Comparing Distributions of Color Words: Pitfalls and Metric Choices

Abstract

Computational methods have started playing a significant role in semantic analysis. One particularly accessible area for developing good computational methods for linguistic semantics is in color naming, where perceptual dissimilarity measures provide a geometric setting for the analyses. This setting has been studied first by Berlin & Kay in 1969, and then later on by a large data collection effort: the World Color Survey (WCS). From the WCS, a dataset on color naming by 2 616 speakers of 110 different languages is made available for further research. In the analysis of color naming from WCS, however, the choice of analysis method is an important factor of the analysis. We demonstrate concrete problems with the choice of metrics made in recent analyses of WCS data, and offer approaches for dealing with the problems we can identify. Picking a metric for the space of color naming distributions that ignores perceptual distances between colors assumes a decorrelated system, where strong spatial correlations in fact exist. We can demonstrate that the corresponding issues are significantly improved when using Earth Mover's Distance, or Quadratic -square Distance, and we can approximate these solutions with a kernel-based analysis method.

A metric for measurable partitions

“…This allows them to use cluster analysis, and related techniques, in order to let the data reveal commonalities in how different languages tend to cluster different colours together. They partly confirm, and partly add new insights to, earlier universalist proposals (see Vejdemo-Johansson, Vejdemo, & Ek, 2014 for further discussion on suitable choices of metrics). Instead of measuring distance between focal colours/centroids, or measuring distance between extensions of colours, one may use the metric proposed in this paper to directly measure the distance between colour categorizations.…”

Morpho-genetic diversity and anti-bacterial activity in root extracts of nine solanaceous species

“…The genotype Welthibbatu contained prickles only (Table 2). Colour codes are according to Vejdemo-Johansson et al, (2014). Means denoted by the same letters within the columns are not significantly different at P<0.05.…”

“…The colour of the stem was recorded before maturity (prior to onset of bloom) and at maturity according to Vejdemo-Johansson et al, (2014). The presence of hairs and prickles on the stem at maturity was also recorded.…”

“…The presence of hairs and prickles on the stem at maturity was also recorded. The petiole length, colour (Vejdemo-Johansson et al, 2014) length, breadth, angle of the leaf with stem, the presence of prickles on the leaves and hairs were recorded in five leaves per plant at mature stage as the parameters related to leaf morphology. Similarly, for floral morphological parameters; length and breadth of petals and sepals, flower diameter, stamen length, numbers of prickles in the pedicel and petal colour were recorded.…”

“…1). Colour codes are according to Vejdemo-Johansson et al, (2014). Glossiness: 1 (dull), 2 (fairly shiny), 3 (shiny) Genetic Diversity Analysis DNA Barcoding PCR based on the matK DNA barcoding primers yielded an amplicon with the length of 936 bp ( Fig.2A).…”