Advances in Role and Reference Grammar

528 ROBERT D. VAN VALIN, JR. & DAVID P. WILKINS

4. Conclusion

Several observations can be made concerning the above analysis of English

remember and MpA irlpangke- and itelare- and the method for predicting

complement types. First, it demonstrates the difference in the semantic and

syntactic organization of these roughly correspondent cognition predicates

in the two languages and illustrates the need for comparative analyses of

this sort. Second, we were able to present semantic decompositions for the

relevant verbs which explicitly demonstrate the similarities and differences

between them (cf. Table 1). Where relevant, we showed how one lan­

guage's semantic and morphosyntactic rules, such as the ability of activity

predicates in MpA to take purposive core adjuncts, or pragmatic rules,

such as the use of imperative with itelare- to give a psych-action remember

sense, can generate interpretations which are lexically encoded in the other

language. Third, this example shows that the same types of rules can be for­

mulated, and justified, to associate the semantic representation of predi­

cates unambiguously with the complements types that they take in each lan­

guage. Fourth, the IRH affords an explicit and theoretically motivated basis

for cross-linguistic syntactic comparisons: English that-clauses and MpA

-r/e-clauses can be considered comparable, because they instantiate the

same syntactic juncture-nexus type, core subordination, and the same

interclausal semantic relation, cognition/propositional attitude.

As the discussion in Sect. 1 showed, there have been virtually no seri­

ous attempts to develop a fully explicit, rigorous theory which predicts syn­

tactic structure from semantic representations. This paper is the beginning

of just such an attempt, and therein lies its primary theoretical significance.

It provides a concrete example of how syntax and semantics can be related

in a principled and general way, and it shows how many of the phenomena

which are today considered to be ad hoc and arbitrary by many syntactic

theories (and therefore simply listed (stipulated) in lexical entries) can be

remember BECOME think.again (x) about.something.be.in.mind.from.before (y)

irlpangke- have.in.mind.again (x) something.x.knows.be.in.mind.from.before(y)

itelare- think (x) about.something.x.knows.be.in.mind (y)

remind CAUSE BECOME think.again (x) about.something.be.in.mind.from.before (y)

irlpangke-lhile- CAUSE have.in.mind.again (x) something.x.knows.be.in.mind.from.before (y)

Table 1 : Comparison of Verb Decompositions