Abstract
So far, we have looked at the processes that are related to A-movement. Case checking and Case absorption, in particular, have been our primary concerns. In this chapter, we will turn to discuss interactions between Case checking and A-bar movement. There is extensive research reported in the literature on so-called Wh-Agreement. Wh-agreement mainly refers to various morphological modifications having to do with verbs and complementizers that are found in clauses in which wh-movement takes place. Our key idea is that these various kinds of morphological change are reflections of an abstract checking relation that links Case processes and A-bar processes. This checking relation is motivated by what we have been advocating in the preceding chapters as the follow-up to Case checking.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
Chamorro also displays a similar pattern, as described by Chung (1982, 1994) and Chung and Georgopoulos (1988), but we will use Palauan for illustration, since Chamorro is a little more complicated. We will come back to certain aspects of Chamorro in section 4.4.2.3.
The preposed wh-phrase is prefixed by the cleft-marker ng- According to Georgopoulos (1991a, note 17; 1991b, Ch. 3), this morpheme is the same as the third person singular subject agreement.
The mood distinction in Palauan is the following: realis in declaratives and yes/no questions; irrealis in negations, conditionals, commands, and some adverbials. See Georgopoulos (199 lb, 27–28) for some illustrative examples. Georgopoulos (1985, note 19; 199 lb, 89–90) notes that the irrealis context retains irrealis morphology even if the subject is extracted. The opposite situation holds in yes/no questions, according to Georgopoulos (199 lb, 156–159), where realis morphology is retained even when a non-subject is extracted.
Georgopoulos (1985; 1991b) argues that Palauan uses only the resumptive pronoun strategy. But wh-movement is an option in UG which is available as a null option. Thus, there is no reason to believe that this option is prohibited unless there is strong evidence. Georgopoulos (1985; 1991b) does not present such evidence. We will therefore assume that movement takes place when no island intervenes.
There is one complication to (4.6b): when an element is extracted from a sentential subject, the matrix verb shows the pattern of local subject extraction. Consider (i).
See Abe (1995) and Collins (1994, 1995) for the checking relation between C° and an intermediate trace in its Spec.
Chemy (1993) has an account which is closer to ours. She claims that the landing site of wh-movement is Spec of MP, which lies between CP and Agr-sP. The wh-moved item is licensed by Spec-head relation at MP or at Agr-sP, which corresponds to non-subject extraction and subject extraction, respectively. Chemy also notes the similarity between Palauan wh-agreement and English cb-support, to which we will turn below. Thus, Cherry’s (1993) account can be taken as a direct precursor of our analysis. The crucial step for us, however, is the discovery that Vto-I-to-C movement which is motivated by Case theory also provides an answer to wh-agreement phenomena. We do not set up a separate phrase MP. See below for discussion.
Cherry (1993) claims without evidence that subject extraction in Palauan lacks a step of verb raising which takes place in non-subject extraction. We do not know, however, whether this assertion is factually supported. The task is made difficult by the verb-initial nature of Palauan.
See the references cited there for earlier proposals on the relation between rich verbal morphology and verb movement. They were formulated without distinguishing among the three inflectional heads (Agr-s, Tns, and Agr-o).
This example shows, in addition, that wh-agreement does not take place in the lower clause. This point will be taken up in section 4.2.2. The same is true with (4.19c) below.
This example shows, incidentally, that wh-in-situ does not trigger whagreement.
Compl in the gloss is completive aspect, and cont is continuous aspect.
The examples that contain a modal such as (i) already satisfy the modalization requirement.
We differ from Chomsky (1993) in assuming that the inability of have and be to undergo LF movement is due to their morphological peculiarity rather than their semantic vacuousness. Thus, the fact that the Mainland Scandinavian counterparts of have and be can remain within VP as in (i) is not problematic under our proposal.
See Bobaljik (1994, 1995), Halle and Marantz (1993) and Lasnik (1994) for other recent approaches to the question of dg-support, according to which do-support is essentially a PF process related to adjacency.
The problem about negation is not that damaging to Chomsky’s (1991) claim. Even though there is no counterpart of d7-support applying in Mainland Scandinavian negative sentences, this may be due to the non-head status of the negation marker inte in Mainland Scandinavian.
It is not obvious whether the appearance of cb in VP ellipsis and VP fronting should be treated in the same manner. One indication that it should not comes from the fact that VP fronting in Mainland Scandinavian involves insertion of a dummy verb, despite the absence of the counterpart of cb-support in negation and interrogative.
Haik uses the information supplied by Tuller (1986).
An account of the that-trace effect and related phenomena which is similar to ours is proposed by Branigan (1994).
The points to be made here apply to Lasnik and Saito’s (1992) account as well.
This is Rizzi’s version of the ECP.
