c Biolinguistica Fennica Working Papers 1 (2009) 21–78. Author(s) Extraction islands in Finnish SAARA HUHMARNIEMI University of Helsinki This paper provides an overview of syntactic extraction islands in Finnish, with an emphasis on wh-movement. Several island types are covered: adjunct islands, wh-islands, subject islands, and left branch condition on subject extraction. Two main conclusions emerge: (i) Extraction conditions may be used as a diagnostics for determining the structural position of the phrase at hand. This is the case with e.g. Finnish prepositional phrases and certain infinitival clauses which may be ambiguous between complement and adjunct positions. (ii) Finnish shows (anti-)agreement effect in all subject extraction contexts: if a DP agrees with the syntactic head and undergoes EPP movement to its specifier, then the DP cannot be extracted. Evidence from infinitival clauses, extraction of DPs out of prepositional phrases, possessor extraction, and subject extraction out of finite clauses is examined in order to establish (ii). It follows that the left branch extraction out of DPs, PPs, and infinitival clauses in Finnish parallels subject extraction in sentential domain. 1. INTRODUCTION This paper investigates phrasal movement in Finnish. The main focus is on the wh-movement in content questions such as (1), in which the direct object mitä ’what’ has left its base position inside the verb phrase and landed to the front of the sentence. (1) a. Mitäi Pekka osti ti ? what.PAR Pekka bought ’What did Pekka buy?’ Some syntactic contexts do not allow such fronting. Consider for example the determiner phrase (DP) in (2a), which contains a genitive possessor argument Merjan ’Merja’s’. The genitive argument cannot be moved to the front of the sentence without moving the whole DP (b-c). The DP thus constitutes an extraction island for the genitive argument. (2) a. Pekka lainasi [DP Merjan kirjaai ] Pekka borrowed Merja.GEN book.PAR ’Pekka borrowed Merja’s book.’ 22 SAARA HUHMARNIEMI b. *Kenen Pekka lainasi [DP ti kirjaai ]? who.GEN Pekka borrowed book.PAR c. [DP Kenen kirjaai ] Pekka lainasi ti ? who.GEN book.PAR Pekka borrowed ’Whose book did Pekka borrow?’ In Finnish, the unavailability of extraction of the genitive argument is independent of the structural properties of its host DP and its position in syntactic structure. We say that DPs in Finnish are strong islands for the extraction of genitive arguments. On the other hand, extraction of complement PPs out of DPs is regulated by the syntactic (and semantic) properties of the DP. We say that DPs are weak islands regarding the extraction of PPs. Another example of a strong island in Finnish is any phrase that occupies an adjunct position in the syntactic structure. Whereas extraction out of complement positions is available to a varying extent, extraction out of adjuncts is not allowed in Finnish. The aim of this paper is to provide a survey of the extraction conditions in Finnish, with special attention paid to the structural position of the phrase at hand. Each phrase type is considered in its own subsection. A special emphasis is placed on the extraction conditions for subjects and genitive arguments. In section 3, I suggest that subject extraction is conditioned by (anti-)agreement, specifically φ-feature (person, number) agreement, following the anti-agreement hypothesis by Ouhalla (1993). In section 4 I briefly examine the extraction out of subject arguments in Finnish. Before continuing with the extraction conditions, section 1.1 introduces the basic syntactic structure for a Finnish finite clause assumed in this paper. Section 1.2 provides an overview of the extraction islands that are presented here. Finnish is a nominative-accusative language with fifteen case suffixes, thus the subjects of finite clauses are normally in the nominative and the objects of transitive verbs in the partitive or accusative, both which are object cases in Finnish. In addition, there are several variations of the same accusative suffix (the n-accusative, zero-accusative and the taccusative). The partitive has been argued to constitute a general complement case in Finnish (Vainikka, 2003). The genitive is used in a number of prehead positions (Vainikka, 1989). See Nelson (1998) for a general account of Finnish case. EXTRACTION ISLANDS IN FINNISH 23 1.1 Finite clause in Finnish The derivation of a transitive finite clause, such as (3), begins by forming a verb phrase that contains the predicate and its arguments. In the structure (4a), the verb introduces the object argument to its complement and the transitivizer head v introduces the external subject argument. Sentential adjuncts (temporal and manner adverbials) are in this paper assumed to be adjoined on the top of the vP , although nothing hinges on this assumption.1 In (4b), the sentence subject has moved to the specifier of T(ense). I assume that the movement of the subject is triggered by EPP (Empty Projection Principle, see for Finnish e.g. Holmberg & Nikanne 2002; Brattico & Huhmarniemi 2006). The subject receives nominative case and T agrees with the φ-features (person and number) of the moved DP.2 Nominal subject is said to undergo A(rgument)-movement to the specifier of T. The object argument does not undergo A-movement but stays in its base position and receives partitive or accusative case from the finite transitive verb. Finally, the finite verb undergoes head movement via the little verb v-projection to T. (3) Pekka luki eilen kirjan. Pekka.NOM read yesterday book.ACC ’Pekka read a book yesterday.’ (4) a. vP vP AdvP eilen ’yesterday’ Pekka vP VP v+luki ’read’ tV kirjan+ACC ’book’ 24 SAARA HUHMARNIEMI b. CP C TP Pekka+NOM TP T+φ vP v+luki ’read’ AdvP eilen ’yesterday’ t vP vP kirjan ‘book’ Whereas A-movement is associated with Case and φ-features, A-bar ¯ movement (A-movement, indicating the complement of A-movement) is triggered by discourse properties of the sentence, i.e. interrogative force or information content of the expression. For example, Finnish content questions are formed by fronting one wh-expression; in example (5a) the object argument is moved to the front; in (b), the subject argument and in (c), an adjunct. The A-bar movement of the wh-expression thus contributes to the formation of content questions in Finnish. (5) a. Minkäi Pekka luki ti ? what.ACC Pekka read ’What did Pekka read?’ (ti ) luki kirjan? b. Kukai who.NOM read book.ACC ’Who read a/the book?’ kirjan ? c. Milloini Pekka luki ti sen when Pekka read that.ACC book.ACC ’When did Pekka read the book?’ In Finnish, wh-movement is obligatory in content questions; whenever the wh-expression is left in-situ or some scrambled position other than the beginning of the sentence, it forms an echo question (6ab). Echo-questions differ from content question in both syntax and semantics (see for Finnish Hakulinen et al. 2004, §1211). (6) a. Pekka luki minkä? (echo) Pekka read what.ACC ’Pekka read what?’ EXTRACTION ISLANDS IN FINNISH 25 b. Pekka luki sen kirjan milloin? Pekka read that.ACC book.ACC when ’Pekka read the book when?’ In the Minimalist program fronting is implemented as a Probe-Goal relation between two elements in the structure (Chomsky, 2000, 2001, 2008). In interrogative sentences, the finite C acts as the Probe for a whfeature, and the wh-phrase that holds the wh-feature is called the Goal. The Probe searches for a matching Goal in its c-commanding domain. When the Probe-Goal relation is successfully formed, the Probe feature at C triggers movement of the wh-expression to the specifier of C.3 The A-bar movement of direct object is illustrated in (8). (7) Minkäi Pekka luki ti ? what.ACC Pekka read ’What did Pekka read?’ (8) CP Minkä CP ’what+ACC’ C+wh TP Pekkai TP T v+luki ’read’ ti vP vP t Thematic properties and structural case of the wh-phrase provide information on the position of the wh-phrase prior A-bar movement. For example, the DP minkä ’what’ in (7) is the thematic object of the predicate luki ’read’ and has accusative case, which is assigned inside the verb phrase. Another piece of support for the existence of A-bar movement comes from the referring properties of reflexive anaphors. In Finnish, reflexive pronoun has to be bound by an antecedent in a structurally higher, c-commanding position (Vainikka, 1989; van Steenbergen, 1991; Kaiser, 2003). For example, the reflexive pronoun may not be base-generated in a structurally higher position than its antecedent (9a-b). However, in sentence (c), the reflexive object has undergone topicalization movement and is still able to refer to sentence subject. 26 SAARA HUHMARNIEMI The reference of the reflexive anaphor is thus independent on A-bar movement. In other words, reflexives reconstruct for A-bar movement in Finnish. (9) a. Pekka rakastaa itseään. Pekka loves self.PAR.Px/3SG ’Pekka loves himself.’ b. *Itse(-nsä) rakastaa Pekkaa. self(-Px/3SG) loves Pekka.PAR c. Itseään Pekka rakastaa ti . self.PAR.Px/3SG Pekka loves ’Pekka loves himself.’ Finnish possessive suffix has similar properties as reflexive anaphors: it has to be bound by a DP in a local configuration similar to reflexives (10a-b) (Kanerva, 1987; Vainikka, 1989; Nelson, 1998; Trosterud, 1993). Possessive suffixes reconstruct for A-bar movement as well (c). (10) a. Pekka möi pyöränsä. Pekka sold big.ACC.Px/3SG out ’Pekka sold his bike.’ b. *Pyöränsä kaatui. bike.NOM.Px/3SG fell Pekka möi ti . c. Pyöränsäi bike.ACC.Px/3SG Pekka sold ’Pekka sold his bike.’ A-bar movement thus targets phrases that already have a thematic role and case assigned. In addition, reflexive pronouns and possessive suffixes reconstruct for A-bar movement in Finnish. 1.2 Finnish extraction islands This paper provides an overview of the basic extraction island types in Finnish: adjunct islands, wh-islands, subject islands, and left branch condition on extraction of subjects. Wh-islands cover both phrases headed by a wh-phrase and the unavailability of extraction out of relative clauses. Much of the terminology and a common ground for the island research in generative grammar was established by Ross (1967). First, let us demonstrate the differences between extraction conditions on arguments and adjuncts in MA-infinitival clauses (11)-(12). Whereas the MA-infinitive in inessive case allows extraction (11b), MA-infinitive in adessive constitutes an extraction island (12b). EXTRACTION ISLANDS IN FINNISH 27 (11) a. Pekka näki Merjan [ kirjoittamassa runoja ] Pekka saw Merja.ACC write.MA.INE poems.PAR ’Pekka saw Merja writing poems.’ b. Mitäi Pekka näki Merjan [ kirjoittamassa ti ]? what.PAR Pekka saw Merja.ACC write.MA.INE ’What did Pekka see Merja writing?’ (12) a. Pekka yllätti Merjan [ kirjoittamalla runoja ] Pekka surprised Merja.ACC write.MA.ADE poems.PAR ’Pekka surprised Merja by writing poems.’ Pekka yllätti Merjan [kirjoittamalla ti ] b. *Mitäi what.PAR Pekka surprised Merja.ACC write.MA.ADE The main difference between the two infinitival clauses (apart from the lexical case) is that they are located in different structural positions. Whereas the MA-infinitive in inessive case is an argument of the verb and occupies a complement position, the MA-infinitive in adessive is a manner adverbial and occupies an adjunct position (Vainikka, 1989; Koskinen, 1997). The structure (13a) illustrates the complement position of the MA-infinitive and (b), the adjunct position. Both figures assume a right-branching head-complement order. (13) (a) Extraction out of complement CP (b) Extraction out of adjunct CP CP C CP ... vP v C ... MA.INE VP V kirjoittamassa DP runoja ... vP MA.ADE VP × V kirjoittamalla DP runoja Similarly as adjuncts, extraction out of phrases that occupy a subject position is generally not available. For example, extraction out of the finite clause subject position is not available in (14a), although extraction of the same element is available out of the object position (b): (14) tutkia ti ] sattuu harvoin a. *Mitäi [ tilaisuus what opportunity study happens rarely omalle kohdalle? to oneself ... 28 SAARA HUHMARNIEMI b. Mitäi Pekka sai [DP tilaisuuden tutkia ti ]? What Pekka got opportunity study ‘What Pekka got the opportunity to study?’ When a phrase is headed by a wh-word (15a), extraction out of the phrase is not permitted (b-c). The phrase forms a wh-island. The whislands have been accounted for by assuming a Superiority condition, which states that a search for a wh-phrase cannot cross another whphrase (Kuno & Robinson, 1972; Chomsky, 1973). In case of example (15b), Probe C sees the wh-expression on the top of the complement clause, which is a closer potential goal for movement, and thus the search does not proceed inside the complement clause. Finnish wh-islands are briefly introduced in section 2.1. (15) a. Pekka mietti [ (että) kuka oli lainannut lehteä]. Pekka wondered (that) who had borrowed newspaper ’Pekka wondered who had borrowed the newspaper.’ Pekka mietti [ (että) kuka oli lainannut ti ]? b. *Mitäi what.PAR Pekka wondered (that) who had borrowed c. Mitäi Pekka luuli [ (että) Merja oli lainannut ti ]? what Pekka thought (that) who had borrowed ’What did Pekka think (that) Merja had borrowed?’ Finally, the genitive arguments in DPs (2) and (16a-b) provide examples of the left branch condition (LBC Ross 1967, 114), which states that a DP which is the leftmost constituent of a larger DP cannot be moved out of that DP.4 I will consider extraction of the leftmost DP constituent in DPs, PPs, infinitival clauses and finite clauses and suggest that whenever a genitive DP enters into a configuration with a head that agrees with the φ-features of the DP, the DP is unavailable for extraction. (16) a. Pekka luki [DP hänen kirjansa] Pekka read (s)he.GEN book.ACC.Px/3SG ’Pekka read his/her book.’ b. *Häneni Pekka luki ti kirjansa. (s)he.GEN Pekka read book.ACC.Px/3SG 2. EXTRACTION CONDITIONS ON PHRASES Here we consider wh-movement; other types of A-bar movement include at least relative pronoun movement, phrasal movement in yes/no questions, and movement to topic and focus positions. The extraction conditions for wh-movement nevertheless hold to a large extend for the other A-bar movement types as well. EXTRACTION ISLANDS IN FINNISH 29 2.1 CPs Extraction out of finite complement clauses is often called “long-distance movement”, since the wh-phrase passes a C-domain of the complement clause, a possible landing site, on its way to the front of the matrix clause. Long-distance movement is available from the finite clause complement object and adjunct positions, as illustrated in (17a-b). The presence of overt complementizer että is obligatory in long-distance movement in Finnish (c). (17) a. Keneti Pekka luuli että Merja oli tavannut ti ? who.ACC Pekka thought that Merja had met ’Who did Pekka think Merja had met?’ b. [ Mihin tarkoitukseen ]i Pekka luuli että Merja what.ILL purpose.ILL Pekka thought that Merja osti lapion ti ? bought shovel.ACC ‘For which purpose Pekka thought Merja bought the shovel?’ Merja oli tavannut ti c. *Keneti Pekka luuli who.ACC Pekka thought Merja had met Long-distance movement of nominative subject is not available (18). (18) että ti tapasi Merjan? *Kukai Pekka luuli Who Pekka thought that met Merja ’*Who did Pekka think met Merja?’ Extraction out of the subject position is marginally available in raising constructions, where the subject is in genitive case and there is no φfeature agreement between the main verb and the subject, such as in (19). In addition, the subject extraction out of the nesessive construction in (20) is considered as more grammatical than extraction of the nominative subject in (18). (19) a. Minun kannattaa/*kannatan jäädä odottamaan I.GEN should.3SG/should.1SG stay.A wait.MA.ILL ’I should stay waiting.’ että ti kannattaisi jäädä odottamaan b. ??Keneni luulet who.GEN think.2SG that should.3SG stay.A wait.MA.ILL vuoroaan? turn.Px/3SG ’Who do you think should stay and wait for his turn?’ (20) a. Minun täytyy/*täydyn lähteä matkalle. I.GEN must.3SG/must.1SG leave trip.ILL 30 SAARA HUHMARNIEMI ’I have to go for a trip.’ b. *?Keneni Pekka kertoi että ti täytyy lähteä matkalle? who.GEN Pekka told that must.3SG leave.3SG trip.ILL ’About whom did Pekka tell that he has to go for a trip?’ Passive complement clauses allow extraction of the partitive object DP (21a-b). The subject position in the passive sentence may be left empty, as in (c). In addition, extraction of the zero-accusative object is more restricted (d). (21) a. Pekka kertoi että tontille rakennetaan Pekka told that site.ALL build.PASS talo/taloa. house.ACC(∅)/house.PAR ’Pekka told that they are building a house on the site.’ b. Mitäi Pekka kertoi että tontille rakennetaan ti ? what.PAR Pekka told that site.ALL build.PASS ’What did Pekka tell they are building on the site?’ Pekka pyysi että jätetään tänne ti ? c. Mitäi what.PAR Pekka ask that leave.PASS here ’What did Pekka ask that we should leave here?’ d. *?Mikäi Pekka kertoi että tontille rakennetaan ti ? what.ACC(∅) Pekka told that site.ALL build.PASS Extraction out of a finite clause is not permitted whenever the clause occupies an adjunct position (22a-b). Likewise, finite clause in subject position constitutes an extraction island (23a-b). (22) a. Pekka toi silmälasit, jotta Merja voisi lukea lehteä. Pekka brought glasses for Merja could read newspaper ’Pekka brought glasses so that Merja could read the newspaper.’ silmälasit, jotta Merja voisi lukea ti . b. *Mitäi Pekka toi what Pekka brought glasses for Merja could read (23) a. [ (Se,) että Pekka osti lehden] harmitti Merjaa. (it.NOM) that Pekka bought newspaper annoyed Merja ’That Pekka bought a newspaper annoyed Merja.’ että Pekka osti ti ] harmitti Merjaa. b. *Minkäi [ (Se,) what it.NOM that Pekka bought annoyed Merja The CP forms a strong island if it is headed by a wh-phrase, such as the wh-subject who ’kuka’ in examples (24a-b). The fronted wh-phrase thus transforms the complement clause to a wh-island. EXTRACTION ISLANDS IN FINNISH (24) 31 a. Pekka mietti [ (että) kuka oli lainannut sanomalehteä] Pekka wondered (that) who had borrowed newspaper ’Pekka wondered who had borrowed the newspaper.’ [ (että) kuka oli lainannut ti ]? b. *Mitäi Pekka mietti what Pekka wondered (that) who had borrowed Also other A-bar movement types that target the C-domain may turn the CP into an island. First, movement of the relative pronoun to the specifier of CP turns the CP into a wh-island (25a-b). In addition, the sentential focus particles that require movement to the front of the sentence (-hAn, -pA, and the question particle -kO) prevent whextraction (26). (25) a. Pekka näki miehen [ joka lainasi sanomalehteä] Pekka saw man who borrowed newspaper ’Pekka saw a man that borrowed the newspaper.’ b. *Mitäi Pekka näki miehen [ joka lainasi ti ]? what Pekka saw man who borrowed (26) a. Juha kertoi [ että Merjalle-han Pekka lainasi kirjan]. Juha told that Merja.ILL-han Pekka borrowed book ’Juha told that it was Merja, to whom Pekka had borrowed the book.’ b. *Mitäi Juha kertoi [ että Merjalle-han Pekka lainasi ti ]? what Juha told that Merja.ILL-han Pekka borrowed The Superiority condition on wh-movement (Kuno & Robinson, 1972; Chomsky, 1973) does not directly account for the wh-island phenomenon observed in relative clauses and sentences containing sentential focus particles in Finnish. The Relativized Minimality principle by Rizzi (1990) provides a more general account for the ban on wh-movement in the above examples. According to Relativized Minimality, it suffices for the constituent Z to prevent the P-G relation between X and Y in configuration (27) if Z has certain characteristics in common with X. In Finnish, the intervening constituent Z could thus be a relative pronoun or a constituent hosting a sentential focus particle. In fact, any constituent that has undergone A-bar movement to the C-domain suffices to cause the intervention effect on wh-movement.5 (27) ... X ... Z ... Y ... To summarize, extraction out of finite clause complement is available apart from the nominative subject. When the finite clause occupies an adjunct or subject position it forms a strong island. Finally, wh-islands may be introduced by wh-phrases, relative pronouns, and constituents hosting a sentential clitic particle. 32 SAARA HUHMARNIEMI 2.2 PPs Finnish prepositions may be divided in two classes according to their extraction properties. Prepositions that take a genitive case complement are strong islands for movement. On the other hand, prepositions that take a partitive case complement allow extraction if the PP occupies a complement position. In this section, I briefly introduce the basic syntactic properties of the two types of PPs, concentrating on the properties of PPs that take partitive case complement. 2.2.1 PPs with partitive case complement The prepositions that take partitive case complement show free word order inside the PP (28a-c). In addition, they never take possessive suffix (Vainikka, 1989, 1993; Manninen, 2003a).6 (28) a. kohti taloa / taloa kohti towards house.PAR house.PAR towards ’towards a/the house’ b. ilman taloa / taloa ilman without house.PAR house.PAR without ’without a/the house’ c. ennen taloa / taloa ennen before house.PAR house.PAR before ’before a/the house’ Manninen (2003b) assumes that Finnish PPs such as those above are VP-adverbials, with a low position inside the VP. The extraction data suggests that PPs may be located in a complement position as well. The examples (29)-(30) illustrate extraction out of a PP located in the direct complement of the verb.7 Furthermore, the presence of a direct object does not always block extraction, as in (31a-b). (29) a. Pekka ei pärjää ilman televisiota. Pekka not.3SG survive without TV.PAR ’Pekka doesn’t survive without a TV.’ Pekka ei pärjää ilman ti ? b. Mitäi what.PAR Pekka not.3SG survive without ’What doesn’t Pekka survive without?’ (30) a. Pekka juoksi kohti puistoa. Pekka ran towards park.PAR ’Pekka ran towards the park.’ EXTRACTION ISLANDS IN FINNISH 33 b. Mitäi Pekka juoksi (suoraan) kohti ti ? what.PAR Pekka ran (straight) towards ’(Straight) towards what did Pekka run?’ (31) a. Pekka jätti Merjan [P P ilman sadetakkia] Pekka left Merja.ACC without raincoat.PAR ’Pekka left Merja without a/the raincoat.’ Pekka jätti Merjan [P P ilman ti ]? b. Mitäi what.PAR Pekka left Merja.ACC without lit. ’What did Pekka leave Merja without?’ To illustrate further the availability of extraction, the PP may be located in the complement of an infinitival clause, such as A-infinitive in sentence (32a), and extraction remains to be available (b). (32) a. Pekka ei anna [ Merjan lähteä [P P ilman Pekka not let Merja.GEN leave.A without sadetakkia]] raincoat.PAR ’Pekka doesn’t let Merja to leave without a/the raincoat.’ b. Mitäi Pekkai ei anna [ Merjan lähteä [P P what.PAR Pekka not let Merja.GEN leave.A ilman ti ]]? without ’What doesn’t Pekka let Merja to leave without?’ I would like to propose that the PP is base-generated to a complement position in examples (29)-(32). Let us compare manner adverbials in (33a-b) to PPs in (34a-b). In manner adverbials, the possessive suffix is not able to refer to the direct object from inside the adverbial phrase. Manner adverbials are thus located above the direct object in Finnish (e.g. on the top of the vP , as illustrated in (4)). However, the possessive suffix is able to refer to the direct object from inside the PP in the examples (34a-b). This indicates that PPs may occur in a lower position than the direct object. (33) autollaani/∗j a. Pekkai löysi Merjanj Pekka found Merja.ACC car.ADE.Px/3SG ’Pekka found Merja with his car.’ kiikareillaani/∗j b. Pekkai näki Merjanj Pekka saw Merja.ACC binoculars.ADE.Px/3SG ’Pekka saw Merja with his binoculars.’ (34) a. Pekkai löysi Merjanj [P P ilman autoaani/j ] Pekka found Merja.ACC without car.PAR.Px/3SG 34 SAARA HUHMARNIEMI ’Pekka found Merja without his/her car.’ b. Pekkai vei Merjaaj [P P kohti autoaani/j ] Pekka took Merja.PAR towards car.PAR.Px/3SG ’Pekka took Merja towards his/her car.’ The proposed configuration where the PP is located to a lower position than the object is illustrated in (35). (35) vP DP Pekka vP VP v+löysi tV VP DP Merjan PP P ilman DP autoa-an However, PPs that take partitive case complements may be ambiguous regarding their structural position and occupy a higher position as well. For example, when the PP is semantically a manner adverbial, as in (36a-b), extraction is not allowed. Binding data suggests the adjunct interpretation as well (c).8 (36) a. Pekkai katselee elokuvia [P P ilman silmälaseja] Pekka watches movies without glasses.PAR ’Pekka watches movies without his glasses.’ b. *Mitäi Pekka katselee [P P ilman ti ]? what.PAR Pekka watches without c. Pekkai katseli Merjaaj [P P ilman without Pekka watched Merja.PAR silmälaseja-ani/∗?j ] glasses.PAR ’Pekka watched Merja without his glasses’ In addition, some partitive-case assigning prepositions, such as ennen ’before’ do not allow extraction at all (37a-b). The example from reflexive binding in (38) suggests that the PP is in fact located on the top of the direct object, and thus does not occupy the complement position. EXTRACTION ISLANDS IN FINNISH (37) a. Pekka pysähtyi [P P ennen puistoa] Pekka stopped before park.PAR ’Pekka stopped before the park.’ b. *Mitäi Pekka pysähtyi [P P ennen ti ]? what.PAR Pekka stopped before (38) Pekkai ohitti Merjanj [P P ennen autoaani/∗j ] Pekka passed Merja.ACC before car.PAR.Px/3SG ’Pekka passed Merja before his car.’ 35 Regardless of the islandhood of the PP, wh-questions may be formed by pied-piping the PP together with the wh-constituent to the front of the sentence: (39) kohti]i Pekka käveli ti ? a. [P P Mitä what.PAR towards Pekka walked t lit. ’Towards what did Pekka walk?’ To summarize, extraction of a partitive DP is available when the PP is located in a structural complement position. In addition, I suggested that the PPs may be ambiguous regarding their structural position, and that the position is sensitive to meaning. 