• Keine Ergebnisse gefunden

The theoretical background of Voice morphology

2.3 The morphology of transitivity alternations

2.3.3 The theoretical background of Voice morphology

Within the Government and Binding framework, the standard analysis of the passive is based around three principles (Baker et al., 1989; Embick, 1997):

1. the passive morphology absorbs the accusative case 2. the passive morphology absorbs the external theta-role

2.3 The morphology of transitivity alternations 31 3. the internal argument that functions as the grammatical subject moves to the subject

position and gets nominative case

This analysis is captured in (15):

(15) [T/I PThe articleiwas [V Pwritten tiby Eleni]].

In Chomsky’s (1981) Government-Binding analysis (see also Jaeggli 1986), verbal pas-sives are derived through the movement of the NP to subject position and through the absorption of the accusative case. The passive morpheme absorbs the theta-role of the external argument. As Demuth puts it, "the formation of verbal passives critically involves movement of the NP from [NP, VP] position (as the object of the verb) to [NP, S] position (as the subject of the sentence), where it then receives nominative case. When movement of the NP occurs, a coindexed traceeis left behind, thus constructing an argument-chain (A-chain)" (Demuth, 1989, p. 57). This analysis is illustrated in (16a) and (16b).

(16) a. Lena tore the doll

b. The dolliwas tornei(by Lena) (17) The doll was torn/white (*by Lena)

With regard to adjectival passives, the most prominent analysis is lexical and not syntactic (see Bresnan (1982); Wasow (1977); Williams (1981) for more information). In other words, adjectival passives do not have a thematic subject (they also do not generally take by-phrases), the accusative case is eliminated (not absorbed like in the case of verbal passives), and the theta-role of the NP is externalised (moves to subject position). Unlike the verbal passive in (16b), the adjectival passive in (17) has no coindexed A-chain (which means that no movement takes place) and does not allow a by-phrase. In other words, torn functions just like any other adjective (Demuth, 1989, p. 58).

Kratzer (1996) in her seminal paper developed a neo-Davidsonian association of external arguments in the syntax. Based on this account, external arguments are not

arguments of the verb; they are introduced by a special VoiceP. Kratzer does not delineate the semantics for the passive Voice head, but her discussions suggest that it is the same as that of the active Voice head. What distinguishes the two is the fact that active Voice typically takes an overt NP as its specifier whereas passive Voice is existentially bound.

The presence of Voice in the syntax of passives is substantiated through languages such as Greek which employs Non-active morphology to express the passive (Rivero, 1990):

(18) [VoiceP Non-active [VP ]]

This analysis holds for Greek reflexives as well according to Alexiadou and Schäfer (2012) and Embick (1997, 2004) among others. These authors support that the morphology of Voice, and whether or not it will be realised as Nact, depends on whether there is no overt external argument that is realised as a Nact Voice head.

However, Chierchia (2004); Reinhart and Siloni (2004); Tsimpli (2006), propose a different analysis: they treat reflexives separately and have been pursuing an unergative analysis.

(19) [v/VoiceP DP [v/Voice <1> v/ Voice [VP V<2> ]]]

According to this analysis, the single argument is generated as a deep subject through a de-transitivisation process that eliminates the internal argument. According to the analysis in (5), v has an agentive feature which is attracted by the DP subject in the specifier of vP and is thus the true external argument. The remaining theta- feature is attracted by Voice. Given the ban on lexicalisation by a DP, the LF interface has two theta-features to interpret in the verbal domain. The reflexive interpretation is the result of the DP attractor interpreting both features in this domain (Tsimpli, 2006). What is more, in their analysis, the single argument is generated as a deep subject. Crucially, this means that in languages such as Greek, passives and anticausatives share morphology, but also an intransitive syntax. The issue remains whether naturally reflexive verbs are unergative or unaccusative.

What is more, the analysis of the German passive and reflexive structures is quite different, as all alternations involving sich have been argued to be ordinary transitive

2.3 The morphology of transitivity alternations 33 constructions. As summarised in Alexiadou and Schäfer (2012), evidence for this claim comes from the fact that i) the reflexive elementsichbehaves like an object pronoun, not like a verbal affix or a clitic (Fagan, 1992; Sells et al., 1987; Steinbach and Gärtner, 2000;

Steinbach, 2002) and ii)sichhas case (Fanselow, 1991). In ordinary reflexive constructions sichcarries an independent theta role and it is also semantically an argument (Doron and Hovav, 2007). That is to say, both an agent and a theme can be focused independently in reflexive constructions in German. This argues against an intransitivity account as well as a bundling account whereby one syntactic argument gets assigned two thematic roles.

(20) Morgens

‘In the mornings she washes herself (always/or emphasis on herself)’1

To sum up, the theoretical literature on Voice has not yet reached a consensus as to the proper characterisation of the structures that surface with special morphology. In addition to this, languages show significant syntactic variation concerning Voice formation; this further complicates the matter of reaching a consensus over the proper analysis of Voice structures. The approaches towards the analysis of Voice delineated above, raise questions for language acquisition, especially bilingual, which are under-researched. The cross-linguistic differences discussed here suggest that a bilingual child must learn subtle facts about the presence vs. absence of SE-anaphors and Non-active morphology, and how these work in the three linguistic systems namely, English, German, and Greek. The questions this study partly seeks to answer are associated with these different Voice systems. That is, assuming that both SE-anaphors (German, English) and Nact morphology (Greek) are involved in Voice alternations, but are subject to different syntactic analyses, a number of questions arise with regard to the acquisition of Voice. The research questions will be presented in the next chapter in detail.

1agent focus: She washes herself, no-one else washes her. (context: She is a disabled patient) theme focus: She washes herself, she washes no-one else. (context: She is a nurse)

2.4 Literature on the acquisition of passives and