A Control Analysis of Partial Pro-Drop { Philip Shushurin } Control Workshop, Berlin, 14 October 2016 Introduction • In many languages null pronouns are licensed in inite embedded clauses but not in matrix clauses. This phenomenon is often referred to as partial pro-drop. • In these positions pronouns may demon- strate PRO-like behavior (1) but are distinct from control constructions in many ways, e.g. they can be far enough from the con- troller (2): (1) Boris i Boris skazal said čto that (on i ) (he) pridet. will.come ‘Boris said that he would come’ (2) Boris i Boris uznal, learnt č̌to what nužno necessary delat’ do.INF kogda when (on i ) he pridët come.FUT domoj. home ‘Boris has found out what is to be done when he comes home’ The data The environments in which null pronouns in embedded clauses (NPEC) appear are of two types: • Complements of attitude verbs, as in example (1): null pronoun is bound, can only be interpreted de se, its antecedent must be in next clause up. • Other environments containing an overt complementizer: wh-complements, adjuncts clauses: the null pronoun is not bound, can be interpreted de se and de re, its an- tecedent can be more tham one clause away, topic-sensitivity: Common properties of two types of contexts: no temporal dependence on the clause containing the antecedent; no split antecedents; optionality; Embedded null pronouns are sensitive to the presence of an overt complementizer and not matrix- embedded distinction: they are impossible in comp-drop environments (3); they are licensed in matrix clauses which contain a complementizer (4). (3) Vova i Vova skazal, said [*(on i ) (he) zajdët will.come več̌erom] in.the.evening ‘Vova said that he would come over in the evening’ (4) Jesli If Vova Vova tuda there pojdët will.go.3sg *(to) then vyjdet will.get.3sg k to stancii station metro subway ‘If you go there you will get to the subway station’ References 1. Holmberg, Anders. Sheehan, Michelle. (2010). Control into inite clauses in partial null-subject languages. Parametric variation: Null subjects in minimalist theory, 125-152. 2. Livitz, Inna. (2014). Deriving Silence through Dependent Reference: Focus on Pronouns. PhD dissertation, NYU. 3. Tsedryk, Egor. (2012) Finite control as movement: the case of Russian. Contact Information Linguistic Department, NYU Email phshushu@gmail.com Phone +1 (917) 3626689 Proposal • NPEC is a null clitic which needs an overt complementizer to cliticize on; • NPEC is a phonologically null minimal pro- noun which is endowed with φ-features and a D feature which is responsible for indexical index: ∅ [uφ: uD: ]; • NPEC valuates its φ features by agreeing with a local verb. Null copulas are devoid of φ-features and thus a NPEC cannot be con- tained in a clause with a null copula; • The D feature can be valuated in 2 ways: by agreeing with the controller or by entering a topic chain: in this case the relation between the null pronoun and its antecedent is not syntactic. Goals • To show that constructions in (1) and (2), although having several important similari- ties, difer in several important respects; • To propose that both constructions contain the same deicient null pronoun which gets its features in two diferent ways: via an agree relation or via a topic chain; • To propose that this pronoun is a null clitic which needs an overt complementizer to cliti- cize on; • To show how the proposed system accounts for the facts; • To propose that the relation between the null pronoun and its antecedent is perhaps an- other type of control relation. D-control vs. φ-control • Similarities between φ-control and D- control: nullness of the pronoun, boundness of the pronoun (this is expected since the null pronoun values its feature via agree with its antecedent), subjecthood; • Diferences between φ-control and D-control: no featural dependency in the controlled clause in phi or temporal features (this is in- deed expected since no phi-agree takes place, φ-features of the minimal pronoun are val- ued within the clause), optionality, depen- dence on overt complementizers, crosslin- guistic distribution: D-control is found only in a limited set of languages (while φ-control is perhaps universal or nearly universal); • I leave the question open whether D-control is another type of control or must be treated independently. Previous accounts • Livitz (2014): φ-agreement of NPEC with its antecedent. NPEC is null because it is reduced as a ’defective goal’. • Tsedryk (2012): A movement derivation: NPEC are traces of their antecedents. Ms. • Holmberg Sheehan (2010): [for Finnish and Brazilian Portuguese]: the null pronoun is a minimal pronoun which has an unvalued D feature. This feature gets valued when NPEC enters a topic chain. NPEC are null by chain reduction. Crucially, all three accounts fail to account for the diference between (1) and (2), for the sensitivity to comp-drop and to overtness of the copula.