2.2.2 PPs with genitive case complement Prepositions that assign genitive case have somewhat different structural properties than prepositions that take partitive case complements. The preposition may bear a possessive suffix and the word order is more restricted (40a-b). In addition, the possessive suffix may be licensed by a c-commanding antecedent, such as the sentence subject in (c).9 (40) a. nuotion lähellä / *lähellä nuotion fireplace.GEN near / near fireplace.GEN ’near the fireplace’ / minun lähelläni b. hänen lähellään (s)he.GEN near.Px/3SG I.GEN near.Px/3SG ’near him’ / ’near me’ [P P lähellääni/∗j ] c. Pekkai näki Merjanj near.Px/3SG Pekka saw Merja.ACC Pekka saw Merja near him. I follow Nikanne (1989) and Manninen (2003b) and assume that Finnish PPs are built so that the DP is always the right-branching complement of the P head. According to Manninen, the genitive DP 36 SAARA HUHMARNIEMI undergoes A-movement to the specifier of P, and the movement is triggered by the EPP-feature on P. I return to the EPP-movement in section 3. Prepositions that assign genitive case are strong islands for extraction (41)-(42). (41) a. Pekka istui [P P nuotion lähellä] Pekka sat fireplace.GEN near ’Pekka sat near a/the fireplace.’ b. *Minkäi Pekka istui [P P ti lähellä] what.GEN Pekka sat near (42) a. Pekka kulki [P P talon ohi] Pekka walked house.GEN past ’Pekka walked past a/the house.’ b. *Minkäi Pekka kulki [P P ti ohi] What.GEN Pekka walked past Nevertheless, wh-questions may be formed by pied-piping the whole PP to the front of the sentence along with the wh-constituent (43): (43) a. [P P Minkä lähellä]i Pekka istui ti ? what.GEN near Pekka sat t ’Near what did Pekka sit?’ b. [P P Minkä ohi]i Pekka käveli ti ? what.GEN past Pekka walked t ’Past what did Pekka walk?’ The unavailability of extraction of genitive arguments is discussed in section 3. 2.3 DPs The basic structure of Finnish DPs consist of a nominal element N0 and optional pre-nominal modifiers: determiner or demonstrative pronoun (D), quantificational expression, numeral, adjectival phrases (AP), and genitive modifiers (DP) (44a-b), see e.g. Brattico (2008, 2009b) for details. (44) a. ne kolme Pekan tärkeää oivallusta those three Pekka.GEN important.PAR insights.PAR ’those three important insights by Pekka’ b. ne kaikki Pekan tärkeät oivallukset those all Pekka.GEN important.PL insights ’all those important insights of Pekka’ EXTRACTION ISLANDS IN FINNISH 37 DPs may contain up to two pre-nominal genitive modifiers, which are conceived as possessives (45a) or thematic arguments of the head noun (b). (45) a. Pekan koira Pekka.GEN dog ’Pekka’s dog’ b. isän auton ostaminen father.GEN car.GEN buying.NOM ’buying of a car by the father’ In addition, DPs may have different post-nominal modifiers: relative clauses and A-infinitives, which always follow the nominal element (46ab); PPs (47a-c) and MA-infinitives (48a), which usually follow the nominal element but may occur pre-nominally. Finally, infinitival adjunct clauses may modify nominal heads with event structure (48b). (46) a. oivallus, [CP joka auttoi häntä] insight which.NOM helped he.PAR ’an insight, which helped him.’ b. päätös [V P auttaa Merjaa] decisions help.A Merja.PAR ’decision to help Merja.’ (47) a. suuntaaminen [P P kohti tulevaisuutta] heading towards future.PAR ’heading towards the future’ b. pala [P P leivästä] piece bread.ELA ’piece of bread’ c. illallinen [P P kahvilassa] dinner cafe.ILL ’dinner at the cafe’ (48) a. lähteminen [V P auttamaan Merjaa] leaving help.MA.ILL Merja.PAR ’leaving to help Merja.’ b. Pekan kaatuminen [AdvP auttaessaan help.ESSA.PX/3SG Pekka.GEN falling.NOM Merjaa] Merja.PAR ’Pekka’s falling when helping Merja’ As examples (47b-c) already indicate, DPs in lexical case are here analyzed as prepositional phrases. I thus adopt the proposal by Nikanne 38 SAARA HUHMARNIEMI (1989, 1993), that Finnish locative case markers are empty prepositions taking complements of various categories, NP (DP in this work), PP, AP or VP. This view is also adopted by Vainikka (1989) and Manninen (2003a). Example (49) illustrates these assumptions. (49) Hän pysyi saunassa he.NOM stayed sauna.INE ‘She/he remained in the sauna.’ PP P DP INE sauna ’in’ DPs are strong islands for extraction of pre-nominal modifiers. Example (50) illustrates the unavailability of extraction of the genitive argument (b), quantifier (c), determiner (d), and AP (e). (50) De a. Minä luin sen Hotakaisen uusimman I.NOM read that.ACC Hotakainen.GEN newest.ACC kirjan. book.ACC ’I read the latest book from Dawkins.’ b. *Keneni sinä luit ti kirjan? who.GEN you.NOM read book.ACC c. *Montako sinä luit ti kirjaa? how many you.NOM read book.PAR sinä luit ti kirjan? d. *Minkäi which.GEN you.NOM read book.ACC sinä luit ti kirjan? e. *Millaiseni what kind of.ACC you.NOM read book.ACC Determiners and demonstratives, quantifiers, and numerals are syntactic heads and form syntactic projections of their own, so they do not undergo phrasal movement and are out of the scope of this paper. I return to the unavailability of extraction of genitive arguments in section 3. Finnish DPs allow a limited extraction from post-nominal positions, such as elative and partitive complements of certain nominals (51ab) (so called kasa-constructions by Brattico 2008), PP-modifiers of nominals with verbal root (54), the object argument and adjuncts of the A-infinitival complement (53) and the MA-infinitival modifiers of nominals with verbal root (54).10 EXTRACTION ISLANDS IN FINNISH (51) a. Kenestä Pekka otti [ kuvan ti ]? who.ELA Pekka took picture.ACC ’Who did Pekka take a picture of?’ b. Mitäi Pekka lapioi [ kasan ti ]? what.PAR Pekka shovelled stack ’What did Pekka shovel a stack of?’ (52) Mihini Pekka suositteli [ kirjoittamista ti ]? where.ILL Pekka recommended writing.PAR ’Where did Pekka recommend writing to?’ (53) Merja sai [ luvan ostaa ti ]? Mitäi what.PAR Merja got permission.ACC buy.A ’What did Merja get a permission to buy?’ (54) [ Juhan [ Ketä auttamaan]i Merja esti who.PAR help.MA.ILL Merja prevented Juha.GEN lähtemisen ti ]? leaving.ACC ’To help whom did Merja prevent Juha from leaving?’ 39 Before continuing, it should be noted that DPs with lexical case do not allow extraction. For example, extraction out of the A-infinitival complement is not possible (55a-b), whereas it is acceptable when the DP hosting the infinitival clause is in accusative, as in example (56). (55) a. Pekka kertoi [DP suunnitelmasta kirjoittaa kirjaa] Pekka told plan.ELA write.A book.PAR ’Pekka told us about his plan to write a book.’ b. *?Mitä Pekka kertoi [DP suunnitelmasta kirjoittaa ti ]? what.PAR Pekka told plan.ELA write.A (56) a. Pekka sai [DP tilaisuuden kirjoittaa kirjaa] Pekka got opportunity.ACC write.A book.PAR ’Pekka got an opportunity to write a book’ b. Mitäi Pekka sai [DP tilaisuuden kirjoittaa ti ]? what.PAR Pekka got opportunity.ACC write.A ’What did Pekka receive an opportunity to write?’ Thus, although the DP in elative case seems to occupy the complement position in (55), extraction is impossible. Since there seems to be no other structural difference between the two sentences (55a) and (56b), it is possible that the existence of the prepositional head (illustrated in (49)) causes the DP to form an extraction island. I leave the question aside in this paper and concentrate on extraction out of DPs in structural object cases: partitive and accusative. 40 SAARA HUHMARNIEMI PP modifiers expressing location are discussed in section 2.3.1 and partitive and elative PPs in section 2.3.2. Extraction out of MA-infinitives is briefly introduced in section 2.3.3. Finally extraction out of Ainfinitival complements in discussed in section 2.3.4. 2.3.1 Locative PPs I suggest next that extraction of locative PPs out of DPs is available for nominals with event structure. Consider first examples (57); whereas it is not possible to extract a PP-modifier out of a DP headed by kahvila, ’cafe’ (a-b), the nominal head yöpyminen, ’sleeping’ allows extraction (c). (57) a. Asko suositteli erästä kahvilaa tuon Asko recommended certain.PAR cafe.PAR that.GEN sillan lähellä. bridge.GEN near ’Asko recommended a cafe near that bridge.’ b. *[ Minkä sillan lähellä] Asko suositteli which.GEN bridge.GEN near Asko recommended kahvilaa ti ? cafe.PAR c. [ Minkä sillan lähellä] Asko suositteli which.GEN bridge.GEN near Asko recommended yöpymistä ti ? sleeping.PAR ’Near which bridge did Asko recommend spending the night?’ The main difference between the nominals in the examples (57a-b) is that the noun kahvila ’cafe’ in (a) does not have verbal root, but the noun yöpyminen ’sleeping’ in (b) has. According to Hakulinen et al. (2004, §587) , Finnish nominals with a verbal root differ from other nominals in that the noun phrase seems to contain an extra pre-nominal position for an adverbial phrase (e.g. a PP). In Finnish, DP-internal PP-modifiers are generally post-nominal (58a-b). However, in (58c), where the noun head has a verbal root, the PP Demarissa ’in Demari’ may occupy a prenominal position as well, whereas the noun juttu ’story’ does not allow the pre-nominal PPs (d). (58) a. kirjoittaminen Demarissa writing.NOM Demari.INE ’writing in the Demari (newspaper)’ EXTRACTION ISLANDS IN FINNISH 41 b. juttu Demarissa story.NOM Demari.INE ’story in the Demari’ c. [ Demarissa kirjoittaminen] on tuonut minulle uuden Demari.INE writing.NOM has brought me new näkökulman politiikkaan. viewpoint to politics ’Writing in Demari has given me a new viewpoint for politics.’ d. *Demarissa juttu Demari.INE story.NOM According to Cinque (1980) (and many researchers thereafter, see e.g. Szabolcsi (1994) for Hungarian), extraction out of DP proceeds through an intermediate landing site at the specifier of DP. Lack of the intermediate landing site would thus explain the ungrammaticality of (57b). In section 2.2, I showed that preposition textit kohti ’towards’ allows extraction of its DP complement. When the PP modifies a noun phrase, the PP may be moved as a whole, as in (59b). Nevertheless, extraction of the DP-complement is not possible (c). This suggests that the PP cannot be located in the complement of the nominal, but instead, is an adjunct. The distribution of PPs inside DPs would thus differ from finite clauses. (59) [ vaeltamista [ kohti tuota a. Asko suositteli Asko recommended hiking.PAR towards that.PAR tunturia]] mountain.PAR ’Asko recommended hiking towards that mountain.’ b. [ Mitä tunturia kohti]i Asko suositteli which.PAR mountain.PAR towards Asko recommended [vaeltamista ti ] hiking.PAR ’Towards which mountain did Asko recommend hiking?’ c. *[ Mitä tunturia]i Asko suositteli which.PAR mountain.PAR Asko recommended [vaeltamista [ kohti ti ]] hiking.PAR towards ’Towards which mountain did Asko recommend hiking?’ The existence of an overt determiner or quantifier (a), or a genitive argument (b) has an effect to the availability of extraction, which supports the intermediate landing site analysis.11 The elements in the left 42 SAARA HUHMARNIEMI periphery of the DP have an effect to the extraction conditions of other elements, a matter I will return in the following sections. (60) a. *[ Minkä sillan lähellä] Asko ehdotti which.GEN bridge.GEN near Asko.NOM suggested tätä yöpymistä ti ? this.PAR sleeping.PAR ehdotti lähellä] Asko b. ?[ Minkä sillan which.GEN bridge.GEN near Asko.NOM suggested Pekan yöpymistä ti ? Pekka.GEN sleeping.PAR It thus seems that nominals with verbal root allow extraction of PPs expressing location, but extraction out of other nominal types is more restricted. In the following sections, I use mostly the derivational ending -minen in examples that require nominals with event structure.12 2.3.2 Elative and partitive PPs Let us now turn to quantifying nominal heads that take a partitive or elative PP modifier, as in examples (61) and (62), discussed in Brattico (2008). Extraction of the PP is available in both cases (see more examples also in Hakulinen & Karlsson (1979); Hakulinen et al. (2004)).13 (61) a. yksi kasa [ pieniä autoja] one stack small.PL.PAR cars.PL.PAR ’one stack of small cars’ on yksi kasa ti ? b. [ mitä autoja]i what.PAR car.PL.PAR is one stack ’Which cars is there one stack?’ (62) a. pala [ siitä isosta leivästä] slice that.ELA big.ELA bread.ELA ’a slice from that big bread’ b. [ Mistä leivästä] Pekka leikkasi palan.ACC ti ? which.ELA bread.ELA Pekka cut slice.ACC ’Which bread did Pekka cut the slice from?’ Brattico (2008) calls above structures “kasa constructions”, and suggests that the partitive case is assigned by a prepositional head as in (63). According to Brattico, Kasa constructions allow the PP to be extracted, as we saw in (61)-(62). EXTRACTION ISLANDS IN FINNISH (63) 43 DP D nP N kasa ’stack’ PP P PRT ’of ’ DP N autocars The extraction of elative PPs out of DPs extends beyond quantifying nominals, and resembles extraction out of NPs, for example, in the English datum (64a-b), where the nominal complement of the preposition has presumably undergone wh-movement. The DP has an underlying structure of (65a). I assume the same structure for Finnish DP kuva Merjasta, ’picture of Mary’ (b). It is probable that the PP moves as a whole in Finnish (64) a. I saw a picture of Mary. b. Whoi did you see a picture of ti ? (65) (a) (b) DP D a NP N picture D PP P of DP DP N Mary NP N kuva ’picture’ PP P STA ’of ’ DP N Merja This type of extraction out of the NP is restricted both syntactically and semantically. Example (66a) illustrates the influence of the verb: whereas some verbs allow extraction, others do not. The example (b) shows the ‘definiteness effect’ by Fiengo & Higginbotham (1981): the presence of determiner pronoun or definite article blocks extraction. (66) luki/osti/*?pinosi kirjoja ti ? a. Kenestäi Asko who.ELA Asko.NOM read/bought/stacked books.PAR ’Who did Asko read/buy/*stack books of? b. *Kenestäi Asko luki tämän/sen kirjan ti ? who.ELA Asko.NOM read this.ACC/the.ACC book.ACC 44 SAARA HUHMARNIEMI The unavailability of extraction in the presence of genitive arguments and possessive suffix is illustrated in (67a-b) (see examples from other languages (Abney, 1987; Stowell, 1989)). The sentences (c-d) provide examples in which extraction is available when the genitive argument is interpreted as the writer of the book, and not the owner (see Davies & Dubinsky (2003)). (67) a. *Kenestäi Asko luki Merjan kirjan ti ? who.ELA Asko.NOM read Merja.GEN book.ACC luki kirjansa ti ? b. *Kenestäi Asko who.ELA Asko.NOM read book.ACC.Px/3SG ti ? kirjaa luki Paasilinnan c. ?Kenestäi Asko who.ELA Asko.NOM read Paasilinna.GEN book.PAR ’Who did Asko read Paasilinna’s book?’ kirjoitti kirjansa ti ? d. Kenestäi Asko who.ELA Asko.NOM read his book.PAR ’Who did Askoi wrote hisi book about?’ Chomsky (1977) suggests that PP-extraction out of an NP may not be an instance of A-bar movement. He proposes that the PP is basegenerated inside the DP but moved out prior A-bar movement via extraposition or reconstruction rule. This may be true for Finnish as well. In addition to general limitations in (66) and (67a-b), there are contexts in which extraction out of the kasa-construction differs from general conditons for A-bar movement. First, in example (68a) the DP is embedded to a complement of the preposition kohti. Extraction of the PP is not available (b). There is no principled reason why the A-bar movement should not be available from the embedded complement. (68) a. Pekka eteni tutkimuksissaan [P P kohti Pekka proceeded investigations.INE.Px/3SG towards totuutta Merjasta ] truth.PAR Merja.ELA ’Pekka proceeded in his investigations towards a/the truth about Merja.’ b. *?Kenestäi Pekka eteni tutkimuksissaan who.ELA Pekka proceeded investigations.INE.Px/3SG [P P kohti totuutta ti ]? towards truth.PAR ti Second, kasa-constructions do not allow pre-nominal PPs at all (69), and thus do not provide an intermediate landing site for the A-bar movement. This suggests that the structural conditions for extraction are EXTRACTION ISLANDS IN FINNISH 45 different for the PP-complements expressing location. In addition, there is no requirement for the nominal to have a verbal root. (69) ti ] a. *Pekka katseli [DP hiekkaai kasaa sand.PAR stack.PAR Pekka watched b. *Pekka katseli [DP leivästäi palaa ti ] Pekka watched bread.ELA piece.PAR Third, data from reflexive binding in (70a-b) shows that the PP does not reconstruct inside of the DP. In sentence (a), the correlate for the reflexive pronoun may be either Paasilinna or Pekka. However, when the PP is fronted, the only available reading is the one in which the correlate is Pekka. (70) a. Pekkai näki Paasilinnanj kirjan itsestääni/j Pekka saw Merja.GEN book.ACC himself.ELA ’Pekka saw Merja’s book about himself.’ b. Itsestääni/∗j Pekkai näki Paasilinnanj kirjan. himself.ELA Pekka saw Merja.GEN book.ACC The presence of the genitive argument has an effect to the availability of extraction, as seen in example (67), so let us consider other examples of reflexive binding as well. The reflexive may be bound by an element in the adjectival phrase in Finnish, as in (71a).14 In addition, a suitable adjectival phrase does not affect the grammaticality (71c). However, the DP-internal genitive argument fails to bind the reflexive in both cases (b-c). (71) a. Pekkai näki Merjanj ottaman kuvan itsestääni/j Pekka saw Merja.GEN take.MA picture.ACC himself.ELA ottaman kuvan. b. Itsestääni/∗j Pekkai näki Merjanj himself.ELA Pekka saw Merja.GEN take.MA picture.ACC c. [ Omasta kurssistaan]i/∗j Pekkai kirjoitti opettajanj Pekka wrote teacher.GEN own.ELA course.ELA pyytämän palautteen. ask.MA feedback.ACC ’About his own course, Pekka wrote the feedback the teacher asked.’ An account for the misplacement of partitive and elative case complements which does not assume A-bar movement is proposed by Hakulinen & Karlsson (1979). They recognize a class of nominals that take partitive or elative case PP-modifiers and express quantity, part or proportion (see also Hakulinen et al. 2004, §592). Hakulinen & Karlsson 46 SAARA HUHMARNIEMI analyze the structures as NP-split constructions (“lohkominen”), where the NP occupies two structural positions at the same time. This analysis presumes that the PP has not undergone A-bar movement and does not necessarily reconstruct to its base position.15 2.3.3 MA-infinitival In this section, I examine the MA-infinitives that modify a nominal head with derivational suffix -minen. The MA-infinitival in inner locative cases may be selected by nouns and verbs with the same root form: (72) and (73). Extraction out of the MA-infinitive from the complement of a finite verb is available, as discussed in section 2.5.3. (72) a. pysyminen [V P Merjaa auttamassa] staying.NOM Merja.PAR help.MA.INE ’staying helping Merja.’ b. palaaminen [V P Merjaa auttamasta] returning.NOM Merja.PAR help.MA.ELA ’returning from helping Merja.’ c. lähteminen [V P auttamaan Merjaa] leaving.NOM help.MA.ILL Merja.PAR ’leaving to help Merja.’ (73) a. pysyä [V P auttamassa Merjaa] stay help.MA.INE Merja.PAR ’to stay in helping Merja.’ b. palata [V P auttamasta Merjaa] palata help.MA.ELA Merja.PAR ’to return from helping Merja.’ c. lähteä [V P auttamaan Merjaa] leave help.MA.ILL Merja.PAR ’to leave to help Merja.’ When the MA-infinitive modifies a noun, it may be extracted as a whole (74a-c). The MA-infinitival complement behaves in this respect similarly to ordinary locative PPs discussed in section (2.3.1). (74) [DP lähtemisen ti ]? a. [Ketä auttamaan]i Juha esti leaving.ACC who.PAR help.MA.ILL Juha prevented ’To help whom did Juha prevent leaving?’ [DP Pekan b. [ Ketä auttamaan]i Juha esti Pekka.GEN who.PAR help.MA.ILL Juha prevented lähtemisen ti ]? leaving.ACC EXTRACTION ISLANDS IN FINNISH 47 ’To help whom did Juha prevent Pekka from leaving?’ [DP Pekan c. [ Ketä auttamaan]i Juha esti Pekka.GEN who.PAR help.MA.ILL Juha prevented lähettämisen ti ]? sending.ACC ’To help whom did Juha prevent sending Pekka?’ The MA-infinitival verb may be stranded, as in (75a), but the construction is limited; (b) for example is not a very good sentence. It thus seems that the MA-infinitive has different properties inside of the noun phrase than in the complement of the verb. In the complement of a finite verb, the MA-infinitive is never an extraction island (see section 2.5.3). (75) a. ?Ketäi Pekka esti [DP Juhan lähtemisen who.PAR Pekka prevented Juha.GEN leaving auttamaan ti ]? help.MA.ILL ’To help whom did Pekka prevent Juha from leaving?’ b. *?Ketäi Pekka esti [DP Juhan lähettämisen who.PAR Pekka prevented Juha.GEN sending auttamaan ti ]? help.MA.ILL ’To help whom did Pekka prevent sending Juha?’ The unavailability of extraction and availability of pied-piping in examples (74) on the other hand suggest that the MA-infinitive could occupy an adjunct position inside the noun phrase instead of the complement position. The structural position of the MA-infinitive inside DPs would thus be similar to PPs, which were considered to be adjuncts in section 2.3.1. Nevertheless, further research is required to verify this prediction. 2.3.4 A-infinitival complement A-infinitival complement allows extraction of arguments and adjuncts regardless of the type of the nominal head (76a-b). Extracted adjuncts may be ambiguous between sentential reading and A-infinitival reading, but it is possible to grasp an interpretation where the adjunct originates inside the DP (c). However, extraction of the genitive subject is not available (d-e). (76) a. Mitä Pekka sai [DP tilaisuuden tutkia ti ]? what Pekka got opportunity.ACC investigate.A ’What did Pekka receive an opportunity to investigate?’ 48 SAARA HUHMARNIEMI b. [ Millä tutkimusasemalla]i Pekka sai which.ADE research station.ADE Pekka got [DP tilaisuuden [A tutkia lintuja ti ]]? opportunity.ACC investigate.A birds.PAR ’In which research station did Pekka receive an opportunity to investigate birds?’ Pekka sai [DP tilaisuuden tutkia c. Missäi where.INE Pekka got opportunity.ACC investigate.A lintuja ti ]? birds.PAR ’Where did Pekka get an opportunity to investigate birds?’ d. ?Pekka näki [DP tilaisuuden [A Merjan todistaa Pekka saw opportunity.ACC Merja.GEN proof.A kykynsä]] competence.ACC.Px/3SG ’Pekka saw an opportunity for Merja to proof her capabilities.’ [A ti todistaa e. *Keneni Pekka näki [DP tilaisuuden who.GEN Pekka saw opportunity.ACC proof.A kykynsä]] competence.ACC.Px/3SG In certain contexts the case of the direct object of the A-infinitive alternates between the zero-accusative and -n-accusative case, as in (77). Interestingly, the case of the object has an effect to extraction, as observed by Hakulinen & Karlsson (1975). The direct object in the -n-accusative case may be extracted, but not in zero-accusative (78a-b). Hakulinen & Karlsson conclude that the infinitival clause forms an island in (78b), and the islandhood may be observed in the lack of case-inflection (zeroaccusative) and unavailability of extraction. I leave the question open for further research, but see Vainikka & Brattico (2009) for an account of zero-accusative case in infinitival complements. (77) Yrjö laati [ suunnitelman [ kaataa karhun/ karhu]] Yrjö made plan.ACC kill.A bear.ACC(n)/ bear.ACC(∅) ’Yrjö made a plan to kill a bear.’ (78) Yrjö laati [ suunnitelman kaataa ti ]? a. Minkäi what.ACC(-n) Yrjö made plan.ACC kill.A ’What did Yrjö made a plan to kill?’ b. *Mikäi Yrjö laati [ suunnitelman kaataa ti ] what.ACC(∅) Yrjö made plan.ACC kill.A EXTRACTION ISLANDS IN FINNISH 49 It should be pointed out that the recursive embedding of the DP to the complement of a partitive case assigning preposition shows that the extraction out of the A-infinitival complement is not limited by the level of embedding (79)-(80): (79) a. Pekka jatkoi kilpailussa [P P kohti [DP Pekka.NOM proceeded competition.INE towards mahdollisuutta [A voittaa auto]]] possibility.PAR win.A car.ACC(φ) ’Pekka proceeded in the competition towards a/the possibility to win a car.’ Pekka jatkoi [P P kohti [DP b. Mitäi what.PAR Pekka proceeded towards mahdollisuutta [A voittaa ti ]]]? possibility.PAR win.A ’Whati did Pekka proceed towards the possibility to win ti ? (80) a. Pekka pääsi [P P lähelle [DP tilaisuutta Pekka got near.ALL opportunity.PAR [A tutkia lintuja]]] investigate.A birds.PAR ’Pekka got near to an opportunity to investigate birds.’ Pekka pääsi [P P lähelle [DP tilaisuutta b. Mitäi what.PAR Pekka got near opportunity.PAR [A tutkia ti ]]]? investigate.A 2.4 APs Adjectival attributes constitute strong islands for extraction. However, extraction out of adjectival phrases in predicative clauses is not completely impossible. For example, the adjectival näköinen ’looking’ seems to allow limited extraction of genitive arguments in predicative clauses such as (81). However, the presence of the possessive inflection on the adjectival head prevents extraction (82).16 (81) a. Poika on [ Pekan näköinen] boy.NOM is Pekka.GEN looking ’The boy is looking like Pekka.’ b. Kenen on poika [ti näköinen]? who.GEN is boy.NOM looking ’Who does the boy look like?’ 50 (82) SAARA HUHMARNIEMI a. Poika on [AP hänen näköisensä] boy.NOM is he.GEN looking.Px/3SG ’The boy is looking like him.’ [ti näköisensä] b. ??Häneni on poika he.GEN is boy.NOM looking.Px/3SG Neither MA-participial (83) nor VA-participial (84) allows extraction of arguments in predicative clauses: (83) ostamia] eilen a. Nuo pyörät ovat [AP Pekan Pekka.GEN yesterday buy.MA.PAR These bikes are ’These bikes are bought by Pekka yesterday.’ b. *Keneni ovat nuo pyörät [AP ti ostamia] who.GEN are these bikes buy.MA.PAR ti ostamia] c. *Milloini ovat nuo pyörät [AP Pekan when are these bikes Pekka.GEN buy.MA.PAR (84) a. Nämä annokset ovat [ ruokahalua herättäviä] these portions are appetite.PAR provoke.VA.PL.PAR ’These portions are appetizing.’ b. *Ruokahaluai ovat nämä annokset [ti herättäviä] appetite.PAR are these portions provoke.VA.PL.PAR To summarize, although adjectival attributes are strong islands for extraction in Finnish, extraction out of predicative APs is not completely ruled out. 2.5 Infinitival clauses This section provides basic extraction conditions for infinitival clause constructions in Finnish. Infinitival clauses may occur in the complement and adjunct positions. Whereas infinitival clause adjuncts are strong islands for movement, complement clauses allow extraction to a varying extent. Infinitival clause complements are discussed in sections 2.5.1-2.5.2: A-infinitive, VA-construction. MA-infinitival complements, as well as MA-adjuncts are considered in section 2.5.3. Sections 2.5.4-2.5.5 investigate infinitival clause adjuncts: temporal and rationale adjuncts and the E-infinitive. Special attention is on adjunct clauses which may nevertheless occupy a structural complement position and allow limited extraction (MA-adjuncts and E-infinitive). The basic syntactic properties of infinitival clauses are described in Hakulinen & Karlsson (1979); Vainikka (1989, 1995); Koskinen (1997). The classification and terminology for infinitival clauses is mostly borrowed from Vainikka & Brattico (2009). EXTRACTION ISLANDS IN FINNISH 51 2.5.1 A-infinitive A-infinitive occupies the direct complement of the verb.17 It may appear without subject (85a), in which case the thematic subject is the matrix clause subject. When the A-infinitive contains a subject, it is in genitive case (b). (85) a. Pekka halusi [ syödä omenoita] Pekka wanted eat.A apples.PAR ’Pekka wanted to eat some apples.’ b. Pekka käski [ Merjan syödä omenoita] Pekka ordered Merja.GEN eat.A apples.PAR ’Pekka ordered Merja to eat some apples.’ A-infinitive allows extraction of the subject, object and adjuncts (86ac): (86) a. Mitäi Pekka halusi [ syödä ti ] what.PAR Pekka wanted eat.A ’What did Pekka want to eat?’ b. Keneni Pekka käski [ti syödä omenoita]? who.GEN Pekka ordered eat.A apples.PAR ’Who did Pekka order to eat some apples?’ [ syödä ti omenoita] c. Milloini Pekka aikoi when Pekka intended eat.A apples.PAR ’When did Pekka intend to eat some apples?’ The presence of zero-accusative does not have an effect to the extraction conditions (87a-b): (87) a. Meidän käskettiin [ ostaa lahja Merjalle] we.GEN ordered.PASS buy.A present.ACC(φ) Merja.ILL ’They ordered us to buy a present for Merja.’ b. [ Mikä lahja] meidän which.ACC(φ) present.ACC(φ) we.GEN käskettiin [ ostaa ti Merjalle]? ordered.PASS.PAST buy.A Merja.ILL ’Which present did they order us to buy for Merja?’ A-infinitive is thus a prototypical example of an infinitival complement: it does not restrict extraction of any element. Some of the extraction properties of A-infinitives in the complement of a nominal were already examined in 2.3.4. 52 SAARA HUHMARNIEMI 2.5.2 VA-construction Similarly as the A-infinitive, the VA-construction occupies the complement position of finite verbs. However, whereas the A-infinitive may occur in the complement of a noun as well, this is not true of the VAinfinitive.18 The VA-construction has morphological forms for past and present/future (88a-b), as well as active and passive (c-d). When there is no overt subject in the active forms, the thematic subject is the matrix clause subject and the infinitival verb receives possessive inflection (a). However, when the infinitival clause has a genitive subject, the possessive suffix is not used (b). (88) a. Pekka muistaa [tavanneensa hänet aikaisemmin] Pekka remembers met.VA.Px/3SG him earlier ’Pekka remembers having met him earlier.’ b. Minä tiedän [ Juhan tapaavan Merjan] I know Juha.GEN meet.VA/PRES Merja.ACC ’I know (that) Juha will meet Merja.’ c. Minä huomasin [ kakkua maistetun] noticed cake.PAR taste.PASS.VA/PAST I ’I noticed (that) the cake had been tasted.’ d. Minä tiesin [ kakkua maistetttavan] I knew cake.PAR taste.PASS.VA/PRES ’I knew (that) the cake would be tasted.’ VA-construction allows extraction of complements and adjuncts (89ab), as well as subjects (c). The presence of the genitive argument does not have an effect to extraction (d) and extraction out of the passive variant is possible as well. The extraction data thus verifies the structural position of the VA-construction in the complement of the verb. (89) a. Keneti Pekka muistaa [ tavanneensa ti aiemmin]? who.ACC Pekka remembers met.VA.Px/3SG earlier ’Who does Pekka remembe having met earlier?’ Merjan ti ]? b. Milloini Pekka muisteli [ tavanneensa When Pekka recalled met.VA.Px/3SG Merja.ACC ’When did Pekka recall having met Merja?’ [ti tapaavan Merjan]? c. Keneni luulet who.GEN think.2SG meet.VA Merja.ACC ’Who do do you think will meet Merja?’ luulet [ Merjan tapaavan ti ]? d. Keneti who.ACC think.2SG Merja.GEN meet.VA ’Who do you think (that) Merja will meet?’ EXTRACTION ISLANDS IN FINNISH 53 e. Mitä sinä huomasit [ti maistetun]? what.PAR you.NOM noticed tasted.PASS.VA ’What did you noticed (that) was tasted?’ Let us briefly look into structures in which the VA-construction is embedded in the complement of A-infinitive, as in (90a-b). The case of the object alternates between the -n-accusative and zero-accusative and the extraction is available in both cases (Vainikka & Brattico, 2009). (90) a. Sinun täytyy uskoa [ voittaneesi you.GEN must believe.A won.VA.Px/2SG kilpailu] competiton.ACC(∅) ’You have to believe (that) you won the competition.’ b. Sinun täytyy uskoa [ voittaneesi you.GEN must believe.A won.VA.Px/2SG kilpailun] competiton.ACC(∅) ’You have to believe that you won the competition.’ (91) sinun täytyy a. [ Mikä kilpailu]i which.ACC(∅) competition.ACC(∅) you.GEN must uskoa voittaneesi ti ? believe.A won.VA.Px/2SG ’Which competition you have to believe that you won?’ b. [ Minkä kilpailun]i sinun täytyy uskoa which.ACC competition.ACC you must believe.A voittaneesi ti ? won.VA.Px/2SG ’Which competition you have to believe that you won?’ The VA-infinitive has identical properties to A-infinitives regarding extraction: there are no restrictions on movement out of the VAconstruction. 2.5.3 MA-infinitive MA-infinitive has several case variants. In the three internal locative case suffixes, inessive, elative, and illative, the MA-infinitive is an argument of a verb and occupies the complement position (Nikanne, 1989; Vainikka, 1989). In adessive and abessive case, the MA-infinitive is interpreted as a manner adverbial (Hakulinen & Karlsson, 1979), thus occupying an adjunct position (Koskinen, 1997). As a general rule, MA-infinitives have 54 SAARA HUHMARNIEMI the same distribution as PPs with lexical cases (Vainikka, 1989). I will first investigate MA-complements and then turn to MA-adjuncts. I assume, following Koskinen (1997), that the MA-infinitive contains a PRO-element that is bound by the closest c-commanding DP in the matrix clause. Whenever the matrix object is present, it binds the PROelement in MA-complement (92a-c). When the matrix object is missing, as in (93), PRO is bound by the matrix subject. (92) a. Pekka näki Merjan [ ostamassa kirjaa] Pekka saw Merja.ACC buy.MA.INE book.PAR ’Pekka saw Merja buying a book.’ b. Pekka kielsi Merjaa [ ostamasta kirjaa] Pekka denied Merja.ACC buy.MA.ELA book.PAR ’Pekka saw Merja buying a book.’ c. Pekka lähetti Merjan [ ostamaan lahjaa] Pekka send Merja.ACC buy.MA.ILL present.PAR ’Pekka went to buy a present.’ (93) Pekka lähti [ ostamaan ruokaa] Pekka went buy.MA.ILL food.PAR ’Pekka went to buy some food.’ The different internal locative case variants of MA-complement were illustrated in (92a-c). All these case variants have similar properties regarding extraction. The extraction out of the complement position is illustrated in examples (94a-c). For now on examples are provided interchangeably from the three cases. (94) Pekka näki Merjan [ ostamassa ti ]? a. Mitäi what.PAR Pekka saw Merja.ACC buy.MA.INE ’What did Pekka see that Merja was buying?’ Pekka kielsi Merjaa [ ostamasta ti ]? b. Mitäi what.PAR Pekka denied Merja.PAR buy.MA.ELA ’What did Pekka go to buy?’ c. Mitäi Pekka lähti [ ostamaan ti ]? what.PAR Pekka went buy.MA.ILL ’What did Pekka go to buy?’ Example (95a) illustrates extraction of an adjunct out of the MAcomplement. The infinitival clause may be pied-piped as a whole as well (b). As predicted, the direct object of the matrix clause may be freely moved to the front of the sentence (c). (95) a. Kenellei Pekka näki Merjan [ ostamassa lahjaa ti ]? who.ILL Pekka saw Merja.ACC buy.MA.INE EXTRACTION ISLANDS IN FINNISH 55 ’Who did Pekka see that Merja was buying a gift for?’ b. [ Mitä ostamassa]i Pekka näki Merjan ti ? what.PAR buy.MA.INE Pekka saw Merja.ACC ’What did Pekka see Merja was buying?’ c. Keneti Pekka näki ti [ ostamassa kirjaa]? who.ACC Pekka saw buy.MA.INE book.PAR ’Who did Pekka see buying a book?’ When the MA-infinitive is in the complement of A-infinitive, the object receives zero-accusative case. Extraction of the zero-accusative object is possible (96a-b): (96) a. Pekan täytyi lähteä [ hakemaan kirja Pekka must go.A fetch.MA.ILL book.ACC(∅) koulusta] school.ELA ’Pekka had to leave to get the book from the school.’ b. [ Mikä kirja]i Pekan täytyi lähteä what.ACC(∅) book.ACC(∅) Pekka.GEN must go.A [hakemaan ti koulusta]? fetch.MA.LL school.ELA ’Which book Pekka had to leave to fetch rom the school?’ Let us now turn to the MA-adjuncts. MA-adjuncts have two case variants: adessive and abessive. For most speakers the MA-adjunct in adessive case cannot contain an overt subject.19 According to Koskinen (1997), MA-adjuncts have the same syntactic structure as MA-complements, but their structural position is different. First, the thematic subject of the MA-adjunct is always the sentence subject. In addition, whereas the reflexive is able to refer to the direct object from inside MA-complement (97a), reference is not available from MA-adjuncts (b-c). It thus seems that whenever the matrix sentence contains a direct object, MA-adjuncts are located above it. (97) koiraansai/j ] a. Pekkai pyysi Merjaaj [ vahtimaan Pekka asked Merja.PAR watch.MA.ILL dog.PAR.Px/3SG ’Pekka asked Merja to watch his/her dog.’ b. Pekkai auttoi Merjaaj [ löytämälläi/∗j Pekka helped Merja.PAR find.MA.ADE kiikarinsa] binoculars.ACC.Px/3SG ’Pekka helped Merja by finding his pinoculars.’ 56 SAARA HUHMARNIEMI c. Pekkai auttoi Merjaaj [ löytämättäi/∗j Pekka helped Merja.PAR find.MA.ABE kiikariaan] binoculars.PAR.Px/3SG ’Pekka helped Merja without finding his binoculars.’ Nevertheless, extraction data suggests that MA-adjunct can occupy a complement position as well. Extraction out of the MA-adjunct in adessive is possible when the matrix object is not present (98a), but otherwise impossible (b).The interrogative sentence may be formed by pied-piping the whole constituent to the front (c). (98) a. ?Ketäi Pekka löysi kotiin [ seuraamalla ti ]? who.PAR Pekka fond home.ILL follow.MA.ADE lit. ’Who did Pekka find home following?’ Merjan b. *?[DP Mitä kirjaa]i Pekka yllätti which.PAR book.PAR Pekka surprised Merja.ACC [ lukemalla ti ]? read.MA.ADE lukemalla]i Pekka yllätti c. [ Mitä kirjaa which.PAR book.PAR read.MA.ADE Pekka surprised Merjan ti ? Merja lit. ’Which book reading did Pekka surprise Merja?’ Let us now turn to the MA-adjunct in abessive case. The infinitival clause may contain an overt subject in genitive case (99b). A pronominal subject causes possessive inflection to the infinitival verb. Otherwise, the thematic subject is the matrix subject, as in (b). In addition, MA-adjunct in abessive may contain possessive suffix bound by the matrix subject (c). This form is restricted to cases where the infinitival verb does not contain an object argument (d). (99) a. Pekkai käveli kotiin [ hänenj näkemättään∗i/j ] Pekka walked home.ILL (s)he.GEN see.MA.ABE.Px/3SG ’Pekka walked home without him seeing.’ b. Pekka käveli kotiin [ näkemättä Merjaa] Pekka walked home.ILL see.MA.ABE Merja.PAR ’Pekka walked home without seeing Merja.’ [ huomaamattaani ] c. Pekkai käveli kotiin Pekka walked home.ILL notice.MA.ABE.Px/3SG ’Pekka walked home without noticing.’ EXTRACTION ISLANDS IN FINNISH 57 d. *Pekka käveli kotiin [ huomaamattaan Pekka walked home.ILL notice.MA.ABE.Px/3SG Merjaa] Merja.PAR Similarly as with the MA-adjunct in adessive (in (98a)), when there is no direct object in the matrix clause and no overt subject in the infinitival clause, extraction out of the infinitival clause is possible (a). However, extraction is impossible in the presence of direct object (b) or a genitive argument (c). In addition, extraction of the genitive subject is not possible at all (d). (100) Pekka käveli kotiin [ näkemättä ti ]? a. Ketäi who.PAR Pekka walked home.ILL see.MA.ABE ’Who didn’t Pekka notice when he walked home?’ b. *?Mitäi Pekka korjasi pyöränsä [ huomaamatta ti ]? what.PAR Pekka fixed bike.ACC notice.MA.ABE Pekka nauroi [ Merjan huomaamatta ti ]? c. *Mitäi what.PAR Pekka laughed Merja.GEN notice.MA.ABE [ti näkemättä] d. *Keneni Pekka käveli kotiin who.GEN Pekka walked home.ILL see.MA.ABE MA-adjunct undergoes wh-movement as a whole (101): (101) ti ? näkemättä]i Pekka käveli kotiin [DP Kenen who.GEN see.MA.ABE Pekka walked home.ILL ’Who didn’t see that Pekka walked home?’ MA-adjunct in adessive occupies a direct complement position of certain finite verbs, such as jättää ’leave’ (Hakulinen et al., 2004, §1627). In this case, the MA-adjunct cannot have a subject of its own (b). A-bar movement is freely available (c). (102) a. Pekka jätti katsomatta elokuvan. Pekka left watch.MA.ABE film.ACC ’Pekka didn’t watch the film.’ b. *Pekka jätti Merjan katsomatta elokuvan. Pekka left Merja.GEN watch.MA.ABE film.ACC ti ? c. Minkäi Pekka jätti katsomatta what.ACC Pekka left watch.MA.ABE ’What Pekka didn’t watch?’ The examples (102), (98), and (100) show that the MA-infinitives in adessive and abessive case, which are generally conceived as manner 58 SAARA HUHMARNIEMI adverbials and hence adjuncts, may sometimes occupy a complement position and allow extraction. Thus, whereas MA-infinitives in the inner locative cases are not islands for extraction, the islandhood of MAadjuncts depends on the structural position of the infinitival clause. 