Augmenting Basic Colour Terms in English

In an unconstrained colour naming experiment conducted over the web, 330 participants named 600 colour samples in English. The 30 most frequent monolexemic colour terms were analyzed with regards to frequency, consensus among genders, response times, consistency of use, denotative volume in the Munsell and OSA colour spaces and inter-experimental agreement. Each of these measures served for ranking colour term salience; rankings were then combined to give a composite index of basicness. The results support the extension of English inventory from the 11 basic colour terms of Berlin and Kay to 13 terms by the addition of lilac and turquoise. Introduction The human visual system is able to discriminate millions of different colours, but for practical purposes and everyday communication we tend to organize them into a smaller set of colour categories and give them common names such as ‘red’ or ‘light blue’. Psychophysical colour-naming experiments offer the most direct and legitimate method of determining the mapping between colour names and the corresponding regions of perceptual colour spaces. Over recent years, colour naming data derived from such experiments has been used for image processing, computer vision and gamut mapping. Colour names are used to signify regions of colour space with empirical significance, and have been found to play an important role in long-term memory and to enhance colour recognition. Most languages have a large number of names to describe colours. Like all words, they are subject to fashion and may change their meanings over time. Yet, there exists a small number of basic colour terms (BCTs) that are shared and comprehended well by most speakers in each language. Unconstrained colour naming experiments are able to capture a great deal of the richness of colour language, including single and multiple word descriptions, but establishing the basic terms in each language has proved to be a difficult task that requires multiple criteria and appropriate tests. 12 The gamut of colours perceived by a normal trichromatic observer is a manifold in a three-dimensional colour space. Each colour from the visual environment can be mapped onto a point within this colour space, and a colour term can be defined by the extent of the applicable region. We are interested in finding the distribution of each colour term, and also the location within each region of the point representing its center-of-mass, called centroid. Because perceived colour space is a continuum, with no intrinsic restrictions on how it can be mapped into a lexicon of colour terms, it would seem that any number of arbitrary mappings would be equally valid. In practice, speakers of diverse languages show a surprising degree of consensus, 13 especially for focal colours, with the inter-language differences being less than the intra-language differences among speakers. It has been hypothesized that languages gravitate to an optimal set of categories and return to them despite departures from the norm by individual speakers. The main question underlying this study is what is meant by ‘basicness’. This goes to the heart of the meaning of categories, and why people seek to classify an object or a perceived characteristic of an object as being in one category or another. The tendency to discriminate one thing from another is inextricably linked with learning and recognition. To categorize a situation as ‘A’ not ‘B’ may mean the difference between safety and danger, even life and death, for example the identification of the colour of the traffic lights on the road ahead. Dummett noted the distinction between categories of names (linguistic designations) and categories of the entities to which the names refer. He argued that categorical concepts are necessary for us to single out ‘things’ in every situation. So the category differences between the names we use are, ipso facto, also category differences between the things singled out by these names. In this way, the connection between the category of an expression used to refer to a given entity (the ‘reference’) and the category of the entity referred to (the ‘referent’) is ensured. For communication between people in a society, however, more is needed beyond the categorical association of a name with an entity. Each individual might have his or her own idiosyncratic system of categories, but for meaningful exchange with others there has to be some shared understanding and commonality of categories throughout the social group. At the simplest level, this begins with monosyllabic utterances referring to physical entities in the surrounding environment familiar to every individual, such as earth, sea, sky, snow, etc. These are the basic terms upon which human communication and language are built. In the case of colour, the referent is not a physical object but a sensation, something experienced by an individual. The colour name is a label, shared by members of a language group, on the assumption that when individuals look at the same object they experience a similar sensation, or at least that the individual sensations they experience can be categorized with a common name. In many cases, colour names arise by familiar association with food. Thus, orange is not only the name of an edible fruit, but also a signifier for the colour of that fruit. By abstraction then orange becomes a metonym for all objects that give rise to the same group of visual sensations. Basicness in the context of colour names means the minimal set of linguistic signifiers that individuals within a language group can use to communicate their categories of colour sensation. This depends on the responses of the human visual system, the referents in the environment of the social group and the stage of development of the language. To qualify as a BCT, a colour name in our view should: a. be widely used in a population of speakers; b. have a shared meaning for the associated colour stimulus; c. be salient in the sense that the colour is easily identifiable in an array; and d. be reliably distinguishable from its neighbours in colour space. Criterion (d) means that a basic colour name should retain its identity when inverted, i.e. that the centroid of a colour name should not be subsumed by larger neighbour categories. This article explores the notion of basicness using a large set of empirical responses from an online colour naming experiment and with a set of appropriate behavioural measures. The collected names are analyzed by the frequency of usage of colour terms, consensus among genders, consistency of responses, response times, denotative volume in colour space and inter-experimental agreement. We explore, in particular, whether the number of BCTs in English should be extended beyond the 11 established by Berlin and Kay. This would have useful applications for improving the precision of colour naming in colorimetric colour spaces and for facilitating colour communication within and between different cultures over global networks.