Recent findings of colour semantics can be divided into:Single-Colour Semantics
There have been two main subjects in single-colour semantics: classification and quantification. The classification of colour semantics is typically concerned with whether a large number of colour-semantics scales can be reduced into a smaller number of categories, so as to discover underlying factors of colour semantics. The principal component analysis has been widely used for such work, including studies by Kobayashi (1981), Sato et al. (2000) and Ou et al. (2004a).
The first effort for quantifying colour semantics was made by Sato et al. (2000) who claimed that such models had a structure similar to a colour difference formula. Using similar techniques, Ou et al. (2004a) developed a number of colour-semantics models for scales like warm-cool (WC), heavy-light (HL) and active-passive (AP):
where L*, 
and 
represent CIELAB lightness,
chroma and hue angle, respectively.
The diagram below shows the experimental settings in Ou’s study. Click here to see a simulation of the experiment.
Xin et al. (2004) also developed colour-semantics equations which focused on the correlation between scale values and the three colour-appearance attributes, i.e. hue, lightness and chroma, assuming that these 3 attributes are independent of each other in the models.
Colour-Pair Semantics
Ou et al. (2004b) revealed a simple relationship that relates single-colour semantics and colour-pair semantics. The following shows the experimental settings of this work.
Ten colour-semantics scales were used in this study: warm-cool, heavy-light, modern-classical, clean-dirty, active-passive, hard-soft, tense-relaxed, fresh-stale, masculine-feminine and like-dislike. The following equation was found to apply to all these scales except "like-dislike":
where E is a colour-semantics
value for a colour pair made by colours 1 and 2; 
and 
are
colour-semantics values for these two colours.
This equation suggests an additive relationship in colour semantics between single colours and colour combinations, as illustrated in the following.
As has been verified by Sueeprasan et al. (2005), this model is useful in particular for combinations of more than two colours, but this requires further psychophysical evidence.
Cross-Cultural Issues
A study was conducted recently to investigate how cultures influence the viewers’ emotional responses to single colours and to colour pairs (Ou et al., 2005). In this study, a psychophysical experiment was carried out in 6 countries: Britain, France, Germany, Spain, Sweden and Taiwan. A total of 20 single colours and 190 colour pairs were used as the stimuli, presented individually on a calibrated cathode-ray-tube (CRT) monitor in a darkened room at each experimental site. Four scales, “warm-cool”, “heavy-light”, “active-passive” and “like-dislike”, were used in the experiment to measure colour semantics and preference using the method of categorical judgement. In this experiment, the CRT monitors for different countries were chracterised to achieve accurate colour reproduction using the GOG model (Berns, 1996) with a colour sensor, as shown in the following.
The experimental results show that there was little cultural effect on all the scales except “like-dislike”. The Spanish data were found to disagree with the other 5 observer groups on “like-dislike”. The experimental results show that the Spanish observers tended to prefer colour pairs with small lightness difference between constituent colours in each pair, while the other 5 groups tended to prefer colour pairs with large lightness difference.
Single-Colour Preference
Perhaps the simplest definition of colour preference is “one colour being liked more than another”. (Nemcsics, 1993) While early studies of colour preference focused mainly on ranking of hue preferences, reports have shown that colour preference is affected by all the three colour-appearance attribute, i.e. hue, lightness and chroma. A quantitative model for single-colour preference was developed (Ou et al., 2004c) based on psychophysical data for British and Chinese observers, as shown below:
Single-Colour Preference =
where L*, a* and b* are the three CIELAB coordinates for the colour stimulus.
This model relates single-colour preference with the colour difference between the colour under test and a reference colour (i.e. the least liked colour), a medium-lightness yellow at a hue angle of 105º and chroma of 31. This model implies blue to be the most liked colour and yellow the least liked, provided that the colours under comparison are of the same lightness and chroma.
While it has been difficult to model colour-combination preference, researchers turned to look at the relationships between preference and harmony in colour combinations, and the results were positive. The problem is, modelling colour harmony too is challenging for scientists.
Here we introduce a colour harmony model for two-colour combinations developed recently on the basis of psychophysical data.
Colour-Pair Harmony
Despite
a great number of colour harmony theories available
today, a perhaps most objective definition of colour harmony is: “when two or more colours seen in neighbouring areas
produce a pleasing effect,
they are said to produce a colour harmony.” (Judd
and Wyszecki, 1975) Based on categorical-judgement data
using Chinese observers, a quantitative model of harmony for two-colour combinations
was developed recently by Ou et al., (2006) with
a CRT monitor for presenting the visual stimuli, as shown on the right.
The model was found to have good predictive performance for a number of independent
experimental data. A simplified outline of the model is shown in the following:-
Equal-chroma. Any two colours identical/similar in chroma tend to appear harmonious when combined together.
Unequal lightness values. Small lightness variations between the constituent colours in a colour pair may reduce the harmony.
High lightness values. The higher the lightness value of each constituent colour in a colour pair, the more likely it is that they appear harmonious.
This model means that colour harmony can be determined by the interrelationships between constituent colours in a colour combination, in terms of similarity/dissimilarity in each colour attribute. Note, however, that the model may apply only to colour pairs presented on a grey background, thus in its current form the model is not necessarily useful in design practices.
