Colour - Cognitive Universalism Flashcards
Basic term requirements (Berlin & Kay, 1969)
Monolexemic, application not restricted to narrow class of objects, psychologically salient, not source-descriptors
Berlin and Kay (1969) Study
Obtained naming data and focal data (best-example) for speakers of 20 languages
Systematic sequence of colour term development across languages
Best-example data and ranges of colours seemed to cluster together in colour space, typically to six Hering primaries
Berlin and Kay (1969) Criticisms
Out of 7000 languages worldwide, only 20 sampled and all around San Francisco Bay area
All pps bilingual, so perhaps it was speaking English that augmented categorical perception of colour to be like English
Only tested 1-2 pps per language
Anglo-centricity
Roberson et al (2000) / Davidoff et al (1999)
Colour categories not organised around universal foci, but determined by naming distinctions at category boundaries (vary cross-linguistically).
Viewed foci as epiphenomena - once categories defined by language-determined boundaries, best examples derived secondarily as centres of these categories.
Refuting universalism through: (1) Best-example choices for some Berinmo colour terms diffuse and not falling at proposed universal foci; (2) Berinmo-English differences in boundaries influenced non-linguistic memory for colour
However, non-industrialised > perhaps Universalism (Berlin & Kay, 1969) findings resulted from global spread of industrialisation, not genuinely universal forces
Regier et al (2005)
Aimed to delineate between Berlin and Kay (1969) and Roberson et al (2000)
World Colour Survey which addressed Berlin and Kay (1969) weaknesses: 110 unwritten languages of non-industrialised, 25 speakers per language
Naming data and focus data
Pooled all focus data from all speakers and calculated how many best-example choices fell on each colour chip, with most following sequence of six Hering foci (Berlin & Kay, 1969)
Found best examples of colour categories cluster more tightly across languages of all societies (compared WCS and Berlin and Kay (1969) datasets) than do centres of categories’ extensions > supports claim that best examples reflect universal structure around which colour categories are formed
Roberson et al (2000) versus Regier et al (2005)
Roberson et al (2000) reported 5 colour terms for Berinmo, and distribution similar to several WCS five-colour-term languages.
Roberson et al (2000) stimulus array didn’t include any achromatic chips (couldn’t test if similar to WCS and English for black and white), leaving best-example choices peak near WCS peaks for red, yellow, green, but Roberson et al (2000) collected data for only even-numbered columns in array, so red, yellow, green universal foci not themselves available as selections to Berinmo pps.
Criticisms of Regier et al (2005)
Unable to assess source of universally favoured regions
Unable to assess linguistic relativity, but doesn’t dispute it as long as one agrees that variation of category boundaries itself is constrained by universal forces
Roberson et al (2005) showed that best examples do fall outside cluster of six Hering, and that category boundaries do vary, suggesting that category boundaries determined by more than six proposed universal foci
Regier et al (2007) - Well-Formedness
Test new view that colour naming reflects optimal or near-optimal divisions of an irregularly shaped perceptual colour space
Took modal colour term for each chip in Munsell space for each WCS language, and converted it to CIELAB colour space where distance between colours corresponds to perceptual dissimilarity
Simulated theoretically optimal colour-naming systems, maximising well-formedness > for 3, 4, 5, 6 colour terms, most languages close to optimal categorisation
Tested prediction in Berinmo, shown to counterexemplify universal tendencies in colour naming (Roberson et al., 2000) by systematically rotating actual data of naming system by hue columns in stimulus array, finding unrotated naming system highest well-formedness
For most of 110 languages in WCS, rotation showed same results
Well-Formedness (Regier et al., 2007)
Categories constructed to maximise similarity within categories and minimise it across categories
