The Proto-Tallic root for ’colour’ was √kwa. Signum was applied to it whereby the signum-positive vowel a denoted a bright colour, the signum-negative vowel ι a dark colour. Hence T. kawá originally stood for anything from yellow to white and red, while kιwá stood for anything from black to deep green and blue. This was later differentiated, so that kawá became ’red’ (cf. the cognates K. achva, H. hakua ’red’ < *akwa), kιwá became ’black’; and mulkwá was invented to mean ’colourless, white, pale’ (but not ’shining white’).
In fact, ’red’ seems to have just been a synonym for ’colour’ – compare also Spanish colorado ’coloured, red’, or the Slavic languages where ’red’ and ’beautiful’ are almost the same. ’Red’ was connected with the colour of ripe fruits (cf. kwásta ’fruit’) and a vigorous, lively state.
More colours entered the language through the sound-symbolic root √n-kh (see sound-symbolism: http://sindanoorie.net/glp/phonosymb.php) which described the growing of plants. By regular vowel gradation, *nekhe denoted a small plant, and *nokho a large one. The commonest derivatives were adverbial *enkhe, *onkho conveying the idea of a vigorous growth of small and large plants respectively.
In Talmit, however, these words shifted in meaning and were also applied to humans – the former describing a young child, the latter an old person. When applied to plants, *onkho then shifted from ’grown tree’ to ’bare tree in winter’. The idea was of course the analogy between a year’s cycle and a human lifetime.
The regular reflex of *enkhe was T. éχne with metathesis of *ŋχ, and came to mean ’grellow’ [sic], that is the range of colours between green and yellow (which is the colour of grass and leaves). It can still be used in the sense of ’leaves’ (collective) in phrases like kátu-mo éχne ’a tree’s green’. A very curious word is aϕálχne, meaning ’sunlight seen through the leaves of a tree’, a compound with afál ’state of brightness’.
Instead of expected *óχno one finds the slightly irregular form óχon which came to mean ’white’ by an association with aged hair and winter’s snow.
The root √kun originally denoted swampy ground or shallow water. It formed a pair with √ka denoting solid ground, so that the two often appeared together. Hence the dvandva compound kássekun ’ka and kun’ appearing in the phrase kássekun-nu prangánun lit. ’to cross ka and kun’ – meaning something like ’go at great lengths, achieve something through a lot of trouble’. √kun also survives in kúrmen ’large swamp without trees’ < *kun + augmentative *-men.
Both roots yielded colour words and interacted along the way. First, *-ŋwa ’-hued’ was appended to *ka and regularly yielded *kama ’brorange’ [sic], i.e. any shade between brown and orange (the colour of the soil and sand). The word *kun was often expanded to *kunu by analogy to *kama and came to mean ’grue’ [sic], i.e. any shade between green and blue (the colour of water). It remains with this meaning as kún(u) in Talmit. The nasal n, in its turn, influenced *kama to change to kána (also aided by compounds like kantwésta ’rowan’, lit. ’redberry’, kanθrébe ’beet’, lit. ’redheart’, kaχánta ’gooseberry, strawberry’; where m > n before t), making †káma an archaic variant. The latter remains fossilized in the poetic expression kam-nóimo bállin ’the earth-coloured moon’ (cf. Ancient Greek γαιοφανής).
In recent times, kána seems rather to shift to a more reddish, brick-coloured type of brown by the influence of kawá. It is the colour of foxes and copper – the idiom kanaχór ’red-brown skin’ describes a sneaky and insidious person. The ’brorange’ shade is steadily being replaced by the a-grade kéan. Finally, the initial consonant is often altered to h: hána, héan.
Vowel superposition was applied to colour words in Talmit more than anywhere else in the language2:
As an originally sound-symbolic word, *onkho had an intensive form by i-infixion: *onkhoi ’vigorous flourishing of a large plant’. However, since it left the class of sound-symbolic roots, the vowel i instead became associated with negative signum and the meaning ’black’. Although no separate form is recorded, the vowel-superposed form oχóini does appear; and means ’black and white’, metaphorically ’good and evil’ (as Jap. kokubyaku, shirokuro, kuroshiro).
The superposition of éχne ’grellow’ and kún(u) ’grue’ becomes ukéone, uχéone ’grullow = yellow∼green∼blue’, elliptically ’nature’.
Adding oχóini into the mix, one gets oiχáune ’multi-coloured’.
All these words are of course very artificial elements within the language, first constructed by poets, then taken over into the living speech for the lack of other terms. This is not unheard of, compare the invented pseudo-Latin word conundrum in English.
Summarizing, the colour terms in Talmit are (5 basic ones out of possible 6, see http://wals.info/chapter/133):
The obscure suffix *-ŋwa ’-hued’ becomes -wa after consonants and -ma after vowels in Talmit, and can be used to derive all sorts of additional non-basic hues and general apperances: tóχorwa ’snow-white’, tálzama ’blood-red’, kérkama ’wine-coloured’, bámnema ’sky-blue’, bíkwa ’glossy, looking like fat’ and so on. As an independent word amá < ə-má it has become the evidential particle of first-hand experience, not restricted to sight any longer. In Kymna, *-ŋwa appears in the word hengva ’blue’).
Regarding hair colour, kιwá is used for black, and kéan, héan for blond, brown and ginger. Regarding eye colour, kún(u) is used for blue, green and grey; and kéan, héan for brown.