Optionality in complying with EPP in Dutch: a quest for a syntactic justification

(1)

Optionality in complying with EPP in Dutch:

a quest for a syntactic justi ication

Is insertion of a dummy subject in existential constructions

syntactically driven? Jessica Biermann (10633197) Master thesis General Linguistics University of Amsterdam Supervisor: Dr. P. Sleeman 16 June 2016

(2)

3.4 Licensing of Spec-TP 35 3.4.1 Subject raising to Spec‐TP 36 3.4.2 vP raising to Spec‐TP 38 3.4.3 Failed pied‐piping 41 4. The study 45 4.1 Participants 45 4.2 Method 45 4.3 Procedure 48 4.4 Analysis 48 4.5 Results 49 4.5.1 Head directionality 50 4.5.2 Predicate type 51

(3)

4.5.3 Subject type 53

5. Discussion 55

5.1 Overt expletives in existential constructions 55 5.2 Ramifications of covert expletives 57

5.3 Complying with EPP 58

5.4 Correction of the interface 59

6. Conclusion 61

References 63

Appendix I Analysis of subjects in existential constructions 66

Appendix II Analysis of subjects in existential constructions (contextually defined) 73

Appendix III Instruction and final survey 87

(4)

Abstract

Optionality in insertion of a dummy subject in existential constructions is the main topic of this thesis. This study further reflects Biberauers (2003) claim that not all languages may be equally in need of expletive elements. The main question of this study reads To which extent is the presence of a dummy subject in Dutch existential constructions

syntactically motivated following the requirements to comply with the Extended Projection Principle according to the Generative paradigm? To answer this question first a

descriptive analysis was conducted to seek for variances in the distribution of subject‐ DPs in active sentences and the use of dummy subjects in existential constructions in English, German and Dutch. After this comparative analysis of the use of existential constructions it became clear that the optionality of dummy subjects in postfinite position in Dutch was the most peculiar phenomenon and worth exploring by conducting an experimental study. The most relevant principle which accounts for the insertion of a dummy subject in existential constructions is the Extended Projection Principle. Biberauer (2003) discusses two approaches of how languages comply with EPP. The first strategy, which we see in English, is subject raising or insertion of a dummy subject in the specifier position of TP. Since the minimalist era it is assumed that a finite T carries an EPP feature which, in conjunction with agreement, drives A‐movement. The phi‐features of the specifier (subject) must agree with the head in T and the external argument in Spec‐vP is raised to the Spec‐TP. In case the subject remains in situ a dummy subject must be inserted in Spec‐TP to satisfy EPP, as (1) and (2) show. (1) John does not eat chocolate (2) There is not a man eating chocolate. The second strategy, which Biberauer discusses and which German employs to satisfy EPP is a verb raising strategy. Following this vP raising strategy the specifier of vP is attracted to Spec‐TP and the complete vP is pied‐piped to Spec‐TP. Therefore the insertion of an expletive in Spec‐TP is blocked since there is no available landing site.

(5)

This explains the ungrammaticality of clauses with a postfinite dummy subject in German, demonstrated in (3). (3) Ich glaube dass (*es) heute ein Brief gekommen ist. I believe that there today a letter arrived is. “I believe that there has arrived a letter today.“ The there-type es in German is introduced in the clause to act only as a default filler for Spec‐CP in the absence of a suitable frontable element (Biberauer, 2003) to ensure the preverbal slot of the clause is overtly filled in declarative sentences. In Dutch, however there are systems in which in postfinite position a covert expletive appears to be able to alternate with its overt counterpart, as (4)‐(5) shows. (4) Ik denk dat er morgen een man een huis koopt. I think that there tomorrow a man a house buys. “I think that there is a man buying a house tomorrow.” (5) Ik denk dat morgen een man een huis koopt. I think that tomorrow a man a house buys. “I think that there is a man buying a house tomorrow.” Biberauer (2003) claims that Dutch mostly follows the vP raising strategy, but that in some cases this vP raising operation fails and a dummy subject is inserted to save the syntactic structure, causing the optionality in Dutch existential constructions. From the experimental study in this thesis it became clear that only a minimal preference can be detected for existential constructions with an overt dummy subject compared to existential constructions with a covert dummy subject. All clauses with a covert expletive were accepted as grammatical. Therefore the claim from Biberauer must be rejected, since it does not explain why in this study all clauses with a covert dummy subject were accepted. Dutch thus exceptionlessly follows the vP raising strategy. However, the presence of the dummy subject has a consistent effect to raise the acceptability of existential constructions. This infers that the necessity of a dummy subject in existential constructions is motivated by the interface to concretize that the

(6)

subject‐DP is not the topic of the sentence. The expletive is postsyntactically inserted in the topic position in PF to mark the sentence as non‐topical. For the reader it becomes more apparent that the subject in the clause must be interpreted as focus material. In this study the correlation between this optionality and the role of the interface has not been fully explored and merits further research.

(7)

1. Introduction

Alternation of the basic word order of constituents is a universal trait of natural languages. There is a general consensus that the linear order of constituents, at least to some extent, is determined by the information structure. The relation between syntax and the information structure becomes apparent as the presupposed information precedes new information. Following this, a clause could be binary separated into topic: the given information which is contextually defined and the prosodic unmarked focus, the new information. (Erteshik‐Shir, 2007, Broekhuis, 2007). In certain languages a subject‐DP can freely move from its base position at the end of a clause into the domain of given information in an initial position of a clause. In the majority of sentences the subject occurs in an initial position, showing the subject is the topic of the sentence. However, across languages, the subject‐DP can occur in a different position than the default syntactic position of topics. In some clauses the subject should syntactically be interpreted as focus material. One approach of marking is to have a non‐topic element occupy the left‐peripheral position, as can be seen in

existential constructions. Following the Expletive Insertion Rule a dummy subject needs to be inserted to indicate the sentence as a non‐topic sentence, i.e. an all focus sentence. In the existential construction in (6) the expletive there co‐occurs with a subject in situ.1

(6) There is a man walking in the park. main clause

I hear there is a man walking in the park. embedded clause

In existential constructions with a topicalized subject‐DP the EPP feature, which triggers movement to Spec‐TP, is relevant and merits further exploration. Chomsky (1981) formulated that following the Extended Projection Principle all clauses must have subjects. Since then the Extended Projection Principle has been widely adopted as the default assumption for all languages (Biberauer, 2003; 1) and it was assumed that an overt 1_{Other strategies to mark the subject as non‐topic material, such as clefting constructions and subject inversion, are} not discussed in this thesis.

