31
1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the Ultra-colour System Li-Chen Ou*, M. Ronnier Luo*, Guihua Cui*, Angela Wright** * : Colour & Imaging Institute, University of Derby, Kingsway House, Kingsway, Derby, DE22 3HL, United Kingdom **: Colour Affects, 908, Keyes House, Dolphin Square, London SW1V 3NB Abstract This report includes two psychophysical experiments: 1) the relationship between colour combinations and adjective combinations and 2) the verification of the Wright theory on colour harmony. Both experiments were carried out in the 5 countries: Britain, France, Germany, Spain, and Sweden. A total of 47 females and 72 males took part in Experiment 1; 48 females and 71 males in Experiment 2. In the first experiment, 100 adjectives and 32 colours were used as stimuli presented on a calibrated Cathode Ray Tube (CRT) display. All the colour stimuli were selected evenly from the 4 colour groups, CG 1 to CG 4, defined by the Ultra-colour system, and so were the adjectives, from AG 1 to AG 4. Five-colour combinations and 5-adjective combinations were generated from the stimuli. In the experiment, for each judgement an adjective combination and 4 colour combinations were presented simultaneously on screen and observers were asked to select one of the 4 colour combinations that appeared in correlation with the adjective combination presented. Experimental results show good agreement between observer responses and the Wright theory in AG 2. In Experiment 2, the colour stimuli were the same as those used in Experiment 1. From these colours 2 sets of five-colour combinations were generated, each containing 224 colour combinations. In Set 1 each colour combination was generated by the guideline of Wright theory, i.e. the colours in a colour combination were all from the same CG. In Set 2 each colour combination was a copy of a Set 1 colour combination where a component colour was replaced by one from a different CG. Experimental results show good agreement between observer response and the Wright theory in all the CGs. The experimental results were compared between the observer groups. The Spanish data were found in best agreement with the other observer groups, followed by German, Swedish, Chinese, French, and British-2. The British-1 data were found in poor correlation with all the other groups.

A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

  • Upload
    hahanh

  • View
    220

  • Download
    2

Embed Size (px)

Citation preview

Page 1: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

1

Final Report for the OKI Project

A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with

the Ultra-colour System

Li-Chen Ou*, M. Ronnier Luo*, Guihua Cui*, Angela Wright**

* : Colour & Imaging Institute, University of Derby, Kingsway House, Kingsway, Derby, DE22 3HL, United Kingdom

**: Colour Affects, 908, Keyes House, Dolphin Square, London SW1V 3NB

Abstract This report includes two psychophysical experiments: 1) the relationship between colour combinations and adjective combinations and 2) the verification of the Wright theory on colour harmony. Both experiments were carried out in the 5 countries: Britain, France, Germany, Spain, and Sweden. A total of 47 females and 72 males took part in Experiment 1; 48 females and 71 males in Experiment 2.

In the first experiment, 100 adjectives and 32 colours were used as stimuli presented on a calibrated Cathode Ray Tube (CRT) display. All the colour stimuli were selected evenly from the 4 colour groups, CG 1 to CG 4, defined by the Ultra-colour system, and so were the adjectives, from AG 1 to AG 4. Five-colour combinations and 5-adjective combinations were generated from the stimuli. In the experiment, for each judgement an adjective combination and 4 colour combinations were presented simultaneously on screen and observers were asked to select one of the 4 colour combinations that appeared in correlation with the adjective combination presented. Experimental results show good agreement between observer responses and the Wright theory in AG 2.

In Experiment 2, the colour stimuli were the same as those used in Experiment 1. From these colours 2 sets of five-colour combinations were generated, each containing 224 colour combinations. In Set 1 each colour combination was generated by the guideline of Wright theory, i.e. the colours in a colour combination were all from the same CG. In Set 2 each colour combination was a copy of a Set 1 colour combination where a component colour was replaced by one from a different CG. Experimental results show good agreement between observer response and the Wright theory in all the CGs.

The experimental results were compared between the observer groups. The Spanish data were found in best agreement with the other observer groups, followed by German, Swedish, Chinese, French, and British-2. The British-1 data were found in poor correlation with all the other groups.

Page 2: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

2

1. INTRODUCTION

This is a final report for a project sponsored by OKI Europe Ltd. In this report the Ultra-colour system, a computer controlled colour selection system developed by Angela Wright and the Colour & Imaging Institute (CII), is verified by psychophysical experiments in the 5 countries: Britain, France, Germany, Spain, and Sweden. In Britain, the two languages, Chinese and English, were used in the experiments by two groups of observers.

Central to this project is the research of colour emotion and colour harmony. Colour emotion indicates the connection between colours and affective words, such as “warm” and “active”, and can be predicted by the knowledge of CIE colorimetric colour appearance attributes such as hue, lightness, and chroma (Ou et al., 2004a-b). Colour emotion is essential for designers to perform colour planning. Colour harmony correlates colour combinations with “pleasant feelings” (Judd and Wyszecki, 1975).

The Ultra-colour system is a colour selection tool with which users can generate harmonious colour combinations according to a specific colour emotion. The theory behind the system was originated by Angela Wright (Wright, 1995). It features the following novel thinking, which is called the Wright Theory in this report.

All colours are categorised into four groups. Each group has a natural affinity to a unique personality type. Colours within each group harmonise. Each colour-group or personality-type is universal across different cultures. There are a number of adjectives associated with each group.

The goals of this project include

1) To verify the underlying theory of the Ultra-colour system. 2) To investigate the difference of colour harmony response between cultures. 3) To demonstrate the effectiveness of colour reproduction using CRT colour sensor

technology.

For Goal 1, software was developed by the CII to verify the Ultra-colour theory. It includes the two experiments: colour adjective and colour harmony. The former investigates the connection between colour combinations and adjective combinations. The latter examines the performance of the Wright theory on colour harmony.

For Goal 2, the experiments were conducted using 6 languages (Chinese, English, French, Germany, Spanish, and Swedish) in 5 countries. The results were analysed to compare colour harmony response between different cultures.

Page 3: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

3

For Goal 3, the experiments were conducted on CRT monitors. Each monitor was calibrated with a colour sensor. The calibration software was developed by the CII.

2. METHODS

The two experiments were conducted on a CRT display at a dark room. A Sequel Chroma-4 colour sensor was used for CRT calibration by applying the GOG model (Berns, 1996), which ensures consistency of colour reproduction on the display over experimental sessions and over different sites. Basic design of the experiments is given below.

a) Experiment 1

The objective of this experiment is to investigate the connection between adjective combinations and colour combinations. Four groups of adjectives and 4 groups of colours were selected in accordance to the Wright theory, i.e. the 4 adjective groups, labelled AG 1 to AG 4, were selected in association with 4 personality type, and so were the 4 colour groups, CG 1 to CG 4. The Wright theory assumes the correlation between an adjective and a colour/colour combination if they are in association with the same personality type. The present experiment is to examine whether this theory applies to adjective combinations and colour combinations.

