Debrecen Workshop on Argument Structure May 25 – May 27, 2012 University of Debrecen Debrecen, Hungary Locations: Talks: Room 111, Main Building Lunch and coffee breaks: Room 109, Main Building

Program Friday, May 25 9:00 – 10:00 10:00 – 10:15 Session 1 10:15 – 11:15 11:15 – 12:00 12:00 – 13:30 Session 2 13:30 – 14:15 14:15 – 15:00

15:00 – 15:45 15:45 – 16:15 Session 3 16:15 – 17:00 17:00 – 17:45

registration opening remarks Chair: György Rákosi invited talk by Louisa Sadler (University of Essex): Dative arguments in Maltese Júlia Bácskai-Atkári (Research Institute for Linguistics of the Hungarian Academy of Sciences): Arguments of adjectives in degree expressions lunch break Chair: István Kenesei Niina Ning Zhang (National Chung-Cheng University): Times, arguments, and the projection of VP shells Artemis Alexiadou, Gianina Iordachioaia, Fabienne Martin, Florian Schäfer (University of Stuttgart), and Mariangeles Cano (University of Madrid): External arguments in derived nominals Mikhail Knyazev (Utrecht Institute of Linguistics): Sentential arguments are DPs: evidence from nominalization coffee break Chair: John Beavers Nora Boneh (The Hebrew University of Jerusalem) and Léa Nash (Université Paris 8 / CNRS UMR 7023): Paths to datives Héctor Fernández-Alcalde (Universidad Autónoma de Madrid / CCHSCSIC): Dative case, prepositions, and argument structure in Spanish

i

Saturday, May 26 Session 4 9:30 – 10:30 10:30 – 11:15 11:15 – 12:00 12:00 – 13:00 Session 5 13:00 – 13:45 13:45 – 14:30 14:30 – 14:45 Session 6 14:45 – 15:30 15:30 – 16:15

16:30

Chair: Katalin É. Kiss invited talk by Balázs Surányi (Research Institute for Linguistics of the Hungarian Academy of Sciences and Pázmány Péter Catholic University): On the configurationality of argument positions in Hungarian Inna Tolskaya (CASTL, UiT): Verbal prefixes in Russian: conceptual structure versus syntax Víctor Acedo-Matellán and Cristina Real-Puigdollers (Universitat Autónoma de Barcelona): Aspectual and quantificational properties of locative verbs lunch break Chair: Anna Kibort Christopher Piñón (Université de Lille 3 / STL UMR 8163): The reflexive impersonal construction in Polish Alexis Dimitriadis and Martin Everaert (Utrecht Institute of Linguistics OTS): Characterizing reflexivization: semantic and syntactic perspectives coffee break Chair: Louise Mycock Attila Cserép (University of Debrecen): Idiom passivization from a cognitive linguistic viewpoint Tetsuya Kogusuri (Daito Bunka University): The passive of the gesture expression construction: event structure and discourse function departure for wine tasting and dinner in Tokaj/Bodrogkisfalud

Sunday, May 27 Session 7 9:30 – 10:30

10:30 – 11:15 11:15 – 12:00 12:00 – 13:30 Session 8 13:30 – 14:30 14:30 – 15:15

Chair: Valéria Molnár invited talk by Katalin É. Kiss (Research Institute for Linguistics of the Hungarian Academy of Sciences and Pázmány Péter Catholic University): Ways of licensing external possessors Eduardo Soares (Federal University of Rio Grande do Sul): Nonaffected incremental themes: the case of verbs of creation Razieh Shojaei, Gholamhossein Karimi-Doostan, and Ali Safari (Tehran University): A-structure alternation in Persian LVCs lunch break Chair: Christopher Piñón invited talk by John Beavers (The University of Texas at Austin): On non-agentivity and causer suppression in colloquial Sinhala Joseph Potashnik (Tel Aviv University): Constraining the unaccusative alternation ii

15:15 – 15:30 coffee break Chair: Péter Pelyvás Session 9 15:30 – 16:15 Orsolya Tánczos (Pázmány Péter Catholic University): Causative constructions in the Udmurt language 16:15 – 17:00 Balkiz Ozturk and Eser Erguvanli Taylan (Bogazici University): Transitivity in Pazar Laz

Alternates Enikő Kovács: The structure of the mental lexicon, a lexicalist viewpoint Ana Paula Scher (University of São Paulo), Alessandro Boechat De Medeiros (Universidade Federal do Rio de Janeiro) and Rafael Dias Minussi (University of São Paulo): Argument Structure in Distributed Morphology: evidence from Brazilian Portuguese Maryse Grône (Paris Diderot University): The influence of volitionality on the acceptability of unaccusatives and passives in the resultative construction in English

iii

Acknowledgments We would like to express our gratitude to the following colleagues for their assistance in reviewing abstracts for the Debrecen Workshop on Argument Structure: Gábor Alberti Alex Alsina József Andor Huba Bartos John Beavers Miriam Butt Liz Coppock Mary Dalrymple Martin Everaert Berit Gehrke Julia Horvath István Kenesei Anna Kibort Andrew Koontz-Garboden Helge Lødrup Valéria Molnár Louise Mycock Péter Pelyvás Christopher Piñón Adam Przepiórkowski Günter Radden Louisa Sadler Florian Schäfer Peter Sells Peter Svenonius Ida Toivonen Stephen Wechsler

iv

Contents

Louisa Sadler Dative arguments in Maltese …………………………………………………………………..7 Júlia Bácskai-Atkári Arguments of adjectives in degree expressions ……………………………………………….8 Niina Ning Zhang Times, arguments, and the projection of VP shells …………………………………………..10 Artemis Alexiadou, Gianina Iordachioaia, Fabienne Martin, Florian Schäfer, and Mariangeles Cano External arguments in derived nominals ……………………………………………………..12 Mikhail Knyazev Sentential arguments are DPs: evidence from nominalization ………………………………14 Nora Boneh and Léa Nash Paths to datives ……………………………………………………………………………….16 Héctor Fernández-Alcalde Dative case, prepositions, and argument structure in Spanish ……………………………….19 Balázs Surányi On the configurationality of argument positions in Hungarian …………………………….. 21 Inna Tolskaya Verbal prefixes in Russian: conceptual structure versus syntax ……………………………..22 Víctor Acedo-Matellán and Cristina Real-Puigdollers Aspectual and quantificational properties of locative verbs …………………………………24 Christopher Piñón The reflexive impersonal construction in Polish ……………………………………………..26 Alexis Dimitriadis and Martin Everaert Characterizing reflexivization: semantic and syntactic perspectives ………………………...28 Attila Cserép Idiom passivization from a cognitive linguistic viewpoint …………………………………..31 Tetsuya Kogusuri The passive of the gesture expression construction: event structure and discourse function ……………………………………………………………………………32 Katalin É. Kiss Ways of licensing external possessors ……………………………………………………….34 v

Eduardo Soares Nonaffected incremental themes: the case of verbs of creation ……………………………...35 Razieh Shojaei, Gholamhossein Karimi-Doostan, and Ali Safari A-structure alternation in Persian LVCs ……………………………………………………..37 John Beavers On non-agentivity and causer suppression in colloquial Sinhala ……………………………39 Joseph Potashnik Constraining the unaccusative alternation …………………………………………………...41 Orsolya Tánczos Causative constructions in the Udmurt language …………………………………………….43 Balkiz Ozturk and Eser Erguvanli Taylan Transitivity in Pazar Laz ……………………………………………………………………..45 Enikő Kovács The structure of the mental lexicon, a lexicalist viewpoint ………………………………….47 Ana Paula Scher, Alessandro Boechat De Medeiros and Rafael Dias Minussi Argument Structure in Distributed Morphology: evidence from Brazilian Portuguese ……..49 Maryse Grône The influence of volitionality on the acceptability of unaccusatives and passives in the resultative construction in English …………………………………………………….51

vi

Dative Arguments in Maltese Louisa Sadler (University of Essex)

This paper focusses on the syntactic expression of three argument verbs in Maltese and specifically on the class of ditransitive predicates in which the third argument is typically a beneficiary or recipient argument. Despite the relatively large literature on classes of ditransitive predicates crosslinguistically, the expression of such event types in Maltese has received very little attention. Maltese is from the South Arabic branch of Central Semitic, with a Maghrebi/Siculo- Arabic stratum, a Romance (Sicilian, Italian) superstratum and an English adstratum. Working within the framework of LFG's Lexical Mapping Theory, Kibort (2008) offers a typology of the patterns of alignment of arguments to syntactic grammatical functions for such ditransitive predicates, distinguishing between the dative shift construction (in which the beneficiary or recipient argument outranks the theme/patient argument), a canonical dative construction and an oblique recipient/beneficiary construction. We provide an extensive description of a range of ditranstive constructions in Maltese arguing that the language has both canonical dative constructions and dative shift constructions, with the latter construction being available for only a small number of predicates, and show how the approach to ditransitive constructions which Kibort (2008) develops provides a suitable framework for modelling the Maltese data. This talk builds on joint work with Maris Camilleri (University of Essex).

Arguments of Adjectives in Degree Expressions The aim of my talk is to provide a unified analysis for the structure of degree expressions, with special attention to arguments of adjectives; the survey will be centred on – though not limited to – English data. Using a Minimalist framework, I will show that absolute, comparative, and superlative degree expressions can be analysed in the same way and that the arguments of adjectives not only fit into the established framework but they also provide additional evidence for it. Certain adjectives are known to take arguments of their own. Consider: (1) Liz is proud [PP of her husband]. In this case, the PP of her husband is an argument of the adjective proud. I will discuss three major questions in connection with this. First, such adjectives can be attributive modifiers in nominal expressions only without their arguments: (2) Liz is a proud woman. (3) *Liz is a proud [PP of her husband] woman. Second, it must also be clarified how arguments of adjectives can be accommodated into degree expressions containing other obligatory elements, such as a comparative subclause (a CP introduced by than): the PP must precede the CP. Compare: (4) Liz is prouder [PP of her husband] [CP than Mary is]. (5) *Liz is prouder [CP than Mary is] [PP of her husband]. Third, in some cases the PP argument may be separated from the adjective head, unlike in (5). In (6), enough appears between the adjective and the PP; in (7), the subclause contains the PP of Peter without the AP it is supposed to be part of (i.e. the AP x-proud, x referring to a certain degree of pride): (6) Liz is proud enough [PP of her husband]. (7) Liz is prouder [PP of George] than Mary is ___ [PP of Peter]. (___ = x-proud) As for the syntax of degree expressions, I will adopt the following structure, using the example of the QP far more interesting than the first one: (8)

QP Q’

QP far

Q much+ -erj

DegP Deg’