Rizzi himself is fully aware of this point, judging from the discussion in Rizzi (1990, 52). See also Frampton (1991) for a critical discussion of Rizzi’s account.
Modem Hebrew, discussed by Shlonsky (1988), provides a similar paradigm.
Note, however, that English contrasts with Moore and the relevant dialect of Hausa in that intermediate traces of subject extraction require morphological realization of wh-agreement distinct from non-extraction cases in English. In Moore and the relevant dialect of Hausa, only the head of a wh-chain requires morphological realization distinct from non-extraction.
Frampton (1991) distinguishes between subject-oriented agreement and specifier-oriented agreement. In the former, local subject extraction shows a distinct morphology from extraction of other elements, extraction of non-subjects usually using the same complementizer morphology as non-extraction cases. Specifier-oriented agreement does not distinguish between subject extraction and non-subject extraction. Although Frampton limits his discussion to the Comp morphology, the typological distinction between subject-oriented agreement and specifier-oriented agreement holds for verbal morphology, too, as we have seen above.
See McCloskey (1979, 1990) for wh-agreement in Irish. Cf. also Noonan (1992).
This leads to ambiguity in certain cases. McCloskey (1977, 1979) notes that examples like (i) are ambiguous.
We may add weak pronouns in the Germanic languages to our list. The relevant fact is that weak subject pronouns, but not weak object pronouns, can appear sentence-initially in German and Dutch. Travis (1984, 1991a) and Zwart (1993a,b) take this fact to indicate absence of movement to Spec of CP in subject-initial sentences. Alternatively, we can interpret this asymmetry as due to the differencein C°’s ability to license weak forms, depending on whether the local subject is extracted or not, since the subject/non-subject asymmetry is characteristic of wh-agreement. We will shortly see positive motivation in the next section to
The complementizer agreement of West Flemish (Bennis and Haegeman 1984, Haegeman 1992) and other Dutch dialects, discussed in depth by Zwart (1993a,b), also displays the subject/non-subject asymmetry. This phenomenon should be included as a species of wh-agreement.
The material discussed in this section is taken from Watanabe (1994b).
The inversion asymmetry in English will be taken up later.
See Watanabe (1994b) for why only movement parameters are constrained by Degree-0 Learnability.
The developmental sequence for children learning German is consistent with this idea. See Poeppel and Wexler (1993) for a recent discussion.
Déprez (1990) also notes this problem and suggests that Agr° can be raised to C° while T° cannot, assuming that Agr-sP dominates TP in Romance while TP dominates Agr-sP in Germanic. This kind of parametric variation, however, is not allowed in the Minimalist syntax.
In Italian, indicative embedded questions require inversion, while subjunctive embedded questions do not. See Watanabe (1994b) for discussion of this contrast.
In view of the fact that some Spanish verbs allow the sequence complementizerwh phrase in their complements (Rivero 1994b and Suner 1993), one might be tempted to say that this inversion takes place below the CP level. Inversion takes place, however, even in the complements to the verbs which do not allow the sequence complementizer-wh phrase. Saber ‘know’, used in the text examples, is such a verb, as can be seen from (i).
Suner (1994, 364) judges similar examples in (i) as acceptable.
Torrego (1984) also suggested this possibility, but left the question open.
English is an exception. We will deal with it in the next section.
We abstract away from embedded Topicalization. See Watanabe (1994b) for discussion.
Pace Travis (1984, 1991a) and Zwart (1993a, b), who claim that subject initial sentences do not involve verb raising to C°. See Vikner and Schwartz (1991) for arguments against this type of analysis. Cf. also Vikner (1994) and Thrßinsson (1994).
Absence of specification for any operator feature in declarative clauses indicates that no movement takes place even at LF.
Since we are assuming checking theory, this property of Agr-s is also reflected on the verb. Absence of an agreement marker on the verb should be understood in this way.
Suffer (1992) claims that specificity is the factor that determines the appearance of a subject clitic: the subject clitic is a marker of specificity of the subject, regardless of its syntactic position. She does not give a satisfactory answer, however, to the question why the subject clitic disappears even for the postverbal subject which is a name such as Mario.
Here, there are two possibilities: either PF crashes or LF crashes. Chomsky (1993) adopts the former possibility, while Chomsky (1994) takes the latter. The choice is immaterial for the present concern.
The material in this section is drawn from Watanabe (1994c, to appear). See also Collins and Branigan (1995) for the analysis of Quotative Inversion, which displays essentially the same property as French Stylistic Inversion and Japanese Nominative-Genitive Conversion.
Pronominal objects shift even in those languages which do not allow a full DP object to shift. The analysis of Romance elides proposed in section 3.2.2.2 of the previous chapter provides a solution here. Pronominal object shift is allowed even when Spec of TP is unavailable in overt syntax, because these pronominal objects adjoin to Agr-oP.