In the experiment, each observer was presented with 4 colour combinations at the 4 corners of screen and an adjective combination at the centre. Each adjective combination consisted of 5 adjectives and was supposed to generate a unique emotional feeling that represented the 5 adjectives in that group. Each colour combination comprised 5 colours and was presented in the form of a wheel. In the experiment observers were asked to pick one out of the 4 colour wheels that agreed best with the adjective combination presented.

b) Experiment 2

The objective of this experiment is to examine the Wright theory on colour harmony under the 6 cultures: Chinese, British, French, German, Spanish, and Swedish. The Wright theory suggests that there are 4 categories of colours and the colours of the same category create colour harmony.

The 4 colour groups used in Experiment 1, CG 1 to CG 4, were also used in the present experiment. In the experiment, two colour wheels were presented simultaneously on two sides (left and right) of a CRT monitor. Observers were asked to pick one colour wheel that appeared to harmonise better than the other. One of the 2 colour wheels was made by colours of the same CG, while the other colour wheel

Page 4: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

4

was made by colours all from the same CG but one. The Wright theory suggests that the former is more harmonious than the latter.

2.1 Adjective combinations

Four groups of adjectives (in English) were selected in association with the 4 personality types, Spring, Summer, Autumn, and Winter, as defined by Wright (1995). The 4 adjective groups were labelled AG 1 to AG 4. Each AG contained 25 adjectives, which were divided into 5 subgroups. Each subgroup formed a 5-adjective combination.

There were 20 adjective combinations used in the experiment, as summarised in Tables 1 (a) to (d), for AGs 1 to 4, respectively. The adjective combinations “1-a” to “1-e” were in association with the personality type Spring; “2-a” to “2-e” in association with Summer; “3-a” to “3-e” in association with Autumn; “4-a” to “4-e” in association with Winter.

The adjectives were selected originally in English and were translated into the 5 languages by native speakers: Chinese, French, German, Spanish, and Swedish, as presented in Tables 1 (a) to (d).

Table 1. Adjective combinations in 6 languages for AGs 1 to 4, presented from (a) to (d), respectively.

(a) English Chinese French Germany Spanish Swedish 1-a Fun ��� amusant spass divertido rolig Adaptable ���� adaptable anpassumgsfähig armonioso anpassningsbar Bright ��� intelligent hell llamativo klar Charming ��� charmant charmant encantador charmig Cheerful ��� joyeux freudig alegre gladlynt 1-b Lively ��� allègre lebendig vivo livfull Active ����� actif handelnd activo aktiv Enthusiastic ��� enthousiaste begeistert entusiasta entusiastisk High-spirited ��� plein de vie temperamentvoll animado uppåt Optimistic ��� optimiste optimistisch optimista optimistisk 1-c Clever ��� intelligent klug inteligente klipsk Cooperative ��� coopératif hilfsbereit integrador samarbetsvillig Friendly ��� amical freundlich amigable vänlig Informal ���� informel informell informal informell Practical ��� pragmatique praktisch práctico praktisk 1-d Fresh ��� frais frisch fresco fräsch New �� neuf neu nuevo ny Outgoing ��� extraverti kontaktfreudig extrovertido utåtriktad Impulsive ��� impulsif impulsiv impulsivo impulsiv Communicative ����� communicatif gesprächig comunicativo meddelsam 1-e Fanciful ����� capricieux einfallsreich ocurrente fantasifull Imaginative ����� plein d'imagination fantasievoll imaginativo uppfinningsrik Creative ����� créatif kreativ creativo kreativ Inspired ������ inspiré inspirierend inspirado inspirerad Mischievous ��� malfaisant boshaft travieso skälmsk

Page 5: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

5

(b)

English Chinese French Germany Spanish Swedish 2-a Calm ��� calme Ruhe tranquilo lugn Peaceful ��� paisible friedlich pacífico rofylld Nurturing ��� nourricier Nahrung maternal fostrande Courteous ���� courtois höflich amable hövlig Compassionate ����� compatissant mitfühlend dulce barmhärtig 2-b Formal ��� cérémonieux formell formal formell Elegant ��� élégant elegant elegante elegant Gracious ��� bienveillant gütig bonito älskvärd Modest ��� modeste anständig modesto anspråkslös Reserved ��� réservé reserviert discreto reserverad 2-c Cool ��� cool zurückhaltend sereno sval Dignified ���� digne würdevoll digno värdig Fair-minded ��� juste gerecht honrado rättvis Patient ��� patient geduldig paciente tålmodig Polite ��� poli höflich correcto artig 2-d Reassuring ����� rassurant beruhigend tranquilizador uppmuntrande Responsible ��� responsible verantwortungsbewusst responsable ansvarsfull Ordered ����� ordonné ordnung ordenado välordnad Diplomatic ��� diplomate diplomatisch diplomático diplomatisk Conservative ��� conservateur konservativ conservador konservativ 2-e Soft ��� doux weich suave mjuk Understated ����� discret untertreiben reservado nedtonad Sensitive ��� sensible sensibel sensible känslig Graceful ��� gracieux reizend agraciado behagfull Sensible ��� sensé vernünftig sensible förståndig

(c)

English Chinese Germany French Spanish Swedish 3-a Abundant ��� reich abondant sólido riklig Energetic ����� energisch énergétique enérgico energisk Rebellious ����� rebellisch rebelle rebelde upprorisk Visionary ����� Visionär visionnaire visionario visionär Idealistic ����� idealistisch idéaliste idealista idealistisk 3-b Earth ����� erdig terre à terre básico jord Mature ��� ausgereift mature maduro mogen Natural ��� natürlich naturel natural naturlig Sensual ��� sinnlich sensuel sensual sensuell Generous ����� großzügig généreux espléndido generös 3-c Colourful ����� farbenprächtig bigarré colorido färgstark Brave ��� tapfer brave valiente modig Original ��� original original original originell Protective ����� fürsorglich protecteur protector beskyddande Adventurous ����� abenteuerlich aventureux aventurero äventyrlig 3-d Authentic ��� authentisch authentique auténtico autentisk Ethical ����� ethisch moral ético moralisk Intense ��� intensiv intense intenso intensiv Sincere ��� aufrichtig sincère sincero innerlig Honest ��� ehrlich honnête honrado ärlig 3-e Environmental ��� umgebungsmässig écologiste ecológico miljö- Traditional ��� traditionell traditionnel tradicional traditionell Substantial ��� kräftig substantiel substancioso omfattande Solid ��� fest solide sólido stadig Committed ��� verpflichtet engagé comprometido engagerad

Page 6: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

6

(d)