APi interesting Deg tj

RP R’

ti R Ø

CP than the first one

The AP and the CP are generally taken to be arguments of the degree head (cf. Lechner 2004). These are – following den Dikken (2006) – located in a Relator Phrase (RP), the head

of which is empty here. The AP moves up the specifier of the DegP, generated right above the RP, in order to agree with the head. This agreement relationship is necessary to rule out impossible configurations such as a non-scalar adjective combining with a comparative head (e.g. *more pregnant). The Deg head moves up to the Q head: here it combines with much to form more; in morphological comparatives, the Q head is zero and the degree morpheme is morphologically merged with the AP in PF (e.g. good + -er will give better). Optional QP modifiers are located in [Spec; QP]; these also agree with the Q head, i.e. a QP like very can occur in absolute constructions but not in comparatives (very tall vs. *very taller), which is the other way round for far (*far tall vs. far taller). If the adjective has a PP argument of its own, the adjective establishes a predicative relationship with it. I will show that in this way the adjective acquires a predicative feature [+pred], which percolates up to the QP and which is interpretable on the whole QP as well. Though QPs can generally be modifiers in nominal expressions, a [+pred] QP cannot agree with a [–pred] noun phrase, which would be necessary for an attributive relationship involving agreement. Hence the ungrammaticality of (3), as opposed to the grammatical predicative usages in (1) – or, as postnominal (adjunct) modifiers, in (9): (9) Liz is a woman proud [PP of her husband]. The feature-based approach also explains the behaviour of certain adjectives that are inherently predicative, i.e. they cannot be used as attributes even on their own: (10) *I saw an afraid girl. (11) *I saw an afraid [PP of snakes] girl. (12) I saw a girl afraid [PP of snakes]. I will show that the proposed analysis is applicable to all degree expressions and that the difference between pronominal vs. postnominal appearance is truly dependent on the [±pred] features and not on the Deg heads; the Deg heads define the subtype of pronominal appearance only (e.g. in the case of too, the QP moves to a higher functional position than the NumP, resulting in structures like John is too proud a man, as opposed to *proud a man). As (4) shows, the CP, if any, is extraposed from within the degree expression; I will adopt the analysis of Kántor (2008), who claims that this is due to the fact that the CP is a phase spelt out on its own, hence extraposition is due to the nature of PF mechanisms and not to syntactic movement. I will show that the same applies to the extraposition of certain PPs (cf. Lee-Schoenfeld 2007), which are invariably predicative (i.e. arguments of adjectives). Nevertheless, the CP and the PP are sent earlier to the interfaces and will thus appear rightmost at PF (the CP sent first will be the very last). Due to separate Spell-Out, QPs can be deleted without affecting the PP, as in (7). Since Deg heads precede PPs anyway, the extraposition of the PP is normally not visible. The only exception is enough, as in (6), which I will show to be a RELATOR head and hence appearing after adjectives, unlike Deg heads (e.g. too, how, more); in addition, enough can appear together with its subclause (e.g. a big house, enough for us to live in), an option not available for Deg heads. References den Dikken, Marcel (2006) Relators and Linkers: The Syntax of Predication, Predicative Inversion, and Copulas. Cambridge, MA: MIT Press. Kántor, Gergely (2008) A Phase-based Approach to Rightward Movement in Comparatives. Newcastle Working Papers in Linguistics 14. 81–99. Lechner, Winfried (2004) Ellipsis in Comparatives. Berlin–New York: Mouton de Gruyter. Lee-Schoenfeld, Vera (2007) Beyond Coherence: The Syntax of Opacity in German. Amsterdam: John Benjamins.

Times, Arguments, and the Projection of VP Shell Based on the homogeneousness of the properties of an argument with respect to time, Klein (2010) discusses certain types of verb stems in their default readings. Heterogeneous Intransitive (HEI) and Transitive (HET) cover telic and degree achievement verb stems, and Homogeneous Intransitive (HOI) and Transitive (HOT) cover the rest. We propose that arguments of the four types of verb stems are base-generated as in (1). In the theta-domain, HE verbs have vP, whereas HO ones do not, regardless of how many arguments occur. In (1b), the two DPs are the specifier and complement of V. Supporting facts are listed below. e.g. sleep, dance, vibrate, be (1) a. HOI: [VP V DP] b. HOT: [VP DP1 [V DP2]] e.g. cost, weigh with a measure phrase, resemble, admire c. HEI: [vP v [VP V DP]] e.g. die, (intransitive) drown, rise, remain d. HET: [vP DP1 v [VP V DP2]] e.g. leave, close, slay, (transitive) drown, observe Structure Integrity All of the arguments to support severing the subject from VP or vP shell projecting (Marantz 1984; Larson 1988; Kratzer 1996) come from HE, rather than HO, constructions. The latter fails in pseudo-cleft (Zucchi 1998: 349) (2a) and replacement by the vP-proform do so (Ross 1970; cf. Stroik 2001: 367) (2b,c). These are covered by (1b), where [V DP2] is an intermediate projection, which lacks visibility (Chomsky 1995), and is not vP. (2) a. *What John did was resemble his father. b. *Mary likes Sam, and Chris does so too. c. *The shoes cost 5 dollars, and the gloves do so too. Structure Richness Why is there no ditransitive or applicative HO verb? If only HE verbs may host their arguments in two layers of VP, the contrast in (3) is explained. (1b) does not have enough positions for three arguments, if no multiple Specs are allowed in theta-domain. (3) a. I gave him the clothes. [HET] b. *I like him the clothes. [HOT] Intended: ‘I like him with respect to the clothes.’ There is no HO object-control construction, which has a DP and a clause complement. (4) a. Mary forced John to feed the baby. b. *Mary admired John to feed the baby. Intended; ‘Mary admired John for his feeding of the baby.’ Many verbs may occur in either HO or HE structures (Dowty 1979: 60; Rosen 1999). It is easy to change an otherwise HO construction into a HE one by adding a delimitable element. But we do not add material to a HE structure to change it into a HO one (Thompson 2006: 218). As in (3) and (4), the data in (5)–(10) show HE structures are richer than HO ones. (5) ADDITION OF DIRECT OBJECT a. Bill ran (*in 5 minutes). [HO] b. Bill ran the mile in 5 minutes. [HE] (6) ADDITION OF INDIRECT OBJECT a. That book costs three dollars. [HO] b. That book has cost me three dollars. [HE] (7) ADDITION OF COGNATE OBJECT a. Terry sang (*in an hour). [HO] b. Terry sang the ballad in an hour. [HE] (8) ADDITION OF X’S WAY EXPRESSION a. Terry sang (*in an hour). [HO] b. Terry sang her way to the Met in 10 years. [HE] (9) ADDITION OF FAKE REFLEXIVE a. Terry sang (*in an hour). [HO] b. Terry sang herself to sleep in an hour. [HE] (10) ADDITION OF RESULTATIVE 1

a. Terry ran (*in an hour). [HO] b. Terry ran us ragged in an hour. [HE] Note that heterogeneous events can be repeated (Mary dried the dishes for hours before being released from duty) (Thompson 2006: 218; Ramchand 2008: 31). It is just like all nominals can be counted if an appropriate unit is identified (two drops of water as well as two books). Certain formatives mark the HE status of the verb, and their absence marks the HO status, e.g., meg- in Hungarian (Hopper & Thompson 1980: 267). Presumably, such formatives are licensed by vP, and thus the HE reading correlates with a richer structure. Intransitives exhibit HO-HE contrasts Only HEI allow expletives. This is explained if the expletive is base-generated in Spec of vP (Deal 2009). (1c), but not (1a), has vP. (11) a. There arrived a train in the station. [HE] b. *There laughed a man in the hallway. [HO] Transitives exhibit HO-HE contrasts In some languages (e.g. Finnish, Hungarian) only HE structures have Acc case marker (Hopper & Thompson 1980: 262). The fact is captured in the (1b)-(1d) contrast: the overt case marking is licensed by vP in the languages (Chomsky 1995). Cross-transitivity HO-HE contrasts Certain rules apply to both HET and HEI, but not HOT and HOI. First, re- occurs with HE verbs only (12) (Horn 1980). Since again behaves differently, the issue is not semantic. Note that re- scopes over either the HE root or the affected DP (Marantz 2005). But the ambiguity is independent of the HE restriction. (12) a. The door reopened. b. I reopened the door. [HE] c. John {*resmiled/smiled again}. d. *John re-admired his father. [HO] Second, time frame PPs occur with HE verbs only (13). Thompson (2006) shows that such PPs are licensed by a projection higher than VP. Their absence in HO constructions indicates that the structures of HO constructions are lower and thus simpler than that of HE ones. (13) a. John walked to the store in two hours. b. John destroyed the toy in two hours. [HE] c. *John slept in two hours. d. *John admired his father in two hours. [HO] The proposal in (1) may also show the following properties: 1 Generalization: Experiencer is never base-generated in vP, regardless of the position of the other argument in the structure (Belletti & Rizzi 1988; Pesetsky 1995). (14) a. Ingrid fears ghosts. (DP1 of (1b)) b. The ghosts threatened Ingrid. (DP2 of (1d)) 2 Generalization: Causative-inchoative alternation occurs with HE verbs only, i.e., vP but not VP. 3 Generalization: Reinhart’s (2002) [+c] roles (agent/causer/instrument) are DP1 of (1d). 4 Simplicity: (1) avoids the complexity and arbitrariness of the stipulation that all verbs project vP but v has different “flavors” (Arad 1999). 5 Integration: Based on Klein’s insight, (1) integrates event structures with argument structures. 6 Generalization: The structures of homogeneous eventuality constructions are simpler than those of heterogeneous ones. This is parallel to Borer’s (2005) analysis of nominals: the structures of mass nominals are simpler than those of count ones. The former shows homogeneousness, whereas the latter does not. Selected references Deal, A. 2009. The origin and content of expletives: evidence from “selection”. Syntax 12: 285-323. Dowty, D. 1979. Word meaning and Montague Grammar, Reidel. Hopper, P. & S. Thompson. 1980. Transitivity in grammar and discourse. Language 56: 251-299. Horn, L. 1980. Affixation and the Unaccusative Hypothesis. CLS 16: 134-146. Klein, W. 2010. On times and arguments. Linguistics 48: 1221-1253. Rosen, S. 1999. The syntactic representation of linguistics events. GLOT International 4.2: 3-11. Ross, J. 1970. Act. In D. Davidson & G. Harmon (eds.) Semantics of natural language, 70-126, Reidel. Stroik, T. 2001. On the light verb hypothesis. Linguistic Inquiry 32: 362-369. Thompson, E. 2006. The structure of bounded events. Linguistic Inquiry 37: 211-228. Zucchi, S. 1998. Aspect shift. In S. Rothstein (ed.) Events and Grammar, 349-370, Kluwer. 2

