PP licensing in nominalizations - CiteSeerX

NELS 38, University of Ottawa

October 26-28, 2007

PP licensing in nominalizations Artemis Alexiadou, Elena Anagnostopoulou & Florian Schäfer (University of Stuttgart, University of Crete & University of Stuttgart) [email protected], [email protected], [email protected]

1.

Introduction

In this paper we investigate the distribution of PPs introducing external arguments in nominalizations in comparison to the behavior of PPs in the verbal domain. We show that the behavior of external arguments in nominalizations does not mimic the corresponding behavior of the verbal domain. We propose that PPs can be introduced in two different ways in the nominal domain: 1)

Presence of functional structure.

2)

In the absence of functional structure, PPs can be licensed via an interplay of the encyclopedic meaning of the root involved and the properties of the preposition itself.

1.1

The licensing of PPs in the verbal domain

In languages like English (and German) virtually any transitive verb can form a passive, where the by-phrase re-introduces absorbed external arguments instantiating all different thematic roles (Jaeggli 1986, Baker, Johnson & Roberts 1989, Fox & Grodzinsky 1998, Collins 2005 among others). (1) a. John destroyed the book b. The book was destroyed by John (2) a. The storm destroyed the window b. The window was destroyed by the storm (3) a. All the students fear that professor b. That professor is feared by all students (4) a. John received the package b. The package was received by John

(Agent) (Causer) (Experiencer) (Recipient)

The Voice Hypothesis (Kratzer 1994): Voice is responsible for the introduction of external arguments. The same head introduces a DP in the active and licenses a PP in the passive. Observation: Causer PPs are special in that they can appear in the absence of Voice. This is the case with anticausatives across languages (Alexiadou, Anagnostopoulou & Schäfer (AAS) 2006): (5)

The window broke from the storm / *from John

(Causer / *Agent)

•

AAS take this as evidence that (anti-)causative verbs are decomposed as follows: 1

(6) a. Active/Passive:

b. Anticausative:

[ Voice [ vCAUS [ Root + Theme ]]]

[ vCAUS [ Root + Theme ]]

•

Agentivity and causation are syntactically represented in terms of distinct functional heads. CAUS is taken to introduce a causal relation between a causing event (the implicit argument of CAUS) and the resultant state denoted by the Root + Theme complex. Voice introduces the external argument.

•

Causatives and Anticausatives have a CAUS head but differ in that only the former have Voice. The causative alternation is a Voice-alternation.

•

Licensing of PPs:

▪ vCaus licenses Causer PPs in anticausatives ▪ Voice licenses Agent PPs and Causer PPs in passives

1.2

VP within nominalizations

Derived nominals include VPs (of variable size): Current syntactic theories of nominalization assume that the structure of derived nominals contains a nominal head, n, that takes a VP (of variable size) as its complement (Alexiadou 2001, Borer 1993, 2003, Fu, Roeper & Borer 2001, Borsley & Kornfilt 2000, van Hout & Roeper 1998 to mention a few). (7)

[n [ (Voice) [ vP ... ]]]

Nominalization = Passivization: Several authors argued that nominalization is akin to passivization in that it absorbs the external argument (Grimshaw 1990). If this is the case, then we would expect nominalizations to behave like verbal passives (Alexiadou 2001, Borer 1993, 2003, Sichel 2007). •

This predicts that PPs realising external arguments in nominalizations strictly mimic the behavior of the verbal domain.

Question: To what extent are these expectations borne out? Observation: The behavior of external arguments in nominalizations does not mimic the behavior of the verbal domain. The differences can be explained in terms of the view that “passive meaning” is not passive in the sense of a process suppressing the external argument, but rather as a by-product of the fact that certain portions of structure are missing in a decomposition approach towards argument alternations (Kratzer 1994, Marantz 1997 and many others). The specific implementation we propose characterizes the properties of nominalizations in terms of the decomposition argued for in AAS (2006) (cf. (6) above).

1

Note that CAUS could also simply be seen as an eventive v of the type proposed in Marantz (2005). In this case the causative semantics would not be directly encoded on any verbal head but would result from the combination of an activity v and its stative complement (see Ramchand 2006 for related ideas).

2

2.

Mismatches between the distribution of PPs in passives and corresponding nominalizations

•

A number of observations from different languages cast doubts on the view that nominalization is akin to passivization.

2.1

English

A.

Nominalizations of alternating verbs seem to behave like their verbal counterparts, with one major difference (ambiguity of the intransitive variant).

•

In the absence of a by-phrase, the nominal in (8a-b) is ambiguous between two interpretations: a passive and a spontaneous, anticausative, reading (see Harley & Noyer 2000, Sichel 2007). In the verbal domain the two interpretations are morphologically encoded (i.e. the passive is clearly distinct from the anticausative structure).

(8) a. the accumulation of wealth (by John) b. the accumulation of dust c. the destruction of the vase (by John/by the wind) •

External arguments can also surface in the pre-nominal genitive position (as in 9a-b). 2

(9) a. John's accumulation of wealth b. John's destruction of the vase B.