English Chinese Germany French Spanish Swedish 4-a Contemporary ��� zeitgenössisch contemporain contemporáneo modern Confident ���� zuversichtlich confident firme självsäker Glamorous ����� zauberhaft glamour sofisticado glamorös Respective ���� angesehen respecté respetado respekterad Stylish ����� stilvoll chic elegante elegant 4-b Decisive ��� entscheidend décisif determinante bestämd Detached ��� abgetrennt détaché desligado fristående Influential ����� einflussreich influent influyente inflytelserik Objective ��� objektiv objectif objetivo saklig Uncompromising ����� kompromisslos intransigeant no comprometido kompromisslös 4-c Efficient ���� leistungsfähig efficace eficiente ändamålsenlig Dry �� trocken sec seco torr Pragmatic ����� pragmatisch pragmatique pragmático pragmatisk Superior ��� überlegen supérieur superior överlägsen Precise ��� exakt précis preciso exakt 4-d Logical ����� logisch logique lógico logisk Articulate ����� artikulieren expressif estructurado tydlig Focused ����� fokussiert concentré centrado fokuserad Perfectionistic ����� perfectionistisch perfectioniste perfeccionista perfektionistisk Productive ��� produktiv productif fructífero fruktbar 4-e Grand ��� großartig grandiose grandioso pampig Wealthy ��� wohlhabend riche rico rik Exotic ����� exotisch exotique exótico exotisk Accomplished ����� erreicht accompli conseguido fulländad Charismatic ������ charismatisch charismatique carismático karismatisk

2.2 Colour Combinations

Thirty-two colour stimuli were selected evenly from the 4 colour groups, CG 1 to CG 4, each in association with one of the 4 personality types: Spring, Summer, Autumn, and Winter, as defined by Wright (1995).

As shown in Table 2, each CG contains 8 colours that give a reasonable coverage of the entire range of colours in hue, lightness, and chroma, as presented in Figures 1 (a) and (b). In each CG, 5 out of 8 colours were selected to generate a 5-colour combination in shape of a wheel. Examples of the colour wheels are shown in the Appendix.

In the experiment, colour wheels were presented on a CRT monitor. The background colour behind the wheels was a uniform yellowish white, with the CIELAB values (L*, a*, b*) of (97.9, -0.9, 9.2). This colour was selected as the background because according to the Wright theory it is located at the borderline between the regions of the 4 CGs in the CIELAB colour space where this colour is not in association with any of the 4 personality types.

Page 7: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

7

Table 2. The 32 Colour stimuli used in the experiments CG Code L* a* b* C* h

1 97.5 -9.8 49.8 50.8 101 2 79.6 -8.7 -23.7 25.3 250 3 85.8 10.9 42.1 43.5 76 1 4 74.2 -0.3 2.8 2.9 97 5 85.3 15.7 -10.7 19.0 326 6 57.8 57.8 36.4 68.3 32 7 85.7 -23.2 -7.7 24.4 198 8 84.0 -34.8 31.9 47.2 137 9 83.6 14.4 6.8 15.9 25 10 83.7 -23.1 7.5 24.3 162 11 92.6 -6.3 32.9 33.5 101 2 12 47.4 -4.5 -25.2 25.6 260 13 45.5 48.4 1.2 48.4 1 14 77.7 -7.4 -15.6 17.2 245 15 70.8 -0.5 2.9 3.0 99 16 69.9 15.7 -10.9 19.1 325 17 50.6 52.5 45.2 69.2 41 18 40.3 -23.5 -9.0 25.2 201 19 85.7 6.7 71.9 72.2 85 3 20 60.1 -39.9 47.9 62.3 130 21 68.3 29.6 56.1 63.4 62 22 35.9 30.2 -13.8 33.2 335 23 83.6 -13.8 -2.7 14.1 191 24 78.9 -46.2 79.9 92.3 120 25 96.9 -3.4 1.5 3.8 156 26 4.9 -0.8 0.8 1.1 134 27 38.6 45.9 -59.7 75.3 308 4 28 57.4 -41.3 9.4 42.3 167 29 96.3 -14.1 92.8 93.9 99 30 47.0 61.9 27.4 67.6 24 31 96.2 -8.7 -6.0 10.6 214 32 41.4 30.3 -72.1 78.2 293

0

25

50

75

100

0 25 50 75 100

C*

L*CG 1CG 2CG 3CG 4

-100

-50

0

50

100

-100 -50 0 50 100

a*

b*CG 1CG 2CG 3CG 4

(a) (b) Figure 1. Colour stimuli of the 4 CGs in the CIELAB colour space: (a) a*- b* diagram and

(b) L* - C* diagram

2.3 Observer Groups

As summarised in Tables 3 (a) and (b), a total of 47 females and 72 males participated in Experiment 1; 48 females and 71 males in Experiment 2. The observers were native

Page 8: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

8

speakers of the 6 cultures: British, Chinese, French, German, Spanish, and Swedish. All observers were asked to perform the experiment twice, and thus the number of observations was as twice as the number of observers.

Note that there were two British observer groups, “British-1” and “British-2”. The former were design students at the Coventry School of Art and Design, Coventry University; the latter were residents in London. “British-1” observers were all from design background, while “British-2” observers were from non-design background. The Chinese observers were students at the University of Derby, UK, and have been in Britain for at least 6 months before the experiments. This suggests that they may have been affected by British culture and thus the experimental data may be biased due to cultural difference. Note that the observer groups “Chinese”, “French”, “German”, and “Spanish” were all students from engineering-related background, while “British-1” and “Swedish” were from design background. All the observers were aged from 20 to 30 and have passed the Ishihara’s Tests for Colour Deficiency (Ishihara, 2003).

Table 3. Observers for (a) Experiment 1 and (b) Experiment 2 (a)

Female Male Description British-1 7 13 Students at Coventry University British-2 3 2 Residents in London Chinese 3 8 Students at University of Derby (UK) French 5 18 Students at Université Jean Monnet German 10 10 Students at Bundesanstalt für Materialforschung und -prüfung Spanish 10 10 Students at Universidad de Granada Swedish 9 11 Students and staff members at Chalmers tekniska högskola All 47 72

(b) Female Male Description

British-1 6 14 Students at Coventry University British-2 3 2 Residents in London Chinese 3 8 Students at University of Derby (UK) French 7 16 Students at Université Jean Monnet German 10 10 Students at Bundesanstalt für Materialforschung und -prüfung Spanish 10 10 Students at Universidad de Granada Swedish 9 11 Students and staff members at Chalmers tekniska högskola All 48 71

2.4 Performance of CRT Displays

Experiments 1 and 2 were both conducted on CRT displays, and thus the accuracy and stability of colour reproduction by the displays were essential. To ensure reliability of CRT display performance, a Sequel Chroma-4 colour sensor was used in

Page 9: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

9

monitor calibration for each display by applying the GOG model (Berns, 1996). What follows shows calibration results for the CRT monitors used in the experiment.

2.4.1 Consistency of Colour Sensors

Five colour sensors were used in the 7 observer groups to process monitor calibration before experimental sessions. The calibration was performed by using a colour sensor to measure colours on screen. The measurement results were analysed to establish a predictive model for XYZ to RGB on that monitor. To ensure consistency of colour sensor performance, the 5 sensors were tested on one CRT monitor before dispatched to experimental sites. Each colour sensor measured 12 test colours in CIE colorimetric values. For each colour, the CIE data were compared between the 5 sensors to compute the CIELAB colour difference values, from which a mean value for the 12 colours was obtained. The mean values indicate the difference of measurement results between the 5 sensors and represent the consistency of the sensors.

