from the nouns they modify. Semantic relations are regulated by an exuberant case system; modifiers are connected
with their heads by a shared case. Subordinate clauses are kept together, but within the morder is also totally free. Of
course, the rich morphological system requires use of the morphosyntax component. (Stephen Anderson has
suggested (p.c.) that Classical Latin too mayfitthis characterization.)
As mentioned earlier, even in the usual cases when phrase structure is demonstrably present, there is wide variation
among languages in how rich a use is made of phrase order and how rich the morphology is.
Turning to grammatical functions, Van Valin and LaPolla (1997) argue that Acehnese (an Austronesian language from
Sumatra) does not make use of grammatical functions. Recall that the argumentforgrammatical functions is that there
are grammatical principles that refer to subject and/or object independent of their semantic roles. Van Valin and
LaPolla show that allanalogous grammatical principles inAcehnese actuallyare dependentonthesemanticroles Actor
and Undergoer (Patient),so there is no justification for syntacticallyindependent subjectand objectroles. They mount
a similar argument for Mandarin Chinese, where the relevantsemantic roles are Topic and Comment. They also argue
that in intransitive sentences, Warlpiri acts like it has grammatical subjects, but in transitive sentences, the relevant
notion is Actor.
In English, of course, grammatical functionsplay a role with respect to the arguments of verbs. But for the arguments
of nouns the issue is not so clear. For instance, there is something that looks rather like a passive in noun phrases.
Parallel to the active and passive sentences (7a) and (7b) are the“active”and“passive”noun phrases (7c) and (7d).
(7) a. The enemy destroyed the city.
b. The city was destroyed by the enemy.
c. the enemy's destruction of the city
d. the city's destruction by the enemyBut such“passive”noun phrases are highly quirky and subject to ill-understood semantic factors (see Grimshaw
1990). For instance, the paradigms in (8) and (9) admit a verbal passive but not a nominal“passive.”
(8) a. John observed Bill.
b. Bill was observed by John.
c. John's observation of Bill
d. *Bill's observation by John
(9) a. John knew the answer.
b. The answer was known by John.