Map 116 is about the lexical choices that APiCS languages make when referring to the “cool” colour sensations green and blue, specifically, whether the basic colour terms for green and blue are identical or different. The map is based on the corresponding WALS chapter by Kay & Maffi (2005), who follow Hering ( 1964) in proposing a universal of six primary colour categories (black, white, red, yellow, green, blue). Languages differ as to how the psychological colour space is covered by their basic colour terms, i.e. words whose “meaning is not predictable from the meaning of [their] parts”, whose “signification is not included in any other color term”, whose “application must not be restricted to a narrow class of objects”, and which are “psychologically salient for informants” (Berlin & Kay 1969: 6).
The data for the World Color Survey (see Kay et al. 2010, http://www1.icsi.berkeley.edu/wcs) that form the basis of Kay & Maffi (2005) WALS contribution were collected using objective psychological methods (to elicit colour terms, a set of 330 Munsell colour chips were shown to native speakers). It has to be pointed out that the APiCS data is more impressionistic and results thus have to be treated with some caution: The contributors were not provided with standardized colours to show to speakers of the contact languages, and in many cases the questionnaire answers are based on the contributor’s post-hoc interpretation of field recordings and may therefore be influenced by the colour system of the contributor’s language. Nevertheless, Map 116 shows some interesting results that call for further research.
Note that Kay & Maffi (2005) “Green and blue” map includes information on whether these colour terms also cover black or yellow. This is not the case in APiCS, which considers the terms for green and blue only, regardless of whether they also refer to other colours. From the comments of some APiCS contributors it is clear, however, that in contact languages the terms referring to green and blue cover other colour sensations too. For example, Sango vuko/voko means ‘be dark/black/blue/green/etc.’ (Samarin 2013) and Korlai pret, in variation with Marathi-derived shahi ‘blue’, refers to ‘black, (dark) blue’ (Clements 2013). That the same term refers to black and blue is a common phenomenon in the world’s languages and is found in the first four of the five evolutionary stages of basic colour term systems proposed by Kay & Maffi (1999: 748); see also §3.
We distinguish between four values for this feature:
Value 1. In differentiation one basic colour term refers to green and another to blue. This is attested in e.g. African American English:
green referring only to green
blue referring only to blue (Green 2013)
With 83% of the languages in our sample showing differentiation, this is the most frequent value.
Value 2. Identity refers to a colour system where one basic colour term covers both green and blue, and no word exists that denotes only ‘green’ or only ‘blue’. A minority of six APiCS languages (8%) have such a system, as illustrated by Fa d’Ambô:
vedyi referring to green or blue
blue referring only to blue (Post 2013)
Value 3. In the situation called overlap there are two different basic colour terms. One refers to the green and the blue sensation and the other only to green or blue. Fanakalo, Nigerian Pidgin, and Pidgin Hawaiian are the three contact languages in our sample (4%) that have green/blue overlap:
grin referring to green or blue
blu referring only to blue
Value 4. The last value, not applicable, contains the three interesting cases of so-called non-partition languages (Kay & Maffi 1999: 744-745, 751-753), which do not have a basic colour term that covers blue and/or green. The basic colour terms of non-partition languages do not completely partition up the psychological colour space. Kikongo-Kituba, Lingala, and Mixed Ma’a/Mbugu have terms for black, white, and red, but the other colour sensations are referred to by fixed phrasal expressions (e.g. kuler ya saka~saka ‘colour of cassava leaves’ in (4a)) or sometimes by loans from languages that do have a basic term for the colour in question (e.g. bule < French bleu ‘blue’ in (4b)):
In fact, the three value 4 languages look exactly like the only non-partition language in the World Color Survey: Yélîdnye, a non-Austronesian language of Rossel Island (Papua New Guinea), has basic terms for black, white, and red, but uses fixed phrasal expressions to refer to the other colours (Kay & Maffi 1999: 751–753).