The test results were summarised in Table 4. The largest mean colour difference was found between Sensors 2 and 4, with a value of 1.58, followed by the colour difference between Sensors 2 and 3 (1.26), Sensors 1 and 4 (1.13), and Sensors 2 and 5 (1.11). Sensor 2 was found to differ most from the others, with a mean value of colour difference 1.21 for all comparisons on Sensor 2. However, this value does not show significant colour difference for human vision. It indicates good consistency of the 5 colour sensors. Notice that Sensor 1 was used in both the observer groups “Spanish” and “British-1” and Sensor 2 in “Chinese” and “British-2”.

Table 4. Mean CIELAB colour difference between the 5 colour sensors for 12 test colours on the same CRT monitor

Sensor 1 Sensor 2 Sensor 3 Sensor 4 Sensor 5 Mean Sensor 1 - 0.88 0.79 1.13 0.97 0.94 Sensor 2 - 1.26 1.58 1.11 1.21 Sensor 3 - 0.84 0.84 0.93 Sensor 4 - 0.93 1.12 Sensor 5 - 0.96 Observer group

Spanish British-1

Chinese British-2 French German Swedish 1.03

2.4.2 Accuracy of CRT Colours

To ensure reliability of the CRT monitors used in the experiment, each monitor was calibrated by performing CRT characterisation with a colour sensor before each experimental session. The characterisation process was performed after one-hour warm up of the monitor.

Page 10: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

10

A colour temperature of 6500K was set on each CRT monitor. In the CRT characterisation, a predictive model for XYZ to RGB was established by colour measurement of a colour sensor, followed by a test session where the mean CMC (1:1) colour difference was calculated between measured and predicted colorimetric values on 12 test colours. The 12 colours were exactly the same ones as mentioned in Section 2.4.1. The mean CMC (1:1) colour difference was a measure of prediction performance of the CRT characterisation as well as a measure of accuracy for colour reproduction on that monitor. The smaller the colour difference, the better the accuracy of CRT colours.

In each observer group, an experimenter was asked to ensure a mean CMC (1:1) colour difference of less than 1 for CRT characterisation before the experiment; a retry of CRT characterisation was necessary if the colour difference value was greater than 1. In this way each CRT display was precisely calibrated for each experimental session and the accuracy of colour reproduction was ensured. Results of CRT characterisation are shown in Table 5 for the 7 observer groups. Note that in French group two monitors were used—one for Experiment 1 and the other for Experiment 2; the two monitors were characterised with the same colour sensor.

Table 5. Results of CRT characterisation in terms of mean CMC (1:1) colour difference between measured and predicted colorimetric values for 12 test colours

British-1 British-2 Chinese French-1 French-2 German Spanish Swedish Mean

CMC (1:1) ΔE 0.62 0.58 0.34 0.56 0.54 0.65 0.61 1.06 0.62

2.5 Experimental Procedure

The 2 experiments were conducted at a dark room. Prior to each experimental session, each observer spent one minute getting adapted to the experimental environment. This was followed by experimental sessions described below.

2.5.1 Procedure for Experiment 1

Experiment 1 was to investigate the connection between adjective combinations and colour combinations. A total of 60 judgements were performed in this experiment, where the 4 AGs comprised 15 judgements each. Every AG contained 5 adjective combinations, as shown in Tables 1 (a) to (d), and each adjective combination was tested 3 times with different colour wheels for each time. The sequence of presenting adjective combinations was randomised and was made different between observers.

Figure 2 shows the experimental layout in which an adjective combination is accompanied by 4 colour wheels. From the 4 colour wheels each observer was asked

Page 11: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

11

to pick one that appeared in association with the adjective combination presented. The 4 colour wheels were such generated that the colours in a wheel were all from the same CG; each colour wheel represented one of the 4 CGs. Locations of the 4 colour wheels were randomised for each adjective combination presented and were made different between observers. At the beginning of an experimental session, observers were given the instruction:

Please regard the 5 adjectives as a whole, which expresses one unique affective feeling. And again regard each colour wheel as a whole, which evokes one unique affective feeling. Please can you indicate which of the 4 colour wheels that appears in connection with the whole adjective combination in terms of affective feelings?

After a colour wheel was clicked and the “OK” button was pressed from screen, the 4 colour wheels and the adjectives vanished. And this completed one judgement. A period of one second was taken as a delay before the experiment went on the next judgement. This delay was to eliminate the afterimage effect created by colour wheels. It took about 15 to 20 minutes for each observer to complete all the 60 judgments.

2.5.2 Procedure for Experiment 2

Experiment 2 was to examine the Wright theory on colour harmony for 5-colour combinations under the 6 cultures: Chinese, British, French, German, Spanish, and Swedish. As shown in Figure 3, in the experimental session each judgement was a comparison of colour harmony between two colour wheels. The instruction was given to each observer at the beginning of judgements:

Please indicate which of the 2 colour wheels that appears more harmonious than the other. Note that the word “harmonious” here is defined as “not conflicting” between colours within a wheel.

In comparison between the 2 colour wheels, all component colours except one in one wheel were exactly the same as the corresponding colours in the other wheel. One of the wheels contained colours all from the same CG, while the other wheel contained all but one from the same CG. The Wright theory suggests that the former is more harmonious than the latter.

Each observer performed a total of 224 comparisons, where the 4 CGs were tested 56 times each. As shown in Table 2, each CG contained 8 colour stimuli, from which 5 colours were selected to form a colour wheel. The sequence of the comparisons was randomised for each observer. The locations (left and right) of 2 colour wheels in each comparison were also randomised.

Page 12: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

12

Figure 2. Screen layout for Experiment 1

Figure 3. Screen layout for Experiment 2

Page 13: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

13

3. RESULTS AND DISCUSSION

Nominal data were collected from both experiments. Data analysis was carried out within each observer group. Comparisons were made between observer groups to investigate the effect of cultural difference. The following describes the data analysis and the results.

3.1 Experiment 1: Colour Combinations and Adjective Combinations

The main objective of this experiment was to find out the relationship between colour combinations and adjective combinations. It is important to ensure reliability of the experimental data before further data analysis. Here the two measures of data reliability were examined: observer repeatability and accuracy.

3.1.1 Observer Repeatability and Accuracy

Observer repeatability indicates how well an experimental data set agrees with a replicated one. In this experiment each observer was asked to perform all the judgements twice. For each observer the two data sets were compared to see whether the data were reliable to use in the later data analysis. In this experiment a repeatability value was determined by proportion of the same decision made by each observer in two experimental sessions; the greater the value, the better the agreement between the 2 data sets.

Table 6 shows comparison results of repeatability (in percentage) between female and male observers. The results show better repeatability for female observers than that for males, with a repeatability value of 55.0% for females and 51.5% for males. However, the French and Swedish data indicate that male observers were better than females in repeatability. Among the 7 observer groups, the observer group British-2 had the best repeatability (64.6%) and British-1 the poorest (36.7%). Note that in British-2, only 3 females and 2 males took part in the experiment, and accordingly there is no need to compare repeatability between genders in this group.

