OF NOMINATIVES AND DATIVES 467sional predicates of sound-classes ("features") in terms of which we specify
the phonetics of syllables, must be translatable ultimately into a form com
patible with other extensional languages of science, e.g., that of physics or
physiological psychology. Notice that it does not say that such an extension
is a sufficient characterization of sound systems, only that it is a necessary
characterization of them. The other links in the set of constructs leading to
constraints on possible phonological systems supply the sufficient charac
teristics, and these include number of segments of such-and-such types,
rules of distribution (surface, underlying, or both), etc.
It is not my purpose here to outline modern theories of phonology in
any greater detail. It is merely the purpose to motivate the general conclu
sion, that it is only by constraining systems with formal-functional univer
sal, that the two halves of a theory of some realm of language are mutually
established. There is no such possibility as establishing the one without
assumptions about the other; there is no such thing as a formal universal
without a substantive assumption (i.e., universal also implied); there is no
such thing as a functional or substantive universal without assumptions
about the forms that have these functions or substance. In the realm of
phonology and phonetics, the linkage of these two domains is demonstrated
by the interdependence of the theoretical apparatus necessary: formalisms
of, say, process phonology whose rules "conspire," of course, to produce
certain surface distributions (there is no surprise in this), and substance of,
say, distinctive-feature characterization of sound types which describe cer
tain surface distributions. This interdependence is demonstrated in such
concepts as "markedness," the asymmetric fitness of certain phonetically-
characterizable sound types to serve in certain distributions relative to other
sound types; "hierarchy of features," the strict ordering of intensional
descriptive one-place predicates that define classes of possible sounds in
any particular language, some more inclusive than others as a result; "(in
trinsic and extrinsic) descriptive order" of the apparent rules needed to
state distributions of phonemic segments, in morphophonological alterna
tion conditions particularly; etc.
The apparatus derives its validity from the principle of formal-func
tional linkage. And it has been the experience of phonologists that the
single formal-functional hypothesis has sufficed to bring a great deal of
order out of an apparent diversity of hopeless proportions, even though
limited to referential meaning. It is an unspoken expectation that these uni
versal patterns obtain in any particular language, all other things being