 343
The Colour Design Process in Architecture
Colour is an element of architecture, thus colour design
have to be an integral part of architectural design process.
Smith emphasizes that environmental colour is multifac-
eted, playing a variety of roles in our everyday lives.
5
However, prototypes and models for designs are often
presented in white or monochromatic combinations, irre-
spective of the materials incorporated and the colours that
may be applied in the final constructed building, interior
or object. Questions are therefore raised concerning
design professionals’ perceptions of the importance of
colour in relation to space and form, and to the experi-
ence of place. According to Mikellides, an architect wants
or considers it desirable to use colour in his design on the
one hand, but fears or is ignorant of its possibly undesir-
able effects on people on the other hand.
6
Mikellides
explains this conflict as an architectural dilemma: what
we know about colour as scientists and psychologists and
what we do as artists and architects. This conflict can
only be overcome in architecture by bridging colour
theory and application.
This topic continues to be discussed not only in practice
but also in education. Janssens and Mikellides investigated
architectural students’ knowledge about the perceptual and
psycho-physiological aspects of colour, colour nomencla-
ture, existing myths and beliefs and how colour is used in
their everyday work in the studio.
7
The findings show there
is a severe lack of knowledge about colour research. Read-
ing colour literature and attending lectures or conferences
was rare. However, attitudes toward colour research were
positive, especially for studies on colour perception and
colour preferences. Students perceived colour design as
their own future responsibility and basic design education
seemed to be the main source of colour information.
Although this study is over 10 years old, it is likely that
the architectural education has not changed much. Accord-
ing to Akbay,
8
during the progression of colour education
in basic design courses, students’ original colour decision
tendencies—toward subjective and intuitive attitudes—
decreased and their tendencies towards knowledge-based
and analytical attitudes increased.
On the other hand, design students are asking for the
guidance of a methodology in their colour design, and
they need clarification of the ambiguity and generality of
the findings of emperical research. At this point, it is im-
portant to investigate how to implement theoretical colour
knowledge planted via abstract compositions in basic
design into the architectural design process. The product
of design is coloured. The aim of the designer is to end
up with a satisfactory design product supported by a suc-
cessful colour scheme. Particularly in architecture, aes-
thetical and functional criteria are determinative over
evaluating colour design. The colour design is expected to
be compatible with functional appropriateness and
requirements, visual comfort and satisfaction of users’
needs for sure, nevertheless colour harmony and colour
emotions are seemingly the points of departure for achiev-
ing a harmonious, pleasing result, spatial quality and
desired mood. However, the link between harmony
(which is a measure of aesthetics) and pleasantness
(which is a measure of emotion) of colour combinations
is still being discussed.
9
Colour Harmony. In architecture, colour is not an iso-
lated entity; it always exists with other colours of the
form or the space. Thus, harmony of colour combinations
is a major concern for colour designers. On condition that
harmony is a ‘‘measure of aesthetics’’, colour harmony is
tried to be explained on the order of colours.
According to Arnheim,
10
the first attempt to define har-
monic colour combinations dates back to 1839, to Chev-
reul’s statement about harmony of similarities and har-
mony of contrasts. The 20th century has seen several
attempts at explain colour harmony.
11–14
Although these
are not guaranteed formulae, they can be considered clues
to understanding which colours work well together. These
general rules are mostly known as colour schemes, and
are broadly based on Chevreul’s early statement on colour
harmony. According to Burchett,
15
no acceptable model
exists for explaining the concept of colour harmony and
the author identifies key attributes and related terms con-
sistently used to describe it.
In recent years, studies of the dynamics of colour combi-
nations and colour harmony have been done using different
methods. Ha
˚rd and Sivik
16
developed a theoretical model
for colour combinations described on the Natural Color
System. Nemcsics
17
examined the harmony content of
scales found in different locations in various positions of
the axial sections of the Coloroid colour system. U
¨nver and
O
¨ztu
¨rk
4
used contrast quantity limits calculated by the ele-
ments of the Munsell Color System. Ou and Lou
18
devel-
oped a quantitative model for two-colour harmony. Ural
19
examined colour harmony in terms of the colour dynamics
of two-colour combinations by analysing the relations
between three dimensions of colour. However, there is still
no common and generally accepted method for determining
the dynamics of colour combinations.
Colour Emotions. Literature on colour emotions has
been authored mostly by experimental psychologists. The
method of semantic differential introduced by Osgood
et al.
20
is the generally accepted method for colour emo-
tion research. In this method, results are interpreted from
ratings on semantic scales and categories devised by fac-
tor analysis. Lack of control of the stimulus material and
the viewing conditions, and using isolated colour chips
are two points of criticism for such research. From this
point of view findings of studies investigating the consis-
tencies and/or contradictions between evaluations of
abstract/isolated and contextual situations are of great im-
portance. Such investigations are also carried out by com-
paring the profiles of semantic ratings and/or the extracted
factors obtained from different data sets.