External arguments in derived nominals 1. We offer a syntactic analysis of an interpretative restriction on the external argument of derived nominals. Sichel (2011) recently (re)formulated this restriction as follows: while accomplishment verbs allow both direct and indirect participants/causers as external arguments, nominals derived from them require direct participation (irrespectively of the ±animate nature of the external argument). Based on cross-linguistic evidence (English, German, Greek, Romanian, Spanish and French), we propose that this restriction is related to the absence of Voice in derived nominals and the thematic properties of the preposition introducing event participants in the nominal domain. 2. As is well known, English nominalizations are morpho-syntactically deficient in comparison to their verbal sources. For instance, they do not allow ECM, double objects, and particle shift (Kayne 1984, Abney 1987), that is, constructions that have been argued to require a VP shell structure (Larson 1988, Harley & Noyer 1999). A further property that characterizes only a subset of English nominalizations, namely, derived nominals, and on which we focus here, is the restriction on the nature of their external argument. While verbal forms allow both direct and indirect participants/causers, direct participation is required in derived nominals, illustrated in (1) with respect to the by-phrase realizing the external argument. (We concentrate on by-phrases; the same effect holds with English transitive nominals (see Marantz 1997, Harley & Noyer 2000, Pesetsky 1995, Grimshaw 1990, Sichel 2010/2011), but the other languages we will discuss don’t permit transitive nominals for independent reasons.) Following Sichel (2011), we call this the direct participation effect. Derived nominals contrast with ing-of gerunds, which allow indirect causers as external arguments (2), although they share properties in terms of morpho-syntactic deficiency (e.g. no ECM, etc.; cf. Chomsky 1970): (1) a. The expert/the results verified the initial diagnosis b. The verification of the initial diagnosis by the expert/#by the results c. ?the verifying of the initial diagnosis by the experts/by the results 3. To our knowledge, Sichel (2010) is the first to offer an account of this phenomenon. She claims that this behavior cannot follow from the general morpho-syntactic deficiency of derived nominals, as ing-of gerunds are equally deficient. To account for the contrast in (1a, c) vs. (1b), she proposes that derived nominals of accomplishments do not inherit the event structure of their source verb. While accomplishments are complex events (Grimshaw 1990), their derived nominals denote simple events, and as such require direct participation of the external argument, which must be co-temporal with the unfolding simple event. By contrast, ing-of gerunds can denote complex events, and thus permit indirect causers as external arguments, which are not co-temporal with the unfolding event. Sichel does not offer an explicit syntax-event structure mapping, and phrases her account in the framework of Rappaport Hovav & Levin (1999). From this perspective, unlike their verbal source, derived nominals of accomplishment verbs have a simple event structure, e.g. [x ACT ]. By contrast, ing-of gerunds maintain the complex event structure of the verb, e.g. [[x ACT] CAUSE [BECOME [y ]]]. In syntactic accounts of event structure such as e.g. Ramchand (2008) and Harley (2011), the former have the structure of activities, i.e. lack a Result State component and simply contain a Process/v head, while the latter contain such a component [v + ResultP]. But note that the morpho-syntactic composition of English nominals such as in (1b) does not seem to match the semantic interpretation attributed to them by Sichel. Following Harley (2011), the verbalizing affix -ify in verify realizes the Process/v head and combines with the verbal root -ver that realizes a semi-compositional result head in (2). Thus the structure in (2) necessarily corresponds to the structure of an accomplishment and it nevertheless builds derived nominals. Furthermore, derived nominals permit reprefixation, e.g. the re-verification of the diagnosis. Re- has been argued to be compatible with accomplishments only (Wechsler 1990). This makes Sichel's account untenable. (2) [vP ify [ResP ver]] 4. We propose that the effect observed by Sichel is indeed related to a structural difference between the two nominal types, but of a different kind, since both are accomplishments. It is a difference relating to the structural level introducing the external argument (Voice). Following Kratzer (1996), we attribute the difference between derived nominals and ing-of gerunds to the absence of Voice in the former, but not in the latter. Voice introduces and thematically licenses the external argument in the active and in the passive. Here we view

nominalization as a type of passivization (Grimshaw 1990). If the nominalization contains Voice, then the thematic role it carries should be similar in the nominal and the verbal domain. In the absence of Voice, languages appeal to their inventory of lexical prepositions to introduce event participants. In some languages, these prepositions are thematically restricted. For English, this means that derived nominals lack Voice, while ing-of gerunds have Voice. Evidence for this comes from the observation that English ing-of gerunds are not compatible with a self-action interpretation, while derived nominals are: in (3), registration, but not registering allows an interpretation where the children registered themselves. (3) The report mentioned the painfully slow registering/registration of the children Following Kratzer (2003), we conclude from (3) that ing-of gerunds require the implicit realization of the verb’s external argument under Voice, just as verbal passives do (see 4a). The by-phrase acts as a purely formal licenser of the external argument, whatever thematic properties Voice assigns to it. The interpretation of by-phrases in derived nominals is restricted, however. In (4b) the by-phrase is not related to a Voice-head but acts as a nominal modifier; in the absence of Voice, derived nominals behave similarly to non-derived nouns (the book by Chomsky). Crucially, the type of by involved can only introduce an agent or direct causer, see Fox & Grodzinsky (1998), Alexiadou & al. (AAS, 2009). (4) a. [ing [VoiceP [vP [Root]]] b. [ation [vP [Root]]] 5. Support for our approach comes from comparing the English data with the corresponding derived nominals in German, Greek, Spanish, French, and Romanian. While Romanian (5) and Spanish show the direct participation effect, the other three languages don’t (6, German): (5) justificarea masurilor de austeritate de catre Emil Boc/#ratingul scazut al SUA justification measures of austerity by Emil Boc/#the rating down-graded of US (6) Die Bestätigung der ursprünglichen Diagnose durch die Ergebnisse des Tests The confirmation of the initial diagnosis by the results of the test Our structural analysis predicts that: i) Romanian and Spanish nominals should lack Voice and their prepositions should be thematically restricted to direct participants (as they are similar to English derived nominals); ii) as Greek, German and French nominals are similar to ing-of gerunds (i.e. no restrictions with respect to the interpretation of the by-phrase), they should either have Voice (option 1) or compensate the absence of Voice through a semantically powerful lexical preposition which is flexible enough to thematically license direct as well as indirect participants (option 2). These predictions are borne out: i) Romanian and Spanish nominals are compatible with self-action (see (7) for Romanian), so they lack Voice, as predicted. In both languages, the prepositions for external arguments, de catre and por, introduce only agents/direct causers, i.e., they are similar to English by. (7) anuntarea oaspetilor (Ag = Th) (8) die Anmeldung der Gäste (Ag = Th) the announcement guests.Gen the announcement the.Gen guests ii) Option 1 is exemplified by Greek, and Option 2 by German and French. Greek nominals are incompatible with self-action, so they include Voice (see AAS 2009; apo and me in (6b) introduce direct and indirect causers, respectively). But German (8) and French nominals can have a reflexive interpretation, suggesting that they lack Voice. (Note that German -ung nominals are possible only from verbs that are accomplishments, qualifying thus as complex events, e.g. *Laufung ‘running’ (Roßdeutscher 2010); this is a further counter-argument to Sichel's account.) We explain these cases via the semantic flexibility of the prepositions involved: in German the preposition durch isn’t restricted to direct participants and is compatible with other thematic roles in non-derived nouns, cf. *the danger by the storm vs. its German counterpart die Gefahr durch den Sturm. French par is also unlike English by: it can co-occur with a direct participant, suggesting that it is not restricted to a direct participant interpretation, see Kayne (1977:228) Sa vérification du diagnostic par l'expérience 'His verification of the diagnostic by the experiment'. 6. Our analysis supports the view that not all types of nominalizations share the same internal structure (ing-of gerunds vs. derived nominals, Abney 1987, and others). Moreover, it implies that there are two types of by-phrases: an argumental one in the presence of Voice, and a modifier one in its absence. We will show that this is confirmed by the Romanian supine and French -age nominals which always project Voice and whose by-phrases are argumental (see Cornilescu 2001, for Romanian).

Grimshaw (1990) notes an interesting fact about she calls complex event nominals, namely that they do not license sentential complements, as can be seen from examples like (1a), in contrast to result nominals, as in (1b). (1a) (1b)

*Their frequent announcement that they were the greatest eventually became tiresome. ?Their frequent announcements that they were the greatest eventually became tiresome.

The explanation she proposes for the contrast in (1) is the following. First, she argues that derived nominals fall into two major classes. Complex event nominals, on the one hand, and simple event nominals and result nominals, on the other. What distinguishes the two is that only the members of the former class have argument structure in the sense of projecting syntactic arguments. What looks like arguments of the latter class is actually complements, which correspond to arguments in the lexical conceptual structure of the lexical item but are not syntactic arguments. The crucial difference between complements and syntactic arguments is that only the latter are subject to theta-role assignment. Second, she assumes that nouns have no direct theta-marking capacity and hence require prepositions to transmit their theta-role to the argument. Thus, in (2a), money can only get its Theme role via the preposition of while in (2b), it gets its role directly from the verb. (2a) (2b)

donation *(of) money to hospitals They donated money to hospitals.

Unacceptability of (1a), then, follows from the fact the CP being an argument of a complex event nominal has to get a theta-role but fails to do so due to nominals’ inability to (directly) theta-mark. I believe, however, that this account cannot be correct. Consider the data on Russian nominalizations in -nie/-tie (henceforth, nouns in -nie), which are usually taken to be able to project argument structure (see, e.g., Tatevosov 2012 and references therein). These nominalizations routinely allow their internal arguments to be marked with genitive case, as in (3). Given Grimshaw’s assumptions, ‘the letter’ has to be theta-marked but since there is supposedly no preposition in the structure, we may conclude that nouns in -nie have direct theta-marking capacity. (3)

na-pisa-n-ij-e prf-write-n/t-noun-nom ‘writing (of) a letter’

pis´m-a letter-gen

Now let’s look at nouns in -nie derived from verbs taking sentential complements. A case in point would be opoveščat’ ‘notify’, which takes both an internal accusative argument and a sentential complement (4). (4)

Ona opoveščala znakomyx, čto vyxodit zamuž. she notified acquantances that goes.out married ‘She notified her acquantances that she is going to get married’.

In the corresponding nominalization the internal argument can be realized and marked with genitive case, which shows that argument structure is indeed projected. The crucial thing, however, is that the sentential complement cannot be realized unless embedded in an aboutPP (5).

(5a) Administracijej provoditsja opovešče-nie naselenija o tom, authorities.INS is.organized notify-NIE population.GEN about it čto iz zooparka sbežal tigr. that from zoo escaped tiger ‘Notification of the population is being made by the authorities about the fact that a tiger escaped from the zoo’. (5b) *Administracijej provoditsja opovešče-nie naselenija, authorities.INS is.organized notify-NIE population.GEN čto iz zooparka sbežal tigr. that from zoo escaped tiger The contrast in (5) is striking because there is evidence that nouns in -nie do have ability to directly theta-mark their arguments so the sentential complement in (5b) should be able to get its theta-role. This suggests that something else is wrong with (5b) and presumably with the English (1b) as well since it would be implausible if the same fact received different accounts in two different languages. I would like to argue that the source of unacceptability of (1b)-(5b) is the lack of structural (accusative) Case. This possibility was considered but rejected by Grimshaw based on the assumption that clauses do not need Case. This assumption, however, has been questioned most notably by Bošković (1995), who argued that clauses optionally do have Case features. In his story, what forced clauses to have a Case feature was their need to move. Clearly, in the cases under discussion it must be something else. I would like to propose that clauses that function as syntactic arguments are DPs and hence have to be licensed by a Case-assigning head. This account leads to an interesting prediction that appears to be confirmed, namely, that when verbs like ‘notify’ are passivized with their internal argument moved to the subject position the sentential complement becomes degraded (since there is no Case-licensor for them). This account also has the strong consequence that clauses in Case-less positions (and here I depart from Bošković in assuming that clauses co-occurring with an accusative internal argument do receive Case, although it cannot be realized on another DP with a nominal head) are not syntactic arguments but rather modifiers or complements in Grimshaw’s sense. This might be true for sentential complements to adjectives, which also do not assign accusative Case, under the assumption that adjectives do not project syntactic arguments. Further consequences have to be empirically testified. What might the proposed requirement on argumental clauses follow from? It might be the case that syntactic argumenthood is intrinsically linked to DP/NP-hood, which is perhaps an interesting hypothesis to explore. References Bošković, Ž. 1995. 'Case Properties of Clauses and the Greed Principle', Studia Linguistica 49, 32-53. Grimshaw, Jane. 1990. Argument structure. Cambridge, Mass.: MIT Press. Tatevosov, Sergei. 2011. Severing perfectivity from the verb. Scando-Slavica 57(2).