Kayne (1972, note 6) and Kayne and Pollock (1978, note 12) mention in passing the impossibility of Stylistic Inversion in the presence of a direct object. See also Kayne (1979, 717).
Exceptions are object idioms, as in (i).
Collins and Branigan (1995) propose an account similar in spirit but based on a somewhat different defmition of domains.
Valois and Dupuis (1992) claim that cliticization of the direct object does not save the structure. Kayne (1972, 106) gives a marginal status to the examples in (i). We have no account of the degraded status in case of cliticization.
Ura (1993) claims that there is no such choice for substitution. This makes a distinction between adjunction and substitution that Chom sky (1994) tries to capture.
Viviane Déprez and Dominique Sportiche (personal communications) observe that Stylistic Inversion disallows participle agreement.
The expletive pro hypothesis was proposed by Brandi and Cordin (1989) and Rizzi (1986a) for the northern Italian dialects discussed above, too. This hypothesis runs into a non-trivial problem in the face of the fact, noted by Safir (1985), that failure of subject raising is disallowed in non-subject extraction. In the Trentino example (i), the presence of a subject clitic is obligatory, indicating that the subject must be raised to Spec of Agr-sP.
Spec of TP should be available in overt syntax in this language, since the postverbal subject option is possible in a transitive clause.
Why the pairing of the complementizer shape and the subject position is not the other way round (i.e., the presence of the overt complementizer forcing the preverbal subject extraction) is the question that we are not ready to answer. The pattern in Bani-Hassan Arabic is the same as the classical that-trace effect.
See the references cited in Watanabe (1994c, to appear) for previous approaches to the phenomenon. Let us emphasize here that the transitivity restriction was never treated adequately in the past literature.
Use of lower case letters for nominative and genitive here is deliberate. See below in the text.
Hiroyuki Um (personal communication) suggests that the genitive of negation in Russian, discussed in Pesetsky (1982), may be analyzed as failure of raising in overt syntax, this time, to Spec of Agr-oP. Given the scope sensitivity displayed by the Russian phenomenon, this is plausible.
We are abstracting away from the interaction with the double object construction. See Gibson (1980).
The distribution of agreement in Chamorro is as follows (from Chung (1982, 44)):
We will also abstract away from the referentiality factor discussed in Chung (1994).
This is restricted to definite DPs: oblique indefinites are not marked.
The same is true with French examples, too.
Oblique phrases in Chamorro are subject to other interpretations as well. The important point is that construal as a direct internal argument is among the possibilities allowed.
Postal (1977) discusses antipassive in French, but these cases crucially do not involve oblique marking of internal arguments. French thus lacks the oblique insertion startegy, too.
The data come from the Ixtahuac5n dialect of Mam.
The form here abstracts away from the effects of morphophonemic rules.
See Smith-Stark (1978) for the earliest discussion. See also Cooreman (1994).
Ordinary antipassive in K’ekchi is illustrated in (i).
Ayres (1983) notes that there are some morphological processes that group focus antipassive and ordinary antipassive together. For us, however, agreement is the most important property that decides the structural Case marking status of DP arguments.
In focus antipassive of K’iche’, at least one of the arguments must be 3rd person. There is no way, therefore, to tell whether 1st person is higher than 2nd person.
According to the survey by Dayley (1981), we may add Pocomam, Tzutujil, Aguacatec, Jacaltec, Tzotzil, and Yucatec to the list in (4.89ii). In Tzutujil (Dayley 1981, 1985), the oblique strategy is also possible, conditioned by the person
We will shortly turn to the question why this morphology does not appear in extraction of intransitive subjects.
Later, we come back to the question why adjunction to Agr-sP is not an option in this derivation.
In K’ekchi, we have to assume that oblique marking of direct internal arguments is used only for focus antipassive. Remember that ordinary antipassive in this language does not involve oblique marking of direct internal arguments.
On the other hand, it is not clear how to accommodate our account in the recent theories of Ergativity by Bittner (1994), Bittner and Hale (1994), Bok-Bennema (1991), Campana (1992), Johns (1992), Murasugi (1992).
In ordinary clauses, both the subject and the object are crossreferenced by agreement.
See also Koizumi (1994), Takahashi (1994a), and Ura (1994b) for similar ideas.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1996 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Watanabe, A. (1996). WH-Agreement. In: Case Absorption and WH-Agreement. Studies in Natural Language and Linguistic Theory, vol 37. Springer, Dordrecht. https://doi.org/10.1007/978-94-015-8615-3_4
Download citation
DOI: https://doi.org/10.1007/978-94-015-8615-3_4
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4744-1
Online ISBN: 978-94-015-8615-3
eBook Packages: Springer Book Archive