The studies investigating the consistencies and/or con-
tradictions between evaluations of abstract/isolated and
contextual situations are compared in detail by Taft,
21
who states that some of the research supports that colour
is an attribute of an object, preferences of colours are
344 COLOR research and application
linked to the objects and colour chip preferences are not
generalizable (e.g., Norman and Scott; Wise and Wise;
Whitfield and Wiltshire; and Davidoff, cited in Taft
21
)
The findings of research which looks at colour-meaning
associations in a wider sense rather than merely at prefer-
ence, point out consistencies between abstract/isolated and
contextual situations to some extent. Osgood et al.
20
found that potency and activity ratings are largely unaf-
fected by the object, while evaluations are dependent on
the object. Taft compared semantic ratings of colour sam-
ples (chips) with those same colours applied to a variety
of familiar objects.
18
Analyses performed on the data
indicated that generally few significant differences existed
between chip and object ratings for the same colour. The
results of this study have implications for the use of col-
our chips in colour planning, as colour chips may serve
as an economical and expedient medium. In Sivik’s
22
research comparing colour chip ratings with colour ratings
of two types of simulated building exteriors, descriptive
scales are largely unaffected by context, while evaluative
scales are most affected by context. In research by Ural,
19
a subject group of architects first generated two-colour
combinations in an abstract medium and then, after apply-
ing those combinations to architecturally defined contexts,
subjects were asked to re-evaluate the new situations.
Findings showed that subjects generally agreed that the
colour combinations had to be reconsidered and they pro-
posed significant colour changes for the new conditions.
The findings also showed that contextual characteristics
are important components of the decision-making process
in architectural colour design. Nevertheless, in an archi-
tectural context, space has its own characteristics which
are quite different from those that an object has. Billger’s
article dealt with the way colours in a room affect each
other through contrast effects and reflections.
23
The men-
tioned study was based on the visual observations of col-
ours in a sequence of experimental situations. It was
found that the effects of both contrasts and reflections
were evident in the spatial appearance of colours.
In recent research, the use of colour chips has also been
examined by making comparisons with the spatial con-
text.
24–26
Destefani and Whitfield questioned how people
select colours for real settings and how they make such
aesthetic decisions.
24
According to the authors, ‘‘after more
than a century of empirical research in psychology, answers
to such elementary question remain elusive and the main
reason is that psychology has focused upon questions of
theoretical interest within psychology, and used colour as a
means to this end. In doing so, typically researchers have
investigated people’s responses to more abstract stimuli—
mainly colour chips—and neglected colour selection for
real-world objects.’’ When Hogg et al.
25
compared seman-
tic ratings of colour samples and a simulated interior space,
they found generally high correlations. However, Stahre
et al.
26
compared the results of a colour-chip study to one
of a full-scale room, and found that the colours were per-
ceived as more distinct and stronger and that they aroused
much stronger emotions in the full-scale room.
Problems and Aims
The literature review reveals the fact that theoretical
knowledge about colour harmony and colour emotions is
based mostly on isolated colour samples and abstract
combinations, and it seems that such knowledge needs to
be reconsidered within an architectural context. Addition-
ally, during the architectural colouring process, colour
chips/samples are widely used at least for initial colour
decisions. Research investigating the consistency of col-
our evaluations in isolated/abstract media and contextual
situations in the field of architecture has great importance
both for knowledge implementation in the field and for
developing architectural colouring methods. Although
there is supplementary research that sheds some light on
the question, any research investigating the consistency of
colour decisions throughout all stages—from isolated
chips to abstract media and finally to the form/space sim-
ulation(s)—has not been found in the literature.
The aim of this study is thus to analyze the assembly
of the stages of the colour design process within an archi-
tectural context and to assist and sustain designers/archi-
tects by sparking a discussion on colour design method.
The scope of the study is circumscribed as comparing the
different kinds of visualizations with each other, rather
than comparing them with real-life situations. Therefore,
this research was designed to explore:
1. Whether the evaluations (semantic ratings) of the col-
our schemes have associations or differences when the
medium changes from colour chips to abstract compo-
sitions, from abstractions to contextual situations and
from 2D to 3D media,
2. Which constructs (semantic scales) go together and how
these constructs are categorized (factorized) when the me-
dium that the colour scheme is presented in differs, and
3. Whether the same colour and scheme characteristics
predict the evaluations (semantic ratings) of the colour
schemes when presented in different media.
METHODS
To investigate the above research questions, the method of
semantic differential was used. In the questionnaire, seven
bipolar five-step scaled adjective pairs were applied. For
investigating constructs related to an architectural context,
evaluative adjectives were preferred and descriptive adjec-
tives were not found to be relevant to the scope and objec-
tives of the research. The chosen adjective pairs were cate-
gorized under the three primary factors of Osgood et al.
20
;
evaluation, activity and potency (pleasure, arousal and
dominance in Mehrabian).