Paths to datives Background. According to lexicalist theories, predicates select from one up to three arguments. Until the late 80s, three place predicates have posed a problem for their syntactic structuration based on the binary branching. The two types of solutions that were proposed: (i) small clause approach, where the predicate selects a clause in which two internal arguments are locally linked (Kayne 1984 a.o.) and (ii) the predicate is decomposed into two (or more) syntactic heads each selecting one argument (Larson 1988 a.o.) cannot predict why the number of internal arguments should be limited to two, and cannot account for the fuzzy argument/adjunct status of the third participant either. These issues are well-illustrated by ditransitive verbs: cross-linguistically, beyond core cases such as give, sell, hand, it is difficult to decide whether the Recipient/Goal is a true argument of verbs such as write, throw. Moreover, many languages treat Recipients on a par with possessors and beneficiaries (e.g. from German: (weil) Anne ihmDAT den Teller zerbrach ‘Anne broke a dish for him’; (weil) ich schrieb ihrDAT die Antwort ‘I wrote her the answer’ vs. (weil) ich ihmDAT ein Buch gab ‘I gave him a book’ (examples from McIntyre 2006), (see Malchukov et al. 2010). Claim. We propose a new look at the distinction between lexical and non-lexical – or core and noncore – third participants marked with dative case (or by a dative-like preposition), which is largely based on Beavers’ (2011) analysis of ditransitive verbs as one aspectual class including a scalar change along a path. Assuming that predicates are associated with a single scale dimension, we claim that “core-datives” realize an initial or final point on a path scale on which the theme argument is located as the result of change, whereas “non-core datives” are optional event participants that may be associated with predicates denoting other types of scalar change: property scales, extent/volume scales; they are not event delineators, directly engaged in a specific scale construal. We show that this semantic divide can be traced by various syntactic and morphological means in typologically unrelated languages. Data I. Georgian a- dative constructions. Canonical ditransitive verbs obligatorily contain the verbal affix a- (or its ø-allomorph), whose presence licenses a dative marked DP, unambiguously interpreted as a recipient: (1) Vanom a-txova/a-chuka /a-chvena C’igni gogos Vano.erg lent/granted/showed book.abs girl.dat ‘Vano lent/granted/showed the girl a book.’ The a- affix may appear, systematically, with a class of otherwise transitive verbs (e.g. creation verbs, but not only), (2a), yielding a ditransitive construction, where the dative marked DP denotes a surface from which or to which the theme is caused to move (2b). (2) a. Vanom da=C’era leksi b. V. da=a-C’era leksi gogos / tixas Vano.erg wrote poem.abs V.erg wrote poem.abs girl.dat / clay.dat ‘Vano wrote a poem.’ ‘V. wrote a poem on a girl (as a surface) / on clay.’ (2b) does not mean that the dative marked DP is a benefactor of any kind. It is not restricted to humans. The dative DP is locational, the poem was applied on the girl/clay by means of writing. Importantly, in Georgian, change of state predicates (associated with a property scale) are not compatible with these “ditransitive constructions”: while deadjectival predicates are completely excluded, in spite of the presence of the affix a- (3a), canonical verbal change of state predicates such as break, cut, pulverize give rise to a reading where the verbal root is interpreted as denoting the manner component of a bi-eventive predicate, not its result component (3b), (cf. Levin & Rappaport-Hovav 2010). (3)a. *Vanom da=a-mrgvala comi tapas b. Vanom da=a-t’exa kvercxi papas Vano.erg rounded dough.abs pan.dat Vano.erg broke egg.abs porridge.dat Intended: ‘Vano applied dough to the pan ‘Vano applied (added) an egg to the by rounding it.’ porridge by breaking it.’ It ensues that the morpheme a- licenses a dative marked DP with a change of location construal, but fails to do so with a change of state (result state) construal. Georgian possesses another means for 1

licensing dative marked DPs: the affix u-, which is considerably less selective as to verbs it can attach to. The dative marked DP is interpreted as a concerned participant of the underlying event (‘benefactive’ dative), (4): (4) a.V. da=u-C’era leksi ninos b. V. da=u-mrgvala comi dedas V. wrote poem.abs Nino.dat V. rounded dough.abs mother.dat ‘Vano wrote poems for Nino.’ ‘Vano rounded dough for his mother.’ Data II French. A reminiscent contrast is attested in French: With change of location verbs variable binding between the dative marked DP and the theme is symmetrical (5), while with change of state verb, variable binding is asymmetric (6), the dative being hierarchically higher than the theme. (5) a.Elle a rendu à chaque élève son cartable. b. Elle a rendu à son propriétaire chaque cartable She gave-back to every pupil his schoolbag She gave-back to its owner every schoolbag (6) a.Elle a peint à chaque habitant sa maison. b. *Elle a peint à son locataire chaque maison. She painted to every inhabitant his house She painted to its tenant every house. Other distinguishing syntactic properties between the two types of French dative concern the possibility to appear in nominalizations, word-order differences, restrictions on determiners of the direct object DP (Boneh & Nash to appear). Both French and Georgian present a similar divide between the event type of the underlying verb and interpretation of the dative marked DP. They differ as to the formal means expressing this divide: morphological in Georgian (there are no differences in c-command asymmetries between Georgian a-/u- dative constructions), syntactic in French. Outline of the analysis. Ditransitives are bi-eventive constructions consisting of a cause event and a small clause (Marantz 2011) realizing a path-scale construal. A dative DP/PP in these constructions denotes an initial or final point on a simplex/complex path-scale: the resutling change that the theme argument undergoes is coming into/off contact with this point. Other bi-eventive verbs that express change of state are also decomposed into a causing subevent and a lower subevent involving a property-scale construal where the theme gets located at some point of the scale denoted by the predicate as a result of the event. The impossibility to interpret the dative DP as a recipient with property-scale related predicates follows if we assume that each predicate is associated with a single type of scale. “Non-core” datives are not related to a particular type of scalar dimension lexicalized by the verb; these will be shown to be secondary subjects, taking the lower subevent as their predicate by a process of lambda-abstraction. Theoretical consequences. Ditransitives have been repeatedly analyzed as analogous to possessive configurations and involving the presence of HAVE predicate (Harley 2002, Kayne 2010). While many ditransitive constructions contain an intended recipient, verbs of resemblance, comparison, connection, with an obligatory dative argument have always posed a problem to a unifying view of dative ‘havers’ (cf. McIntyre 2006). (7) a. Isa a comparé Paul *(à une limace) Fr. b. Vanom se=a-dara pavle *(Ch’ias) Georgian ‘Isa compared Paul to a slug’ ‘Vano compared Paul to a worm’ These examples beg for a more global treatment of ditransitive constructions, such as ours, according to which the theme undergoes a change by coming into contact with another entity interpreted as a point of the scale. If this is indeed so, the possessive flavor of ditransitives does not arise from the presence of a specific predicate but from specific syntactic configurations, namely those involving scale-related construals, where the point on the scale reached by the theme is expressed by a dative argument. An additional major consequence of our approach is that goals are not ranked higher than themes on the thematic hierarchy, as generally accepted, pace Baker (1996). This implies that universally, there is a single basic configuration to constructions containing ‘coredatives’ in which the theme is mapped higher than the goal, but languages vary according to the structural means at their disposal to re-organise this order. Lastly, our approach leads to consider dative case as a structural case dependent on the case-licensing of the theme rather than related to a particular semantic role (cf. Baker & Vinokurova 2010). 2

References. Baker, Marc. 1996. “On the Structural Position of Themes and Goals”. In J. Rooryck and L. Zaring (eds.) Phrase Structure and the Lexicon. Kluwer, Dordrecht, 7-34. Baker, Mark & Nadya Vinokurova. 2010. Two Modalities of Case Assignment: Case in Sakha. Natural Language and Linguistic Theory 28:593-642. Beavers, John. 2011. “An Aspectual Analysis of Ditransitive Verbs of Caused Possession in English”. Journal of Semantics, 28: 1-54. Boneh, Nora & Léa Nash. To appear “Core and non-core datives in French”. In Beatriz Fernández & Ricardo Etxepare (eds.) Variation in Datives. Oxford: Oxford University Press. Harely, Heidi. 2002. Possession and the double object construction. Yearbook of Linguistic Variation 2: 29–68. Malchukov, Andrej, Martin Haspelmath & Bernard Comrie. 2010. “Ditransitive construction: a typological overview”. In Andrej Malchukov, Martin Haspelmath & Bernard Comrie (eds.) Studies in Ditransitive Constructions: A Comparative Handbook. Berlin: De Gruyter Mouton, pp. 1-64. Kayne, Richard S. 1984. Connectedness and Binary Branching. Dordrecht: Foris. Larson, Richard, K. 1988. “On the double object constructions”. Linguistic Inquiry 19: 335-391. Marantz, Alec. 2011. “Restitutive reand the first phase syntax/semantics of the VP”, handout of the talk presented at Paris workshop "Structuring the Argument". McIntyre, Andrew 2006. “The interpretation of German datives and English have”. In D. Hole, A. Meinunger and W. Abraham (eds.) Datives and Other Cases. Amsterdam: Benjamins. 185-211. Rappaport-Hovav, M. & B. Levin. 2010. “Reflections on Manner/Result Complementarity”, in E. Doron, M. Rappaport Hovav, and I. Sichel, eds., Syntax, Lexical Semantics, and Event Structure, Oxford University Press, Oxford, UK, 21-38.

3

Dative case, prepositions, and argument structure in Spanish

Spanish dative constructions have been widely regarded as equivalent to English double object constructions (Demonte 1995; Bleam 1999; Cuervo 2003). C-command asymmetries are observed in an alternation-like pattern, with the dative clitic being responsible for it: (1) a.*Juan entregó una foto de sí misma a María. Juan handed a picture of herself to María b. Juan le entregó una foto de sí misma a María. Juan CL.DAT handed a picture of herself to María „Juan handed María a picture of herself. Cuervo (2003) assumes a Low Applicative structure à la Pylkkänen (2002) for dative clitic structures like (1b), including those in which the dative encodes a beneficiary (2a), a location (2b) or a possessor (2c) rather than a goal/recipient: (2) a. Juan le construyó un mausoleo al emperador. Juan CL.DAT built a mausoleum to.the emperor „Juan built the Emperor a mausoleum. b. Juan le colocó cortinas al salón. Juan CL.DAT placed curtains to.the living.room „Juan fitted the living room with curtains‟. c. Juan le rompió la bici a María. Juan CL.DAT broke the bike to Mary „Juan broke María‟s bike‟. Pujalte (2009) notes, however, that not all the sentences in (1) and (2) behave alike. They respond differently to a battery of tests, including nominalisation (3) and passivisation (4): (3) a. La entrega de la foto a María. the handing of the picture to María b.*La construcción del mausoleo al emperador. the construction of.the mausoleum to the emperor c.*La colocación de cortinas al salón. the placement of curtains to.the living.room d.*La rotura de la bici a María. the breaking of the bike to Mary (4) a. A María le fue entregada una foto. to María CL.DAT was handed a picture „Mary was handed a picture‟. b.*A María le fue rota la bici. to María CL.DAT was broken the bike „María‟s bike was broken‟. These results lead her to distinguish between what she calls “core datives” (i.e. those appearing with bona fide triadic verbs), which are both selected for by the main verb optionally featuring the clitic, and “non-core” datives, which are introduced by a Pylkkänen-style Appl head hosting the dative clitic and adding an argument to an otherwise dyadic verb. Building on Marantz‟s (1991) case hierarchy, Pujalte considers the dative case assigned by Appl to be a “dependent” case which needs to be licensed in a nominative assignment environment, preventing non-core datives from appearing in nominalisations and passive. Core datives are allowed because the dative case assigned by V is a “lexical” case. However, this proposal does not account for the c-command asymmetries displayed by “core” dative verbs: if entregar and entregarle project the same structure, the facts in (1) remain unexplained. In addition, the syntactic distinction between core and non-core datives is