2.5.4 Strong extraction islands: Rationale adjunct and Temporal adjunct This section considers two strong extraction islands in Finnish: rationale adjunct and temporal adjunct. The common denominator between these two infinitival clause types is their adjunct position and obligatory possessive inflection in the presence of matrix subject. The structural details for both constructions are available in Vainikka (1989, 1995) and Koskinen (1997). The infinitival verb in rationale adjunct contains a case marking -kse and the possessive suffix. The construction does not contain an overt subject, and the thematic subject of the infinitival clause is the matrix subject. In addition, the verb receives possessive inflection in the presence of matrix subject (103a). Rationale adjunct does not allow extraction of the direct object (b) or adjuncts (c). The non-existence of the matrix object does not affect the extraction conditions (d). Nevertheless, whmovement of the whole adjunct phrase is possible (e) (103) Rationale adjunct a. Pekka osti omenoita [ valmistaakseen piirakkaa Pekka bought apples.PAR prepare.KSE.Px/3SG pie.PAR tässä uunissa] this.INE oven.INE ’Pekka bought apples in order to make some pie.’ b. *Mitäi Pekka osti omenoita [ valmistaakseen ti ] what Pekka bought apples.PAR prepare.KSE.Px/3SG omenoita [ piirakkaa c. *[ Missä uunissa]i Pekka osti what.INE oven.INE Pekka bought apples.PAR pie.PAR valmistaakseen ti ]? prepare.KSE.Px/3SG ti ]? d. *Minkäi Pekka kiirehti [ voittaakseen what.ACC Pekka hurried win.KSE.Px/3SG e. [ Mitä valmistaakseen]i Pekka osti omenoita ti ? what.PAR prepare.KSE Pekka bought apples.PAR ’In order to prepare what did Pekka buy some apples?’ EXTRACTION ISLANDS IN FINNISH 59 Temporal adjunct has two tense variants, present and past (Vainikka, 1989; Vainikka & Brattico, 2009), and it may have a genitive subject or be subjectless. When the subject is missing, the thematic subject is the matrix clause subject and the infinitival verb has a possessive suffix (104a-b). A pronominal subject causes possessive inflection on the verb (c-d). The temporal adjunct has a passive form as well (e). (104) Temporal adjunct a. Pekka kompastui [ kävellessään kotiin] Pekka fell walk.ESSA/PRES.Px/3SG home ’Pekka fell when walking home.’ b. Pekka lepäsi [ käveltyään kotiin] Pekka rested walk.ESSA/PAST.Px/3SG home.INE. ’Pekka rested after walking home.’ c. Pekka lähti [ Merjan laitettua ruokaa] Pekka left Merja.GEN cook.ESSA/PAST food. ’Pekka left after Merja had cooked.’ d. Pekka lähti [ hänen laitettua-an ruokaa] Pekka left (s)he.GEN cook.ESSA/PAST-Px/3SG food. ’Pekka left after he/she had cooked.’ e. [ Autettaessa vanhuksia] täytyy olla kohtelias. help.ESSA/PASS seniors.PAR must.3SG be.A polite. ’It is important to be polite when helping elderly people.’ The temporal adjunct is a strong island for movement: Extraction is not available from the complement (105a), nor from the subject position (b). The only way a content question can be formed is pied-piping of the whole infinitival clause to the front of the sentence (c). (105) tj ]? a. *Ketäj Pekka kompastui [ auttaessaan who.PAR Pekka fell help.ESSA/PRES.Px/3SG t ruokaa]? b. *Keneni Pekka lähti [ti laitettua who.GEN Pekka left cook.ESSA/PAST food c. [ Ketä auttaessaan]i Pekka kompastui ti ? who.PAR help.ESSA/PRES.Px/3SG Pekka fell ’Who Pekka was helping when he fell?’ Both temporal adjunct and rationale adjunct are strong islands as a consequence of their position in the syntactic structure: they always occupy an adjunct position. The fact that the possessive suffix never refers to the direct object supports the conclusion that they cannot be realized in a complement position (106): 60 (106) SAARA HUHMARNIEMI a. Pekkai tapasi Merjanj [ pyytääkseeni/∗ apua] Pekka met Merja.ACC ask.KSE.PX/SG3 help.PAR ’Pekka met Merja in order to ask for help.’ b. Pekkai näki Merjanj [ kävellessääni/∗j kotiin] Pekka saw Merja.ACC walk.ESSA.PX/SG3 home.ILL ’Pekka saw Merja when walking home.’ 2.5.5 E-infinitive The E-infinitive is a manner adverbial (Hakulinen & Karlsson, 1979) and generally occupies an adjunct position in the sentence. When the E-infinitive does not contain an overt subject, the thematic subject is the matrix subject and there is no possessive inflection (107a). When an overt subject is present, it is in genitive case. In addition, pronominal elements cause possessive inflection on the verb (b-d) (Koskinen, 1997, pp. 315). E-infinitive allows an impersonal subject as well (d). (107) E-infinitive a. Pekka käveli kotiin vihellellen. Pekka walked home.INE whistle.E ’Pekka walked home whistling.’ b. Virve pudotti maljakon [ Lasun nähden] Virve.NOM drop.PAST.3SG vase.ACC Lasu.GEN see.E ’Virve dropped the vase, Lasu seeing it’ c. Maljakko putosi [ minun nähteni] vase.NOM fall.PAST.3SG I.GEN see.E.Px/1SG ’The vase fell, me seeing it’ d. Pekka onnistui hyvin [ ottaen huomioon hänen Pekka succeeded well take.E account.ILL his kokonsa] size.ACC.Px/3SG ’Pekka succeeded well considering his size.’ Extraction out of the E-infinitive is marginally available when the matrix clause does not contain a direct object (114a). However, extraction is not available in the presence of the direct object (109) and when the thematic subject of infinitival clause is impersonal (109b). Genitive arguments may not be extracted (c) (108) a. ?[ Mitä laulua]i Pekka käveli kotiin [vihellellen ti ]? which.PAR song.PAR Pekka walked home whistle.E.n ’Which song Pekka walked home whistling?’ EXTRACTION ISLANDS IN FINNISH (109) 61 a. *?[ Mitä laulua]i Pekka tapasi Merjan which.PAR song.PAR Pekka met Merja.ACC [vihellellen ti ] whistle.E b. *Minkäi Pekka onnistui hyvin [ ottaen huomioon ti ]? what Pekka succeeded well take.E account.ILL ’Pekka succeeded well considering what?’ c. *Kenen Pekka kaatui [ti nähden]? who.GEN Pekka fell.PAST.3SG see.E Similarly as with other infinitival clause adjuncts, the E-infinitive may be fronted as a whole (110). (110) ti ? [ Mitä laulua vihellellen]i Pekka käveli kotiin which.PAR song.PAR whistle.E.n Pekka walked home.ILL ’Which song Pekka was whistling when he walked home?’ To summarize, the E-infinitive generally occupies an adjunct position. However, E-infinitive seems to be able to appear also in the complement of a verb, in which case it allows limited extraction. 3. φ-FEATURE AGREEMENT AND EXTRACTION 3.1 Introduction Finnish DPs, PPs, and certain infinitival clauses contain a genitive argument that cannot be extracted although extraction of complements and adjuncts is possible. First, it is never possible to extract the genitive argument out of DPs or PPs, as seen in examples (41d) and (50), repeated here as (111) and (112). In addition, the MA-adjunct in abessive does not allow extraction of the genitive subject, (100)), repeated here as (113a-b), although extraction of the object is available. The same is true for the Einfinitive (107), repeated here as (114). (111) a. Minä luin [ Hotakaisen kirjan] I.NOM read Hotakainen.GEN book.ACC b. *Keneni sinä luit [ti kirjan]? who.GEN you.NOM read book.ACC (112) a. Pekka istui [ nuotion lähellä] Pekka sat fireplace.GEN near ’Pekka sat near a/the fireplace.’ b. *Minkäi Pekka istui [ti lähellä]? what.GEN Pekka sat near 62 SAARA HUHMARNIEMI (113) a. Pekka käveli kotiin [ näkemättä Merjaa]? Pekka walked home.ILL see.MA.ABE Merja.PAR ’Pekka walked home without noticing Merja.’ [ti näkemättä]? b. *Keneni Pekka käveli kotiin who.GEN Pekka walked home.ILL see.MA.ABE (114) nähden]. a. Pekka kaatui [ Lasun Lasu.GEN see.E Pekka fell ’Pekka fell, Lasu seeing it.’ b. *Kenen Pekka kaatui [ti nähden]? who.GEN Pekka fell see.E The unavailability of extraction of genitive arguments may be considered as an extension of the left branch condition (LBC) (originally Ross, 1967), which prohibits extraction of a leftward-branching specifier. In this case, LBC would cover genitive arguments inside DPs, PPs and infinitival clauses. However, the extended left branch condition would not account for the Finnish data. For example, genitive subject arguments of the VA-construction and the A-infinitive are able to move freely (115ab): (115) a. Keneni Pekka arveli ti lähtevän kotiin? who.GEN Pekka thought leave.VA home.ILL ’Who did Pekka think was going home?’ b. Keneni Pekka käski ti lähteä kotiin? who.GEN Pekka ordered leave.A home.ILL ’Who did Pekka order to go home?’ I would like to suggest that the LBC in Finnish is regulated by (anti)agreement. Specifically, I suggest that the extraction of genitive arguments is permitted only in constructions in which there is no φ-feature agreement between the genitive argument and the syntactic head. The lack of agreement in subject extraction is called anti-agreement (Ouhalla, 1993). The anti-agreement phenomenon is widely recognized in the subject extraction out of finite clauses. In addition, the anti-agreement effect has been argued to exist at least in possessor extraction (Boeckx, 2003). The syntactic configuration in all of the left branch extraction islands is very similar: The subject DP has undergone EPP-movement to the specifier of a syntactic head (N, P, V or T), and there is φ-feature agreement between the DP and the head, as illustrated in (116).20 The φ-feature agreement is established before the EPP-movement in many minimalist models (Chomsky, 2000, 2001), even so that φ-feature agreement is a prerequisite for the EPP. For the purposes of this paper, it suffices to EXTRACTION ISLANDS IN FINNISH 63 say that in Finnish both EPP-movement and φ-feature agreement exist in a left branch extraction island, regardless of the order in which these operations take place. (116) (a) EPP-movement XP (b) φ-feature agreement XP DP DP XP X +φ X +φ YP ... t XP ... YP ... t ... In the following, I go through the example cases which do not allow extraction of the genitive argument: DP, PP, and MA-adjunct in abessive, and show that they all fall under the configuration in (116), in section 3.2. In section 3.3 continue by showing that there is no φ-feature agreement in structures where the subject may be extracted: VA-constrution and Ainfinitive. Finally, I show that the my observations on LBC in Finnish may be extended to subject extraction out of finite complement clauses, in section 3.4. First of all, I adopt here the approach proposed by Kanerva (1987), where the possessive suffix is φ-feature agreement marker.21 Possessive suffixes may attach to a wide variety of elements: nominals, prepositions, adjectivals and infinitival verbs (117a-d). (117) a. minun kirja-ni (NP) I.GEN book-Px/1SG ’my book’ b. minun lähellä-ni (PP) I.GEN near-Px/1SG ’near me’ c. minun kaltaise-ni ihminen (AP) I.GEN alike-Px/1SG person ’a person like me’ d. minun näkemättä-ni (MA in abessive) I.GEN leave.MA.ABE-Px/1SG ’without me seeing’ In addition, the word order between the head and the genitive DP is fixed (118). This supports the view that φ-feature agreement is connected with EPP-movement to the specifier position (Brattico, 2009a; Brattico & Leinonen, 2009). 64 (118) SAARA HUHMARNIEMI a. *kirja-ni minun (NP) ’book-Px/1SG I.GEN b. *lähellä-ni minun (PP) near-Px/1SG I.GEN c. *kaltaise-ni minun (AP) alike-Px/1SG I.GEN d. *näkemättä-ni minun (MA in abessive) see.MA.ABE-Px/1SG I.GEN Common nouns do not cause overt φ-inflection in Finnish (119a), whereas personal pronouns do (b-c). (119) a. Pekan pyörä Pekka.GEN bike.NOM ’Pekka’s bike’ b. hänen pyöränsä (s)he.GEN bike.NOM.Px/3SG ’his/her bike’ c. (minun) pyöräni I.GEN bike.NOM.Px/1SG ’my bike’ 3.2 DPs, PPs and infinitival clauses that show φ-feature agreement Brattico & Leinonen (2009), Brattico (2008) and Brattico (2009a) propose that genitive arguments that modify nominal head undergo EPPmovement from the complement of a nominalizer head -n to the specifier of -n, and the φ-feature agreement is a result of the specifier head configuration, as illustrated in (120). The derivation of the DP thus falls under the structural configuration for LBC proposed in (116). (120) EPP-movement of the DP ment nP DPi nP DPi nP n φ-feature agree- ti nP n +φ ti EXTRACTION ISLANDS IN FINNISH 65 Consider next PPs that take the genitive argument. In section 2.2, I showed that PPs do not allow extraction of genitive arguments. The φfeature agreement pattern is equivalent to that of NPs, as can be seen in (121). Examples (122) illustrate the usage of uninflected form ohi, ’past’, which has a special form ohitse that takes φ-inflection 22 . (121) a. Pekan edellä Pekka.GEN before ’before Pekka’ b. hänen edellä-än (s)he.GEN before-Px/3SG ’before him/her’ c. (minun) edellä-ni I.GEN before-Px/1SG ’before me’ (122) a. Pekan ohi/ohitse Pekka.GEN past ’past Pekka’ b. hänen ohitse-en (s)he.GEN past-Px/3SG ’past him/her’ c. (minun) ohitse-ni I.GEN past-Px/1SG ’past me’ The word order inside the PP is fixed.23 According to Manninen (2003b), there is a φ-feature agreement between the DP and the prepositional head. In addition, the DP undergoes EPP-movement from the complement of P to the specifier of P. The EPP movement is illustrated in (123a). In this paper, the φ-feature agreement correlates with the specifier-head configuration (b), which produces the same configuration as (116).24 (123) EPP-movement PP DP hänen φ-feature agreement PP DPi hänen PP P ohitseen t PP P ohitseen+φ ti 66 SAARA HUHMARNIEMI Let us now turn to the unavailability of extraction of genitive arguments out of the MA-adjunct in abessive. The infinitival verb inflects with the φ-features of its subject, as in (124b-c). (124) a. Pekan huomaamatta Pekka.GEN notice.MA.ABE ’without Pekka noticing’ b. hänen huomaamatta-an (s)he.GEN notice.MA.ABE-Px/3SG ’without him/her noticing’ c. (minun) huomaamatta-ni I.GEN notice.MA.ABE-Px/1SG ’without me noticing’ I leave the exact syntactic structure of the MA-infinitive open here. However, there are at least a couple of ways to derive the EPP-movement based on the existing analyses. For example, it is possible that MAinfinitive contains a nominalizer head -n, and the movement to the specifier of -n would already suffice for the φ-inflection, as suggested for nPs by Brattico (2009a). Another option would be to adopt the approach by Koskinen (1997, 357) and take it that the MA-adjunct contains a TPprojection. In Koskinen’s approach, the infinitival verb would raise to T and the subject to its specifier, resulting in a spec-head configuration and φ-feature agreement. Finally, the φ-inflection is present in both strong islands, rationale and temporal adjunct, which were presented in section 2.5.4. The φinflection in the temporal adjunct patterns with NPs and PPs (125). (125) a. Merjan auttaessa Merja.GEN help.ESSA/PRES ’when Merja was helping’ b. hänen auttaessa-an (s)he.GEN help.ESSA/PRES-Px/3SG ’when he/she was helping’ c. (minun) auttaessa-ni I.GEN help.ESSA/PRES-Px/1SG ’when I was helping’ Another strong island, rationale adjunct always contains a possessive suffix, but no overt genitive argument. EXTRACTION ISLANDS IN FINNISH 67 3.3 Structures that show anti-agreement First, prepositions that do not inflect in φ-features allow extraction, as was illustrated e.g. in (29), repeated here as (126). These prepositions show free word order inside the PP. All PP-complements which take the partitive case complement belong to this class. (126) Mitäi Pekka ei pärjää [ ilman ti ] what.PAR Pekka not.3SG survive without ’What doesn’t Pekka survive without?’ I suggest that the essential difference between the extraction conditions on different types of prepositional phrases is that in anti-agreement structures, there is neither EPP-movement of the complement DP that would change the word order nor φ-feature agreement (127). The free word order inside the PP would be in this case a consequence of topicalization movement. (127) a. ilman häntä / häntä ilman without (s)he.PAR / (s)he.PAR without ’without him/her’ b. *ilman-sa häntä / *häntä ilman-sa without-pxsg3 (s)he.PAR / (s)he.PAR without-pxsg3 Let us now turn to infinitival clauses that allow extraction of genitive arguments. As seen in section 2.5.2, VA-construction allows extraction of any element. VA-infinitival may inflect with φ-features, but the inflection does not pattern with DPs, PPs, nor MA-adjuncts. Importantly, the φinflection in the presence of personal pronoun is ungrammatical (128a). It is clear that there is no φ-agreement between an overt subject and verb. However, I do not have an explanation on why the agreement is not triggered in VA-construction. (128) a. Pekka tiesi hänen lähtevän(*-sä). Pekka knew (s)he.GEN leave.VA/PRES(-Px/3SG) ’Pekkai knew that (s)he was going to leave.’ b. Pekka tiesi lähtevän-sä. Pekka knew leave.VA/PRES-Px/3SG ’Pekkai knew that hei was going to leave.’ Another example of genitive argument extraction is provided by the A-infinitive, which never shows φ-feature inflection and the extraction of genitive argument is always available. See Vainikka (1989, Ch.5) for discussion on the properties of the A-infinitival clause subject. 68 (129) SAARA HUHMARNIEMI Keneni Pekka käski ti lähteä? who.GEN Pekka ordered leave.A who did Pekka order to leave? 3.4 Finite complement clauses The left branch condition under φ-feature agreement is observable also in finite clauses. As seen in (18), extraction of subjects from complement clauses is not available, although object and adjunct arguments may be extracted (130): (130) *Kukai Pekka luuli että ti tapasi Merjan? Who Pekka thought that met Merja ’*Who did Pekka think met Merja?’ In the case of the nominative subject it is widely held that the subject undergoes EPP movement to the specifier of T and T agrees with the φ-features of the nominative subject (see e.g. Brattico & Huhmarniemi 2006 for Finnish). (131) CP C TP DPi TP n +φ TP ...ti ... However, when the verb does not agree with the φ-features of the subject, as in raising construction (19) , repeated here as (132), the extraction is marginally available. My informants evaluated (132) as only marginally acceptable, but nevertheless better than the sentence (130).25 (132) a. Minun kannattaa/*kannatan jäädä odottamaan I.GEN should.3SG/should.1SG stay.A wait.MA.ILL ’I should stay waiting.’ b. ??Keneni luulet että ti kannattaisi jäädä odottamaan who.GEN think.2SG that should.3SG stay.A wait.MA.ILL vuoroaan? turn.Px/3SG ’Who do you think should stay and wait for his turn?’ EXTRACTION ISLANDS IN FINNISH 69 To summarize, although the genitive argument in (132) seems to occupy the position of the nominal subject, it does not cause φagreement on the verb. It thus seems that φ-agreement together with EPP correlates with the unavailability of the subject extraction. 4. SUBJECT ISLANDS As discussed in the previous section, subjects that cause φ-feature inflection on the head cannot undergo A-bar movement to the front of the sentence. These DPs do not allow any elements to be extracted out of them either. Nevertheless, there are two infinitival clause types which allow A-bar movement of genitive subjects: the VA-construction and the A-infinitive. In this section, I will briefly examine the extraction conditions out of the subject position in these two infinitival clauses as well as finite clause. The subject position of the A-infinitive is a strong island for extraction. However, the subjects of VA-infinitivals are weak islands: they allow extraction of adjuncts but not complements (pointed out by Pauli Brattico, p.c.). The same is true for finite clauses: whereas adjunct extraction out of a finite clause subject seems to be available, complement extraction is not. It should be noted that the grammaticality estimations and analyses proposed in this section are tentative; further research is required for establishing a proper account of extraction out of the subject position(s) in Finnish. In the following subject extraction tests, I use the three types of DPs that allow extraction when the DP occupies a direct object position. First, kasa-constructions; second, PP-modifiers of nominals with event structure, in particular, minen-nominal; and third, nominals with Ainfinitival complement. Consider first the extraction out of the VA-infinitival subjects. The extraction out of kasa-construction is marginally available (133). In addition, the locative PP modifiers of minen-nominals may be extracted (134). However, the complement DP of a PP modifier cannot be extracted (135). (133) a. Pekka arveli [[ kuvan Merjasta] hävinneen] Pekka imagined picture.GEN Merja.ELA disappeared.VA ’Pekka imagined that the picture of Mary had disappeard.’ [[ kuvan ti ] hävinneen]? b. ??Kenestäi Pekka arveli who.ELA Pekka imagined picture.GEN disappeared.VA 70 SAARA HUHMARNIEMI (134) a. Pekka arveli [[ matkustamisen Thaimaahan] Pekka imagined traveling.GEN Thailand.ILL vähenevän ajan myötä] reduce.VA/PRES time along ’Pekka imagined that traveling to Thailand would reduce over time.’ b. ?Mihini Pekka arveli [[ matkustamisen ti ] where.ILL Pekka imagine traveling.GEN vähenevän ajan myötä]? reduce.VA/PRES time along lit. ’Where did Pekka imagine traveling would reduce over time?’ (135) a. Pekka arveli [[ juoksemisen kohti lavaa] Pekka imagined running.GEN towards stage.PAR aiheuttavan häiriötä] cause.VA/PRES disorder.PAR ’Pekka imagined that running towards the stage would cause disorder.’ Pekka arveli [[ juoksemisen kohti ti ] b. *Mitäi what.PAR Pekka imagined running.GEN towards aiheuttavan häiriötä]? cause.VA/PRES disorder.PAR The direct object of the A-infintival complement cannot be extracted (136b). However, extraction of adjuncts seems to be available (c). (136) a. Pekka arveli [[ suunnitelman kantaa vettä Pekka imagined plan.GEN carry.A water.PAR kaivoon] olevan huono] well.ILL be.VA/PRES bad ’Pekka imagined that the plan to carry water to the well was bad.’ [[ suunnitelman kantaa ti ] b. *?Mitäi Pekka arveli what.par Pekka imagined plan.GEN carry.A olevan huono]? be.VA/PRES bad c. ?Mihini Pekka arveli [[ suunnitelman kantaa where Pekka imagined plan.GEN carry.A vettä ti ] olevan huono]? water.PAR be.VA/PRES bad ’Where did Pekka imagine it was a bad idea to carry water?’ EXTRACTION ISLANDS IN FINNISH 71 If we adopt the approach where DP-internal PP-modifiers may occupy an adjunct position, then the island tests demonstrate that the genitive subject position of the VA-infinitive constitute a weak extraction island: whereas extraction out of complement is not possible, extraction of adjuncts is allowed at least to a certain extent. Let us now turn to genitive subjects of A-infinitivals. First, extraction out of the kasa-construction in genitive subject position is not possible (137). Similarly, extraction out of the nominal minen-complement (138) nor A-infinitival complement or adjunct is not available (139). The genitive subject of the A-infinitive thus seems to be a strong island for extraction. (137) a. Pekka käski [[ ryhmän koululaisia] kokoontua Pekka ordered group.GEN students.PAR meet.A uimahallin edessä] swimming pool.GEN front ’Pekka ordered a group of students to meet in the front of the swimming pool.’ ti ] kokoontua b. *Ketäi Pekka käski [[ ryhmän who.PAR Pekka ordered group.GEN meet.A uimahallin edessä] swimming pool.GEN front (138) a. Pekka antoi [[ yöpymisen luksushotellissa ] päättää Pekka let sleeping.GEN hotel.INE end.A hienon matkan] great.ACC journey.ACC ’Pekka let spending the night in a luxury hotel to end the great journey.’ b. *Missäi Pekka antoi [[ yöpymisen ti ] päättää where.INE Pekka let sleeping.GEN end.A hienon matkan]? great.ACC journey.ACC (139) a. Pekka antoi [[ ehdotuksen ostaa maata] lopettaa Pekka let proposal.GEN buy.A land.PAR end.A keskustelun] discussion.ACC ’Pekka let the proposal to buy some land to end the discussion.’ 72 SAARA HUHMARNIEMI b. *Mitä Pekka antoi [[ ehdotuksen ostaa ti ] lopettaa what.PAR Pekka let proposal.GEN buy.A end.A keskustelun] discussion.ACC c. *Mistä Pekka antoi [[ ehdotuksen ostaa maata where.PAR Pekka let proposal.GEN buy.A land.PAR ti ] lopettaa keskustelun] end.A discussion.ACC To conclude, consider extraction out of the finite clause subjects. Extraction out of the kasa-constructions seems to be available (140). In addition, whereas extraction of PPs out of minen-nominals is available (141), extraction of the prepositional complement is not (142). (140) a. [ Kuva Merjasta] on aiheuttanut kohua. picture.NOM Merja.ELA has caused fuss.PAR ’The picture of Merja has caused some fuss.’ ti ] on aiheuttanut kohua? b. *?Kenestäi [ kuva who.ELA picture.NOM has caused fuss.PAR ti ] aiheuttanut kohua? c. ?Kenestäi on [ kuva who.ELA has picture.NOM caused fuss.PAR ’About whom has a picture caused some fuss?’ (141) a. [ Yöpyminen [ joen lähellä]] ei houkutellut. sleeping.NOM river.GEN near not.3SG tempted ’Sleeping near the river was not tempting.’ b. [ Minkä lähellä]i yöpyminen ti ei houkutellut? what.gen near sleeping not.3SG tempted ’Near what wasn’t sleeping very tempting?’ c. [ Minkä lähellä]i ei [ yöpyminen ti ] houkutellut? what.gen near not.3SG sleeping tempted ’Near what wasn’t sleeping very tempting?’ (142) a. Ryntääminen kohti lavaa ei houkutellut. rushing.NOM towards stage.PAR not.3SG tempted ’Rushing towards the stage was not tempting.’ b. *?Mitäi [ ryntääminen [ kohti ti ]] ei what.PAR rushing.NOM towards not.3SG houkutellut? tempted EXTRACTION ISLANDS IN FINNISH 73 c. *?Mitäi ei [ ryntääminen [ kohti ti ]] what.PAR not.3SG rushing.NOM towards houkutellut? tempted In addition, extraction out of the A-infinitival complements and adjuncts is not availble at all (143). (143) a. [ Suunnitelma lakkauttaa kirjastoja] on aiheuttanut plan.NOM discontinue.A libraries.PAR has caused vastustusta. resistance ’The plan to discontinue libraries has caused resistance.’ b. *?Mitä [ suunnitelma lakkauttaa ti ] on aiheuttanut what.PAR plan.NOM discontinue.A has caused vastustusta? resistance ti ] c. *?Mihini on [ suunnitelma viedä kirjoja where.PAR has plan.NOM take.A books.PAR aiheuttanut vastustusta? caused resistance Extraction from finite clause subjects is thus very restricted. If PPmodifiers are considered as adjuncts, we could say that adjunct extraction is available and that the finite clause subjects are weak islands for extraction. All in all, the extraction out of the subject position seems to be very limited in all the three cases. REFERENCES Abney, Steven P. 1987. The English Noun Phrase in Its Sentential Aspect. Ph.D. thesis, MIT, Cambridge, MA. Boeckx, Cedric. 2003. Resumption and asymmetric derivation. In Anna Maria Di Sciullo (ed.), Asymmetry in Grammar, volume 1: Syntax and semantics of Linguistik Aktuell, Amsterdam/Philadelphia: John Benjamins. Brattico, Pauli. 2008. Kayne’s model of Case and Finnish nominal phrases. Nordic Journal of Linguistics 31.2, 135–160. Brattico, Pauli. 