27
The list of adjectives and the
related factors are given in Table I.
Coloured materials used for this research were: colour
schemes presented as colour chips, abstract compositions,
perspective drawings and 3D models—the sequential tools
of colour-design process. These four media were prepared
for a simple architectural space (4 m 35m33m)
assumed to have a study/work function.
Volume 35, Number 5, October 2010 345
Colour chips were prepared as 3 cm 33 cm samples
[Fig. 1(a)]. Abstract compositions were generated as a 9
by 9 checkerboard pattern of 3 cm 33 cm squares, con-
sidering both proportional relations and the adjacency of
coloured areas [Fig. 1(b)]. Perspectives were drawn as
one point on A4 drawing paper [Fig. 1(c)]. In order not to
cause value differences that might affect the appearance
of colour schemes, light, shade and shadow effects were
eliminated in the perspective drawings. 3D models were
produced at a scale of 1/20 [Fig. 1(d)]. For homogeneous
illumination and in order to prevent the effects of glare,
shade and shadow daylight from the north was the pre-
ferred light source.
In an architectural context and particularly in spatial
design, the number of colours combined in a scheme dif-
fers depending on various factors. In this research, the
visual differentiation of surfaces was taken into considera-
tion and the number of colours limited to five. Colours
were chosen for the ceiling, floor, back and left walls,
right wall and the furniture. In the research, 15 different
colour schemes were tested against semantic scales.
Scheme characteristics of colour combinations were
defined by the three variables of the Munsell Color Sys-
tem (Hue, Value, Chroma). Maximum contrast quantity
coefficients
4
were calculated; coefficients under 0.34 were
defined as ‘‘similar,’’ coefficients over 0.66 defined as
‘‘contrasting’’ and coefficients between the two defined as
‘‘moderate.’’ For the cases defined as ‘‘similar,’’ the domi-
nant descriptive characteristics of colours were stated as
follows: ‘‘warm’’ or ‘‘cold’’ for Hue, ‘‘light’’ or ‘‘dark’’
for Value, ‘‘weak’’ or ‘‘strong’’ for Chroma. For the cases
defined as ‘‘contrasting’’ and ‘‘moderate,’’ the descriptive
quality was stated as ‘‘both’’ for data analyses. Corre-
sponding Munsell notations, schemes and the dominant
descriptive characteristics of the colour combinations can
be seen in Table II.
Subjects consisted of 40 second- and third-year interior
architecture students, who were asked to evaluate 15 col-
our schemes presented in four different media. First, sub-
jects were informed about the given space and assumed
function, then materials were presented on separate desks
and their order changed after each five subjects. In the
experiment room, only indirect daylight (from the north)
was allowed as illumination. The same viewing and illu-
mination conditions of coloured materials were provided
during the experiment. Data consisted of subject ratings
on seven bipolar, five-step semantic differential scales of
15 colour schemes, each presented in four different media
and colour-scheme characteristics (similarity or contrast
and dominant descriptive characteristics over three colour
dimensions). Each subject rated a total of 60 cases against
all scales. The collected data were analysed by SPSS 8.0
for Windows.
RESULTS
This research aimed to examine the effects of context
throughout the architectural colour-design process by
means of different media that, represented the sequential
steps of the process. Data analysis based on the three
research questions is as follows:
i. To see whether there are any associations between
the evaluations of colour schemes presented as chips,
abstract colour graphics, perspective drawings and mod-
els, data sets were analyzed by correlating the semantic
ratings of each medium. Table III shows the correlations
(Pearson’s rand P) between colour chips, abstract
graphics, perspective drawings and 3D models over seven
semantic scales.
Results showed that there is always a significant posi-
tive relationship between semantic ratings of three of the
media—abstract colour graphics, perspective drawings
and models. However, this cannot be said in every case
for colour chips. Colour chips showed significant associa-
tions with all the other media only for the adjective pairs
harmonious/discord and exciting/calming. On the con-
trary, there is no significant association between chips and
TABLE I. Bipolar adjectives used in the questionnaire
and the related factors identified by Osgood et al.
17
Adjectives Factors
a1 Harmonious/discord Evaluation
a2 Pleasant/unpleasant Evaluation
a3 Comfortable/uncomfortable Evaluation
a4 Spacious/confined Potency
a5 Static/dynamic Potency þactivity
a6 Exciting/calming Activity
a7 Introverted/extroverted Activity
FIG. 1. An example of the coloured materials: Colour chips (a), 2D abstract compositions (b), perspective drawings (c),
and 3D models (d).
346 COLOR research and application
all the other media for the adjective pair pleasing/unpleas-
ing. To investigate medium effect on the semantic ratings
of the four media, multiple comparisons in ANOVA
(analysis of variance) were performed. Table IV shows
the medium effect on the semantic ratings. Results
showed an overall significant effect for the type of me-
dium for six of the adjectives. Only for one pair of adjec-
tives (exciting/calming) was no significant effect of me-
dium observed. In Table IV, Scheffe’s range shows where
the medium effect appears.