not as clear-cut as Pujalte‟s analysis predicts, since beneficiary and some locative dative verbs do allow passivisation: (5) a. Al emperador le fue construido un mausoleo. to.the emperor CL.DAT was built a mausoleum „The Emperor was built a mausoleum‟. b. Al animal le fue colocado un collar satelital. to.the animal CL.DAT was placed a collar satellite „The animal was fitted with a satellite collar‟. From the semantic point of view, only goal, beneficiary, and locative imply a transfer of possession of any kind: handing a picture, building a mausoleum or installing and antivirus result in Mary having the picture, the Emperor having a mausoleum, and the animal having a satellite collar. Conversely, breaking a bike does not result in Mary (not) having a bike; it encodes a change of state undergone by the bike, its possession remaining unaffected. The interaction between the tests of nominalisation and passive and the semantic interpretation of the structure gives a tripartite distinction of Spanish datives: (6) Test entregar construir/colocar romper Nominalisation    Passive  /?  Change of possession    This distinction can be accounted for if a finer-grained argument structure is proposed: 1. Verbs of the entregar-type display an alternation akin to English dative alternation. The non-doubling variant, equivalent to the to-variant in English, encodes a locative resultative construction headed by the preposition a: this variant can be nominalised, as Spanish locative constructions generally can. The doubled construction, equivalent to English DO, contains a possessive resultative phrase (Harley 2002) which features argument inversion (accounting for the c-command facts in (1)); the clitic heads an aspectual projection Asp (cf. Bleam 1999) responsible for the affected interpretation and structural dative assignment, spelled out as a. 2. Construir/colocar-type verbs only combine with the possessive resultative, therefore lacking a prepositional variant with a. Being a structural case marker, a cannot appear in the nominal construction; however, actual prepositions such as para „for‟ and en „on‟ are allowed. 3. Romper-type verbs do not combine with the possessive resultative construction, hence the lack of transfer-of-possession interpretation. They select for a DP internal argument containing a possessor, which can undergo “possessor raising” (Landau 1999). This DP-internal possessor argument cannot be targeted by passive, as it is not an argument of the event. vP vP (7) a. vP b. c. Juan

Juan

Juan v

v

VP

Asp le

V PP entregar ACC una foto P a

María

v

AspP VP

HAVEP V entregó/construyó María/ DAT el emperador HAVE ACC

Asp P Asp le

VP

ACC

DP V romper DAT María D una foto un mausoleo la

References: Demonte,V. 1995. Datives in Spanish. Probus 7; Bleam, T. 1999. Spanish and the syntax of clitic doubling. PhD diss., U. Delaware.; Cuervo, C. 2003.Datives at Large. PhD Diss., MIT; Barss, A. & H. Lasnik. 1986. A note on anaphora and DO. LI 17; Pylkkänen, L. 2002. Introducing Arguments. PhD Diss., MIT. Pujalte, M. 2009. Condiciones sobre la introducción de argumentos. MA Diss, U. Comahue. Marantz, A. 1991. Case and Licencing. In Proceedings of ESCOL 9. Harley, H. 2002. Possession and the double object construction. Linguistic Variation Yearbook 2; Landau, I. 1999. Possessor raising and the structure of VP. Lingua 107.

bici

On  the  configurationality  of  argument  positions  in  Hungarian     Balázs  Surányi     Research  Institute  for  Lingusitics  HAS  /  Pázmány  University     In  the  (non)configurationality  debate  concerning  Hungarian  in  the  80s  and  90s,   Hungarian   was   shown   to   be   characterized   by   an   articulate   and   hierarchical   preverbal   domain,   with   distinct   A-­‐bar   positions   dedicated   to   discourse   functions   such   as   topic   and   focus.   What   this   debate   did   not   conclusively   settle,   however,   is   the   question   whether   or   not   the   structure   of   argument   positions   inside   the   predicate  phrase  is  also  configurational.       The   most   prevalent,   and   indeed   empirically   most   well-­‐argued   and   elaborated   analysis   that   has   emerged   is   that   of   É.   Kiss’   (1981,   1987a,   b;   1991;   1994a,   b;   2002;   2003),   according   to   which   the   answer   is   negative:   arguments   are   base-­‐ generated  in  the  verb  phrase  in  a  free  order  in  a  flat  structure.  In  É.  Kiss  (2008)   this   analysis   is   amended   by   the   proposal   that   nonconfigurationality   of   the   predicate   phrase,   instead   of  obtaining   at   the   level   of   base   structure,  arises  in  the   course  of  the  syntactic  derivation.     Drawing  on  a  careful  evaluation  of  data  involving  anaphor  binding,  Condition  C,   Weak   Crossover,   idioms,   CED   islands,   Q-­‐scope   and   others,   the   present   talk   offers   an   alternative   approach,   based   on   a   hierarchical   verb   phrase   (vacated   by   the   raised   verb)   and   a   local   scrambling   movement   that   operates   in   the   post-­‐verbal   domain  of  the  clause.  This  local  scrambling  is  of  the  Japanese−Korean  type  (e.g.,   Saito   2000),   with   no   discernible   discourse-­‐semantic   effects   in   terms   of   specificity.   It   is   argued   that   the   account   based   on   Japanese-­‐type   scrambling   movement   makes   available   a   better   descriptive   coverage   by   accurately   accounting  for  a  broad  range  of  structural  symmetries  and  asymmetries  holding   between  the  subject  and  the  object  argument.                                      

Verbal Prefixes in Russian: Conceptual structure versus syntax I address the problem of widespread polysemy of Russian verbal prefixes and argue that different uses of a single prefix share a core conceptual meaning, and this conceptual meaning combines with another meaning component which is a function of its syntactic position 1. Background. I adopt a neo-constructivist viewpoint where the meaning rests partly on what is brought from the lexicon and partly on the syntactic structure (cf. Borer (2005) and Ramchand (2008)). I also use Ramchand’s first phase syntax approach, where a verb may contain initiation, process and result projections, which host the corresponding thematic arguments. I also rely on the lexical/superlexical distinction between prefixes (Svenonius (2004), Babko-Malaya (1999)), where the lexical prefixes are located in the res (for result) part of the syntax, while the superlexical prefixes are above aspect. I propose a finegrained substructure inside each prefix class, where both res and proc (process) heads are a possible locus of prefix insertion (cf. intermediate prefixes do- and pere- by Tatevosov (2008)). 2. Argument structure. I start with the hypothesis that each prefix heads a small clause where it takes two arguments (optionally distinct from the verbal arguments), which may move up to the verbal thematic argument positions. The complex structure of a prefix (cf. split PPs in Svenonius (2003)) allows for the asymmetric relationship between the arguments, where the external argument corresponds to figure and the internal argument corresponds to ground, while either may surface as the verb’s direct object (2). Admitting both nominals and events as prefixal arguments allows us to preserve the uniformity of the prefix meaning. For example the prefix pro- with the basic meaning ‘through’ contributes the relational content in any of the following situations: the resultee moving through ground, the initiator moving through event, the event happening through the resultee, and so on. The inner argument structure of the verb determines the position where the prefix is inserted and the arguments that are assigned to it. 3. Time-space. The core meaning of the prefix is compatible with the scalar structures provided by either time or space: the same prefix refers to spatial domain with directional verbs and to the temporal domain with non-directional verbs. For example, the prefix pro- refers to the length of the path in (1a), while in (1b) the same prefix refers to duration of the activity. (1)

a.

b.

Pro-jti pjatj kilometrov. dir pro-walk five kilometers. ‘to walk five kilometers. Pro-xoditj pjatj chasov. pro-walknon-dir five hours ‘to walk for five hours.’

When added to a directional motion verb, prefixes behave as lexical, and the arguments correspond to resultee and path. The result of a directional motion verb is the displacement of a figure along the path specified. Thus, the prefix describes the change of location of the resultee in respect to ground.

When the same prefix is added to a non-directional motion verb, there is no result implied, and it behaves as a superlexical prefix, so the arguments are initiator and event. The superlexical prefix refers to the initiator ’s involvement in the event, i.e. the activity described by the main verb. The presence of the res vs. init projections and availability of the corresponding arguments determines where the prefix is inserted. 4. Figure-ground reversal. After closer examination of transitive verbs we may notice that sometimes the direct object of the verb looks more like figure (2a) and sometimes like ground (2b). (2)

a.

b.

pro-bitj dyrku v stene through-hit hole in wall to make a hole in a wall pro-bitj stenu. through-hit wall to breach a wall

In the first case, the hole is through the wall. In the second case hitting happens through the wall. In both cases the wall acts as the path traversed (Rheme), parallel to ‘five km’ in (1a). Like path, the wall is a proper object of the verb ‘hit’, while in (2a) the prefix introduces an unselected object (hole) (*bitj dyrku), which is created as the result. The matching of argument structure to conceptual content allows for a certain flexibility of interpretation. The verbs may contain or not contain the res projection depending on the argument structure. Some surface direct objects are resultees, others are undergoers, and some others turn out to be rhemes, and prefixes may scope over each depending on insertion site, deriving polysemy. 5. Conclusion Thus, the analysis allows us to preserve a single lexical entry for cognate prefixes. The meaning variation is determined by the syntactic position of prefixes, and their argument structure is predicted depending on the inner structure of the verb. While the lexical entry specifies the kind of transition (lexical distinctions, idiomatic usages), the lexical entry is unspecied for the domain and scope of the transition. This information is provided by the syntactic structure. References Babko-Malaya, Olga. 1999. Zero Morphology: a study of aspect, argument structure and case. New Brunswick, New Jersey: Rutgers dissertation. Borer, Hagit. 2005. In name only (structuring sense, vol. i). Oxford: Oxford University Press. Ramchand, Gillian. 2008. Verb meaning and the lexicon. Cambridge: Cambridge University Press. Available at http://ling.auf.net/. Svenonius, Peter. 2003. Limits on P: filling in holes vs. falling in holes. Nordlyd, Tromsø Working Papers on Language and Linguistics. Proceedings of the 19th Scandinavian Conference of Linguistics 31(2). 431–445. Svenonius, Peter. 2004. Slavic prefixes inside and outside VP. In Peter Svenonius (ed.), Nordlyd, tromsø working papers on language and linguistics 32.2: Special issue on Slavic prefixes, 205– 253. Tromsø: University of Tromsø. Available at www.ub.uit.no/munin/nordlyd/. Tatevosov, Sergei. 2008. Intermediate prexes in russian. In A. Antonenko C. Bethin & J. Baylin (eds.), Formal approaches to slavic linguistics. the stony brook meeting 2007, 423 442. New York: Ann Arbor, MI: University of Michigan Press.

