Basic color categories and Mandarin Chinese color terms

Lü [4], who surveyed 1,815 local respondents in Taiwan and collected more than 1600 color terms, reported that there were terms for each of the eleven basic color terms proposed by Berlin and Kay [1]. Lin et al. [5, 6], applying a similar unconstrained method as Berlin and Kay [1] and Lü [4], suggested that there were eleven basic color names in Mandarin Chinese. These were used in the same way as the corresponding terms in British English and therefore matched the eleven found by Berlin and Kay [1]. However, Gao and Sutrop [7], using a similar approach, showed that there were nine basic color terms in Mandarin Chinese with orange and brown missing from Berlin and Kay’s [1] eleven. Using a more quantitative approach, Uchikawa [8] found that of the eleven basic color terms, those used most consistently are gray, blue, green, red, yellow, purple, and pink; less consistently used are brown, orange, white and black.

The discrepancies in what comprises the basic Mandarin color terms may be resulted from a particular characteristics of Mandarin Chinese in that there is more than one synonym for each color. Different respondents may just use different terms for the same color. This variety of color terms makes it difficult to determine the basic color terms according to the criteria suggested by Berlin and Kay [1]. For instance, Uchikawa [8] and Gao and Sutrop [7] identified that while there were two terms for brown in Mandarin (Zong and He), each respondent may consistently use just one or the other to name the brownish chips. Hence, in the total population there was little consistency in the naming of brownish chips even though each of the two terms may be salient for an individual. As a result, brown was excluded from basic color terms for failing Berlin & Kay’s [1] criteria of saliency. Later researchers who were either native Mandarin speakers themselves or had native Mandarin speakers as colleagues corrected their data, based on their first-hand understanding of the language [4, 5, 7], suggested more basic color terms than those of Berlin and Kay [1]. The discrepancies among those research findings could be caused by multiple color terms used in Mandarin for a single category of color. However, although factoring in multiple color terms for the same perceived colors may be a linguistically sound process, it still requires empirical justification. Thus the purpose of this study is to test whether these terms refer to the same color categories.

Many studies have described the basic color terms according to color categories [9–15] without distinguishing between the two. These studies showed that basic color terms could be just labels for basic color categories, which were common cognitive structures among users of a language. In the studies of Mandarin color terms, Lin et al. [6] used the basic color terms as indexes to categorize color samples. Thus, while Mandarin speakers may show inconsistency in the use of color terms, their reference to color categories may be consistent. However, assessing this is not possible by the conventional method for color terms, which is to collect the terms used for various color samples. For instance, one respondent may use only Zong for brown while the other, He. The researcher would not know whether each was using a different word for the same color category or two overlapped color categories with the test sample as the intersection. In the current study, we used a different approach. Instead of asking participants to give color terms for a color sample, we gave them color terms and measured what color samples were associated with each term. If two color terms were synonyms for the same color category, the participants should choose similar sets of color samples as representations for these terms. Thus, the participants’ response to these terms would be highly correlated. Such correlation should be able to be detected by a factor analysis. Thus, our paradigm should be more suitable for establishing the association between color samples and color terms when individuals use a variety of words.

For the test, thirty-two single-character Mandarin words were selected from 4430 of the most frequently used characters in the Academia Sinica Balanced Corpus [16] (Table 1). We used only single-character words to ensure the test material met the monolexical criterion of Berlin and Kay [1] and we used only medium- to high-frequency words to fulfill the saliency criterion. Eleven of these were Lü’s [4] Mandarin basic color terms. Some words did not appear on the list of basic color terms but were considered to be synonyms of the eleven terms. For example, Chi (赤), Zhu (朱), and Dan (丹) all mean RED in Mandarin, and Zong (棕) and Zhe (褚) both mean BROWN. Some terms are names of common objects but are also commonly used to describe the color of another object. An example is Cha (茶), or ‘tea’, which is often used by Mandarin speakers to describe a brownish color. The rest are modifiers that can be associated with many colors and often are not used as an independent term. One example is the word Dan`(淡) or ‘pale’ which can be linked to many other color terms to form a compound word for the unsaturated version of that color. With basic color terms, we expected that the participants would have a highly concentrated selection in a handful of samples. However, if the number of selected samples is limited, it would be difficult to discern whether it is due to a lack of motivation for the participants to pick up more samples or the specialization of the color category. On the other hand, the modifiers, which can be associate with many colors, should allow the participants to select a wide range of color samples and thus can be used a diagnostic tool for the former issue. In indeed, as reported in Result, the participants did make a widespread choice of samples for the modifiers and concentrated ones for the basic color terms. Thus, it is an evidence that our participants made their best effort for the task.