Numerical distribution. Values 2 (identity) and 3 (overlap) both involve identity in that there is a basic colour term that covers both the green and the blue sensations. Even though Kay & Maffi (2005): 542-543) WALS sample is biased in that Meso and South America are over-represented and Sub-Saharan Africa and Asia are under-represented, a comparison with the APiCS sample may be enlightening. In the WALS map, language systems relying on some kind of green/blue identity form a great majority (85 languages, 71%), so the fact that only 9 APiCS languages (13%) have such systems needs an explanation.
There is a strong tendency for APiCS languages to derive the terms for both green and blue from their lexifiers. Only three languages derive one of the colour terms from a substrate/adstrate: Fanakalo (luhlaza ‘green/blue’ < Zulu, bluwan ‘blue’ < English; Mesthrie 2013), Korlai (verd ‘green’ < Portuguese, shahi ‘blue’ < Marathi; Clements 2013), and Sri Lanka Malay (ijo ‘green’ < Malay, niila ‘blue’ < Tamil; Slomanson 2013). Some contact languages have colour terms from two European languages, such as Papiamentu bérdé ‘green’ from Portuguese or Spanish and blau ‘blue’ from Dutch (Kouwenberg 2013b). Similarly Pichi, which has grin and blu from English as well as verde ‘green’ and azul ‘blue’ from Spanish (Yakpo 2013).
In the vast majority of cases, the meanings of the lexifier colour terms remain unchanged in the contact language, so that the colour systems are the same as those of the lexifier. Since there is a strong bias in APiCS towards European-lexified contact languages (60 of 76 languages have a European lexifier) and all European lexifiers are value 1 (differentiation) languages, the proportion of differentiation languages is much higher in APiCS than in Kay & Maffi (2005) sample—of the 60 European-lexifier APiCS languages, only five are value 2 (identity) or value 3 (overlap) languages: Angolar, Casamancese Creole, Fa d’Ambô, Saramaccan, and Nigerian Pidgin.
Geographical distribution. With the exception of Chinuk Wawa and Saramaccan, the value 2 (identity) and value 3 (overlap) languages are all located in Africa, as are the three non-partition languages (value 4). As far as the European-lexified contact languages among these are concerned, this points towards strong substratal/adstratal African influence, a point that could also be made with regard to the “radical” creole Saramaccan across the Atlantic. With regard to the African-lexifier APiCS languages, this seems to be simply a case of the transfer of the lexifier colour systems to the contact languages, just as with the transfer of green–blue differentiation to the European-lexifier contact languages, discussed above.
The evolution of colour term systems. Kay & Maffi (1999, 2005) propose a five-stage model of the evolution of colour term systems in the world’s languages. The model is only concerned with the six primary colour categories (black, white, red, yellow, green, blue) and basically predicts a development from a hypothesized two-term Stage I to Stage V, where each of the six categories has its own distinct basic colour term.
Diachronic evidence in APiCS supports the Kay & Maffi model. For example, Early Sranan blakka/brakka covered both black and blue (van den Berg & Bruyn 2013), while in Modern Sranan blaka ‘black’ no longer refers to blue (Winford, p.c.). That is, Modern Sranan has entered the last evolutionary stage, where all six primary colour categories are referred to by separate basic colour terms.
Diachronic evidence in APiCS also suggests that in terms of the feature values, the evolution of colour systems may develop from identity (value 2) via overlap (value 3) into differentiation (value 1). Nigerian Pidgin, for example, has moved from identity to overlap, probably under lexifier influence: “Formerly, the stative verb blak was used to refer to the spectrum from black to purple to blue to green, but now the separate stative verbs blu ‘be blue’ and grin ‘be blue/green’ are much more commonly used” (Faraclas 2013). Similarly, Fanakalo started out with the Zulu-derived luhlaza covering green and blue (identity), then added English-derived bluwan ‘blue’ (overlap) (Mesthrie 2013), and may at some point end up with luhlaza ‘green’ and bluwan ‘blue’ (differentiation). This last stage seems to have been reached by Principense, presumably via the same trajectory: “Modern Principense differentiates between ‘green’ and ‘blue’; however, only blue has a corresponding ideophone, which suggests that formerly, zulu [now ‘blue’] might have referred to ‘blue’ as well as to ‘green’” (Maurer 2013c). To what extent these developments are influenced by lexifier/adstrate pressure is a matter for future research.