The PPs in the nominalization cannot realise all thematic roles, unlike the verbal passive (Hornstein 1977, Rappaport 1983, Jaeggli 1986, Grimshaw 1990, Fox & Grodzinsky 1998 and others):

(10) a. the imprisonment of refugees by the government b. the destruction of the city by lightning (11) a. The fear of Harry (*by John) b. The respect for Mary (*by John) (12) The receipt of the package (*by John)

(agent) (causer) (*experiencer) (*recipient)

• The reasoning of section 1.1 suggests that in (10a-b) VoiceP is present, as the Agent/Causer PPs are licit. • On the other hand, (11)-(12) suggest that Voice is absent as the recipient/experiencer PP is out. • NB: the nominalization in (12) is eventive and is interpreted as passive in the absence of a by phrase. (13) The receipt of the package takes place at the agreed date of departure and is carried out by your local transport officer together with radiation protection 2

Here we do not go into the debate concerning the availability of transitive nominalizations from alternating verbs (see e.g. Pesetsky 1995, Marantz 1997, Harley & Noyer 2000, Borer 2003, Sichel 2007 and others).

3

2.2

German

A. In German we find a different type of mismatch between verbal passives and nominalized infinitives. B. Further, German shows a mismatch between two types of nominalizations, nominalized infinitives and ung-nominals. •

These mismatches become obvious if we look at nominalizations of verbs undergoing the causative alternation, more specifically nominalizations of verbs forming anticausatives marked with the reflexive pronoun ‘sich’.

•

The verbal behaviour is as in English: ƒ the active licenses Agent and Causer DPs (14a) ƒ the passive licenses Agent and Causer PPs (14b) ƒ the anticausative licenses only Causer PPs (14c)

(14) a. Hans / der Wind öffnete die Tür Hans / the air opened the door b. Die Tür wurde von Peter / durch einen Windstoss geöffnet The door was by Peter / through a gust-of-wind opened c. Die Tür öffnete sich (durch einen Windstoß) The door opened REFL through a blast-of-wind •

In German verbal passives, agents are introduced by preposition 'von' and causers by preposition 'durch'. In the nominalization, only 'durch' is licit, as 'von' shifts to the function of a genitive/possessor marker.

•

ung-nominals of alternating verbs never combine with the reflexive pronoun and they behave thematically similar to English -ation nominals; they license both Agent and Causer PPs:

(15) a. die the b. die the •

Öffnung open-ung Öffnung open-ung

der the-gen der the-gen

Tür door Tür door

durch through durch through

Peter Peter den Wind the wind

(agent) (causer)

To the extent that psych verbs undergo -ung nominalization in German, the PP can introduce experiencers, unlike English:

(16) die Bewunderung des Künstlers durch das Publikum the admiration the artist-gen through the public •

(experiencer)

Similarly, a recipient PP is licit, unlike English: 3

(17) der Empfang des Pakets durch Hans the receipt the package-gen through Hans 3

(recipient)

-ung nominalizations in German are not possible with all verb classes, see Rossdeutscher (2007).

4

Æ

ung-nominals (and zero derived nominals) seem to behave as verbal passives; they show full productivity.

•

Nominalized infinitives of alternating verbs show an interesting deviation. They have a passive version and an anticausative version, the latter morphologically marked by the reflexive pronoun ‘sich’. The nominalized passive differs from the verbal passive in only licensing Agent PPs. ƒ The passive pattern: No reflexive morphology present; the internal argument has genitive and an agent but no causer can be introduced via a durch PP.

(18) a. das Öffnen der Türen durch Peter the open-en the-gen doors through Peter b. *das Öffnen der Türen durch den Wind the open-en the-gen doors through the wind

(agent) (*causer)

ƒ The anticausative pattern: Reflexive morphology present; the internal argument has genitive and a causer but no agent can be introduced via a durch PP. (19) a. ?*das the b. (?)das the

sich REFL sich REFL

Öffnen open-en Öffnen open-en

der the-gen der the-gen

Türen doors Türen doors

durch through durch through

Peter Peter den Wind the wind

(*agent) (causer)

•

The sich pattern (anticausative) behaves as in the verbal domain.

•

The non-sich pattern (passive) differs from the verbal passive in that it excludes causerPPs and only licenses agent PPs.

•

This is not unique to nominalization; this pattern is found in the verbal domain in languages such as Greek (and Hebrew, Icelandic).

2.3 Greek 2.3.1 Verbal passives in Greek (and Hebrew) •

De-adjectival verbs illustrate the Greek passive: With this class, passives bear nonactive morphology, while the anticausative bears active morphology. Causer PPs are only licit with the anticausative, Agent PPs are only licit with the passive. (Hebrew is similar, cf. Doron 2003, Alexiadou & Doron 2007).

(20) a. To The b. to the

pukamiso shirt pukamiso shirt

stegnose me dried-Act with stegnothike apo dried-Nact by

ton the to the

aera wind Jani John

/ / / /

apo mono tu / *apo to Jani by itself / by the John *me ton aera with the wind

5

2.3.2 Nominalizations in Greek •

Nominalizations in Greek present further mismatches between active and passive.