(8)

pleonastic pronoun must be inserted in Spec‐TP to satisfy the Extended Projection Principle in case the subject‐DP is absent in the topic position. However, across languages differences between the use of expletives in existential constructions and thus variances in complying with the Extended Projection Principle could be detected. More recently, it has even been proposed that the EPP does not always require an overt or covert element to occupy Spec‐TP (Biberauer, 2003).2_{This raises more questions on the} universality of the EPP. These variances become apparent in a further analysis of the omission of overt expletives in existential constructions. Dutch shows a remarkable optionality with respect to the insertion of a dummy subject in postfinite position as illustrated in (7). (7) a. Er koopt waarschijnlijk morgen een man een auto. There buys probably tomorrow a man a car. “There is probably a man buying a car tomorrow.” b. *pro koopt waarschijnlijk morgen een man een auto. pro buys probably tomorrow a man a car. “There is probably a man buying a car tomorrow.” c. Waarschijnlijk koopt er morgen een man een auto. Probably buys there tomorrow a man a car. d. Waarschijnlijk koopt morgen een man een auto. Probably buys tomorrow a man a car. “Tomorrow there is a man buying a car” (7) shows that in Dutch the subject een man must be considered to be focus material regardless whether the expletive er is inserted in the sentence or not.3_{Thus, it seems} possible to omit the expletive in postfinite position without the sentence becoming ungrammatical. Furthermore, it follows that the clause has the same core meaning 2_{On this analysis, languages with rich verbal morphology satisfy the EPP without the need for the literal extension of} a lexically required structure. Since rich morphology languages are not in scope in this thesis, this will no further be discussed. 3_{Embedded clauses show a similar pattern as main clauses. Passive constructions are not considered. (7) also shows} that a dummy subject is required as a CP iller. The preverbal slot cannot remain empty in main clauses.

(9)

whether or not the expletive is inserted. The Dutch examples suggests that some languages don’t necessarily need to comply with the Extended Projection Principle. In this paper I will explore the possibilities of this observation and what the ramifications are for the syntax‐discourse interface. My main empirical question is as follows:

To which extent is the presence of a dummy subject in Dutch existential constructions syntactically motivated following the requirements to comply with the Extended Projection Principle according to the Generative paradigm?

In this study I will compare the optionality in the use of an overt expletive and the subsequent syntactic implications of omitting an expletive in Dutch existential constructions. Though the main focus in this thesis is on Dutch I will parallel the observations from Dutch to German, as a morphologically richer V2 language and to English, as a non‐V2 language. In this thesis I will answer the subsequent questions. 1. To what extent is the use of an overt expletive in existential constructions required to comply with EPP in preverbal and postfinite position of a clause in V2 and non V2 languages? 2. What are the ramifications of focusing a subject‐DP without the use of an expletive with respect to the syntax in Dutch? 3. How can the optionality in existential constructions with respect to complying with EPP be accounted for from a Generative standpoint? The first question will be answered in the theoretical background of this thesis. For the second question an experimental study will be conducted. The third question will assess how the optionality of the insertion of a dummy subject in Dutch could be explained following Biberauer’s (2003) theory, which outlines two different strategies to comply with EPP. The main claim of this study is that Dutch existential constructions with an overt as well as with a covert expletive are syntactically grammatical. It must be assumed that since no syntactic motivation explains the optionality of the presence of a dummy subject, an eventual insertion of the expletive is done after spell‐out since the core meaning of the clause hasn’t changed.

(10)

This thesis is organized as follows. The first two chapters outline the theoretical background and summarize previous findings that are relevant to this current study. The second chapter gives a brief overview of the distribution of the subject‐DP in a clause in Dutch, compared to English and German following the minimalistic paradigm. The third chapter discusses the structure of existential constructions and explores the Extended Projection Principle which supposes to trigger movement of an subject‐DP to or insertion of an expletive into the left periphery. In chapter 4 I will present the methodology and results of the experiment. Finally, in chapter 5 the discussion and in chapter 6 the conclusions are outlined.

(11)

2. Movement of the subject-DP

One of the core constituents in a sentence is the subject. As Stowell (1983) has proposed all lexical phrases may contain a subject, but nonetheless the position of the subject in a clause is far from fixed. A sentence has a canonical word order and variation in the position of argument DPs is derived from syntactic movement. Movement is a pervasive characteristic of natural language and in many languages the subject can remain in situ but could as well, due to morphosyntactic or discourse informational reasons, be moved to the left periphery. The traditional subject could be deconstructed into three components, one thematic‐ aspectual, one morphosyntactic and one discourse‐informational component (Svenonius, 2001). In most straightforward cases all these components converge on a single DP, in other cases they diverge. The interaction between these components vary from language to language. In this section I will discuss each component. In section 2.1 I will outline the thematic role of the subject and in section 2.2 the morphosyntactic component. In this section I will also compare Dutch as a V2 language to English as a non V2 language.4_{Section 2.3 discusses the discourse informational component, with} notions such as focus, topic, subject scrambling and specificity. Lastly, section 2.4 discusses the structure of the left periphery and the movement operations which are syntactically required for discourse purposes.

2.1 Thematic role of the subject-DP

All arguments in a clause have a thematic role. This thematical role determines the position of the argument in the clause. An important notion in this matter is subcategorization. It denotes the necessity for lexical items to require the presence and types of the syntactic arguments with which they co‐occur. For example the intransitive verb walk subcategorizes for a single argument, which is the subject of the clause. Each verb may have one or more internal arguments, as examples (8)‐(10) demonstrate. 4_{The analysis of German is similar to Dutch. Therefore, for German no morphosyntactic analysis is presented. All claims made for} Dutch are also valid for German.

(12)

(8) The woman walked walk, Vi [ ___ ]5 (9) John loves Mary love, Vt [ ___ DP ] (10) Mary gave Peter flowers give, Vdt [ ___ DP DP ] A distinction must be made between internal and external arguments. An internal argument of a verb has to be realized inside the maximal projection of that verb (VP). The argument which is closest to the verb is called the direct internal argument, while the others are called the indirect internal arguments. An external argument is the argument outside the maximal projection of the verb. This is mostly the subject of the clause. In contrast to intransitive, transitive and ditransitive verbs, ergative verbs, such as break or melt, do not require an external argument. In this case the internal argument must be moved to the canonical position of the external argument or a dummy subject must be inserted to save the syntactic structure. Arguments are merged according to a thematic hierarchy. Direct objects are low thematic roles, merged first as a sister‐to‐V, with indirect objects being merged next in Spec‐VP, and agentive argument‐type subjects being merged last in Spec‐vP (Biberauer, 2003). This could be represented in the following syntactic structure. (11) I gave John a book vP DP VP I DP V’ John V DP gave a book

Figure 1. Syntactic representation of the verbal domain (head initial language)

(13)

The thematic role of the argument also determines its case assignment and agreement with verbal features. For example, the subject has the agent theta role, which represents the acting constituent in the clause. The subject agrees with the verb in person and number and requires nominative case. 2.2 Morphosyntactic component Subjects in various languages demonstrate different behaviour in their distribution in clauses. Movement of the subject is a core trait of natural languages and this movement is triggered by the necessity to check morphosyntactic features, which contain properties for tense and agreement. (figure 2.) INFL’ INFL VP AGR T [PERSON] [±FINITE] [NUMBER] [±PAST] [CASE]

Figure 2. Overview of the morphosyntactic features within the functional domain. The Inflection Phrase (INFL’) represents Agreement and Tense.