Aspectual and quantificational properties of locative verbs Denominal locative verbs are of two kinds: locatum and location verbs, exemplified in ((1)a) and ((1)b) respectively: (1) a. Sue saddled the horse ( “Sue fit the horse with a saddle”) b. John shelved the books ( “John put the books on a shelf”) These verbs have been analyzed within syntactic theories of argument structure in relation with aspect and its connection with argument structure. Hale and Keyser (1993, 2002) endorse that aspect is orthogonal for argument structure, and propose that locative verbs involve two different structures, one of them containing a preposition that expresses a central coincidence relation, locatum verbs, and the other cointaining a preposition that expresses a terminal coincidence relation, a location verb, independently from their aspectual properties. Harley (2005) puts forth an analysis whereby the aspectual properties of both types of verbs derive from the (conceptual) (un)boundedness of the root they embed, following the hypothesis that aspect derives from nonconfigurational semantic properties. Finally, Mateu (2008) follows the idea that prepositions determine the telicity/atelicity of the predicate and proposes that telic locative verbs, either location or locatum, feature a preposition of terminal coincidence, and that atelic locative verbs, which are only locatum, feature a preposition of central coincidence. In this work we revamp the classic Halekeyserian analysis of locative verbs whereby location and locatum verbs involve two different structures, but we take into account the (a)telicity properties of these verbs. Our proposal departs from the observation that locatum and location verbs do have different aspectual properties (contra Mateu 2002, Harley 2005). The aspectual interpretation of locatum verbs depends on the mass/count interpretation of the embedded root: locatum verbs are atelic if the embedded root is construed as a mass ((2)a) and are telic if the root is construed as a count entity ((2)b). (2) a. L‟Elna ha en-vinagrat l‟amanida durant/#en cinc minuts the=Elna has in-vinegared the=salad for in five minutes „Elna has put vinegar on the salad for/in five minutes‟ b. L‟Elna ha en-sellat el cavall en/#durant cinc minuts the=Elna has in-saddled the horse in/ for five minuts „Elna has saddled the horse in/for five minutes five minutes‟ On the contrary, location verbs are telic irrespective of the embedded root ((3)a) and ((3)b). (3) a. En Jan ha em-botellat l‟aigua en/#durant cinc minuts The Jan has in-bottled the=water in/for five minutes „Jan has bottled the water in/for five minutes‟ b. En Jan ha en-vinagrat els cogombres en/#durant cinc minuts the Jan has in-vinegared the cucumbers in/ for five minutes „Jan has put the cucumbers into vinegar in/for five minutes‟ The proposal states that although all locative verbs contain a prepositional head in their inner structure, the nature of the preposition varies in location and locatum verbs. While location verbs are construed on a bounded preposition, locatum verbs contain an unbounded preposition. In our account the notion of terminal/central coicindence relation does not yield (a)telicity. Instead, (a)telicity is better understood as linked to the notion of (un)boundedness, orthogonal to the central/terminal distinction. The notion of (un)boundedness allows us to naturally derive the relation between (a)telicity and quantification, a relation which is clearly manifested in the behavior of locative verbs. We set off from Bosque & Masullo‟s (1999) observation that some verbs in Spanish may allow adverbial quantifiers to scope into the embedded nominal complement. The phenomenon is called i(nherent)-quantification, and is illustrated by examples such as those in (4).

(4) En Jan ha sagnat molt the Jan has bled a_lot „Jan has bled a lot‟ ( “Jan has produced a lot of blood”; cf. sang „blood‟) Crucially, B&M point out that locatum verbs involving a mass entity, like enmantecar „grease‟ (cf. manteca) are i-quantifiable ((5)a), while location verbs like enjaular aren‟t ((5) b ). (5) a. María enmantecó demasiado el molde. (Spanish) María greased too_much the mould „Marta greased the mould too much.‟ b. *Juan enjauló mucho el pájaro. Juan encaged a_lot the bird We only partly concur with their position. Thus, while we accept that locatum verbs may not accept i-quantification if they embed a root construed as a bounded entity (see (7)), we reject B&M‟s contention that the non-i-quantifiability of a verb like enjaular „cage‟ depends on the fact that it embeds a bounded root (cf. jaula „cage‟). Instead, we show that location verbs never allow i-quantification, irrespective of their embedding a bounded or an unbounded root (see (6)a and b). Thus, we claim that the bounded nature of the preposition in location verbs blocks iquantification, while the unbounded nature of the preposition in locatum verbs is transparent for i-quantification and allows a quantifier like massa „too much‟ to quantify the embedded root. (6) a. #En Jan ha embotellat massa l‟aigua „Jan has bottled the water too much‟ b. #En Jan ha envinagrat massa els cogombres (vinagre = „vinegar‟) „Jan has put the cucumbers too much into vinegar‟ (7) a. #L‟Elna ha ensellat massa el cavall „Elna has saddled the horse too much‟ b. L‟Elna ha envinagrat massa els cogombres „Elna has provided the cucumbers with too much vinegar‟ The pattern of i-quantification found in locative verbs has a striking parallel in the domain of quantification of PPs (Oltra-Massuet & Pérez Jiménez 2011). Thus, while Spanish preposition con „with‟ does not allow degree quantification irrespectively of the (non-)gradability of its nominal complement (see (8)), preposition sin does, depending on the (non-)gradability of its nominal complement (see (8)b). The data show that bounded prepositions like con make their complement opaque to quantification. (8) a. *Muy con luz/corbata vs. b. Muy sin luz/*corbata very with light/tie vs. very without light/*tie In conclusion, we observe that location verbs are always both telic and non-i-quantifiable and locatum verbs can be atelic and i-quantifiable, on the one hand, or telic and non-i-quantifiable, on the other, depending on the conceptual properties of the embedded root. The data therefore show that the boundedness of the preposition ensures telicity, and that an unbounded preposition does not contribute to the aspectual and i-quantification properties of the event. The conclusion has the welcome side effect of supporting the view that telicity and quantification constitute one and the same transcategorial notion (see Borer 2005, among others). Selected references: HALE, Kenneth & Samuel J. KEYSER. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge, Massachusetts: MIT Press. HARLEY, Heidi. 2005. How do verbs get their names? Denominal verbs, Manner incorporation and the ontology of verb roots in English. In Nomi Erteschik-Shir & Tova Rapoport (eds.), The Syntax of Aspect. Deriving Thematic and Aspectual Interpretation, 42-64. Oxford: Oxford University Press. MATEU, Jaume. 2002. Argument Structure. Relational Construal at the Syntax-Semantics Interface. Bellaterra: Universitat Autònoma de Barcelona dissertation.

The reflexive impersonal construction in Polish In the study of argument linking, two kinds of alternations or operations are commonly distinguished under one guise or another: morphosyntactic alternations and morphosemantic alternations (see e.g. Ackerman and Moore 2001, p. 5). A morphosyntactic alternation – the paragon example is the passive voice – does not affect meaning and so is considered to be a purely syntactic ‘relinking’ of arguments, whereas a morphosemantic alternation both affects meaning and involves a relinking of arguments. It is safe to say that morphosyntactic alternations are generally better understood than morphosemantic alternations. A possible example of a morphosemantic alternation is the reflexive impersonal construction in Polish (Kibort 2001, sect. 7.4, Ackerman and Moore 2001, sects. 6.2.2–6.2.3, Kibort 2008, sect. 5.2), which applies to both transitive and intransitive verbs: (1)

a.

b.

Budowało si˛e szkoł˛e. (Kibort 2008, (53)) built.3SG.NEUT REFL school.ACC ‘A/The school was built.’ Tutaj si˛e ta´nczyło. (Kibort 2001, (49)) here REFL danced.3SG.NEUT ‘There was dancing here.’

Observe that in (1) what would be the subject argument in the active voice is unexpressed, the verb shows singular neuter agreement, and the object argument (if there is one) has accusative case-marking (just as it would in the active voice). What would be the subject argument in the active voice may sometimes be realized as an indirect object argument with dative case-marking in the reflexive impersonal construction: (2)

a.

b.

Łatwo si˛e kapralowi biło Piotra. (Kibort 2001, (51)) easily REFL corporal.DAT beat.3SG.NEUT Peter.acc ‘The corporal found it easy to beat Peter.’ A mi si˛e wła´snie fajnie czytało! (internet) and I.DAT REFL in.fact well read.3SG.NEUT ‘And I in fact had a good read!’

Ackerman and Moore 2001, sects. 6.2.2–6.2.3 propose what may be the most explicit analysis of the reflexive impersonal construction, arguing that it is a morphosemantic alternation. (That it is a morphosemantic alternation is a view also taken by Kibort (2001, p. 181) but curiously not by Kibort (2008, p. 274), who treats it as a morphosyntactic alternation, diverging from her earlier view, though without saying why.) Ackerman and Moore’s idea is that this construction exhibits (what they call) attenuated agentivity, which basically means that what would be the subject argument in the active voice is less agent-like in the reflexive impersonal construction (whether it is left unexpressed or expressed as an indirect object with dative case-marking). They make use of Dowty’s (1991) proto-role properties for Proto-Agent and Proto-Patient, proposing (p. 152) essentially the following analysis for the alternation between czyta´c ‘read’ and czyta´c si˛e ‘read REFL’: (3)

a.

b.

czyta´c ‘read’: < x1 , x2 >, where x1 has the Proto-Agent property VOLITIONAL (and may have other ProtoAgent properties) and is linked to SUBJ, and x2 has various Proto-Patient properties and is linked to DIRECT OBJECT. czyta´c si˛e ‘read REFL’: < x1 , x2 >, where x1 lacks the Proto-Agent property VOLITIONAL (but may have other ProtoAgent properties) and has the Proto-Patient property CAUSALLY AFFECTED and is 1

linked to INDIRECT OBJECT, and x2 has various Proto-Patient properties and is linked to DIRECT OBJECT. It is clear that, in the ‘active frame’ in (3a), the argument x1 is volitional and not causally affected, whereas in the ‘reflexive impersonal frame’ in (3b), it is causally affected and lacks volition. The argument x2 remains constant in both frames. Ackerman and Moore offer the unacceptability of the following sentence as evidence that the argument x1 lacks volitionality: (4)

*Jankowi czytało si˛e t˛e ksia˛z˙ k˛e John.DAT read.3SG.NEUT REFL this.ACC book.ACC dobrowolnie/celowo/ch˛etnie. (Ackerman and Moore 2001, (16b), p. 152) voluntarily/on.purpose/willingly

If x1 (linked to Jankowi) in (4) were volitional, we would expect adverbs expressing intentionality to be compatible with the reflexive impersonal verb, but this is not the case, which suggests that x1 lacks intentionality. There are two difficulties with Ackerman and Moore’s account, a theoretical one and an empirical one. The theoretical difficulty is that on their analysis, the relation between the active frame and the reflexive impersonal frame is not semantically monotonic. This is because either the property VOLITIONAL is eliminated from the reflexive impersonal frame or the property CAUSALLY AFFECTED is eliminated from the active frame. The choice of a non-monotonic analysis should be a last resort, even in the case of a morphosemantic alternation. The empirical difficulty is that it is hard to see how the argument x1 should always lack volition. For example, x1 may be an argument of the reflexive impersonal verb chcie´c si˛e ‘want REFL’ and control the volitional subject argument of an infinitive: (5)

Chciało si˛e wyjecha´c. wanted.3SG.NEUT REFL leave.INF ‘There was eagerness to leave.’

