470 MICHAEL SILVERSTEIN
1. Léxico- and grammaticosemantic categories
So the "relatively restricted and transparent subject matter" of the domain
of phonology is really a result of the inspired formal-functional hypotheses
at first implicitly, later explicitly enunciated by all good phonological prac
titioners. It is the organizing power of a simple, unifunctional assumption
about the ultimate articulation of speech into an integral (whole) number of
syllables, that leads to the successes in explanation of what languages are up
to in this realm.
It would appear that at the other end of language, where we are con
cerned with constraints upon subsystems of the second plane of the double
articulation of language, i.e., those of coding "meaning," we must be no
less rigorous in exposing fallacious attributions and assertions about so-
called universals of a strictly formal, or a strictly substantive type.
To take a very simple example, one that has captivated cognitive
anthropologists, consider the phenomenon of coding "color" in language.
Here, Berlin & Kay (1969; cf. also Kay & McDaniel 1978) have made a
most reasonable assertion about lexicalization properties of languages,
based on what can only be called the happenstance of an inspired guess of
researchers whose own language is fortunately a maximal system from this
point of view. The universal of color coding must be expressed as a relation
ship between monolexemic coding of phenomena of certain sort and those
visual phonomena which can be described with the extensional language of
the psychophysics of hue. The original researchers had little idea of what
they were postulating, let alone little idea of the nature of the data they
were using, but subsequent reworking has clarified the situation somewhat.
The universal order of monolexemic coding of hue, then, states as follows:
if a language has N monolexemic terms denoting maximally bright hues,
then, for N—2,...,11, for each number we can specify a focal hue that is
specifically and differentially coded by each of the N terms. Furthermore,
the focal hue for the jth item is locatable in color-space with respect to the
kth item, for each j,k < N < 11. Thus, two term systems will always dif
ferentiate a "white" from a "black"; three-term systems will further dif
ferentiate the "black" area into "black" and "red"; four-term systems will
further differentiate the "white" area as well into "white" and "yellow";
and so on, through the last specified "pink", "brown" and "gray", as shown
in Figure 1. (Kay & McDaniel [1978:639] revise this somewhat). Observe
here that while in many languages (and the societies where they are spo-