Scheffe’s range test found that the evaluations of the
colour chips differed from the abstract graphics only for
the adjective pair introverted/extroverted (P¼0.045).
Evaluation of the colour chips differed from the perspec-
tive drawings for the adjective pairs pleasing/unpleasing
(P¼0.013), spacious/confined (P¼0.000) and comforta-
ble/uncomfortable (P¼0.000). Evaluation of the colour
chips differed from the model for the adjective pairs com-
fortable/uncomfortable (P¼0.002) and static/dynamic (P
¼0.008). Semantic ratings of the abstract graphics dif-
fered from the perspective drawings for the adjective pairs
spacious/confined (P¼0.000) and introverted/extroverted
(P¼0.000). They differed from the model for the adjec-
tive pair introverted/extroverted (P¼0.002). Differences
between the ratings of the perspective drawings and the
model appear for the adjective pairs harmonious/discord
(P¼0.036), spacious/confined (P¼0.028), comfortable/
uncomfortable (P¼0.000) and static/dynamic (P¼
0.042).
ii. Next, factor analysis was performed in order to
observe which constructs (semantic scales) go together
and how these constructs are categorized (factorized)
when the medium that the colour scheme is presented in
differs. Table V shows the variance percentages
accounted for the factors and the orthogonal factor load-
ing matrices for seven adjectives of colour chips, abstract
graphics, perspective drawings, and 3D models. On the ta-
ble the four media arranged as columns. Classifications
given on top of them show the characteristics of the four
media (e.g., colour chips are presenting 2-dimensional,
abstract media where colours are loosely combined, while
abstract graphics are also 2-dimensional and abstract, but
colours are organized as compositions). On the rows the
categorization of the constructs are listed under the order
of extracted factors.
For the colour chips the first factor accounted for
28% of the variance, the second factor 26% and the third
factor 25%. The first factor gathers the adjective pairs
TABLE II. Corresponding Munsell notations and scheme/colour characteristics.
Schemes N1 N2 N3 N4 N5 HUE Value Chroma
1 5Y9/2 5Y9/6 2,5GY8/3 25GY3/1 7,5Y8/7 Similar-warm Contrasting Contrasting
2 10Y9/6 7,5R3/4 5YR6/14 5R4/12 10Y8/10 Similar-warm Contrasting Contrasting
3 7,5RP9/2 7,5R2/6 7,5R6/2 6,5R4/12 7,5RP4/8 Moderate Contrasting Contrasting
4 10B7/4 7,5GY8/4 5B7/4 5P7/6 7,5BG7/4 Contrasting Similar-light Similar-weak
5 9/0 5B5/1 5B9/2 2,5P4/4 7,5B5/8 Similar-cold Moderate Contrasting
6 5Y8,5/8 5Y8/10 3,75YR6/12 2,5PB7/8 2,5GY7/8 Contrasting Similar-light Similar-strong
7 5YR9/2 10R7/8 10B8/4 10R6/12 5PB5/10 Contrasting Similar-light Contrasting
8 5Y9/2 5P8/4 5YR9/6 5P7/6 5Y8/6 Contrasting Similar-light Similar-weak
9 7,5Y9/2 10B8/4 5GY8/6 2,5PB6/10 7,5Y9/4 Contrasting Similar-light Contrasting
10 10R9/2 10R8/4 10R6/6 7,5R10/4 10R5/8 Similar-warm Moderate Moderate
11 7,5BG8/2 7,5BG4/2 7,5B6/6 7,5BG4/4 2,5BG4/4 Similar-cold Moderate Similar-weak
12 7/0 10YR8/10 10R6/12 8,75R5/12 2,5Y8,5/12 Similar-warm Similar-light Contrasting
13 5B7/2 5B4/4 5P5/6 10GY4/6 10PB4/4 Contrasting Similar-dark Similar-strong
14 5/0 7,5P5/2 7,5P3/2 3,75R4/12 10RP2/6 Similar-warm Similar-dark Contrasting
N1, ceiling; N2, floor; N3, back and left walls; N4, right wall; N5, furniture.
TABLE III. Associations among colour chips, abstract graphics, perspective drawings, and 3d models over
seven semantic scales.