The constituents in the lexical domain share a feature with a XP in the functional domain (Rizzi, 2004). This implies that one of the features of the DP, such as case in the verbal domain must agree with the case feature in the functional domain.6_{In case these} features do not agree the syntactic derivation is ill‐formed. Depending on the strength of the functional domain a XP moves overtly or covertly when the features are checked. According to the Minimalist Program (Chomsky, 1995) the thematic and discourse corresponding features are checked to a constituent at the moment the constituent 6_{Functional domain is TP/ IP.}

(14)

enters the syntax.7_{The subject must check morphosyntactic features to ensure it agrees} with the verb. The thematic features of subjects are case checked through spec‐head agreement against the features of the verb (Rizzi, 2004). A constituent in the lexical domain shares a feature with an XP in the functional domain and depending on the strength of these features overt movement takes place. This implies that there is a parametric variation in the way morphosyntactic features are checked. Chomsky (1995) and following studies have demonstrated that case and agreement could be checked without an overt movement (Boeckx, 2003). A relevant notion concerning the movement of the subject and verb in Germanic languages is the Verb Second (V2) constraint. In the following subsections I will discuss the difference between V2 languages, such as German and Dutch and non‐V2 languages such as English. 2.2.1 V2 languages With verb second the main restriction on word order is that, in main clauses, the constituents may appear in essentially any order, as long as the finite verb is in the second position in the clause. (12) a. Ik heb dat boek niet gelezen. Ich habe das Buch nicht gelesen. I have that book not read. b. Dat boek heb ik niet gelezen. Das Buch habe ich niet gelesen. That book have I not read. c. Gelezen heb ik dat boek niet. Gelesen habe ich das Buch nicht. Read have I that book not. “I have not read that book.” To explain this phenomenon a special verb preposing rule Verb Second is needed which adjoins the inite verb to whatever constituent happens to be in irst position in the

(15)

declarative sentence. (Den Besten, 1983). The V2‐constraint in main clauses, due to which inite verbs move to the C‐domain, attract an overt movement of V.8_{Following the}

Head Movement Constraint the verb must successively move via T to the C‐domain (Travis, 1984).9_{Consecutively the subject moves to Spec‐CP to ensure it appears in a} preverbal position. (13) illustrates the movement operations:10 (13) a. De kinderen hebben deze film gezien b. Die Kinder haben diesen Film gesehen c. The children have this movie seen “The children have seen this movie.”

[CP De kinderenj [C’ hebbeni [TP tj [T’ [VP tj [V’ deze film gezien]] T ti]]]]

However, V2 languages, show numerous structures that challenge the idea that an overt movement to Spec‐TP takes place to check nominative case. (Biberauer, 2003). Alexiadou & Anagnostopoulou (2006) argue that subjects can be licensed within the verbal domain, since they can follow any adverbial elements in a clause which mark the left side of the vP. The c‐commanding requirement states that a constituent can only move to a structural higher position and lowering is prohibited (Zagona, 2002). From this analysis it must be concluded that Dutch and German are parametrically set on licensing within the verbal domain. Nominative case is thus assigned to the subject‐DP in Spec‐vP. (Bennis, 1987).11_{Movement of the subject occurs due to topic assignment or} following the V2 constraint. In Dutch and German, due to the V2‐constraint, the subject can appear in a preverbal as well as a postverbal position. The insertion of an adverbial element in (14) and (15) 8_{I follow the assumption of Biberauer (2003) that the V2 constraint necessarily involves a inite verb in C.} 9_{I will refer in this thesis to the inflectional domain via the cover‐term TP and assume functional categories are head‐} initial and that functional verbs (i.e. temporal, passive and modal auxiliaries) are either merged in or moved to T. 10_{Embedded sentences show no evidence for the strength of verbal features in Dutch and German, since the verb is} always in the inal position of the syntactic structure. Nonetheless, I assume that also in embedded clauses V is moved to T. With embedded sentences it is possible that after the verbal cluster a PP occurs. This must be explained by extraposition of PP or VP‐fronting (Koster, 1987, Barbiers, 1995). 11_{Chomsky (2000) assumes that a “Criterial Probe” (T) in a structure attracts a ”Criterial Goal” (Spec‐vP) which} carries a similar criterial feature (Rizzi, 2007).

(16)

shows that the word order is not fixed and thus that the subject can remain in a position in the verbal domain. This means that, as the subject can occur in its base position, the nominal features are weak. Zagona (2002) claims that when a language has weak nominal features a preverbal subject is not obligatory. In Dutch, as (14) and (15) demonstrate, the preverbal position could also be filled by an adverbial element. (14) a. Vanavond leest waarschijnlijk een student een boek. b. Heute abend liest wahrscheinlich ein Student ein Buch. c. Tonight reads probably a student a book. “Tonight probably a student reads a book.”

[CP Vanavondj [C leesti [TP [T’ [VP waarschijnlijk [VP tj [VP een student [V’ een boek ti]]] T ti]]]]] (15) a. Vanavond leest Jan waarschijnlijk een boek. b. Heute abend liest wahrscheinlich Jan ein Buch c. Tonight reads Jan probably a book “Tonight probably Jan reads a book.”

[CP Vanavondj [C leesti [TP Jank [T’ [VP waarschijnlijk [VP tj [VP tk [V’ een boek ti]]] T ti]]]]] (14) illustrates that clauses containing an indefinite (non‐specific) subject are grammatical in case the subject remains in situ. _{A clause with a definite subject as} demonstrated in (15) requires the subject to move to Spec‐TP. These examples show that movement is dependent on the information structure of the subject. This will be further discussed in 2.3. 2.2.2 Non-V2 languages However, non‐V2 languages behave differently, as can be seen by analysing English. English has a weak functional domain which means movement of the verb to the functional domain is not triggered. Assuming that the adverb remains in its base position as an adjunction of vP, it becomes clear that the verb does not move to the Inflection Phrase since in English the adverb does not intervene between the lexical verb and the complement. This is illustrated in (16) where the finite verb eat remains in situ

(17)

following the adverb often ensuring that the lexical verb and the complement are adjacent to each other. Due to the fact that in English no verb raising occurs, often auxiliary verbs are inserted in T, as (17) and (18) illustrate.12_{(19) shows that in case the} clause contains a (Wh) question an auxiliary verb must be inserted to save the syntactic structure. (16) I often eat apples. * I eat often apples. (17) I have often eaten apples. (18) I do like to eat apples. (19) * Eat I often apples. Have I often eaten apples?

