The Morphosyntactic Profile of Wh-Agreement
In languages such as Chamorro, Hausa, Irish, Kikuyu, Moore, and Palauan, wh-constructions exhibit a special type of morphosyntactic flagging, which has come to be known as wh-agreement (see Haı¨k 1990; Watanabe 1996; Chung 1998; and the references cited there). This agreement phenomenon has the following typological profile:
It occurs only in the classic wh-constructions, such as constituent questions, relative clauses, and focus constructions, which we assume, following ‘‘On Wh-Movement’’ (Chomsky 1977), are derived by wh movement. In some of these constructions, Chomsky’s diagnostics reveal that wh-movement has occurred even though there seems to be no overt element at all that instantiates the wh-phrase. We take no firm position on whether such constructions involve a deleted wh-phrase (as Chomsky originally proposed) or a null operator (Browning 1987). For convenience, we will sometimes refer to the moved element in such constructions as a null operator.
B. The morphological effects of wh-agreement surface on V or on T, C (see Zaenen 1983), or on some other designated head in the functional layer of the clause, such as Focus (see Green 1997). C. In general, wh-agreement is not sensitive to person, number, or gender (f-features).
D. The agreement explicitly registers the presence of a moved wh-phrase. In some languages, such as Hausa (Tuller 1986) and Irish (McCloskey 1979, 1990, 2001), this is all that wh-agreement registers; in other languages, such as Palauan (Georgopoulos 1985, 1991), it registers additional features of the moved wh-phrase, such as category type or Case. E. In long-distance constructions, in which wh-movement crosses a clause boundary, wh-agreement comes in two flavors. Either (i) the agreement is recursive, in the sense that it surfaces on every designated head on the path of wh-movement; or else (ii) the agreement is nonrecursive, meaning that it surfaces only on the highest designated head on the path of wh movement. As far as we know, no other patterns are attested.
We illustrate this typological profile via the two types of wh-agreement found in Chamorro, an Austronesian language of the Mariana Islands. Chamorro has a standard inventory of wh-constructions, including constituent questions, relative clauses, focus constructions, and comparatives, which display all the earmarks of constructions derived via wh-movement. The language also has not just one but two types of wh agreement. The first type of agreement, called ‘‘wh-agreement’’ by Chung (1998), surfaces on V (or A) and registers the Case of the moved wh phrase. Compare (1), which is a simple declarative clause, with the constituent questions in (2a) and (2b). In (2a), the nominative Case of the wh-subject is flagged by the wh-agreement morpheme-um-; in (2b), the objective Case of the wh-object is flagged by the morpheme-in- plus pos sessor agreement. These morphemes are given in italics below.
(1)
(2)
The same agreement pattern reappears in the context of relative clauses, which are externally headed and can either precede or follow the head NP. We consider here only postnominal relative clauses. Notice that in these and all other relative clauses, there is no overt relative pronoun dis tinct from the head. For simplicity’s sake, we will refer to the moved element in such constructions as a null operator, although other analyses are possible.
(3)
The second type of wh-agreement is called ‘‘operator-C agreement’’ by Chung (1994, 1998). This agreement surfaces on C and registers the cate gory type of the moved wh-phrase, as well as its operator feature (i.e., the feature that identifies it as an interrogative phrase, a relativized DP, or a focus). The realizations of ‘‘operator-C agreement’’ are most transparent when the wh-phrase is a DP denoting a location in time or space, as in the examples below. In the question (4a), which is formed on the locative phrase manu ‘where?’, the agreement is realized as the complementizer na (in the Guam dialect), ni,ornai (in the Saipan dialect). In the focus construction (4b), in which the focus is the locative gi gima’mami ‘in our house’, the agreement has the same realization.
(4)
But in the relative clause (5), the null operator corresponding to the relativized DP is a locative, and the agreement is realized as the complementizer a¨nai.
(5)
We turn now to the patterning of these types of Chamorro wh agreement in long-distance constructions. Recall from point E above that wh-agreement comes in two flavors: either it surfaces (i) on every relevant head on the path of wh-movement, or it surfaces (ii) only on the highest relevant head. The agreement that Chung called ‘‘wh-agreement’’ comes in flavor (i): it appears on every V on the path of wh-movement, as can be seen from the italicized verbal morphology in the long-distance constructions in (6) and (7).
(6)
(7)
But ‘‘operator-C agreement’’ comes in flavor (ii): it appears only on the highest C on the path of wh-movement, as can be seen from the italicized complementizer alternations in (8) and (9).
(8)
(9)
(10)
‘Where is the house where they told the priest that he should live?’
The agreement patterns just illustrated are not limited to constituent questions and relative clauses, but instead occur in the full range of Cha morro wh-constructions, exactly as expected if both types of agreement track the history of wh-movement. At the same time, their different distribution in long-distance constructions serves to reinforce our point that there can be parametric variation in the way the wh-agreement phenome non is realized. This variation can occur not just across languages, but— strikingly—within a single language, as Chamorro shows quite clearly. We will have occasion to return to this point later.
