No language has the term *'reen' meaning ‘red and/or green, but nothing in between’, or the term *'yellue' meaning ‘yellow and/or blue, but nothing in between’. There would seem to be no a priori reason for the absence of these terms, for it is easy to construct not only meanings but also possible neurological substrates for them. Thus, *'reen' would be the representation of the activity in the red-green receptors (and no others), while *'yellue' would be the representation of the activity in the blue-yellow receptors (and no others). However, we know that lexicalization does not simply represent outputs of particular neuronal sets, for two reasons. First, many languages have a term equivalent to 'grue' ‘green and/or blue’, which represent partial outputs of two opponent sets, rather than the full output of one opponent set. Second, the factor that allows grue to exist, while blocking *'reen', seems to be perceived spatial contiguity, which of course corresponds to wavelength contiguity. Green and blue are contiguous on the spectrum; red and green, or yellow and blue, are not.
- Bickerton (1981), Pag. 240

[…] as lexicalization progressively dissects semantic areas, the meanings of the earlier lexical items must change: “black,” which originally embraces half the spectrum, must gradually reduce in scope until it eventually occupies only a narrow band of it. […] Thus, the semantic range of terms in subsystems is determined by the number of terms in such subsystems and by the semantic ranges of the other terms.
- Bickerton (1981), Pag. 243