Verb Movement in English is restricted to have and be (Pollock, 1989). Because English lost most of its verbal inflections the agreement phrase became opaque to theta role assignment and verb movement was restricted to be/have. Pollock (1989) concludes that theta theory is the module of Universal Grammar that is responsible for the lexical restrictions that bear on verb movement to TP. The lexical verbs must assign a theta‐ role. The fact that only have or be can undergo verb movement suggests that there is a potential tension between theta‐role assignment and movement to TP (Pollock, 1989). Theta role assignment is “blocked” by the morphologically “poor” agreement of English and have or be could undergo verb movement because these two verbs do not have to assign theta‐roles. In case auxiliary verbs are not present the sentence is syntactically saved via do-support (Carnie, 2007), as demonstrated in (18). Furthermore, in declarative clauses in English the subject‐DP must always occur in a preverbal position, as (20) and (21) demonstrate.13 (20) John has often eat chocolate. (21) *Yesterday has John eaten chocolate. 12_{Af ixes presented in T must be lowered in English to be united with the verb.} 13_{Sentences that start with only/never are not considered, in these cases the subject occurs postverbally.}

(18)

To attain a preverbal subject the subject‐DP must be raised from the verbal domain to Spec‐TP, also complying with EPP requirements. Figure 3 illustrate the movement of the subject to Spec‐TP to ensure a preverbal position is acquired. CP C’ C TP T’ T vP

Figure 3. Syntactic structure demonstrating movement of the subject from the vP domain to the TP domain. (details of vP domain not represented) (head initial language)

2.3 Discourse-informational component As illustrated in section 2.2 the default syntactic position of the subject is the preverbal position. Constituents in various languages demonstrate different behaviour in their distribution in clauses. Movement of the subject‐DP to the left periphery is relevant for the information structure and it appears that cross linguistically the distribution of the subject varies within a clause. In this section I will first discuss specificity of subject DPs, which is relevant for the position of the subject in a clause. In section 2.3.2 and 2.3.3 I will explore the notions topic and focus which are relevant for the information structure and in 2.3.4 the consequences and possibilities to scramble the subject‐DP to a different position in the clause.

(19)

2.3.1 Specificity of subject DPs Constituents can occur in a variety of word orders without a difference in their core meaning. Specificity is one of the important factors in these movement operations. Changing the word order is relevant for the information structure, as (22) and (23) illustrate. Definite subjects and specific indefinite subjects belong to the topic of a sentence and thus move to the left periphery. A non‐specific subject is to be considered as something new in the discourse and therefore movement to the left periphery could not be motivated.14 (22) Een man koopt waarschijnlijk morgen een auto. A man buys probably tomorrow a car. specific “A man probably buys a car tomorrow.” (23) Waarschijnlijk koopt morgen een man een auto. Probably buys tomorrow a man a car. non-specific “Probably tomorrow a man buys a car.”

In (22) a man has a specific reading, which means the speaker is familiar with the

identity of the constituent and refers to a specific constituent. Therefore it belongs to the presupposition and movement to a higher position in the syntactic structure is required (Zwart, 2012). In (23) the adverbs probably and tomorrow precede the subject. In this case the subject has a non‐specific reading.15_{Because of its non‐specific reading the} subject remains in situ and is marked as non‐topic material belonging to the focus of the sentence. Presupposed information is in many languages positioned at the beginning and new information at the end of a sentence. In the following subsections the concepts topic and focus are discussed from a syntactic perspective. 14_{A constituent is non‐speci ic if it designates a person or matter which has not been introduced so far and therefore}

is not identi iable for the speaker or hearer. An inde inite speci ic constituent is only new to the hearer (Lexicon of Linguistics).

15_{In German specificity is not expressed through grammatical means. However, German does distinguish between}

(20)

2.3.2 Focus Focus is the new information in a sentence and the word order shows what the new information is. 16_{Chomsky (1971), Jackendoff (1972) and Lambrecht (1994) define} focus as the non‐presuppositional information in a clause.17_{Cinque (1993) asserts that} syntax is a potential reason for cross linguistic variances in focal accentuation. In headfinal languages the lowest embedded constituent is accentuated, in head initial languages, however, the lowest embedded constituent is positioned to the right of the verb. (Zagona, 2002, Zwart, 2012). (24) Dat Jan een kind GELD gegeven heeft. (25) John gave a child MONEY. (24) and (25) illustrate that independent of the word order in a clause the same constituent (in this case the object‐DP) is focally marked. This implies that focus is syntactically motivated. In this thesis I will follow the definition of Diesing (1992)18_who associates the difference between topic and focus with their position in the syntactic structure. vP‐internal material belongs to the focus and vP‐external material belongs to the topic of the sentences. This explains why new information in many language is positioned at the end of the sentence. 2.3.3 Topic Topic on the other hand occurs in the front of a clause and is the presupposed information.19_{The presupposition of a sentence is the information that is assumed by} the speaker to be shared by him/ her and the hearer (Erteshik‐Shir, 2007). In a clause two notions are relevant: topic and comment. The topic of a sentence is a constituent referring to an identifiable entity, which the remainder of the clause may be construed as a comment upon (Zwart, 2012). 16_{For the definition of focus the distinction is made between unmarked (wide) and contrastive (narrow) focus.} Contrastive focus gives an entity in a clause a specific information status, which indicates that that constituent is ‘in focus’, in contrast to an alternative (o.a. Zwart, 2012). In this thesis contrastive focus is not considered. 17_{From: Erteshik‐Shir (2007).} 18_{From: Erteshik‐Shir (2007).} 19_{From: Erteshik‐Shir (2007).}

(21)

(26) Mary went to Vienna

In (26) the topic of the sentence is Mary and the comment is went to Vienna. Subjects bear certain interpretive similarities to topics and yet the two notions are clearly distinct. In (26) the subject is similar to the topic of the sentence. A subject shares with a topic the prominence related to the fact that the described event is presented as being about that argument (“aboutness”). But not in all cases the topic is similar to the subject. Therefore, we need to distinguish a sentence topic from a discourse topic. A discourse topic is the central participant or idea of a stretch of connected discourse or dialogue. The topic is what the discourse is about, while a sentence topic is what a particular sentence is about.

(27) Mary awoke at six o clock that morning. Today she had to start with her new job in Vienna. She hurriedly took a shower and had some breakfast. The train would leave at 7:15 and she did not want to come late the first day. She was too nervous to read the newspaper in the train. Just before eight the train finally arrived in Vienna. The office where she had found the job was only a five minutes walk from the station (...)

In (27) the distinction between a discourse (D‐)topic and a sentence (S‐)topic becomes apparent. In this fragment there are a few sentences in which the train is S‐topic, i.e. which say something about the train. Nonetheless, the D‐topic is Mary. A S‐topic differs from a D‐topic in the fact that it is does not require the discourse linked property (Rizzi, 2005).20 S-topic D-topic Aboutness + + D-linking ‐ +

Table 1. Overview of the difference between subject and topic related features.

According to the Strawson‐Reinhart approach S‐topic has three core properties (Erteshik‐Shir, 2007).