Chips/abstract
graphics
Chips/
perspective
drawings Chips/model
Abstract
graphics/
perspective
drawings
Abstract
graphics/
model
Perspective
drawings/
model
rPrPrPrPrPrP
Harmony 0.279 0.000** 0.141 0.005* 0.254 0.000** 0.494 0.000** 0.464 0.000** 0.440 0.000**
Pleasantness 0.049 0.339 0.000 0.997 0.053 0.296 0.310 0.000** 0.353 0.000** 0.371 0.000**
Comfort 0.161 0.001** 0.058 0.251 0.210 0.000** 0.339 0.000** 0.332 0.000** 0.375 0.000**
Spaciousness 0.239 0.000** 0.041 0.416 0.285 0.000** 0.405 0.000** 0.456 0.000** 0.399 0.000**
Dynamism 0.055 0.281 0.249 0.000** 0.221 0.000** 0.323 0.000** 0.283 0.000** 0.454 0.000**
Excitement 0.243 0.000** 0.120 0.018* 0.231 0.000** 0.361 0.000** 0.424 0.000** 0.304 0.000**
Introversion 0.116 0.022* 0.084 0.096 0.159 0.002* 0.365 0.000** 0.288 0.000** 0.358 0.000**
Volume 35, Number 5, October 2010 347
pleasing/unpleasing and harmonious/discord, and seems to
be ‘‘evaluation.’’ The second factor gathers the adjective
pairs static/dynamic, calming/exciting and introverted/
extroverted, and seems to be ‘‘activity.’’ The third factor
gathers the adjective pairs spacious/confined and comfort-
able/uncomfortable. For the abstract compositions, the
first factor accounted for 28% of the variance, the second
factor 26% and the third factor 25%. The first factor gath-
ers the adjective pairs static/dynamic, calming/exciting
and introverted/extroverted. The second factor gathers the
adjective pairs pleasing/unpleasing and harmonious/dis-
cord, and the third factor gathers the adjective pairs spa-
cious/confined and comfortable/uncomfortable.
The perspective drawings and the models showed
nearly the same order of factors and patterns of gathering
adjectives. Differences are observed only in the percen-
tages of the variances and in the order of two adjectives
in the second factor. For the perspective drawings, the
first factor accounted for 38% of the variance and the sec-
ond factor 30%. For the models, the first factor accounted
for 39% of the variance and the second factor 29%. The
first factor gathers the adjective pairs comfortable/uncom-
fortable, pleasing/unpleasing, harmonious/discord and spa-
cious/confined, and the second factor gathers the adjec-
tives static/dynamic, calming/exciting and introverted/
extroverted.
iii. Stepwise multiple regression was performed in
order to see whether the same colour and scheme charac-
teristics predict the semantic ratings of colour schemes
presented in different media. Table VI shows the predic-
tors of adjectives for each medium.
Results showed that different colour and scheme char-
acteristics are determinative over the ratings of adjectives
when the medium changes. In only one case (pleasing/
unpleasing presented in perspective drawings) are no sig-
nificant predictors observed. The only consistency is
between the predictors of ratings of the adjective pair
observed for introverted/extroverted. For this adjective
pair, Value was entered first and explained 6%, 8%, 15%,
and 7% of the variance in introversion for colour chips,
abstract graphics, perspective drawings and models,
respectively. Other characteristics explained a further per-
centage not greater than 2%. In 17 of the total 28 cases,
the explained percentage of variance (Rsquare change)
appeared at under 10%.
DISCUSSION
The present research was conducted to explore the archi-
tectural/spatial colouring process by investigating the con-
sistency of semantic ratings of four sequential stages of
the architectural colour design process, namely, colour
chips/samples, abstract compositions, perspective draw-
ings and 3D models. The results of the study have contri-
butions not only for previous research comparing colour
samples and applied colours in different fields of design,
but also for propounding a methodology for the architec-
tural colour design process.
From the findings it can be concluded that abstract media
can be used as a tool for initial colour-design decisions in
architecture, however, scrutinizing colour design through a
media simulating contextual characteristics and three-
dimensional relations is inevitable during the progression
of the architectural colour-design process. This work con-
tradicts the research indicating a strong correspondence
between colour chips and contextualized colours (e.g.,
Taft,
21
Sivik,
22
Hogg et al.
25
), the results showed that the
colour chips cannot be considered as efficient representa-
tives of architectural colouring. Significant associations
observed between the semantic ratings of abstract graphics,
and perspective drawings and models means that the
abstract graphics were evaluated more similarly with the
contextual media. Therefore, it can be said that abstract
compositions presenting both proportional relations and
adjacency of colours seem to be more appropriate tools
than colour chips.
There is an overall significant medium effect for six of
the seven adjectives: only for ‘‘excitement’’ is no medium
effect observed (see the results of multiple comparisons
in ANOVA—Table IV). This observation may indicate
that this construct is evaluated independently from the
TABLE IV. Media effects on semantic ratings of seven adjectives (multiple comparisons in ANOVA).
FP
Scheffe’s range
Medium P
Harmony 3.28 0.020* Perspective drawing/model 0.036*
Pleasantness 3.81 0.010* Chips/perspective drawing 0.013*
Comfort 15.56 0.000** Chips/perspective drawing 0.000**
Chips/model 0.002*
Perspective drawing/model 0.000**
Spaciousness 10.80 0.000** Chips/perspective drawing 0.000**
Abstract graphic/perspective drawing 0.000**
Perspective drawing/model 0.028*
Dynamism 4.78 0.003* Chips/model 0.008*
Perspective drawing/model 0.042*
Excitement 0.70 0.550
Introversion 10.79 0.000** Chips/abstract graphic 0.045*
Abstract graphic/perspective drawing 0.000**
Abstract graphic/model 0.002*
348 COLOR research and application
TABLE V. The orthogonal factor loading matrices for seven adjectives of colour chips, abstract graphics, perspective drawings and 3D models.