The first factor, explained 21.2% of variation in the data, was the “red” family. The terms Chi, Hong, Zhu, Zhe, Dan and Xie (赤,紅,朱,赭,丹,血) all had high loading on this factor (Fig 5). The term Hong is a typical translation of “red” in Mandarin. The frequency distributions for Chi, Zhu, Dan and Xie, are virtually almost identical to Hong. This provides evidence that Hong, Chi, Zhu and Dan are indeed synonyms for the color of blood (Xie). The frequency distributions for the secondary term Zhe extended to the brown region, as one would expect from its meaning of reddish brown.

The second factor, explained 16.9% of variation in the data, was the “brown” family. The terms He, Zong and Tu (褐,棕,土) showed a strong loading on this factor (Fig 6). These three terms had a very similar frequency distribution to “brown” in English on the WCS chart, suggesting that they are all synonyms for the color of mud (Tu) or the bark of a pine tree (Zong). The secondary terms Zhe (赭 reddish brown) and Cha (茶 tea, thus brownish green) also showed loading on this factor. Their frequency distribution covered extensive regions overlapping the “brown” region but also extended to “red” or “green” for Zhe and Cha respectively.

The third factor, explained 11.8% of variation in the data, was the “green” family. The terms Cai, Cui, Lu, Bi and Qing (菜,翠,綠,碧,青) were heavily loaded on this factor (Fig 7). Lu is a typical Mandarin translation of “green”. The frequency distribution of Cai, Cui, and Lu were very similar, suggesting that they are synonyms for the color of the vegetable (Cai). The frequency distribution for the secondary terms Bi (Jade) and Qing extended to the blue region, suggesting “blue-green”.

The fourth factor, explained 9.9% of variation in the data, was the “black” family. The terms Hei, Mo and Xuan (黑,墨,玄) were heavily weighted on this factor (Fig 8). Hei is a typical translation of “black”. Virtually all participants selected the darkest monochromatic sample when prompted by Hei and Xuan. In addition to black, Mo can be associated with another color term for the dark version of that color. Thus, it is not surprising that its frequency distribution covered a wide range of dark color samples.

The fifth factor, explained 8.1% of variation in the data, was the “orange” family. The terms Cheng and Ju (橙,菊) showed high loading on this factor and had almost identical frequency distributions to the WCS chart (Fig 9). Thus, these two terms are indeed synonyms for the color of an orange (Cheng).

The sixth factor, explained 6.1% of variation in the data, was the “gray” family. The terms Yin, Te, and Hui (銀,鐵,灰) all had a large loading of this factor (Fig 10). Their frequency distribution covered the intermediate achromatic region. Hui, a typical Mandarin translation of gray, covered a wide range of the achromatic region. Yin (silver) covered mostly the bright achromatic region while Te (iron) covered mostly the dark achromatic region. However, since Te can also be used as a modulator, it is not surprising that it was identified with a wide range of dark colors.

The seventh factor, explained 4.7% of variation in the data, was the “yellow” family. The terms Huang and Jin (黃,金) were heavily weighted on this factor and had a similar frequency distribution (Fig 11), suggesting indeed that Huang refers to the color of gold (Jin).

The terms Ying and Tao (櫻,桃) were heavily weighted on the eighth factor, which explained 4.4% of variation in the data,. Their frequency distributions were almost identical (Fig 12) and occupied the same regions as “pink” in English (1). Thus, this factor represents pink.