(22)

a) The topic is what a statement is about.

b) The topic is the presupposed constituent of a clause.

c) The truth-value of a proposition is determined based on the topic of a clause. In the literature there is consensus on the first two properties. Topic describes (a) the aboutness of a clause and (b) is presupposed information for the speaker and the hearer. (Zagona, 2002). The third property is more controversial. Strawson claims that if the topic expresses aboutness and thus the clause is judged in relation to the topic than it must have a reference to define its truth value. (Erteshik‐Shir, 2007). If, however, the topic has no reference, the sentence cannot be evaluated as either true or false. Strawson illustrates this point with the following example. (28) The King of France is bald (29) The exhibition was visited by the King of France In the context of the question in “What is the King of France like?” (28) has no truth value since the sentence cannot be verified. It cannot be about a non‐existing king. In (29) however, “the exhibition” is the topic. The sentence will therefore be assessed as false if the King of France is not among the visitors at the exhibition. It follows from the properties Strawson assigns to topics that the topic is chosen in accordance with the context. (Erteshik‐Shir, 2007). In this thesis I will assume all three characteristics are relevant and thus that all sentences contain a topic, since all sentences have a truth‐ value. This implies that all sentences have a topic. 2.3.4 Subject scrambling As described in the previous sections syntax is responsible for the position of focus constituents as well as the presupposed information. Movement is blocked when the subject belongs to the unmarked focus of a sentence, but is obligatory in case of topichood. Scrambling is a process which ensures that a constituent moves from its base position in the verbal domain to the topic domain. It is a process that occurs in languages with a free word order and is the reason why sentences have a non‐canonical word order. Scrambling was first defined in Ross’ dissertation Constraints on Variables in Syntax (1967), which states scrambling is a stylistic rule. In the following literature

(23)

multiple definitions have been proposed. Corver & van Riemsdijk (199421_{) describe it as} syntactic reordering and Zagona (2002) as a movement rule which permits a constituent to move to a higher position in the syntactic structure. Movement of the subject to the left periphery will be blocked when it belongs to the new information of a sentence, but is required in case of topichood. This fronting is referred to as topicalization. (Zwart, 2012). Scrambling enables constituents to appear in a variety of word orders without a change in the core meaning of the sentence. (30) and (31) show the difference in the word order. (30) Ik hoor dat hoogstwaarschijnlijk morgen een rijke man een nieuwe auto koopt. I hear that probably tomorrow a rich man a new car buys. (31) Ik hoor dat een rijke man hoogstwaarschijnlijk morgen een nieuwe auto koopt. I hear that a rich man probably tomorrow a new car buys. “I hear that probably yesterday a rich man buys a new car.” The analysis of scrambling is usually focused on the behaviour of direct objects (i.e. Zagona, 2002; van Gelderen, 2003; de Hoop, 1997;200022_{). Broekhuis (2007) claims that} the behaviour of subjects is similar to that of direct objects and this implies that the analyses of direct objects is also adequate for subjects. The question whether scrambling should be considered as an A‐movement or A’‐movement is continuously debated among linguists. Broekhuis (2007) marks the distinction between the two processes following these definitions: object/ subject shift for A‐movement and scrambling for A’‐ movement.23_{van Gelderen (2003) also describes scrambling as a A’‐movement. She} defines scrambling as a syntactic movement motivated by the information structure and 21_{From: Erteshik‐Shir (2007).} 22_{From: Erteshik‐Shir (2007).} 23_{Object scrambling and topicalization are not related in the same manner. The direct object will not belong to the} main topic of the clause when it is scrambled. For the direct object scrambling is an optional movement, whereas for the subject it is a required movement to the left periphery for checking the topic feature.

(24)

includes topicalization as a scrambling process: an A’ movement to the left periphery.24 In this thesis I follow the definition of van Gelderen and claim that scrambling is used to topicalize a constituent, in this case the subject, and thus coding the presupposed information.

2.4 Structure of the left periphery

As section 2.2 pointed out requirements of the case‐agreement system are insufficient to motivate every movement to the left periphery (Rizzi, 2005). The information structure of a clause is also determinative for movement of the subject. Within the generative framework, there has been a continuous debate about the impact of discourse‐linked functions upon syntax. In this thesis I will follow Rizzi’s theory on discourse related movement features. Before discussing Rizzi I will first outline the difference between subject movement and topic movement. Syntactic movements are characterized by their local character and movement occurs in small (recursive) steps. This way a chain is formed. The definition of the notion chain was first introduced by Chomsky and is created by the moved element on the landing site and the trace that remains in the base position. As such a chain contains a unique function (for discourse or theta) and a unique referent. The movement of a subject forms an A‐chain. The movement of subjects can be characterized as highly local. Movement to the subject position is basically constrained to affect the closest nominal. Movement to topic is on the other hand much freer forming an A’‐chain. An A’‐chain connects the position where the theta role and case are assigned to an argument and the left peripheral position where discourse related features are assigned to the constituent (Rizzi, 2004). Movement of the subject to the T domain is obligatory, since a subject must be present in the clause. On the other hand movement to check discourse related features in the C domain is optional and present in the structure only when discourse conditions and communicative intentions require them. In Rizzi (1997) the left periphery of the clause is seen as a structural zone defined by a system of functional heads and their projections, along the following lines. 24_{Topicalization is the de inition that is used in this thesis for all reordering processes where a constituent is fronted} to check the Topic feature.

(25)

Figure 5. Rizzi’s hierarchy of the C-domain.

The C‐domain contains two kinds of information: the interface to the next higher clause, here referred to as Force and the interface to information on finiteness, here referred to as Fin. The topic and focus field are ‘sandwiched’ in between these two interfaces (Rizzi, 1997). Discourse‐related left peripheral positions are created by the usual structure‐ building mechanism. There are special functional heads of Topic and Focus which give rise to their own projections, and whose specifiers are positions dedicated to topical or focal interpretation (Rizzi, 2000; 14).25 ForceP Force’ Force TopP Top’ Top TP T’ T vP

Figure 6. Syntactic representation of the C-domain, with a topic position (head-initial language).26

25_{A low IP area as proposed by Beletti (2000) is not considered in this thesis.}

26_{Other possible projections in the C‐domain have been omitted in this structure. The structure below vP is not}

represented.

(26)

Rizzi’s system is embedded in the minimalist approach in which movement to a specifier position is triggered by the satisfaction of feature requirements of the head. (Erteshik‐ Shir, 2007). According to Rizzi (2000) left peripheral positions, such as topic, are positions defined by special, interpretively relevant features. Thus, movement to the left periphery is motivated by a topic assignment (Erteshik‐Shir, 2007). Fronting a constituent ensures that the constituent is moved from the focus‐set to the topic‐position of the sentence. Under the criterial view, movement to the subject position is movement to a position dedicated to a special interpretive property of the scope‐discourse kind (aboutness). It is related to the other cases of criterial movement to the left periphery of the clause, and complies with the “movement as last resort” guidelines.27_{This assignment ensures, according to Rizzi (2007), that topics are raised} to the assigned specifier position. The movement is triggered by a specific feature that needs to be checked. This could be described as follows.:

A topic of a clause must be in a Spec-Head agreement with Xo[TOPIC].

A X0[+TOPIC] must be in a Spec-Head agreement with the topic.