Table 7 shows comparison results of repeatability between the 4 adjective groups (AGs). The results show a consistent pattern across the 7 observer groups. Among the 4 groups, AG 2 had the best repeatability, with a repeatability value of 56.7% for all observers, followed by AG 1 (53.8%), AG 4 (51.9%), and AG 3 (51.8%).

Table 6. Comparison of repeatability (%) between genders British-1 British-2 Chinese French German Spanish Swedish All

Female 39.6 - 74.7 46.6 60.6 61.6 56.1 55.0 Male 34.9 - 53.3 53.6 54.7 54.5 57.6 51.5 All 36.7 64.6 59.1 52.0 57.7 58.1 57.0 53.5

Page 14: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

14

Table 7. Comparison of repeatability (%) between AGs British-1 British-2 Chinese French German Spanish Swedish All

1 35.9 65.0 61.2 52.8 58.7 57 57.3 53.8 2 37.0 63.3 61.2 55.9 62.7 65.3 58 56.7 3 35.2 68.3 55.2 48.1 60.7 52.7 56.3 51.8 4 38.5 61.7 58.2 51.6 49.3 57.3 56.3 51.9

Accuracy indicates the extent to which individual observers agree with the majority decision. In the experiment an accuracy value is determined by proportion of observers that agreed with the majority decision (for choosing between the 4 colour wheels). The greater the accuracy value, the better the agreement between individual responses and the majority decision.

As summarised in Table 8, among the 4 adjective groups, AG 2 had the highest accuracy value for all the observers (55.9%), followed by AG 3 (45.1%), AG 4 (42.1%), and AG 1 (38.4%). This order was found more or less consistent across all the observer groups except British-1, which was found to have the same accuracy value for all the 4 AGs, as shown in Figure 4.

Table 8. Comparison of accuracy (%) between AGs British-1 British-2 Chinese French German Spanish Swedish All

1 28.8 34.8 47.3 36.4 48.3 43.3 42.8 38.4 2 29.8 68.9 60.9 47.5 63.3 63.7 57.5 55.9 3 27.7 66.7 31.2 38.7 52.3 46.7 56.2 45.1 4 29.1 61.5 39.7 33.0 36.7 44.8 55.3 42.1 Mean 28.9 58.0 44.8 38.9 50.2 49.6 53.0 45.4

0

25

50

75

100

1 2 3 4

Adjective Group

British-1British-2ChineseFrenchGermanSpanishSwedishAll

Figure 4. Comparison of accuracy (%) between AGs

3.1.2 Agreement between Adjective Groups and Colour Groups

The experiment assumes the connection between adjective groups (AGs) and colour groups (CGs). To verify this assumption, the experimental data were analysed within each AG, in which observer’s responses to adjective combinations were calculated in terms of percentage of agreement between an AG and a CG, as determined by

Page 15: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

15

Agreement A-C (%) = Cm,n

Cm,i( )i=1

4

∑×100(%), (1)

where Cm,n is the number of observers that agree with the connection between the m-th AG and the n-th CG.

If the assumption is correct, AG 1 will have the highest percentage value on CG 1, AG 2 on CG 2, AG 3 on CG 3, and AG 4 on CG 4. As shown in Figures 5 (a) to (d), the experimental results agree with the assumption, especially in AG 2, where the highest percentage value was found on CG 2 and this percentage value is far higher than those on the other CGs. However, in the other three AGs the highest percentage value for an AG never exceeds 50(%), while this value in each AG is on the corresponding CG. Note that all the 7 observer groups, except British-1, show a consistent pattern with regard to the percentage of agreement between AGs and CGs. This indicates that the British-1 data agreed least with the experimental assumption.

Adjective Group 1

0

25

50

75

100

1 2 3 4

Colour Group

(a) (b) Adjective Group 3

0

25

50

75

100

1 2 3 4

Colour Group

(c) (d) Figure 5. Percentage of agreement for the 4 adjective groups, (a) AG 1, (b) AG 2, (c) AG 3,

and (d) AG 4, in the 4 CGs

3.1.3 Individual Adjective Combinations

The previous section has pointed out that in AGs 1, 3, and 4 the highest percentage value never exceeds 50%, although the highest value of each AG does correspond to the CG that the experiment expected. To find out why these values are not higher than 50%, further analysis was carried out on individual adjective combinations. Table 9

Page 16: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

16

shows the proportion of observer responses to each adjective combination, from which the phenomena were found:

a) Most of the adjective combinations were strongly correlated with the CGs as assumed.

b) Three adjective combinations, 1-d, 3-a, and 4-e, were strongly correlated with the CGs that were not assumed.

c) Five adjective combinations, 1-c, 1-d, 3-d, 4-a, and 4-d, were found in correlation with none of the 4 CGs.

That some adjective combinations cannot be classified by the 4 CGs may be due to the difficulty in the experiment that observers needed to regard 5 adjectives in a combination as a whole concept. Some observers tended to use one of the 5 adjectives to represent the whole combination instead of relating a complete feeling to the entire adjective combination. If the adjective the observer used does not quite agree with the others in that combination, the experimental results will be serious biased. .

Table 9. Proportion of observer responses (%) to individual adjective combination in association with each CG

AG (adjective-group) CG (colour-group) 1 2 3 4 1-a Fun Adaptable Bright Charming Cheerful 62.6 6.5 17.1 13.9 1-b Lively Active Enthusiastic High-spirited Optimistic 50.3 5.0 29.9 14.8 1-c Clever Cooperative Friendly Informal Practical 29.8 18.9 31.2 20.1 1-d Fresh New Outgoing Impulsive Communicative 27.4 7.4 33.0 32.3 1-d Fanciful Imaginative Creative Inspired Mischievous 21.8 8.6 23.8 45.7 2-a Calm Peaceful Nurturing Courteous Compassionate 32.6 48.2 13.6 5.5 2-b Formal Elegant Gracious Modest Reserved 17.1 59.3 6.2 17.5 2-c Cool Dignified Fair-minded Patient Polite 25.0 56.2 6.8 12.0 2-d Reassuring Responsible Ordered Diplomatic Conservative 17.4 55.2 13.8 13.6 2-e Soft Understated Sensitive Graceful Sensible 27.3 60.8 6.1 5.8 3-a Abundant Energetic Rebellious Visionary Idealistic 16.4 7.6 32.4 43.6 3-b Earth Mature Natural Sensual Generous 17.4 17.1 54.1 11.5 3-c Colourful Brave Original Protective Adventurous 19.7 6.5 54.0 19.9 3-d Authentic Ethical Intense Sincere Honest 24.8 19.7 34.4 21.1 3-d Environmental Traditional Substantial Solid Committed 15.9 13.6 50.8 19.7 4-a Contemporary Confident Glamorous Respective Stylish 20.6 28.5 16.2 34.7 4-b Decisive Detached Influential Objective Uncompromising 10.4 10.5 19.5 59.6 4-c Efficient Dry Pragmatic Superior Precise 14.9 23.8 16.9 44.4 4-d Logical Articulate Focused Perfectionistic Productive 18.3 21.6 21.9 38.2 4-e Grand Wealthy Exotic Accomplished Charismatic 13.1 8.7 44.5 33.7

3.1.4 Summary

This experiment investigates the relationship between colour combinations and adjective combinations. Thirty-two colours and 100 adjectives were used in the experiment. A total of 60 judgements were carried out by each observer. The

Page 17: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

17

experiment was conducted with 6 languages in 5 countries. CRT displays were used to present the colour stimuli the experiment.