2009a. The Probe-Goal System in Finnish, manuscript submitted for publication. Brattico, Pauli. 2009b. The Two-Part Models and One-Part Models of Nominal Case: Evidence from Case Distribution. Journal of Linguistics . Brattico, Pauli & Huhmarniemi, Saara. 2006. Finnish Negation, EPP-principle and the valuation theory of morphosyntax. Nordic Journal of Linguistics 29.1, 5–44. Brattico, Pauli & Leinonen, Alina. 2009. Nominalization and case distribution: Evidence from Finnish. Syntax 12, 1–31. Chomsky, Noam. 1973. Conditions on transformations. In Stephen Anderson & Paul Kiparsky (eds.), A festschrift for Morris Halle, 232–286, New York: Holt, Rinehart and Winston. 74 SAARA HUHMARNIEMI Chomsky, Noam. 1977. On Wh-movement. In P. W. Culicover, T. Wasow & A. Akmajian (eds.), Formal Syntax, New York: Academic Press. Chomsky, Noam. 2000. Minimalist Inquiries: The Framework. In Roger Martin, Davic Michaels & Juan Uriagereka (eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–156, Cambridge, MA.: MIT Press. Chomsky, Noam. 2001. Derivation by Phase. In Michael Kenstowicz (ed.), Ken Hale: A Life in Language, 1–52, Cambridge, MA: MIT Press. Chomsky, Noam. 2008. On Phases. In Carlos P. Otero Roger Freidin & Maria L. Zubizarreta (eds.), Foundational Issues in Linguistic Theory, 133–166, Cambridge, MA: MIT Press. Cinque, Guclielmo. 1980. Extraction from NP in Italian. Journal of Italian Linguistics 5, 47–99. Davies, William D. & Dubinsky, Stanley. 2003. On extraction from NPs. Natural Language & Linguistic Theory 21, 1–37. Fiengo, Robert & Higginbotham, James. 1981. Opacity in NP. Linguistic Analysis 7, 395– 422. Hakulinen, Anne & Karlsson, Fred. 1975. Suomen akkusatiivi: funktionaalinen näkökulma. Virittäjä 1, 339–363. Hakulinen, Auli & Karlsson, Fred. 1979. Nykysuomen lauseoppia. Helsinki: SKS. Hakulinen, Auli, Vilkuna, Maria, Korhonen, Riitta, Koivisto, Vesa, Heinonen, T.R. & Alho, I. 2004. Iso Suomen Kielioppi. Helsinki: SKS. Holmberg, Anders & Nikanne, Urpo. 2002. Expletives, Subjects, and Topics in Finnish. In Peter Svenonius (ed.), Subjects, Expletives, and the EPP, 71–106, Oxford: Oxford University Press. Holmberg, Anders, Nikanne, Urpo, Oraviita, Irmeli, Reime, Hannu & Trosterud, Trond. 1993. The Structure of INFL and the Finite Clause in Finnish. In Anders Holmberg & Urpo Nikanne (eds.), Case and other functional categories in Finnish syntax, 177–206, Berlin: Mouton. Kaiser, Elsi. 2003. (Non)Locality in Anaphoric Relations. In Nelson & Manninen (2003), 1–48. Kanerva, Jonni. 1987. Morphological integrity and syntax: The evidence from Finnish possessive suffixes. Language 63, 498–501. Karttunen, Frances. 1975. The Syntax and Pragmatics of the Finnish Clitic -han. Texas Linguistic Forum 1, 41–50. Koskinen, Päivi. 1997. Features and categories: Non-finite constructions in Finnish. Ph.D. thesis, University of Toronto, distributed by Toronto Working Papers in Linguistics. Kuno, Susumu & Robinson, Jane J. 1972. Multiple wh-questions. Linguistic Inquiry 3, 463– 487. Manninen, Satu. 2003a. Finnish PPs and the Phase Inpenetrability Condition. In Nelson & Manninen (2003), 1–48. Manninen, Satu. 2003b. Small phrase layers: A study of Finnish manner adverbials, volume 65 of Linguistik aktuell. John Benjamins. Nelson, Diane. 1998. Grammatical case assignment in Finnish. London: Garland. Nelson, Diane & Manninen, Satu. 2003. Introduction. In Diane Nelson & Satu Manninen (eds.), Generative Approaches to Finninh and Saami Linguistics, 1–48, Stanford, CA: CSLI Publications. Nevis, Joel A. 1986. Finnish Particle Clitics and General Clitic Theory. Ph.D. thesis, The Ohio State University. Nikanne, Urpo. 1989. On locative case marking in Finnish. In Jussi Niemi (ed.), Papers from the 11th Scandinavian Conference of Linguistics, volume 1, Joensuu: Joensuu University. Nikanne, Urpo. 1993. On Assigning Semantic Cases in Finnish. In A. Holmberg & U. Nikanne (eds.), Case and Other Functional Categories in Finnish Syntax, 89–109, Berlin: Mouton. Ouhalla, Jamal. 1993. Subject-Extraction, Negation and the Anti-Agreement Effect. Natural Language and Linguistic Theory 11, 477–518. Reime, Hannu. 1993. Accusative marking in Finnish. In A. Holmberg & U. Nikanne (ed.), Case and Other Functional Categories in Finnish Syntax, 89–109, Berlin: Mouton. EXTRACTION ISLANDS IN FINNISH 75 Rizzi, Luigi. 1990. Relativized Minimality. Cambridge, MA: MIT Press. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegemann (ed.), Elements of Grammar, 289–330, Dordrecht: Kluwer. Ross, John Robert. 1967. Contraints on variables in syntax, mIT Ph.D. dissertation. Stowell, Tim. 1989. Subjects, Specifiers, and X-Bar Theory. In Alternative Conceptions of Phrase Structure, The University of Chicago Press. Szabolcsi, Anna. 1994. The noun phrase. The syntactic structure of Hungarian. In F. Kiefer & K.E.Kiss (eds.), Syntax and Semantics 27, 179–275, New York: Academic Press. Toivonen, Ida. 1995. A study of Finnish infinitives. Master’s thesis, Brandeis University. Toivonen, Ida. 2000. The Morphosyntax and Finnish Possessives. Natural Language and Linguistic Theory 18, 579–609. Trosterud, Trond. 1993. Anaphors and Binding Domains in Finnish. In Anders Holmberg & Urpo Nikanne (eds.), Case and Other Functional Categories in Finnish Syntax, 225– 244, Berlin: Mouton. Vainikka, Anne. 1989. Deriving Syntactic Representations in Finnish, ph.D. dissertation, University of Massachusetts, Amherst. Vainikka, Anne. 1993. The Three Structural Cases in Finnish. In Anders Holmberg & Urpo Nikanne (eds.), Case and Other Functional Categories in Finnish Syntax, 129–159, Berlin: Mouton. Vainikka, Anne. 1995. Functional Projections in Finnish Non-Finite Constructions. Technical report. Vainikka, Anne. 2003. Postverbal case Realization in Finnish. In Diane Nelson & Satu Manninen (eds.), Generative Approaches to Finninh and Saami Linguistics, 235–266, Stanford, CA: CSLI Publications. Vainikka, Anne & Brattico, Pauli. 2009. Accusative case in Finnish, manuscript submitted for publication. van Steenbergen, Marlies. 1991. Long-distance binding in Finnish. In J. Koster & E. J. Reuland (eds.), Long-distance anaphors, 231–244, Cambridge: Cambridge University Press. Vilkuna, Maria. 1995. Discourse Configurationality in Finnish. In Katalin É. Kiss (ed.), Discourse Configurational Languages, 244–268, Oxford University Press. 1 Koskinen (1997) assumes adjunction to T for manner adverbials and adjunction to vP for temporal adverbials. See Manninen (2003b), for an account in which manner adverbials are specifiers of layered vP s. Holmberg et al. (1993) suggest two positions as well, and locate the sentential adverbs on the top of the vP , whereas the rest are on the top of the VP. 2 I thus adopt the view that the EPP-movement of the subject DP is triggered by caseand/or φ-feature assignment, following the basic insights in Brattico & Huhmarniemi (2006). 3 Following (Rizzi, 1997), the CP-layer may be split into several projections, but the split is not relevant for present discussion. 4 NP in Ross’s terms 5 I exclude here the contrastive focus movement, which is sometimes assumed to target the specifier of C Vilkuna (1995), or some other functional head in C-domain. It seems that contrastively focussed constituent does not always cause intervention effect on whmovement. However, further research is needed. 6 To account for free word order variation, Manninen (2003b) postulates an optional EPP-feature on the P-head, which forces the DP to raise to the specifier of P. 76 7 SAARA HUHMARNIEMI The contexts where the extraction could be used is e.g. the following, where the preposition is focussed and therefore occupies the sentence-final position: (144) A: Ihmiset tarvitsevat teollisuutta elääkseen. people.NOM need industries to.live ’People need natural resources in order to survive.’ B: Kyllä, mutta mitä me voisimme elää ilman? yes, but what.PAR I.NOM could live without ’Yes, but what would I survive without?’ 8 If the PP is interpreted as forming a DP together with ’Merjaa’, the possessive suffix may be bound by Merja. However, this interpretation is not preferred here. 9 Some of the prepositions that assign genitive case may assign partitive case as well. In this case they do not take possessive inflection (145a-c). These PPs have similar extraction properties as the prepositions that take partitive case complements in general. (145) a. lähellä häntä / häntä lähellä near (s)he.PAR (s)he.PAR near ’near him/her’ b. *häntä lähellä-än (s)he.PAR near-Px/3SG c. *minua lähellä-ni I.PAR near-Px/1SG 10 The unavailability of extraction out of relative clauses was discussed in section 2.1. 11 The thematic properties of the DP seem to be relevant as well. For example, when the genitive DP is conceived as object argument in (146), extraction is available. (146) [ Minkä sillan lähelle] Asko ehdotti Pekan jättämistä which.GEN bridge.GEN near Asko.NOM suggested Pekka.GEN sleeping.PAR ti ? ’Near which bridge did Asko suggest we should leave Pekka?’ 12 In general, any nominal with event structure permits preposing and extraction of locative PPs in Finnish (147). However, the exact structural properties of these derived nominals is left open here. (147) a. ryntäys kentälle / kentälle ryntäys rush field.ILL field.ILL rush ’a rush to the field’ b. Minnei Pekka näki ryntäyksen ti ? where.ILL Pekka saw rush.ACC ’Where did Pekka see a rush to?’ 13 Hakulinen et al. (2004, §592) 14 The exact structural properties of the DP and reflexive binding are left open here. 15 An example of NP split in Finnish that does not reconstruct is the position of the numeral with respect to its complement, as in (148). In the split construction, the nominal head may be in plural, whereas in the complement of the numeral the nominal is always in singluar (148b-c). EXTRACTION ISLANDS IN FINNISH (148) 77 a. Miehiä oli kolme. man.PL.PAR was three ’There were three men.’ b. *Kolme miehiä three man.PL.PAR c. Kolme miestä three man.SG.PAR ’three men’ 16 According to Nevis (1986), Karttunen (1975) gives the examples (149a-b) where the adverb would be extracted out of the complement AP. However, this movement does not generalize very well (150). (149) a. sinä olet vielä kovin nuori you.NOM are still very young ’You are still very young.’ b. kovin sinä olet vielä nuori very you.NOM are still young (150) a. sinä olet vielä aika nuori you.NOM are still quite young ’You are still very young.’ b. *aika sinä olet vielä nuori quite you.NOM are still young 17 A-infinitive has an adjunct form as well, illustrated in (151). The subject is in the genitive case and the infinitival clause object is part of the main clause, as may be observed in case alternation between (a-b). (151) Pekka toi omenoita/omenan [ lasten syödä] Pekka brought apples.PAR/apple.ACC children.GEN eat.A ’Pekka brought some apples in order for children to eat.’ Movement of the A-adjunct as well as extraction out of it is not permitted (151a-b). The meaning of the infinitival is ’in order to eat’, which points towards an interpretation as adjunct. I leave the extract structure of the A-adjunct for future research. (152) syödä] omenoita [ ti a. *Keneni Pekka toi who.GEN Pekka brought apples.PAR eat.A omenoita ti ? b. *[ Kenen syödä]i Pekka toi Pekka brought apples.PAR who.GEN eat.A 18 VA-construction is also used in a complement of an auxiliary verb in present and past perfect in (a-b) and with negation (c). Extraction is freely available. (153) a. Pekka on kirjoittanut romaanin. Pekka has write.VA/PAST/SG novel.ACC ’Pekka has written a novel.’ b. Pekka oli kirjoittanut romaanin. Pekka had write.VA/PAST/SG novel.ACC ’Pekka had written a novel.’ 78 SAARA HUHMARNIEMI c. Pekka ei kirjoittanut romaania. Pekka not.3SG write.VA/PAST/SG novel.ACC ’Pekka did not write a novel.’ 19 Some speakers may allow a subject (154) (Toivonen, 1995). Due to the marginality of these sentences, the constructions containing a subject are left aside here. (154) Hän jäi henkiin [ meidän auttamalla häntä] (s)he.NOM stay-PAST.3SG life.ILL we.GEN help.MA.ADE (s)he.PAR ’He/she survived by our helping him’ 20 I leave the Case features aside, since they do not seem to have an effect on the extraction conditions. 21 The agreement view for possessive suffixes is adopted e.g. in Vainikka & Brattico (2009); Brattico (2009a). Toivonen (2000) suggest that possessive suffix is an agreement marker in 1st and 2nd person. Nelson (1998), following Reime (1993), suggests that possessive suffixes are a variant of verbal agreement marker, having the same general category AGR. 22 Similarly as with PPs, the possessive suffix is often not used in colloquial speech together with the personal pronoun: hänen edellä, hänen ohi. However, for the purposes of this paper, it suffices if the φ-feature inflection is not ungrammatical, which would indicate a different syntactic configuration. 23 The uninflected forms such as ohi may allow some word order variation when there is no possessive inflection. 24 Manninen (2003a, pp. 313) adopts the view from the Minimalist Program (Chomsky, 2000, 2001), where φ-feature agreement occurs before EPP-movement. 25 There existed a considerable amount of variation among the informants in grammaticality estimations; some informants did not accept long-distance movement at all.
© Copyright 2024