2D Medium 3D Medium
Abstract Contextual
Combination Composition
Colour Chips Abstract Graphics Perspective Drawings 3D Models
Factor
1
EVALUATION
a
‘‘pleasure’’
Pleasing–
unpleasing
Harmonious–
discord
0.88
b
0.86
b
28%
c
ACTIVITY
a
‘‘arousal’’
Static–dynamic
Calming–exciting
Introverted–
extroverted
0.87
b
0.82
b
0.69
b
28%
c
EVALUATIONþ
ACTIVITYþ
POTENCY
a
‘‘agreeableness’’
Comfortable–
uncomfortable
Pleasing–
unpleasing
Harmonious–
discord
Spacious–
confined
0.87
b
0.84
b
0.76
b
0.72
b
EVALUATIONþ
ACTIVITYþ
POTENCY
a
‘‘agreeableness’’
Comfortable–
uncomfortable
Pleasing–
unpleasing
Harmonious–
discord
Spacious–
confined
0.87
b
0.82
b
0.79
b
0.72
b
38%
c
39%
c
Factor
2
ACTIVITY
a
‘‘arousal’’
Static–dynamic
Calming–exciting
Introverted–
extroverted
0.84
b
0.82
b
0.63
b
26%
c
EVALUATION
a
‘‘pleasure’’
Pleasing–
unpleasing
Harmonious–
discord
0.85
b
0.84
b
26%
c
ACTIVITY
a
‘‘arousal’’
Static–
dynamic
Calming–
exciting
Introverted–
extroverted
0.88
b
0.79
b
0.69
b
30%
c
ACTIVITY
a
‘‘arousal’’
Calming–
exciting
Static–dynamic
Introverted–
extroverted
0.84
b
0.82
b
0.66
b
29%
c
Factor
3
EVALUATION
þPOTENCY
a
‘‘estimation’’
Spacious–
confined
Comfortable–
uncomfortable
0.87
b
0.77
b
EVALUATION
þPOTENCY
a
‘‘estimation’’
Spacious–
confined
Comfortable–
uncomfortable
0.87
b
0.74
b
25%
c
25%
c
a
Factors by Osgood et al.
17
b
Orthogonal factor loadings for the adjectives.
c
Variance percentages accounted for the factors.
Volume 35, Number 5, October 2010 349
characteristics of the medium. According to Scheffe’s
range test, the medium effect mostly differentiates abstract
(chips and abstract graphics) and contextual (perspective
drawings and model) visualizations. It can therefore be
said that the subjects’ criteria for evaluating abstract
media differed from their criteria for contextual situations.
Between the perspective drawing and the three-dimen-
sional model, a significant medium effect was observed
for ‘‘spaciousness,’’ ‘‘comfort’’ and ‘‘dynamism.’’ This
finding may indicate that evaluations of these constructs
are more sensitive to a 3D appearance of the colour
scheme and/or architectural/spatial aspects.
Results of the factor analysis support the finding that
abstract media is being evaluated differently than contex-
tual situations. For the colour chips and the abstract com-
positions, ‘‘harmony’’ and ‘‘pleasure’’ extracted a factor
different than ‘‘spaciousness’’ and ‘‘comfort’’ did; how-
ever these four constructs extracted a single factor for
the perspective drawings and the models. The constructs
‘‘ harmony,’’ ‘‘pleasantness,’’ ‘‘spaciousness’’ and ‘‘comfort’’
are accounted for as the first factor for contextualized media
(perspective drawing and model) and this factor seems to
be related to the ‘‘agreeableness/appropriateness’’ of the
colour schemes for the defined context. This indicates that
in an architectural/spatial context, constructs related to
architectural/spatial quality cannot be separated from ones
related to the sense of taste or the factor of evaluation
20
(pleasure
27
) (see the results of factor analysis—Table V).
This finding also supports previous research
23,28
by
revealing the distinctive nature of spatial colour. For
the abstract media, ‘‘spaciousness’’ and ‘‘comfort’’ are
accounted for as the last factor. Different from
‘‘harmony’’ and ‘‘pleasure,’’ these two constructs are
seemingly the determinants of architectural/spatial quality
and relevant to an ‘‘estimation’’ for architectural/spatial
use of colours.
The extraction of ‘‘harmony’’ and ‘‘pleasure’’ for
always the same factor recalls the hypothesis by Ou and
Lou
9
: ‘‘there is a strong link between colour harmony and
the emotion pleasantness’’ and ‘‘a harmonious combina-
tion can always give viewers a pleasing impression.’’