The topic‐criterion ensures that the topic occurs in the specifier of the head that carries the topic feature (Zagona, 2002). A topic‐XP could thus only land in a Spec‐position of the head with a topic feature. A strong topic feature triggers an overt movement ensuring that the constituent reaches a higher position in the syntactic structure. In this thesis I assume that topic is associated with a special Topic projection. The topicalized element must therefore move to the specifier of this TopP, which is used as a discourse‐ configurational position. If the subject is part of the presupposed information of a clause than the topic criterion on TopP attracts movement. Figure 7 illustrates that the subject‐ DP is moved to Spec‐TopP for topic reasons.28_{Subjects belong naturally to the topic of a} sentence and are appointed as unmarked topics by Erteshik‐Shir (2007). In Dutch and 27_{Movement as a “last resort” guidelines determines that there is no optional movement (Chomsky, 1986 and} subsequent work).(Rizzi, 2005) 28_{This movement operation occurs in all three languages which are explored in this thesis, i.e. German, Dutch and}

English. The head‐directionality of igure 7 is head inal, but the relevant movement operations also occur in head‐ initial languages.

(27)

German main clauses due to the V2 constraint the topic of the clause successively moves to Spec‐CP to attain the preverbal position.29 ForceP Force’ TopP’ [+TOPIC] Top’ Top TP T’ T vP [+TENSE]

Figure 7. Topichood of Spec-TopP and a possible movement of the subject-DP to the Topic-position. (details of vP domain not represented) (head initial language)

(28)

3. Existential

constructions and the EPP feature

In section 2 it was outlined that movement of the subject to the left periphery is a configuration for topichood. In case the subject remains in situ an expletive must be inserted to save the syntactic structure following the Extended Projection Principle. However, not all Germanic languages behave similarly. In this section I will first outline the definition of existential constructions. In section 3.2 I will discuss the Extended Projection Principle which triggers movement to Spec‐TP. In section 3.3 I will discuss the optionality of dummy subjects in Dutch clauses. Finally, in section 3.4 I outline the syntactic analysis which explains the dissimilarities in satisfying EPP in Germanic languages. 3.1 Existential constructions In generative syntactic research existential constructions have received a considerable amount of attention. Across languages, in certain clauses the subject‐DP tends not to take the default syntactic position of topics but occurs in the position in the verbal domain. In existential constructions an expletive co‐occurs with a subject in situ. It has the distributional properties of the subject, appearing in the first position in unmarked main clauses, left adjacent to the finite verb (Zwart, 2012). It is assumed that these expletives are semantically empty elements inserted to satisfy the Extended Projection Principle in contexts where this would not otherwise have been possible (Biberauer, 2003). They are defined as genuine ‘placeholders’ and specifically associated with indefinite subject‐DP’s. Expletives have no thematic function, but nonetheless usually appear in subject position (Bennis, 1987).30 Existential constructions are constructions with non‐canonical morphosyntax, which express a proposition about the existence or the presence of someone or something in a context (Francez 2007; McNally 2011: 1830)31_{In Germanic languages the subject‐DP} occurs in existential constructions in a postverbal position. The argument‐DP in clauses 30_{In this thesis I will only investigate the use of pleonastic pronouns in relation to information structure, and not} when used in 0‐argument verbs. 31_{From: Bentley, Ciconte & Cruschina (2013).}

(29)

with an overt pleonastic pronoun remains vP‐internal and an expletive must, according to the Expletive Insertion Rule (Carnie, 2007) be inserted in Spec‐TP.32

Existential constructions are composed by inserting a dummy subject. In Dutch the pronoun er, in English (impersonal) there and in German es is inserted.33_The dummy subject is not an argument‐DP (Koster, 1987, Bennis, 1987), it has no thematic role and insertion of it does not violate the theta‐criterion.

They are formed as in (32).

(32) Er liggen enkele boeken op de tafel. Es gibt einige Bucher auf dem Tisch. There are some books on the table.

Expletive verb DP PP The linear order of the components of the existential construction can also vary across languages (Bentley, 2013).34_{Only the DP is universally available and obligatory in} existential constructions. The other components of the construction may not be present. Generally, in languages with a free word order the variation in word order tends to correlate with the variation in information structure (Bentley et al, 2013).

3.1.1 Dummy subjects as stage topics

An existential construction is a way of marking a sentence as being all focus. (Erteshik‐ Shir, 2007). In such sentences a S‐topic is missing, therefore a stage topic is inserted to occupy the left‐peripheral position. An expletive fulfils the role of the stage topic: they have no thematic function, but nonetheless usually appear in subject position (Bennis, 1987). Erteshik‐Shir (2007) asserts that the pleonastic pronoun has no topical function, but it occurs to fulfil the function of topic with the context as its reference (Erteshik‐Shir, 2007). Gundel (1974) argues that such sentences have a topic which is an element that 32_{The question whether or not EPP must be overtly satis ied by inserting a dummy subject in Spec‐TP will be} answered in the experimental study. 33_{Extraposition dummy pronouns and weather pronouns are not considered.} 34_{The position of the adverbial element could in some languages be fronted.}

(30)

designates the time and/ or place of whatever is asserted.35_{Topic is thus contextually} defined. Since an expletive has a non‐referential character it has no topical function, but is inserted to mark a clause as a non‐topical clause. (33) Er zwemt ’s ochtends een man in de zee. There swims in the morning a man in the sea. “There is a man swimming in the sea in the morning.” In (33) the expletive er is the stage topic of the sentence. Whether the sentence is true or false is determined based on perception. The current time and location again functions as the implicit topic with respect to which the sentence is evaluated (Erteshik‐Shir, 2007). Thus in the sea modifies the locations, so that the sentence will be true only if there is a mean swimming in the sea. An expletive cannot not co‐occur with all kinds of subjects as (35) shows. Quantified, bare nominals and indefinite article DPs can co‐occur with a dummy subject, while definite article DPs and proper names cannot. (34) a. There are three dogs walking in the yard. There are children walking in the yard. There is a man walking in the yard. b. Er lopen drie honden in de tuin. Er lopen kinderen in de tuin. Er loopt een man in de tuin. c. Es laufen drie Hunden im Park. Es laufen Kindern im Park. Es lauft ein Mann im Park . 35_{From: Erteshik‐Shir (2007)}

(31)

(35) a. *There is the man buying a new car. *There is John buying a new car. b. * Er koopt de man een nieuwe auto. *Er koopt John een nieuwe auto. c. * Es kauft der Mann ein neues Auto. *Es kauft Johan ein neues Auto. 3.1.2 Cross linguistic variances

Nonetheless there is also some cross linguistic variation in the use of existential constructions. The first difference is that in Dutch and German, in contrast to English, no auxiliary verb needs to be inserted in the functional domain when a pleonastic pronoun occurs.36_{This could be explained by the movement of the finite verb. Due to the strong} features on V in Dutch and German the finite verb raises to the functional domain.37_A correct syntactic derivation arises with the argument subject in its base position. However, in English an auxiliary must be inserted to save the syntactic structure. Without the auxiliary there would be no finite verb present, which is a requirement for all sentences. Secondly, there is a dissimilarity between the languages in relation to the occurrence of the expletive in the T or C domain. In German, the expletive es can only occur in sentence‐initial position (Drosdowski & Augst, 1984). Due to the V2 constraint which ensures movement of the verb to the C‐domain the expletive es can only occur in Spec‐ CP as illustrated by (36). In Dutch and English however the expletive can occur in non‐ initial (Spec‐TP) position as well. (36) a. Es ist ein Junge gekommen. Er is een jongen gekomen. It/pro is a boy come. “A boy has come.“

36_{In English there are some exceptions such as There comes a man where no auxiliary verb needs to be inserted.}

Nonetheless, in most cases, an auxiliary verb is required.