Colour Preference vs Colour Harmony
To discover the relationships between colour preference (in terms of like-dislike) and colour harmony, Ou et al. (2004a) conducted a psychophysical experiment for two-colour combinations.
Firstly, the two scales "like-dislike" and "harmonious-disharmonious" were found to be highly correlated, with a correlation coefficient of 0.85. This means that basically if a colour pair is liked, it is highly likely that this pair appears harmonious, and vice versa.
For further clarifications, the experimental data were divided into two cases:-
1) when a colour pair was judged as "liked" but "disharmonious";
2) when a colour pair was judged as "harmonious" but "disliked".
As a result, there were about 4% of colour pairs in Case 1 and about 18% in Case 2. This indicates that the colour pairs that were preferred were not necessarily regarded as harmonious, and that those regarded as harmonious were not always preferred. Note that the number of colour pairs in Case 2 was much larger than that in Case 1; this means a higher probability for Case 2 to occur than for Case 1.
The diagram on the right illustrates the relationship between colour preference
and colour harmony according to the experimental data. It shows that the number of colour pairs that were ‘liked
and harmonious’ was almost the same as that of colour pairs that
were ‘disliked and disharmonious’. The former made up 40% of
the total number of colour pairs, and the latter made up 38%. Harmonious
colour pairs comprised 40 + 18 = 58% of the entire range of colour pairs,
and that was a lot higher than the number of ‘liked’ colour
pairs, 44%. This implies that if an observer likes a colour pair, there
will be a 9% chance (4 / 44 = 9%) that the observer finds the same colour
pair disharmonious (Case 1). On the other hand, if an observer decides
that a colour pair is harmonious, there will be a 31% chance (18 /
58 = 31%) that the observer dislikes the same colour pair (Case 2).
The reason why Case 2 occurred more likely than Case 1 is perhaps that colour preference is one of the factors that influence observers’ judgments on colour harmony, i.e. observers tend to judge a colour pair to be harmonious when they like it. However, judgements on colour preference may be dominated by observers’ subjective criteria such as personal taste and the effect of cultural difference, and thus observers more often dislike a colour pair when they find it harmonious.
The above results suggest that although colour preference is highly correlated with colour harmony (with a correlation coefficient of 0.85), there are still areas where they disagree with each other. Further studies are required to clarify this.
References
Berns, R. S., Methods for characterizing CRT Displays, Displays, 16, 173-182 (1996).
Child, I. L., Hansen, J. A. and Hornbeck, F. W., “Age and sex differences in children’s color preferences”, Child Development, 39, 237-247 (1968).
Crozier, W. R., “The psychology of colour preference”, Review of Progress in Coloration, 26, 63-72 (1996).
Eysenck, H. J., “A critical and experimental study of colour preferences”, American Journal of Psychology, 54, 385-394 (1941).
Garth, T. R., “A color preference scale for one thousand white children”, Journal of Experimental Psychology, 7, 233-241 (1924).
Judd, D. B. and Wyszecki, G. , Color in Business, Science and Industry, 3rd Edition, John Wiley & Sons, New York (1975).
Kobayashi, S., The aim and method of the Color Image Scale, Color Research and Application, 6, 93-107 (1981).
Nemcsics, A., Colour Dynamics: Environmental Colour Design, Ellis Horwood, New York (1993).
Norman, R D. and Scott, W. A., “Colour and affect: a review and semantic evaluation”, Journal of General Psychology, 46, 185-233 (1952).
Ou, L., Luo, M. R., Woodcock, A., and Wright, A., A study of colour emotion and colour preference, Part I: colour emotions for single colours, Color Research and Application 29, 232-240 (2004a).
Ou, L., Luo, M. R., Woodcock, A., and Wright, A., A study of colour emotion and colour preference, Part II: colour emotions for two-colour combinations, Color Research and Application 29, 292-298 (2004b).
Ou, L., Luo, M. R., Woodcock, A., and Wright, A., A study of colour emotion and colour preference, Part III: colour preference modelling, Color Research and Application 29, 381-389 (2004c).
Ou, L., Luo, M. R., Cui, G., Woodcock, A., Billger, M., Stahre, B., Huertas, R., Tremeau, A., Dinet, E., Richter, K. and Guan, S., The effect of culture on colour emotion and preference, Proceedings of the 10th Congress of the Association Internationale de la Coleur (AIC), Granada, Spain, 259-262 (2005).
Ou, L., Luo, M. R., A colour harmony model for two-colour combinations, Color Research and Application 31, 191-204 (2006).
Sato, T., Kajiwara, K., Hoshino, H., and Nakamura, T., Quantitative Evaluation and categorising of human emotion induced by colour, Advances in Colour Science and Technology, 3, 53-59 (2000).
Sueeprasan, S., Srimork, P., H, An investigation of colour emotions using two-colour combinations, Proceedings of the 10th Congress of the Association Internationale de la Coleur (AIC), Granada, Spain, 271-274 (2005).
Xin, J. H., Cheng, K. M., Taylor, G., Sato, T. and Hansuebsai, A., “Cross-regional comparison of colour emotions Part I: Quantitative Analysis”, Color Research and Application, 29, 451-457 (2004).
Copyright © 2006-2008 Li-Chen Ou. All rights reserved.