However in consideration of the subject group (interior
architecture students), this finding cannot be generalized.
In all the four media, the constructs ‘‘dynamism,’’
‘‘ excitement,’’ and ‘‘extroversion’’ are always accounted
for together. This result indicates that the evaluations of
these constructs are related regardless of the media in
which the colour scheme is presented. They also seem to
be related to the factor of activity
20
(arousal
27
) or stimula-
tion (see the results of factor analysis—Table V). This
implies that activity ratings are largely independent from
the media and/or the spatial context.
TABLE VI. Stepwise multiple regression of predictors of adjectives (only significant predictors are included).
2D Medium 3D Medium
Abstract Contextual
Combination Composition
Colour chips Abstract graphics Perspective drawings Model
variable
Rsq
change
Sig.
of tvariable
Rsq
change Sig. of tvariable
Rsq
change Sig. of tvariable
Rsq
change Sig. of t
Harmony V 0.068 0.000 V 0.062 0.000 C 0.038 0.000 schH 0.060 0.000
C 0.058 0.000 H 0.048 0.001 C 0.051 0.000
schC 0.012 0.020 C 0.017 0.016 schC 0.015 0.012
Pleasantness V 0.075 0.000 V 0.076 0.000 schH 0.027 0.003
C 0.026 0.001 H 0.016 0.008 H 0.015 0.014
Comfort C 0.123 0.000 H 0.133 0.000 V 0.062 0.000 C 0.108 0.000
schC 0.024 0.001 C 0.012 0.020 schC 0.058 0.000 H 0.034 0.000
schH 0.042 0.000
H 0.011 0.025
Spaciousness C 0.110 0.000 H 0.139 0.000 V 0.298 0.000 V 0.129 0.000
H 0.042 0.000 V 0.094 0.000 schC 0.015 0.003 C 0.048 0.000
C 0.030 0.000 schH 0.015 0.004 H 0.020 0.004
schH 0.016 0.004
schC 0.016 0.007
Dynamism schC 0.063 0.000 C 0.051 0.000 V 0.135 0.000 V 0.069 0.000
V 0.029 0.003 schH 0.036 0.000 C 0.107 0.000 C 0.030 0.000
schH 0.015 0.011 V 0.028 0.001 schH 0.018 0.002 schH 0.024 0.002
schC 0.021 0.003
Excitement C 0.044 0.000 schH 0.084 0.000 C 0.149 0.000 schH 0.066 0.000
schH 0.032 0.000 C 0.057 0.000 V 0.038 0.000 V 0.047 0.000
schV 0.011 0.035 schH 0.015 0.007 schV 0.044 0.000
C 0.039 0.000
schC 0.009 0.034
Introversion V 0.060 0.000 V 0.082 0.000 V 0.154 0.000 V 0.066 0.000
schH 0.020 0.003 H 0.021 0.003
H 0.019 0.004 schH 0.017 0.007
C 0.018 0.005 schC 0.017 0.008
Colour characteristics; H: Hue; V: Value; C: Chroma.
Scheme characteristics; schH: similarity/contrast in Hue; schV: similarity/contrast in Value; schC: similarity/contrast in Chroma.
350 COLOR research and application
Results of the multiple regression analysis showed that
different colour and scheme characteristics are determina-
tive over the ratings of adjectives, however, the percen-
tages explain a maximum total of 30% in the variance of
evaluation of constructs. These results (given in Table VI)
indicate that colour-combination emotions, and media
effects on their evaluation have complex interrelations
and inner dynamics, which are not studied within the
scope of this research. Nevertheless, some rough interpre-
tations can be made about the general pattern. According
to the results, colour characteristics (dominant Hue, domi-
nant Value and dominant Chroma) seem to be determi-
nant for the evaluation of abstract media while scheme
characteristics (similarity or contrast in Hue, similarity or
contrast in Value and similarity or contrast in Chroma)
appear more effective for the evaluations of contextual
media (especially for 3D models). These observations
strengthen the difference between abstract and contextual-
ized media used as different visualization techniques in
the architectural colour design process.
The overall findings of this research have implications
for colour designers who work on architectural/spatial col-
ouring. In architecture, colour chips can be used for initial
colour design decisions, however during the colour design
process the initial decisions have to be reconsidered by
using other media simulating complex colour and colour-
space relations. Abstract compositions presenting both
proportional relations and adjacency of colours seem to
be more appropriate representatives of colour-design com-
pared to colour chips. Perspective drawings may serve to
visualize architectural/spatial configuration of colour-
design, but three dimensional appearance of colour com-
bination can only be observed on 3D models. Thus it can
be concluded that colour chips, abstract compositions,
perspective drawings and 3D models may be used as se-
quential media during the progression of the process.
Light is a very important integral of colour perception
and spatial evaluation. Lighting types (natural and artificial
light) and illumination levels do also affect evaluations of
colour schemes presented in 3D media.