37_{In matrix clauses due to the V2‐constraint the verb moves to C. A dummy subject is inserted in Spec‐CP as a CP‐}

(32)

b. Ich weiß, daß *es ein Junge gekommen ist. Ik weet, dat er een jongen gekomen is. I know, that it/pro a boy come is. “I know that a boy has come.” These observations could be summarized as follows.

Language Word order Expletive in TP Expletive in CP

English SVO 

Dutch SOV + V2 ()* 

German SOV + V2 

Table 2. Summary of the distribution of expletives in presentational clauses in Germanic languages * The is in between parentheses since in Dutch the insertion of the expletive in TP is optional.

3.2 EPP-feature In its traditional form the Extended Project Principle (EPP) is the statement that the subject position is an obligatory component of the form of clauses (Rizzi, 2005). Regardless of the number and type of theta‐roles the verb assigns the subject position in the sentence must be filled. Also verbs without an agentive thematic role require a filler of the subject position. Chomsky (1981) formulated the Extended Projection Principle as follows:

Extended Projection Principle: Clauses must have subjects (Chomsky, 1981)

The Extended Projection Principle is thus a constraint on the output of the merge operations and states that every clause must feature a subject‐related element in the canonical subject position (Biberauer, 2003). All clauses must contain a subject and the specifier position of TP is seen as a systematic landing site for movement. (Rizzi, 2005). In case no subject‐DP or CP is present in Spec‐TP an expletive element compensates the absence of a raised thematic nominal. However, many of the world’s languages do not exhibit an overt movement of the subject to Spec‐TP to satisfy EPP. Chomsky therefore recasted the Extended Projection Principle in 1995. His changed view on EPP states that it can be decomposed into two T‐

(33)

related components, namely a nominal feature and an associated movement trigger. This trigger determines, based on its strength, whether or not movement overtly takes place. This trigger is reformulated into an EPP feature in his work in 2000. (Biberauer, 2003). The EPP feature thus determines whether or not overt movement takes place. In this new view EPP could be satisfied on T both by the nominal element as well as the verb. Under this analysis, the obligatory projection of a TP‐specifier no longer follows, with EPP‐satisfaction being achieved via v‐to‐T movement. A specifier projection is dependent on the presence of a further movement trigger. 38

3.3 Optionality in expletive insertion

Existential constructions are one of the strategies languages use to focus the subject and therefore omit movement of the thematic subject to the EPP‐position. The EPP feature is a relevant feature for the presence of an expletive in existential constructions. Since Chomsky (1982) the Extended Projection Principle has been widely adopted as the default assumption for all languages (Biberauer, 2003). However, if we further look at Dutch we see variances in the possibility to omit an overt expletive in existential constructions. Before discussing the optionality of inserting an expletive in existential constructions first pro, an important notion from the Government & Binding era, will be outlined and how this notion has evolved in minimalism. Many of the world’s languages do not exhibit an overt movement of the subject to Spec‐TP to satisfy EPP. Within the Government and Binding era, null‐subject languages were supposed to satisfy the EPP‐ feature covertly (pro) to ensure all languages behave according to EPP. During the minimalist era two theoretical issues came to the fore. First, it was claimed that v‐to‐T movement in full null subject languages suffices to satisfy the requirement that T attract the subject into its local domain (cf. the Extended Projection Principle of Chomsky 1981). This suggests that rich morphology languages can however satisfy EPP without the need for the movement of the argument‐DP to the specifier position. Secondly, the question was raised whether or not the Extended Projection Principle should be considered as a universal principle. If Spec‐TP does not in fact have to be projected in all languages, not just because there is an alternative means of satisfying the EPP, but 38_{From: Biberauer (2008).}

(34)

because this requirement may not even hold in some languages, it is clear that questions must once again be asked about the necessity of postulating an expletive pro. Alexiadou & Anagnostopoulou (1998) subsequently state that the consequence of a specifierless approach to EPP‐satisfaction implies that in null subject languages no longer pro‐ insertion is required. This implies that pro, whether or not it could occur in Germanic languages, is not a possible solution when the dummy subject is omitted. A striking example that pro is not a possible solution when the dummy subject is omitted is undoubtedly Dutch. In Dutch the dummy subject could occur in postfinite position but this seems not to be obligatory.39_{There are contexts where a covert} expletive appears to be in free variation with its overt counterpart in cases the expletive occurs in postverbal position. (37) Ik denk dat een man morgen een huis koopt. I think that a man tomorrow a house buys. “I think that a man is buying a house tomorrow.” (38) Ik denk dat er morgen een man een huis koopt. I think that there tomorrow a man a house buys. “I think that there is a man buying a house tomorrow.” (39) Ik denk dat morgen een man een huis koopt. I think that tomorrow a man a house buys. “I think that there is a man buying a house tomorrow.” In (37) the subject‐DP is raised to the Spec‐TP position and subsequently to SpecTopP. Such sentences, with the subject as topic of the clause comply with EPP as expected. In (38) and (39) however the adverb precedes the unraised subject, which remains in its base position. As (38) shows the adverb follows the expletive and it would indicate that Dutch is also compliant to EPP. However, in (39) the clause is still grammatically well‐ formed without an overt expletive present. Such constructions cast doubt upon the traditional EPP‐theory and the necessity to fill the Spec‐TP position. In any case it must be concluded that Dutch fails to exceptionlessly behave in accordance with the EPP as 39_{In preverbal position the insertion of a dummy subject is nonetheless obligatory.}

(35)

originally formulated in Chomsky (1981, 1982). This leads us to expect that not all languages may be equally in need of expletive elements as Biberauer (2008) claims and this raises the question as to whether it is necessary or not to have an expletive occupy Spec‐TP and secondly how this optionality is syntactically accounted for. 3.4 Licensing of Spec-TP The behaviour of there‐type expletives raises some problems for the traditional EPP‐ theory. Not all members of the Germanic language family behave similarly with respect to complying with EPP requirement on Spec‐TP. Biberauer (2003) argues that Germanic languages exhibit a range of T‐related EPP‐satisfaction strategies. These could be summarized in the following typology.

