(^482) MICHAEL SILVERSTEIN
As we move up the hierarchy of linkage types in Figure 6, there seems
to be greater and greater formal distinctness of the surface structure
remnants of what could be reconstructed as full "kernel" clauses for the
linked (or dependent) member of the pair under consideration. At one
extreme are possessive NPs, indicating the possessor of some Noun (or
Nominal) in the independent (or relatively more independent) clause; at
the other end are clauses indistinguishable from their unlinked forms. In
between is a whole range of formal de-formations of erstwhile independent
clause surface structure, including such things as reducing the tense-aspect-
mode characteristics of predicates (verbs in particular), forming infinitives,
participles, nouns of action and agency (which are really whole clauses in
many languages, complete with their NP arguments in proper case-marked
form), and so forth. In terms of case-marking systems, it is always the case
that the following kinds of regularities hold: the NP adjuncts of nominalised
constructions never have a greater case-elaboration than the non-
nominalised ones, and generally have far fewer — in fact, the tightest-
linked nominalisations have only genitive adjuncts, as is the case with head
less possessives (e.g., English mine); in all languages, the Agentive hierar
chy, which, if you recall, permitted the appearance of ergative surface case-
markings, is suspended at some point along the hierarchy — and for all
structures higher up — so no ergative case-forms appear in such linked
structures, though accusative ones may, since the Patientive hierarchy is not
suspended at the same point; further, in all languages, there is a reduction
in the elaborateness with which various case-marking distinctions of such
things as direct vs. indirect object are made, in fact consistent across all the
types of plain inflectional systems; etc. So, as it were, case-marking is "tele
scoped" into a reduced or even minimal system, gradually in some lan
guages, more abruptly in others, as we climb the hierarchy of linkage types.
The point is, we cannot interpret the reading for a particular case-
marking relative to a system of case-marking oppositions, without knowing
the placement of the relevant clause in a structure of discourse linkage: as
a coding category of grammar, the genitive in an independent, non-linked
"kernel" clause-sentence is, as an indicator about any underlying proposi-
tional role, something different from the genitive in a tightly-linked noun of
agency construction.
Finally, continuing with the notion of linkage, one of the types of link
age of clauses (or cohesion-devices of discourse) in both multi-sentence
and single-sentence structures, is reference-maintenance. Again here, Ian-
singke
(singke)
#1