The data reliability was tested with the measure of repeatability and accuracy. The test result indicates the repeatability of 53.5% and accuracy of 45.4%. British-2 and Swedish were found the most reliable data in the experiment.

The experimental results indicate the agreement between observer responses and the Wright theory, the latter suggesting a link between colour combinations and adjective combinations in association with the 4 personality types, Spring, Summer, Autumn, and Winter. Among the 4 adjective groups, AG 2 showed the best agreement.

Further analysis was carried out within individual adjective combinations. The results indicate 5 adjective combinations that could not be discriminated between the 4 CGs. Three adjective combinations were found in correlation with the “wrong” CGs. The reason for these adjective combinations may be due to the difficulty observers encountered in the experimental sessions. It would be too difficult for observers to regard 5 adjectives as a whole concept if these adjectives do not quite agree with each other. Some observers tended to use one of the 5 adjectives in a combination presented to do the judgement instead of using all the 5 adjectives to relate a complete feeling to the combination. These problems may lead to great bias of the experimental results.

Page 18: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

18

3.2 Experiment 2: Colour Harmony and the Wright Theory

Experiment 2 was to examine whether the Wright theory of colour harmony applies to 5-colour combinations across the 6 cultures: Chinese, British, French, German, Spanish, and Swedish. According to the Wright theory, there are 4 categories of colours and the colours of the same category harmonise when combined together. The repeatability and accuracy of the experimental data were examined before the analysis on colour harmony.

3.2.1 Observer Repeatability and Accuracy

The repeatability of experimental data is defined as proportion of the same decision made by each observer between two experimental sessions. It was calculated in the same way as described in Section 3.1.1.

Table 10 shows that the repeatability of female observers was slightly better than that of males, with a repeatability value of 70.5% for female and 67.3% for male. This tendency was found in all the observer groups.

As shown in Table 10, German observers showed the best repeatability (77.5%) among the 7 groups, followed by British-2 (74.2%), Spanish (71.3%), Swedish (70.2%), Chinese (65.8%), French (65.1%), and British-1 (58.8%).

In Table 11 is shown the repeatability for each colour group. The group CG 2 was found to have the highest repeatability value, 72.6%, and CG 4 had the lowest, 65.5%. This pattern was found consistent in all the observer groups.

Table 10. Comparison of repeatability (%) between genders British-1 British-2 Chinese French German Spanish Swedish All

Female 60.0 - 76.0 69.4 77.3 74.4 68.1 70.5 Male 58.4 - 62.0 63.3 77.7 68.1 71.9 67.3 All 58.8 74.2 65.8 65.1 77.5 71.3 70.2 68.5

Table 11. Comparison of repeatability (%) between CGs British-1 British-2 Chinese French German Spanish Swedish All

1 58.5 69.2 67.7 68.3 78.8 74.6 69.7 69.8 2 60.0 79.9 71.1 69.1 82.8 74.6 75.6 72.6 3 56.3 74.6 64.9 60.6 75.2 69.0 70.4 66.4 4 60.0 73.2 60.6 62.8 73.4 67.1 65.1 65.5

Accuracy indicates how well individual observers agree with the majority decision. In this experiment an accuracy value is determined by the proportion of majority decision.

Page 19: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

19

Table 12 summarises accuracy values in each observer groups with regard to the 4 CGs. The group CG 2 shows the highest accuracy value, 74.6%, and CG 4 shows the lowest, 60.5%.

Among the 7 observer groups, German data was found the most accurate, with a mean accuracy value of 75.9%, followed by Spanish (72.4%), Swedish (70.2%), British-2 (68.2%), Chinese (67.7%), French (64.3%), and British-1 (56.8%).

Compared to the results in Experiment 1, the present experimental data were far more reliable in both repeatability and accuracy. As shown in Table 10, the repeatability value for Experiment 2 is 68.5%, while in Experiment 1 the reliability value is only 53.5%, as shown in Table 6. In Table 12, the mean accuracy value for Experiment 2 is 67.9%, while in Experiment 1 the accuracy value is only 1, 45.4%, as shown in Table 8.

Table 12. Comparison of accuracy (%) between CGs British-1 British-2 Chinese French German Spanish Swedish All

1 57.6 70.6 71.8 67.2 76.6 75.7 70.2 69.7 2 60.9 78.4 75.2 71.8 83.3 79.6 76.0 74.6 3 56.3 67.7 65.6 59.2 76.4 70.7 71.9 66.6 4 52.4 56.0 57.5 59.0 67.5 63.8 62.5 60.5 Mean 56.8 68.2 67.5 64.3 75.9 72.4 70.2 67.9

3.2.2 Agreement between Visual Results and the Wright Theory

In each judgement of the experiment, observers were presented with two colour wheels on the CRT display. One of the colour wheels is made of colours from the same CG, while the other contains the same colours except one, which is from a different CG. According to the Wright theory, the first colour wheel is more harmonious than the second, because all the component colours in the first colour wheel are of the same CG.

Table 13 shows the agreement (in percentage) between the majority decision of observers and the Wright theory. The values in the table were determined by proportion of judgements in which the majority decision agrees with the Wright theory, as given below.

Agreement W (%) = Di

Ni

×100(%) (2)

where Di is for the i-th CG the number of judgments in which the majority decision agrees with the Wright theory, and Ni is the total number of judgements for the i-th CG.

As a result, CG 2 had the highest agreement value, 92.9%, followed by CG 3 (83.9%), CG 1 (75.0%), and CG 4 (55.4%). This indicates that the Wright theory of colour

Page 20: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

20

harmony worked well in the 4 CGs, especially in CG 2. For all the 224 judgements, the agreement value is 76.8%, which means in 172 judgements the majority decisions agreed with the Wright theory. Among the observer groups, British-2 was found to have the highest agreement value (82.6%), followed by French (80.8%), Spanish (79.5%), Chinese (78.1%), Swedish (77.2%), German (69.2%), and British-1 (67.9%).

Note that British-1 data shows good agreement in CGs 1 to 3 but extremely poor agreement in CG 4 (28.6%). This indicates that British-1 agreed least with the Wright theory in CG 4, while British-2, French and Chinese data show good agreement in the same colour group, with the values of 75.0%, 71.4% and 69.6%, respectively.

Table 13. Agreement W (%) between the Wright theory and the majority decision within the 7 observer groups

British-1 British-2 Chinese French German Spanish Swedish All 1 76.8 76.8 75.0 85.7 69.6 82.1 73.2 75.0 2 87.5 96.4 94.6 91.1 83.9 92.9 92.9 92.9 3 78.6 82.1 73.2 75.0 82.1 80.4 87.5 83.9 4 28.6 75.0 69.6 71.4 41.1 62.5 55.4 55.4 All 67.9 82.6 78.1 80.8 69.2 79.5 77.2 76.8

3.2.3 Comparisons of Colour Harmony Results between Cultures

The agreement values between individual response and the Wright theory in each judgement were compared between observer groups by Pearson product-moment correlation coefficient (r), which represents a measure of cultural difference in colour harmony.