(Kibort 2008, (61))

As for the unacceptability of (4), a plausible alternative hypothesis is that adverbs expressing intentionality require the argument that they modify to be linked to a nominative casemarked noun phrase. Observe that adverbs expressing the lack of intentionality are also not good in this construction, though they should be compatible with a lack of volition: (6)

*Jankowi czytało si˛e t˛e ksia˛z˙ k˛e niechcacy/mimo ˛ voli. John.DAT read.3SG.NEUT REFL this.ACC book.ACC unintentionally/involuntarily

An alternative account of the reflexive impersonal construction does not revoke volition from the argument x1 but rather adds the information that x1 is an Experiencer, thus x1 is both an Agent and an Experiencer. This combination of roles is awkward in thematic role approaches and even in a proto-role approach if Experiencer is defined as ‘sentience without volition or causation’ precisely because Agent entails volition (Dowty 1991, p. 577). Ackerman and Moore’s proposal that x1 is causally affected is more promising but also too general because x1 need not even be sentient in order to be causally affected. I suggest the introduction of a new property for Proto-Patient that aims to capture the notion of Experiencer needed here, namely, ‘undergoes emotional change of state’, which is more specific than Dowty’s ‘undergoes change of state’, which is intended for physical change of states. Moreover, the property ‘undergoes emotional change of state’ should be understood as (aspectually) an activity rather than an accomplishment. The kind of emotional change of state is determined by the kind of event described. For example, in (2a), it is the emotional experience of beating Peter, and in (2b), it is the emotional experience of reading whatever was contextually salient. 2

Characterizing reflexivization: Semantic and syntactic perspectives In addition to argument reflexives such as the reflexive anaphor of English, we encounter in many languages verbal reflexives (Faltz 1977), involving detransitivization of the verb, so that a single argument satisfies what would be two roles in the equivalent transitive predicate. We consider the proper characterization of the argument structure operation behind verbal reflexivization. It has been recognized since Kayne (1975) that the reflexive clitic se in French is not an argument of the verb, but is instead a marker of verbal detransitivization. Syntactically, this can be described as suppression of one argument. Whether this involves suppression of the internal or the external argument is a question that has been addressed at some length. Much of the literature, including Marantz (1984), Pesetsky (1995) and Sportiche (1998), conclude that Romance reflexives are unaccusative, but Reinhart and Siloni (2004, 2005) argue that in fact they are unergative. In the Theta System framework of Reinhart (2000, 2002), reflexivization is reduction (complete elimination) of the internal argument. From the semantic viewpoint, however, it is incontestable that the single argument of reflexives satisfies both thematic relationships of the underlying transitive predicate. For example, John shaves states that John is both the agent and patient of shaving. In other words, at a semantic level the second argument has not so much been removed as identified with the first, and their syntactic realization is the subject John. For Reinhart, thematic roles have semantic content, and the specification of theta roles plays a role in the projection of arguments. So if the syntactic effect of reflexivization is reduction, it necessarily has to extend to the semantics. But we have just explained that semantically there is no reduction, leading to the conclusion that reduction cannot be the right account of reflexivization.1 Accordingly, later versions of the Theta System (Reinhart and Siloni 2005) propose that reflexivization is not reduction but a new operation termed bundling, which combines two theta roles into a compound one that is projected as a single argument. While this neatly captures the semantic side of reflexivization, the very concept of theta role is intermediate between syntax and semantics. The syntactic projection of arguments is partially determined by the semantic (thematic) relationship they bear to the predicate, but it is a given that the semantic categories relevant to syntactic projection are but a subset of each verb’s argument semantics. Putting aside theory-internal considerations of Reinhart’s system, is there evidence that the particular semantics of reflexives are relevant to their syntax? While Reinhart considers theta roles and arity operations to have semantic content, other approaches might allow for a looser coupling of syntax and semantics and might see no need for two arity operations that cannot be syntactically distinguished.

1

Neither can saturation, responsible for passivization and expressed as existential closure, because it allows free reference for the suppressed argument, as with the suppressed subject of passives, and because of theory-internal generalizations about its distribution.

We can address this question by considering syntactic diagnostics for the kinds of arguments available with reflexives. The evidence for whether reflexives are akin to unergatives or unaccusatives - suggesting the presence of an agent or theme, respectively - is not entirely conclusive. Reinhant and Siloni (2004, 2005) present arguments for the suppression of the internal argument. E.g., en-cliticization, which is possible only out of the object position, is incompatible with reflexives: (1) *Il s'en est lavé beaucoup dans ces douches publiques, récemment. there SE of+themcl is washed many in these public showers recently ‘Many of them washed in these public showers recently’ This is a purely structural diagnostic, suggesting that reflexives are unergatives. But what do we know about the thematic properties of the subjects in the verbal reflexives under discussion. In terms of theta roles, the reflexivized verbs pass tests for the presence of an agent, such as allowing impersonal passives (if the language allows one) and allowing modification by adverbs like intentionally or carefully: (2) a. John threw the rock carefully. b. ?? John fell down carefully. [Non-agentive] c. Bill washed carefully. There are fewer tests for the presence of a theme or patient, but reflexives appear to pass these as well (cf. Levin & Rappaport Hovav 2005). The modifier “completely” appears to require a syntactically accessible theme (of an appropriate sort) (cf. 3a-c). It is compatible with suitable reflexives (cf. 3d). (3) a. b. c. d.

John sang the song completely [theme] The fruit was peeled completely [passive theme] *John sang/baked completely [unexpressed theme] John shaved/disrobed completely [reflexive]

Similarly, in English the modifier “painfully” cannot modify the cause or agent (cf. 4a,b), but must refer to an explicit (sentient) patient (cf. 4c): (but, so it seems, not consistently; something that awaits further explanation). Subjects of reflexive verbs appear to patientlike (cf. 4d): (4) a. b. c. d.

# Mary shouted painfully (cannot mean “with pain”) Mary pushed me painfully. Bill fell painfully [unaccusative] Bill shaved, painfully, with a dull razor. [reflexive]

We conclude that reflexive verbs retain both theta roles of the underlying transitive predicate, in the syntactic as well as the semantic sense - to the extent that these might be distinguished – supporting the bundling analysis. This could be important in understanding the mixed results of the diagnostics for unaccusativity, since these too rely to some extent on the theta roles present.

References Faltz, L.M. 1977. Reflexivization: A study in universal syntax. Doctoral Dissertation, University of California at Berkeley Kayne, R. 1975. French Syntax, Cambridge,Mass.s: MIT press. Levin, B & M. Rappaport Hovav 2005.. Argument Realization. Cambridge, Mass.: Cambridge University Press. Marantz, A. 1984. On The Nature of Grammatical Relations. Cambridge, Mass.: MIT Press. Pesetsky, D. 1995. Zero Syntax: Experiencers and Cascades. Cambridge, Mass.: MIT Press. Sportiche 1998. Pronominal Clitic Dependencies, in H. van Riemsdijk (ed.) Language Typology: Clitics in the European Languages, Mouton de Gruyter, Berlin.. Reinhart, T. & T. Siloni 2004. Against the unaccusative analysis of reflexives, in A. Alexiadou, E. Anagnostopoulou, and M. Everaert (eds), The Unaccusativity Puzzle: Explorations of the Syntax–Lexicon Interface. New York: Oxford University Press. Reinhart, T. & T. Siloni 2005. The lexicon–syntax parameter: reflexivization and other arity operations, Linguistic Inquiry 36.3, 389–436. Reinhart, T. 2000. The Theta system: syntactic realization of verbal concepts, UiL-OTS Working Papers, University of Utrecht. Reinhart, T. 2002. The Theta system: an overview, Theoretical Linguistics 28.3, 229–90.

Idiom passivization from a cognitive linguistic viewpoint Early idiom variation research focused on passivizability as one of the key properties of idiomaticity. Theories that try to account for the passive assume that the semantics of the idiom, more specifically the transitivity of the literal and figurative interpretation, as well as the absence or presence of coreference play a significant role (Newmeyer 1974, Akimoto 1983, Langlotz 2006). A representative sample of metaphorical and metonymic V NP idioms taken from idiom dictionaries are studied in a large corpus of American English and passive forms are retrieved to see how a cognitive linguistic approach can explain the attested forms or the non-occurrence of passive variants. The corpus used for this purpose is the COCA (Corpus of Contemporary American English), which consists of over 400 million words. The nominal components of the idiom are preceded by various types of determiners: definite, indefinite, possessive, zero. The investigation focuses on how the metaphorical and metonymic conceptualization associated with the idiom predicts the acceptability of the passive and whether that prediction is borne out. Following Langacker (1991), passivization will be viewed as the reversal of the trajector/landmark alignment, and in accordance with Langlotz (2006), idioms are viewed as complex constructions whose base-form images can be modified in discourse to fit the requirements of the specific contextual meaning. Nunberg et al. (1994) claim that passivizability depends on whether the idiom components have individual figurative senses or not (idiomatically combining expressions vs idiomatic phrases). Previous corpusbased research by the author suggests that on the one hand passivization is possible even if the idiom component is not meaningful at the figurative level (he knew when his leg was being pulled), and, on the other hand, the passive does not always occur with motivated decomposable idioms such as miss the boat or weather the storm. This seems to support Moon’s (1998) claim that strong fossilization is a significant factor in this area of the grammar. Yet, we aim to find motivation for the (non)occurrence of the passive. References Akimoto, Minoji. 1983. Idiomaticity. Tokyo: Shinozaki Shorin. Davies, Mark. (2008-) The Corpus of Contemporary American English: 425 million words, 1990-present. Available online at http://corpus.byu.edu/coca/. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar II. (Descriptive Applications.) Stanford: Stanford University Press. Langlotz, Andreas. 2006. Idiomatic Creativity: A Cognitive-Linguistic Model of IdiomRepresentation and Idiom-Variation in English. Amsterdam and Philadelphia: John Benjamins Publishing Company. Moon, Rosamund. 1998. Fixed Expressions and Idioms in English. Oxford: Clarendon Press. Newmeyer, Frederick J. 1974. The regularity of idiom behavior. Lingua 34/4: 327–342. Nunberg, Geoffrey, Ivan A. Sag and Thomas Wasow. 1994. Idioms. Language 70/3: 491–538.

