Evolution of Colour Terms: 5 Cultural Constraints

Continuing my series on the Evolution of Colour terms, this post reviews studies of cultural constraints on colour naming. For the full dissertation and for references, go here.

This section reviews evidence of cultural constraints on colour terms.  Modelling has shown that cultural transmission can cause individual categorisations of colour space to converge on shared categories, given constraints on perception and the environment (Steels & Belpaeme, 2005) and also without constraints from the environment (Belpaeme & Bleys, 2005).  In Steels & Belpaeme’s (2005) study, cultural transmission was modelled using agents with categorical networks for discrimination, but also a lexicon which associated a word with each category, similar to other models of emergent categories (e.g., de Boer 2000).  No genetic inheritance was modelled.  Agents played a communication game.  The hearer could adjust its categorical network and lexicon to better fit what had been revealed about the speaker’s network and lexicon.  This adjustment could involve adopting the speaker’s name for a category, creating a new category or adjusting the focus of a category.  In this way, language communication stimulates the formation of categories, but also the formation of categories stimulates the language used.  This is a model of two way structural coupling between category formation and language (Maturana & Varela, 1998).  Steels & Belpaeme found that, although category boundaries were not completely shared, agents did have a common repertoire.  Furthermore, agents could be replaced with naive ones without disrupting the communicative success of the population.

This has also been demonstrated in human-based experiments (Dowman, Xu & Griffiths, 2008).  This was done using the transmission-chain paradigm (e.g., Kirby, Cornish & Smith, 2008).  In Dowman et al. (2008), participants were shown colours with made-up names, and asked to memorise them.  The participant was then tested on these names.  However, the participant did not see all colour-name pairs during training, imposing a bottleneck on learning.  The names shown to the first participant are randomised and randomly assigned to colours.  That participant’s output – their recalled pairings of colours and colour names – are given to the next participant as their input.  This process of iterated learning is repeated down the chain.  Chains were initiated with 2, 3, 4 5, and 6 initial colour terms with 13 participants in each chain with.  As with other transmission chain experiments (e.g., Kirby, Cornish & Smith, 2008), structure in the categories emerged in response to a pressure to become more learnable.  With 2 or 3 initial colour terms, the categories changed to be based either on hue or lightness.  With 4-6 categories, categories emerged based on both hue and lightness.

However, these studies have not ruled out perceptual constraints as the major cause of cross-cultural patterns.  This may not be necessary, however:  Kirby, Dowman and Griffiths (2007) have shown that very weak innate biases can be amplified through cultural transmission, leading to universal patterns.  Therefore, it’s possible that the constraints imposed on colour perception by the sensory organs are actually relatively small, and cultural processes are the main driving forces of colour term evolution.  This is supported by showing that simple discrimination rules combined with pragmatic constraints can lead to shared systems, without complex models of the perceptual system (Komarova, Jameson and Narens, 2007).

If culture is a strong influence on colour categorisation, then Relativism is supported, since different cultures can arrive at different ways of comprehending the world, based on their functional needs (e.g., hunting vs. gathering vs. interior decorating).  If this is the case, then we would expect the WCS to contain large variance between cultures, contrary to Berlin and Kay’s original assumption.  There is some evidence for this.  Firstly, there are several systems of colour categorisation in the world that are remarkable outliers in the WCS.  For example, the Tzotzil language of Chiapas, Mexico, studied by MacKeigan & Muth (2006):  Although the WCS lists only 5 basic colour terms for Tzotzil, MacKeigan & Muth argue that there arebetween 900 and 1000 colour compounds.    Below is one speaker’s description of a purple Munsell Chip:

“ ik’ik’+tik s+k’an+al s+yax+al  … This possessive form instructs the viewer to search the overtones and reflections of the black for hints of yellow and then for dark blue, … This colour term is roughly equivalent to “dark’s yellowness (including orange) and blue-greenness”, showing that a literal translation does not arrive at the required “dark purple”. Clearly, in- depth grammatical analysis is required.” (MacKeigan & Muth, 2006, p.27-28)

Note that the system is highly morphologically complex, as predicted by Lupyan and Dale (2009).  The complexity is due to the Tzotzil culture being heavily bound-up with colour.  The colour system is connected to orientation in the landscape, seasonal cycles and personality.    Distinctions are made based not only on hue, saturation and brightness, but also on the size of the object, discreteness, opacity, texture and movement.  For Tzotzil speakers, then, concepts of colour should evoke many kinds of perceptual representations.  Evidence for this would support Relativism.

Furthermore, colour terms have often emerged due to economic pressures (e.g., ‘Azure’ comes from Azurite – a very expensive material used in painting in the Middle Ages; Ball, 2001) or artistic invention (e.g., ‘vermilion’ comes from the insect used to make dye or ‘International Klein Blue’ from the work of Yves Klein; Ball, 2001).  Language contact can also have an effect.  For example, Middle Welsh did not have separate terms for green and blue whereas modern Welsh does, possibly due to a pressure from contact with English, although the original ‘grue’ term can still carry its original connotation (Lazar-Meyn, 2004).  Overall, there is much evidence that cultural constraints can affect categorisation, supporting the Cultural adaptation implication.  However, in order to support a Relativist account, these differences would have to be more than superficial differences in colour words.  The variation caused by cultural processes must reflect fundamental variation in cognition and conceptual approaches to action and the environment.

Next, Categorisation Constraints ->

Steels, L., & Belpaeme, T. (2005). Coordinating perceptually grounded categories through language: A case study for colour Behavioral and Brain Sciences, 28 (04) DOI: 10.1017/S0140525X05000087

Belpaeme, T., & Bleys, J. (2005). Explaining Universal Color Categories Through a Constrained Acquisition Process Adaptive Behavior, 13 (4), 293-310 DOI: 10.1177/105971230501300404

Kirby, S., Cornish, H., & Smith, K. (2008). Cumulative cultural evolution in the laboratory: An experimental approach to the origins of structure in human language Proceedings of the National Academy of Sciences, 105 (31), 10681-10686 DOI: 10.1073/pnas.0707835105

Kirby, S., Dowman, M., & Griffiths, T. (2007). Innateness and culture in the evolution of language Proceedings of the National Academy of Sciences, 104 (12), 5241-5245 DOI: 10.1073/pnas.0608222104

KOMAROVA, N., JAMESON, K., & NARENS, L. (2007). Evolutionary models of color categorization based on discrimination Journal of Mathematical Psychology, 51 (6), 359-382 DOI: 10.1016/j.jmp.2007.06.001