29
Examining the
effects of different lighting conditions (e.g., different light-
ing qualities, quantities and patterns), working with other
simulation techniques (e.g., computerized media, 3D models
at larger or full-scale, and real-life environment) and differ-
ent subject groups (e.g., designers/architects and nondesign-
ers/architects) may offer different results that could be com-
pared with the ones of this study.
1. Green-Armytage P. The value of knowledge for colour design. Color
Res Appl 2006;31:253–269.
2. Lenclos JP. Living in colour. In: Porter T, Mikellides B, editors.
Colour for Architecture. New York: Van Nostrand Reinhold; 1976.
p 73–77.
3. Pile JF. Color in Interior Design. New York: McGraw Hill; 1997.
4. U
¨nver R, O
¨ztu
¨rk LD. An example of facade colour design of mass
housing. Color Res Appl 2002;27:291–299.
5. Smith D. Environmental colouration and/or the design process. Color
Res Appl 2003;28:360–365.
6. Mikellides B. Conflicting experience of colour space. In: Simon JG,
editor. In: Prooceedings of the Forth IAPC Congress. Belgium: Lou-
vain La Neuve; 1979. Vol. 2.
7. Janssens J, Mikellides B. Color research in architectural education—
A cross-cultural explorative study, Color Res Appl 1998;23:328–
334.
8. Akbay S. An analysis of basic design students’ intuitive and analytic
attitudes in colour decisions (MFA dissertation). Bilkent University,
Ankara, Turkey, 2003.
9. Ou L-C, Lou MR. A colour harmony model for two-colour combina-
tions. Color Res Appl 2006;31:191–204.
10. Arnheim R. Art and Visual Perception. Berkeley: University of Cali-
fornia Press; 1974.
11. Birren F. Principles of Color. Pennsilvania: Schiffer Publ. Ltd; 1987.
12. Itten J. The Elements of Color. New York: Van Nostrand Reinhold;
1970.
13. Doyle ME. Color Drawing. New York: Van Nostrand Reinhold;
1981.
14. De Grandis L. Theory and Use of Color. New York: Prentice Hall-
Abrams; 1986.
15. Burchett KE. Color harmony. Color Res Appl 2002;27:38–31.
16. Ha
˚rd A, Sivik L. A theory of colours in combination—A descriptive
model related to the NCS color-order system. Color Res Appl
2001;26:4–28.
17. Nemcsics A. Experimental determination of laws of color harmony.
I. Harmony content of different scales with similar hue. Color Res
Appl 2007;32:477–488.
18. Ou L-C, Lou MR, Woodcock A, Wright A. A study of colour emo-
tion and colour preference. II. Colour emotions for two colour com-
binations. Color Res Appl 2004;29:292–298.
19. Ural SE. Color in Architecture: Effects of Color Dynamics on Color-
ing Artificial Environments (PhD dissertation). Karadeniz Technical
University, Trabzon, Turkey, 1995.
20. Osgood CE, Suci GJ, Tannenbaum PH. The Measurement of Mean-
ing. Urbana: University of Illinois Press; 1957.
21. Taft C. Color meaning and context: Comparison of semantic ratings
of colors on samples and objects. Color Res Appl 1997;22:40–50.
22. Sivik L. Color meaning and perceptual color dimensions: A study of
exterior colors. Go
¨teborg Psychol Rep 1974;4:1–24.
23. Billger M. Colour combination. Effects in experimental rooms. Color
Res Appl 1999;22:230–242.
24. Destefani LRG, Whitfield TWA. Esthetic decision-making: How do
people select colours for real settings? Color Res Appl 2008;33:55–60.
25. Hogg J, Goodman S, Porter T, Mikellides B, Preddy DE. Dimen-
sions and determinants of judgements of colour samples and a simu-
lated interior space by architects and non-architects. Br J Psychology
1979;70:231–242.
26. Stahre B, Ha
˚rleman M, Billger M. Colour Emotions in Larger and
Small Scale. In: Prooceedings of the AIC Interim Meeting: Colour
and Paints, Brazil, 2004.
27. Mehrabian A. Pleasure-arousal-dominance: A general framework for
describing and measuring individual differences in temperament.
Curr Psychol 1996;14:261–292.
28. Ha
˚rleman M, Werner I-B, Billger M. Significance of colour on
room character: Study on dominantly reddish and greenish colours
in north- and south-facing rooms. Colour Des Creativity 2007;1:1–
15.
29. Ural S, Olguntu
¨rk N, Yilmazer S. Qualitative and quantitative
evaluations of colour schemes on 2D graphics, 3D models and artifi-
cial lighting alternatives (in Turkish), 6. In: Ulusal Aydnlatma Kon-
gresi (6th National Lighting Congress), Istanbul, Tu
¨rkiye, 2006.
Volume 35, Number 5, October 2010 351