Language Source of phi-features

(Goal)

EPP Movement

English D(P) in Spec‐vP DP‐to‐Spec‐TP

German Phi‐features on V‐

morphology

vP‐to‐Spec‐TP

Dutch D(P) in Spec‐vP vP‐to‐Spec‐TP

Table 3. Typology of Germanic EPP satisfaction

Following Rizzi (2005) I assume that all clauses must contain a subject and the specifier position of TP is seen as a systematic landing site for movement. This implies that EPP could only be satisfied in Spec‐TP and this requirement could not be satisfied in Spec‐CP. In German and Dutch the expletive in Spec‐CP position is solely introduced in the clause to act as a default filler for Spec‐CP in the absence of a suitable frontable element (Biberauer, 2003) to ensure the preverbal slot of the clause is overtly filled in declarative sentences.40 40_{Since the EPP feature is satis ied in German and Dutch embedded sentences on T, it must be assumed that there is} an uniformed approach for main clauses and also in main clauses EPP is satis ied in the T‐domain.

(36)

In terms of this typology, languages may differ as to the source of the Goal and the following EPP movement.41_{Following Alexiadou & Anagnostopoulou (1998), it is} assumed that the nominal category that agrees with T (the Goal) may be associated with either the Phi‐features of the DP argument in spec‐vP, or (b) the Phi‐ features of the agreement morpheme on the verbal head in rich‐agreement languages. The distinction is determined by the rich morphology of a language. In rich morphology languages such as Spanish or Italian EPP is satisfied solely by v‐to‐T movement. German is similar to these rich morphology languages, with the exception that it is not a pro-drop language and thus the subject‐DP could not be omitted. A second distinction is the EPP movement. Following Biberauer (2003) the Spec‐TP position is not uniquely reserved for subjects, but for checking the EPP feature. In Germanic languages, which are discussed in this thesis, two different strategies are outlined. First, there is the traditional form of the subject‐DP moving to Spec‐TP, which we see in English. Secondly, it is argued by Biberauer (2003) that languages like German and Dutch comply with EPP by moving the complete vP to Spec‐TP. In the following subsections I will discuss these strategies.

3.4.1 Subject raising to Spec-TP

As discussed in section 3.2 the traditional form of the Extended Projection Principle requires movement of the subject‐DP to the specifier position of TP. Following the Attract Closest Principle (Radford, 2006) the phi‐features of the specifier (subject) must agree with the head in T and the external argument in Spec‐vP is raised to the Spec‐TP position in accordance with EPP. In case this external argument remains in situ it was assumed that an overt pleonastic pronoun, according to the Expletive Insertion Rule (Carnie, 2007) must be inserted in Spec‐TP to satisfy the Extended Projection Principle. Since the minimalist era it is stated that a finite T carries an EPP feature which, in conjunction with agreement, drives A‐movement. English is a language which exceptionlessly complies to EPP via subject raising. 41_{The two different strategies – subject raising and vP raising – are discussed from the point of view the subject is} moved from the verbal domain to the left periphery. In case the subject is focused a variation arises in the possibility to insert a dummy subject in the Spec‐TP position, which is the relevant position to satisfy to EPP.

(37)

In English, the subject, if topic of the clause, is raised from its base position in Spec‐vP to Spec‐TP to satisfy the EPP feature and subsequently to the topic dedicated position in the syntactic structure. In case the subject is not raised to Spec‐TP an expletive is inserted in Spec‐TP to comply with EPP and to save the syntactic structure. (40) John does not eat chocolate. (41) There is not a man eating chocolate. As (40) shows the subject precedes negation and is thus raised to Spec‐TP. In case of an existential construction, as illustrated in (41), the EPP‐feature is satisfied by an overt expletive. This dummy subject is required to compensate for the absence of a thematic subject. Figure 8 represents the syntactic structure of this movement operation. CP C’ C TP T’ T vP

Figure 8. Syntactic structure demonstrating movement of the subject from the vP domain to the TP domain. (details of vP domain not represented) (head-initial language).

It is expected that English is almost fully compliant to EPP due to its word order. Because of the SVO word order the subject is recognized by the position in a clause.42 42_{There are two instances when inversion takes place and in which in English a Verb‐Subject word order occurs. In} sentences which start with only or in sentences that begin with a negative phrases the verb precedes the subject.

(38)

This leads to an obligatory overtly realized expletive when the subject is absent (Biberauer, 2003).43 3.4.2 vP raising to Spec-TP A second strategy languages could use to comply with the Extended Projection Principle is via vP raising to Spec‐TP. To satisfy the Extended Projection Principle, according to Biberauer (2003), the complete vP is raised to the specifier position of T.4445_This vP raising strategy gives an explanation why in certain languages no dummy subject could be inserted in the Spec‐TP position. Since the complete vP is raised to Spec‐TP to comply with EPP, this position is no longer available for the insertion of an expletive. A prerequisite for this vP pied‐piping operation is that the language has rich inflectional morphology (Richards & Biberauer, 2005).46_{Rizzi (1986) and Huang (2000) claim a} distinction between English as a non‐null subject language and German as a semi null subject language. In semi‐null languages omission of the expletive in postfinite position is permitted. This phenomenon could be detected in German clauses where an expletive cannot occur in postfinite position. However, in English the insertion of a dummy subject in existential construction is obligatory. The vP raising strategy could be construed as follows. The DP which occurs in the specifier position of vP is probed to the specifier position of TP. The rest of the vP is pied‐piped and subsequently moved together with the DP to Spec‐TP. This implies that in clauses where the subject is also topic of a sentence the complete vP, including the subject, moves to Spec‐TP to comply with EPP. Subsequently, the subject‐DP moves to the dedicated topic position to check the Topic feature and strands the other constituents of the vP. In case the subject is focused, the vP is still moved to Spec‐TP to 43_{For rich morphology languages such as Spanish and Italian this is not applicable, since the subject could be realized} covertly.

44_{This movement occurs for all kinds of subject, regardless whether they are de inite, or de inite, topic or non‐topic.}

45_{The functional verb of the clause is merged into or moved to T (Biberauer, 2003).} 46_{This vP pied‐piping operation should not be considered as an alternative for pro‐drop languages. Within} minimalism, it was claimed that v‐to‐T movement in full null subject languages suf ices to satisfy the requirement that T attract the subject into its local domain (cf. the Extended Projection Principle of Chomsky 1981). This infers that rich morphology languages can however satisfy EPP without the need for the movement of the argument‐DP to the speci ier position.

(39)

comply with EPP. A subsequent movement to the topic position does not follow. This analysis explains why a dummy subject could not be inserted into Spec‐TP, since the position is already filled. Figure 9 shows the syntactic representation of the vP raising strategy. CP C’ C TP dass Spec T’ vP T hat Spec v’ Peter v VP Spec V’ Dem Mann DP V das Auto geschenkt

Figure 9. Syntactic structure representing vP raising to Spec-TP (head final language)

Biberauer (2003) proposes that such a Comp‐to‐Spec movement is permitted since only an element within the complement is targeted, rather than the complement itself. The

Optionality in complying with EPP in Dutch: a quest for a syntactic justification