Advances in Role and Reference Grammar

A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 17

This contrast has important syntactic consequences. Because in the
head-marking pattern the head bears morphemes which indicate its gov­
erned dependents, the dependents can be omitted without affecting the
grammaticality of the phrasal unit; the head alone can count as the whole
unit. Thus in Tzotzil the NPs in (4a) can be freely omitted, and the result is
a clause composed of only the verb with its argument-indexing affixes. This
is illustrated in (8).

(8) a. ?i-0-s-pet lokel ?antz ti tul-e (=4a)

ASP-3ABs-3ERG-carry away woman DEF rabbit-DEF

"The rabbit carried away the woman."

b. Ίί-0-s-pet

ASP-3ABS-3ERG-Carry

"He/she carried him/her/it."

 L-i-s-pet-otik

ASP-lABS-3ERG-Carry-lPL INCL

"He/she carried us (inclusive)."

The same pattern holds for other types of phrases. For example, in posses­
sive constructions, the signal of possession is on the possessed NP, the
head, rather than on the possessor, the dependent, e.g. Tzotzil s-malal li
?antz-e (3ERG-husband the woman-DEF) "the woman's husband" [lit. "herr
husband the woman^"], and here again the dependent can be omitted, yield­
ing a phrase of the same type composed only of the head, e.g. s-malal "her
husband". Thus head-marking constructions are endocentric (in the sense
of Bloomfield 1933). The same cannot be said of the dependent-marking
pattern. In English, a finite verb alone cannot constitute a clause {""carried),
and if the possessor is dropped from a possessive construction, e.g. the
woman's husband —» (the) husband, the result is grammatical but is no
longer a possessive construction, unlike its Tzotzil counterpart. These
dependent-marking constructions are thus exocentric, in Bloomfield's
sense.
This contrast has important implications for syntactic theory, as all
generative theories are based primarily, if not exclusively, on the analysis of
dependent-marking languages. Since it underlies significant differences in
what can count as a possible phrase or clause in the two language types, any
descriptively adequate theory of clause structure must be able to capture it.
In Van Valin (1985, 1987a) it was argued that with respect to clauses in
head-marking languages, the pronominal affixes on the verb are the core