The comparison results are shown in Tables 14 (a) to (d) for the judgements in CG 1 to CG 4, respectively. Table 14 (e) shows the results for overall judgements. The mean correlation coefficient of an observer group indicates how well this observer group agreed with the other groups; the greater the mean value of correlation coefficient, the better the agreement. As presented in Table 14 (e), Spanish data showed the best agreement with the other observer groups, with a mean correlation coefficient of 0.77, followed by German (0.76), Swedish (0.75), Chinese (0.70), French (0.68), and British-2 (0.60). The lowest value was found in British-1 (0.53), indicating that British-1 agreed least with the other observer groups. This tendency was found in each CG, as shown in Tables 14 (a) to (d).

The data in CG 1 showed the highest mean correlation coefficient, 0.78, followed by CG 3 (0.61) and CG 2 (0.60), and the lowest mean correlation coefficient was found in CG 4 (0.48). This indicates that in CG 1, all the observer groups agreed very well with each other, while in CG 4, the data did not agree well between observer groups.

Page 21: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

21

Table 14. Correlation (Pearson r) between observer groups in terms of Agreement W (%) on (a) CG 1, (b) CG 2, (c) CG 3, (d) CG 4, and (e) overall

(a) British-1 British-2 Chinese French German Spanish Swedish Mean

British-1 - 0.50 0.61 0.66 0.66 0.66 0.64 0.62 British-2 - 0.72 0.74 0.72 0.77 0.79 0.71 Chinese - 0.85 0.87 0.91 0.83 0.80 French - 0.83 0.91 0.85 0.81 German - 0.94 0.94 0.83 Spanish - 0.92 0.85 Swedish - 0.83

0.78

(b) British-1 British-2 Chinese French German Spanish Swedish Mean

British-1 - 0.36 0.33 0.29 0.41 0.41 0.40 0.37 British-2 - 0.45 0.56 0.59 0.63 0.54 0.52 Chinese - 0.63 0.76 0.70 0.69 0.59 French - 0.70 0.79 0.66 0.60 German - 0.91 0.90 0.71 Spanish - 0.88 0.72 Swedish - 0.68

0.60

(c) British-1 British-2 Chinese French German Spanish Swedish Mean

British-1 - 0.36 0.37 0.37 0.43 0.46 0.44 0.40 British-2 - 0.55 0.49 0.64 0.60 0.58 0.54 Chinese - 0.64 0.77 0.78 0.75 0.64 French - 0.72 0.72 0.71 0.61 German - 0.75 0.85 0.69 Spanish - 0.79 0.68 Swedish - 0.69

0.61

(d) British-1 British-2 Chinese French German Spanish Swedish Mean

British-1 - -0.03 0.24 0.42 0.36 0.38 0.27 0.27 British-2 - 0.27 0.31 0.41 0.49 0.42 0.31 Chinese - 0.40 0.61 0.64 0.61 0.46 French - 0.65 0.59 0.58 0.49 German - 0.82 0.84 0.62 Spanish - 0.81 0.62 Swedish - 0.59

0.48

(e) British-1 British-2 Chinese French German Spanish Swedish Mean

British-1 - 0.40 0.51 0.52 0.60 0.59 0.56 0.53 British-2 - 0.60 0.60 0.64 0.68 0.65 0.60 Chinese - 0.72 0.79 0.81 0.76 0.70 French - 0.74 0.79 0.72 0.68 German - 0.88 0.91 0.76 Spanish - 0.87 0.77 Swedish - 0.75

0.68

Page 22: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

22

Figure 6. Comparisons of colour harmony results between the 7 observer groups in terms of

Agreement W (%) between the Wright theory and the majority decision within each observer group. This figure corresponds to Table 14 (e).

In Figure 6 is shown the correlation between the observer groups in terms of the agreement value, which was obtained by Eq (2) for each observer group. There are 224 dots in each chart, representing the 224 judgements in the experiment. These dots spread in the four quadrants. In Quadrant 1 (top-right) the dots represent the judgements when both observer groups agreed with the Wright theory. In Quadrant 2 (top-left) the dots represent that the observer group in column agreed with the theory while the group in row did not. In Quadrant 3 (bottom-left) the dots represent both the observer groups disagree with the theory. In Quadrant 4 (bottom-right) the dots represent the observer group in row agreed with the theory while the group in column did not. It follows that the more the dots in Quadrant 1, the better the performance of

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

British-2 Chinese French German Spanish Swedish

r = 0.40 r = 0.59 r = 0.60 r = 0.52 r = 0.51

r = 0.68 r = 0.64 r = 0.60 r = 0.60

r = 0.81 r = 0.79 r = 0.72

r = 0.79 r = 0.74

r = 0.88

British-1

British-2 Chinese

French G

erman

Spanish

r = 0.56

r = 0.65

r = 0.76

r = 0.72

r = 0.91

r = 0.87

Page 23: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

23

the Wright theory for both observer groups; the more the dots in Quadrant 3, the poorer the performance of the theory. For instance, in the chart between Spanish and Swedish, most of the dots are located in Quadrant 1, a few in Quadrant 3, and very few in Quadrants 2 and 4. This indicates that the Spanish and Swedish groups both agreed well with the Wright theory, and the two groups tended to agree with each other.

Figure 7. Comparisons of colour harmony results between the 7 observer groups in terms of

the proportion of majority decision within each observer group

In Figure 7 is shown the correlation between the 7 observer groups in terms of the proportion of majority decision for each colour wheel. There are 448 dots in each chart, representing the 448 colour wheels in the 224 judgements (in each judgement

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

0.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.00.0

0.5

1.0

0.0 0.5 1.0

British-2 Chinese French German Spanish Swedish

r = 0.53 r = 0.66 r = 0.68 r = 0.62 r = 0.61

r = 0.83 r = 0.76 r = 0.77 r = 0.76

r = 0.88 r = 0.85 r = 0.82

r = 0.88 r = 0.82

r = 0.91

British-1

British-2 Chinese

French G

erman

Spanish

r = 0.64

r = 0.80

r = 0.85

r = 0.83

r = 0.93

r = 0.93

Page 24: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

24

were 2 colour wheels simultaneously presented on screen). Note that each chart shows a diagonal symmetry, which is because the proportion values for the two colour wheels in the same judgement always yield a sum of 1.0. The results show very good correlation between each observer group, especially in the three groups: German, Spanish, and Swedish. This suggests that colour harmony responses are of no cultural difference effect between these 3 countries.

