Abstract
The theory developed in this study acknowledges two ways of licensing predicate-argument relations. Associated with every predicate is a domain, defined in terms of m-command, in which thematic roles can be assigned. Thematic interpretation is only possible, however, if guided by the presence of functional markers. This is expressed by the principle in (1) (cf. (5) in chapter 3).
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
The notion of external Θ-role assigner used here and below generalizes over transitive, intransitive and unaccusative VPs. The first two have a lexical external Θ-role; the external Θ-role of an unaccusative VP is motivated by null-operator movement, as argued in chapter 5.
The head-marking affix need not be attached to the main predicate. It can also be attached to an auxiliary in the extended projection of the lexical verb. One way to deal with this would be to say that the external Θ-role of VP may be transferred in its extended projection in case the functional head selecting VP is a verb that lacks an external Θ-role. Possibly, this process of transfer can be analyzed as a case of null-operator movement. In any case, the subject will be Θ-marked by a constituent consisting of the auxiliary and VP, so that it can still be interpreted through head marking. See Ackema 1999 for discussion of the thematic properties of auxiliaries. For ease of exposition, we will abstract away from these issues in the representations that follow.
If we refer to agreement in what follows, we refer to absolutive agreement (spelled out by the external affix). We will argue in section 3.2 that encapsulated affixes, such as ergative agreement, are not able to identify arguments in syntax.
If the thematic linking rules operating within VP associate the least marked Θ-role with the most economical way of marking, as assumed throughout this study, agreement with a possible indirect object is ruled out.
A different type of split ergativity is found in Icelandic, where VP-internal arguments can appear in the nominative in certain circumstances. We believe that the Icelandic head-marking system is not sensitive to aspectuality or phi features, but rather to the argument structure of the verb. More specifically, agreement must always be with the argument that bears the most prominent Θ-role in the verb’s Θ-grid. This will be the subject in transitive, unergative and monadic unaccusative constructions. If an unaccusative or passive verb takes two internal arguments, however, agreement must be with the Theme, even if the Goal is promoted to subject (Icelandic allows promotion of both Theme and Goal). Agreement with an VP-internal Theme leads to what is essentially an absolutive-ergative pattern.
It has been observed that in Arabic VSO order alternates with SVO order. Given that verbal agreement is richer in SVO order than it is when the subject follows the verb, one may think that SVO order is the result of head marking (as seems to be the upshot of Huybrechts 1991). Such an analysis cannot be maintained, however, since A’-fronting of the subject also leads to rich agreement (additional counterarguments are implied by the further discussion in the main text). The alternative analysis proposed by Fassi Fehri (1993) is that SVO order is the result of a general topicalization rule. Indeed, Arabic also allows topicalization of objects. The question that remains is how the distribution of rich agreement in Arabic can be captured. Following Benmamoun (1996), we propose that when the subject and the verb are in the same prosodic phrase a morpho-phonological rule applies by which agreement is reduced. Such a process can only take place when the subject follows the verb, since otherwise it is not in the same prosodic domain as the verb (only its trace is). A similar process of morpho-phonological adaption is found in certain dialects of Dutch (cf. Van Haeringen 1958). If the subject follows an agreeing head (be it a complementizer or a fronted verb) the form of the agreement affix is changed. This duplication of the Arabic pattern in Dutch suggests that morpho-phonological adaption may occur independently of a language being VSO.
Irish is exceptional in that infinitival verbs (so-called verbal nouns) may be preceded by objects. This seems to be due, however, to a particle that accompanies the verb. We speculate that we are dealing with a special instance of head marking, and hence that relevant data do not bear on the OV/VO nature of Irish.
Japanese also has a topic marker wa, which is superimposed on any topicalized DP. Thus, topicalized DPs are not marked for nominative, accusative, etc. (see example (52) below).
As opposed to Japanese, Arabic does not allow double nominatives. One may account for this contrast by manipulating the featural specification of Arabic and Japanese nominatives. An alternative would be to reduce it to the setting of the OV/VO parameter. As argued at length in chapter 2, the verb’s government domain is defined prosodically in VO languages, whereas in OV languages it is defined syntactically. It seems reasonable that in prosodic structures tense is only present where it is spelled out. Hence, the nominative will be licensed by traces in an OV language like Japanese, but not in a VO language like Arabic.
The fact that a wide scope reading of the object is marginally possible in (58a) may be accounted for by a process of quantifier raising. This process does not affect the reading of (58b), as verbs do not seem the undergo it.
At first sight, constructions of the type in (73) exist in pro-drop languages. How the option of pro drop opens up this possibility is briefly discussed in footnote 12.
If expletives are empty categories in pro-drop languages, this may account for the observation that the thematic subject can appear in postverbal position even if no surface subject in the sense of the specifier rule seems to be present (cf. Barbosa 1996 on the general availability of pro). Another account for this fact can be given if strong inflection is argumentai in nature. Leftward verb movement then creates a configuration in which VP can take I as its antecedent, thus satisfying the dependency condition. For a further analysis of expletive constructions along the lines sketched here, see Koeneman & Neeleman 1998.
It is sometimes argued that it in examples like It seems to be impossible to leave this party is an expletive as well. If so, it could be analyzed along the lines of this section. We believe, however, that an analysis along the lines of Bennis 1986 is more plausible. Bennis argues that it is always a full-fledged argument, and that the clause that appears to be a complement in the construction at hand is really an adjunct semantically related to the subject pronoun. Further motivation for an analysis along these lines can be found in Wiltschko 1995.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Neeleman, A., Weerman, F. (1999). Head Marking. In: Flexible Syntax. Studies in Natural Language and Linguistic Theory, vol 47. Springer, Dordrecht. https://doi.org/10.1007/978-94-011-4289-2_6
Download citation
DOI: https://doi.org/10.1007/978-94-011-4289-2_6
Publisher Name: Springer, Dordrecht
Print ISBN: 978-0-7923-7199-1
Online ISBN: 978-94-011-4289-2
eBook Packages: Springer Book Archive