The Passive of the Gesture Expression Construction: Event Structure and Discourse Function 1. GOALS The main goal of this talk is to demonstrate a complex interaction between event structure and a discourse function by examining the condition of the passive of the so-called Gesture Expression Construction (hereafter, GEC), which is exemplified as follows: (1) a. Pauline smiled her thanks. (Levin and Rapoport 1988) b. The doctor nodded agreement. (BNC) As shown in (1), the GEC takes an unergative verb as its main predicate and an NP denoting an expression as its complement. The complement NP normally does not permit passivization, as observed in the literature (e.g., *Her thanks were smiled by Rilla. (Massam 1990:180); see also Levin 1993 and Huddleston and Pullum 2002). However, there are conflicting data attested in corpora: (2) a. …the native would touch one article after another till consent was nodded from the ship… b. … as the time neared when the last line is cast off, the goodbyes are waved, the screw makes the water boil under the stern, and the passage to Alaska is under way. In this talk, I will explain the contradictory behavior of the passive of GECs, by appealing to the interaction between the relevant event structure and the effect of a certain discourse function. 2. CLAIM My claim is two-fold: (i) the passive of GECs is governed by the bound pronoun within the object NP, which is semantically motivated by the inalienable possession relation between the subject and object NPs; (ii) GECs undergo passivization if they serve a presentational function in discourse. These claims explain the unpassivizability and passivizability of the construction: I will provide a syntacto-semantic analysis of the GEC in (i) and then propose a semantic and discourse-functional analysis in (ii). The discussion will proceed in this order. The consequence implies that discourse-functional factors influence the mapping between syntax and event structure. 3. UNPASSIVIZABILITY & BOUND PRONOUN Massam (1990) claims that the object NP of the GEC does not undergo passivization because it contains a bound pronoun. The presence of a bound pronoun within the object NP is supported by the following examples ((b) from Ross 1970): (3) a. Mary smiled {her / * his} thanks. b. * Tom frowned {Ann’s / my} displeasure. Under this analysis, the unpassivizability of GECs naturally follows from the general constraint on the passive subject containing a bound pronoun. As is widely known, the possessive pronoun in the passive subject cannot be bound by the antecedent NP in the by-phrase: (4) a. * Hisi mother is loved by Johni. b. * Hisi role was played by Johni. (Zubizarreta 1985) Then, why is the bound pronoun present in the object NP? Massam does not elaborate on this point. My answer is grounded on the semantics of GECs: the subject and object NPs establish an inalienable possession relation. For example, if Mary smiled thanks, the expression of gratitude cannot be attributed to any other participants but the agent, Mary. In this sense, the agent NP denoted by the subject inalienably possesses the expression described by the object (see Kimball 1973). In fact, the object NP of GECs displays the definiteness effect, just like other inalienably possessed NP: (5) Pauline smiled (*the) thanks. cf. John has {a / *the} sister. Thus, the inalienable possession relation in GECs licenses the presence of a bound pronoun, thereby making the formation of passive impossible. 4. AGENTIVITY & PRESENTATIONAL FUNCTION If the present analysis is correct, the acceptability of the passives in (2) will imply that the bound pronoun is no longer present in the construction. In fact, the subject NP can be marked with a definite article, as in (2b), which in contrast with (5).

The reason for the absence of a bound pronoun, which I propose in this talk, is that, in the passive of GECs, the agent participant is suppressed from the event structure. This proposal is supported by the constraints on the occurrence of the agentive by-phrase in (5a) and agentive adverbs in (5b): (5) a.?? Final goodbyes are waved by the people. b.?? Thanks are nodded {carefully / deliberately}. Although the judgments of (5) vary from speaker to speaker, most of my informants find them degraded, compared with those without agentive elements. The absence of the agent from the event structure means that no inalienable possession is involved in the construction: the expression denoted by the passive subject in (2) is not understood as an entity attributed to a specific participant. Thus, the absence of an agent is responsible for the lack of the bound pronoun and the passivizability of GECs. The agent suppression from the event structure in (2) can be explained by their discourse function: the presentational function. The attested data of the passive GEC are found in the discourse type describing events that continuously take place in a given scene. For instance, example (2b) reports what events happen when ships leave for Alaska and the passivized GEC depicts one of the events taking place in the scene. To the extent that the relevant passive plays such a role in describing a scene, we can say that the passive of GECs serves a presentational function (cf. Lambrecht 1994:144). The presentational function is often associated with non-agentivity. The Locative Inversion Construction (LIC), which is used as a presentational sentence, also suppresses the agent from the event: (6) a.?? Among the guests of honor was seated my mother by my friend Rose. (Bresnan 1994) b.* Out of the room walked a man with long hair deliberately. (Nishihara 1999) As noted by Bresnan, in the LIC, even if the predicate does not lexically describe an event of existence or appearance of an entity, the construction overlays the event on the basic event structure of the predicate; consequently, in (6a), the existence and appearance of the theme is syntactically expressed and the agent is excluded out of the overlaid loc-theme event. As the non-agentivity seems to be pervasive across presentational constructions (cf. presentational there-construction), it can be said that the non-agentive property of (2) is essentially connected to the presentational function in the discourse. Thus, it is the presentational function that brings the non-agentivity to the passive of GECs, which is responsible for the absence of the bound pronoun and licenses the passive form. Interestingly enough, the passivization licensed by a presentational function can be observed with other constructions that otherwise would not be passivizable: Body-Part Gesture Expressions (e.g. John craned his neck) and some type of Cognate Object Constructions (e.g. Mary screamed a terrifying scream.): (7) a. … necks were craned in an effort to get a glimpse of the pair. b. Pictures were taken, laughs were laughed, food was eaten. (Kuno & Takami 2004: 128) (8) a. ?? Screams were screamed by the crowd. / * Screams were screamed deliberately. b.?* Necks were craned by John and Bill. / ?? Necks were craned {carefully / deliberately}. 5. CONCLUSION AND IMPLICATION The conclusion of this talk is that the passive form of the GEC is licensed by the interaction between the event structure and the presentational function. This implies that (i) the consideration of the event structure and its mapping to syntax must incorporate the discourse-functional effect on the former and (ii) the notion of agentivity is determined under syntactic, semantic, and discourse-pragmatic considerations, supporting the view of Van Valin and Wilkins (1996). References: Bresnan (1994) “Locative Inversion and the Architecture of Universal Grammar,” Lg 70-1, 72-131. / Kuno & Takami (2004) Functional Constraints in Grammar, John Benjamins. / Massam (1990) “Cognate Objects as Thematic Objects,” Canadian Journal of Linguistics 35: 161-190.

Ways of licensing external possessors Katalin É. Kiss Institute for Linguistics of the Hungarian Academy and Pázmány P. Catholic University Since Szabolcsi’s (1983) seminal paper, Hungarian external possessors are assumed to be generated and theta-marked in the maximal projection of the possessor, and to be externalized by A-bar movement. This talk will argue that, in fact, they do not form a homogeneous class. Some seemingly external possessors are experiencer or locative binders of an internal pronominal possessor, cf. (1) Jánosnaki fáj John-DAT hurts

a the

proi feje /Jánosnaki pro head-POSS /John-DAT

fáj a hurts the

proi/j proi/j

kudarca. failure

(2) Jánosnaki megvan az proi állása /Jánosnaki megvan a proi,j könyve. John-DAT is the pro job-POSS /John-DAT is the pro book-POSS John has got his job./John has got his book.’ Possessors generated internal to the projection of the possessor can be externalized via movement or reanalysis. In the former case, the remnant noun phrase is a DP; in the latter case, it is a non-specific indefinite NP. For the theme arguments of indefiniteness effect verbs analyzed by Szabolcsi (1983), only the latter option is available. A remnant NP possessum is sometimes semantically incorporated into the verb (e.g., vége van ’end-POSS is [has got an end]’. The different agreement properties of external possessors observed by den Dikken (1999) will be shown to correlate with whether they are generated outside or inside the projection of the possessor, and whether are externalized by movement or reanalysis. The possibility of possessor movement has been related to the argumenthood of the possessum, or the affectedness of the possessor. This talk will argue that the primary criterion is whether or not a pedication relation can be constituted between the possessor and the rest of the sentence. Cf. (3)a. *A bicskánaki megszereltem a zárat [ti a hegyével]. the pocket-knife-DAT fixed-I the lock-ACC the point-POSS-with ’The pocket knife, I fixed the lock with the point of.’ b. A bicskánaki akár zárat is lehet szerelni [ti a hegyével]. the pocket-knife-DAT even lock-ACC possible fix-INF the point-POSS-with ’The pocket knife, one can even fix a lock with the point of.’ c. Csak a SVÁJCI BICSKÁNAKi lehet [ti a hegyével] zárat szerelni. only the Swiss pocket-knife-DAT possible the point-POSS-with lock-ACC fix-INF ’It is only the Swiss pocket knife that one can fix a lock with the point of.’ References Dikken, M. den 1999. “On the structural representation of possession and agreement. The case of (anti-)agreement in Hungarian possessed nominal phrases” . In Crossing Boundaries: Theoretical Advances in Central and Eastern European Languages, I. Kenesei (ed.), 137178. Amsterdam: John Benjamins. Szabolcsi, A. 1983. “ The possessor that ran away from home” . The Linguistic Review 3: 89–102.

Nonaffected Incremental Themes: the Case of Verbs of Creation Affectedness is widely believed to be relevant to many phenomena concerning argument realization, including passive, unaccusativity, linking to objecthood, dative alternation (Beavers 2006, 2010, 2011a,b). A test that is commonly used to diagnose affectedness of an argument is the structure What happened to x is... (Cruse 1973: 13). In this paper, we will investigate some properties of this test, checking whether it actually diagnoses affectedness. Beavers (2010, 2011b) offers a very good description of what is involved in this test. According to him, affected arguments that can be inserted in this structure are potential affected arguments, that is, those that are subject to possible change; in his words, ''the verb lexically specifies that there are specific possible outcomes'' (Beavers 2010: 16). Beavers formulates this idea as follows: (1)

x has potential for change iff ɸ → ◊∃g[result′(x, g, s, e)], where x = a theme, s = a scale, g = a non-quanticized change, e = an event

(2)

What happened to x is ɸ is felicitous iff ɸ is felicitous and entails x has potential for change.

We discuss what Beavers thinks is a counter-argument to the validity of the test: the fact that verbs of creation, as in (3), despite satisfying (1), do not entail that their (incremental) themes exist. This causes the relevant sentences to fail, as shown in (4). (3)

Rebecca built/made a Victorian style house. (Piñon, to appear)

(4)

#What happened to a Victorian style house is Rebecca build/make it.

Given examples like (4), Beavers (2008, 2011b) concludes that the test is not wholly reliable. ''Effected (created) objects'' such as those in (3) produce odd results under the test because the construction What happened to x is... seems to presuppose the prior existence of the theme. Indeed, as Beavers shows, if there is a context in which such a previous existence requirement is satisfied, the resulting sentences do become acceptable; as illustrated in (5): (5)

a. b.

What happened to the model airplane is John built it. What happened to the shed is John rebuilt/refashioned/reconstructed it. (Beavers 2011b: 6)

Nevertheless, Beavers' (2008, 2011b) solution for the problem is not a real solution; rather, it displaces the discussion to the issue of whether the existence of the (incremental) theme is presupposed or not. We propose a different solution: affectedness with verbs of creation is not a lexical entailment on their incremental themes. If this is true, the case of verbs of creation does not represent a failure of the test; rather, the test shows that these verbs, despite having an incremental theme as an argument, do not entail affectedness for this argument. Beavers' observation about the presupposition of existence in Cruse's test is correct; but his conclusion that the test is not wholly reliable is not, since these verbs do not have an affected argument, according to the characterization we propose here: if a thing does not exist, it cannot be affected. Consider, for example, the sentences in (5b): the prefix re- actually changes the meaning of verbs like build, fashion and construct, in such a way as to make the existence of the shed presupposed. That is, when one rebuilds, reconstructs, etc. something, this thing must exist previously.

Theoretically, our aim in this paper is propose a characterization for “affectedness” that is able to capture this idea. In our terms, Cruse's test shows that, for a thing to be perceived as affected in an event e, this thing must exist in a time t previous to the occurrence of e. More formally, we propose the following definition of affectedness (which is based on Parsons 1990, von Fintel & Gillies 2005, von Fintel 2006, and Beavers 2010, 2011b). (6) [[affect y]] c, i = I iff °y & °w' ϵ f: [[°y]] c,