Note that the correlation coefficient between British-1 and British-2 is the smallest among the comparisons. This may be interpreted by the difference in educational background between the 2 groups—one of them was of design background and the other non-design background. However, Swedish observers were also from design background, while their data agreed very well with those from non-design background, such as Spanish and German, both having a correlation coefficient of 0.93 with Swedish data. Note that the British-1 data had the lowest value in both observer repeatability and accuracy, which was found in both Experiments 1 and 2. This suggests that the British-1 observers were in great difficulty performing the experiments. It may also suggest a very wide diversity of opinion on colour harmony among British designers, which also implies a significant effect of cultural difference.

3.2.4 Summary

This experiment was to examine whether the Wright theory of colour harmony works for 5-colour combinations across the 6 cultures: British, Chinese, French, German, Spanish, and Swedish. The Wright theory of colour harmony suggests 4 categories of colours and that the colours of the same category harmonise when combined together.

The reliability of experimental data was tested in terms of repeatability and accuracy. The test results indicate that among the 7 observer groups, German had the highest reliability, followed by Spanish, Swedish, British-2, Chinese, and French. The observer group British-1 was found to have the smallest value in both repeatability and accuracy.

The experimental results show good agreement between observer response and the Wright theory. Among the colour groups, CG 2 was found to have the best agreement and CG 4 the poorest.

In comparisons of colour harmony response between observer groups, Spanish observers showed the best agreement with the other observer groups, followed by German, Swedish, Chinese, French, and British-2. The British-1 data was found to have poorest correlation with the other groups.

Page 25: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

25

Effects of cultural difference and educational-background difference on colour harmony were identified in British-1 data. However, these effects were not found in the other observer groups.

4. CONCLUSIONS

This project is concerned with the following issues:

a) Colour reproduction on a CRT display with a colour sensor

This project contains two psychophysical experiments in 5 European countries. Two common ways of preparing image samples for experiments are physical colour samples (such as paper cards and coloured fabric) and CRT displays. The latter was used in the present project because: 1) by using CRT displays the experiments can be repeated in different countries without worries about likely damages to physical colour samples due to changes in climate and humidity from an area to another, which can occur when physical colour samples are used, 2) the viewing conditions can be well controlled by using CRT displays and 3) using CRT displays is a convenient way of preparing a large number of stimuli with specific colour coordinates. The experimental results show good performance of colour reproduction, which was done with a colour sensor. This indicates a bright future for CRT colour reproduction.

b) Ultra-colour system and colour emotion

Experiment 1 investigates the relationship between colour combinations and adjective combinations. The experimental results show good agreement between adjective combinations and the CGs, as assumed by the Ultra-colour system, especially in CG 2. However, a few adjective combinations failed to correspond to any CG or were found in connection with “wrong” CGs. The reason for this may be that in the experiment observers were asked to correlate a combination of 5 adjectives with one of 4 combinations of 5 colours. Observers found it difficult to do so. It would be suggested that the experiment should be carried out with only one adjective. The future work of Ultra-colour system on colour emotion may focus on the range of affective words in association with the 4 personality types, and the link between these words and colour combinations.

c) Ultra-colour system and colour harmony with a cultural context

The results from Experiment 2 ensure ability of Ultra-colour system in colour harmony prediction. The experiment was conducted in 5 different European countries, and the results indicate the consistency of colour harmony response across the countries. This suggests that colour harmony is independent of cultural context, which

Page 26: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

26

was supported most by Spanish, Swedish, and German data. However, the experimental results indicate effects of cultural difference and educational-background difference in British-1 data, as British-1 was found in poor correlation with all the other observer groups. The colour harmony theory behind the Ultra-colour system is based on colour classifications in association with 4 personality types. However, this theory does not seem to sufficient to develop a comprehensive system of colour design, because the theory does not deal with the interrelationship between colours in a combination. Nor is it concerned with the foreground-background effect, and the effect of colour shape. The future work on colour harmony would focus first on the understanding of interrelationship between colours in a combination.

Page 27: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

27

5. ACKNOWLEDGEMENTS

Many thanks go to our collaborators across 5 European countries for their great effort on the experiments, including Dr Andrée Woodcock, Coventry School of Art and Design, Coventry University, United Kingdom; Dr Monica Billger, Design och media, Sektionen för Arkitektur, Chalmers tekniska högskola, Sweden; Prof Rafael Huertas Roa, Departamento de Óptica, Universidad de Granada, Spain; Prof Alain Tremeau, Université Jean Monnet - LIGIV, France; and Prof Klaus Richter, Bundesanstalt für Materialforschung und -prüfung (BAM), VIII.3901, Germany. Special thanks go to Angela Wright, who has been working with us in the development of the Ultra-colour system and in the present project provided a supplementary observer group and carried out an experimental session in London.

6. REFERENCES Berns, R. S. (1996) Methods for characterizing CRT displays. Displays 16, 173-182.

Ishihara, S. (2003) Ishihara’s Tests for Colour Deficiency. 38 Plates Edition. Tokyo: Kanehara Trading Inc.

Judd, D. B. and Wyszecki, G. (1975) Color in Business, Science and Industry. 3rd Edition. New York: John Wiley & Sons.

Ou, L., Luo, M. R., Woodcock, A., and Wright, A. (2004a) A study of colour emotion and colour preference. Part I: Colour emotions for single colours. Color Research and Application 29, in press.

Ou, L., Luo, M. R., Woodcock, A., and Wright, A. (2004b) A study of colour emotion and colour preference. Part II: Colour emotions for two-colour combinations. Color Research and Application 29, in press.

Wright, A. (1995) The Beginner’s Guide to Colour Psychology. London: Kyle Cathie Limited.

Page 28: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

28

Appendix. Adjective Combinations and the Related Colour Wheels According to the Wright Theory

Group 1

(1-a) Fun Adaptable Bright Charming Cheerful (1-b) Lively Active Enthusiastic High-spirited Optimistic (1-c) Clever Cooperative Friendly Informal Practical (1-d) Fresh New Outgoing Impulsive Communicative (1-e) Fanciful Imaginative Creative Inspired Mischievous

Page 29: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

29

Group 2 (2-a) Calm Peaceful Nurturing Courteous Compassionate (2-b) Formal Elegant Gracious Modest Reserved (2-c) Cool Dignified Fair-minded Patient Polite (2-d) Reassuring Responsible Ordered Diplomatic Conservative (2-e) Soft Understated Sensitive Graceful Sensible

Page 30: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

30

Group 3 (3-a) Abundant Energetic Rebellious Visionary Idealistic (3-b) Earth Mature Natural Sensual Generous (3-c) Colourful Brave Original Protective Adventurous (3-d) Authentic Ethical Intense Sincere Honest (3-e) Environmental Traditional Substantial Solid Committed

Page 31: A Cross-cultural Study on Colour Emotion and Colour · PDF file1 Final Report for the OKI Project A Cross-cultural Study on Colour Emotion and Colour Harmony in Association with the

31

Group 4 (4-a) Contemporary Confident Glamorous Respective Stylish (4-b) Decisive Detached Influential Objective Uncompromising (4-c) Efficient Dry Pragmatic Superior Precise (4-d) Logical Articulate Focused Perfectionistic Productive (4-e) Grand Wealthy Exotic Accomplished Charismatic