•

In Greek, the same preposition that introduces agents and causers in the verbal domain, apo 'from', is also used in the nominal domain.

A.

First, we find passive nominals of verbs that lack a passive: ƒ The example in (21) only has an anticausative but not a passive interpretation, as the licensing of PPs shows:

(21) To dasos kaike apo ti zesti / *apo to Jani the forest burnt-Nact by the heat / *by the John

(anticausative/*passive)

ƒ The corresponding nominalization, surprisingly allows both Agent and Causer PPs. (22) to kapsimo tu dasus apo ton Jani / apo ti zesti the burning the forest-gen by the John / from the heat ƒ A similar phenomenon is illustrated with nominalizations of de-adjectival verbs. These also can license both Agent and Causer PPs (23). The licensing of the Causer PP is not surprising; the licensing of the Agent PP, however is not expected. Recall that the corresponding verbs license an Agent PP but no Causer in the passive (24b). (23) to anigma tis portas apo ton aera / apo to Jani the opening the door by the wind / by the John (24) a. I the b. I the B.

porta door porta door

anikse apo ton aera / opened-Act by the wind / anihtike *apo ton aera opened-Nact by the wind

*apo to Jani by the John / apo to Jani / by the John

(causer / agent) (causer / *agent) (*causer / agent)

Second, nominalizations with experiencer/recipient PPs are OK, unlike their English counterparts. 4

(25) i paralavi tu paketu apo to Jani the receipt the package-gen by the John (26) i latria tu Theu apo tus anthropus the admiration the God-gen from the people

(recipient) (experiencer)

4

Markantonatou (1992) notes that in nominalizations the experiencer surfaces with genitive. (i) a. o fovos tu Jani ja tis kategides the fear the John-gen for the storms b. i agnia tu Jani ja tus kanones tu pehnidiu the ignorance the John-gen for the rules of-the game c. i agapi tu Jani ja ti Maria the love the John-gen for the Mary d. i stenahoria tis Marias ja ta nea su the worry the Mary-gen for your news Markantonatou claims that these nominalizations are stative and explains the impossibility of a by- phrase with such nominals by observing that that no such phrase is licit in stative passives.

6

2.4

Conclusion

• The distribution of PPs in the nominal domain shows a number of mismatches compared to that in the verbal domain. • Verbal domain: 1. Anticausatives: license causer PPs across languages. 2. Verbal passive licenses Agents across languages: In some languages the passive does not license Causer PPs (Greek, Hebrew, …). 3. Recipient/Experiencer/other external arguments are licensed as PPs in the passive across languages. • Nominal domain: 1. English nominalizations allow only Agent and Causer PPs. In contrast to the English passive, other theta roles are out. 2. German nominalized infinitives behave like Greek/Hebrew verbal passives in that they do not license Causer PPs. 3. German ung-nominalizations show full productivity in that they license all kind of theta roles in a PP like English and German verbal passives. 4. Greek nominalizations seem to be less restrictive than the verbal passive in that they do license Causer PPs. Plus they license experiencers and recipients, unlike English. 3.

Towards an explanation

3.1

English nominalization

A.

Note that in English ALL types of external theta roles can appear in the prenominal possessor position, no matter whether they can appear in the by-phrase or not.

(27) a. the destruction of the coral reef system (by the tsunami) b. The tsunami's destruction of the coral reef system

(causer) (causer)

(28) a. the destruction of the vase (by John) b. John's destruction of the vase

(agent) (agent)

(29) a. the accumulation of wealth (by John) b. John's accumulation of wealth

(agent) (agent)

(30) a. The fear of Harry (*by John) b. John's fear of Harry

(*experiencer) (experiencer)

(31) a. The respect for Mary (*by John) b. John's respect for Mary

(*experiencer) (experiencer)

(32) a. *The receipt of the package by John b. John's receipt of the package

(*recipient) (recipient)

7

1.

Voice hypothesis: Kratzer (1994), Marantz (1997) and subsequent work: 'external' arguments are never assigned by the lexical entry, but by Voice. -(at)ion nominals lack Voice and therefore they never have an external argument. 5

•

When the nominalization structure includes Voice, as in the verbal gerund (John's destroying the city), the external argument is realized obligatorily.

•

When the nominalization lacks Voice, then the external argument can be realized as a possessor.

The specific interpretation the processor can get is rather free and it is mainly dependent on the concept expressed by the possessee (Grimshaw 1990, Williams 1985). ƒ In the case of object nouns that lack argument structure, the possessor can be interpreted as owner or author. (33)

John's book a. a book that John wrote b. a book that John lent to me ƒ In the case of destruction, the possessor can be interpreted as an agent/causer, based on our encyclopedic knowledge about destroy. ƒ In the case or fear, the possessor can be interpreted as an experiencer, based on our encyclopedic knowledge about fear. ƒ In the case of receipt, …

2. If -ation nominals lack Voice, by-phrases with such nominals cannot be licensed by Voice. They must be licensed by encyclopedic knowledge about the nominal, similar to what happens in the case of the prenominal possessor. If so, why is their interpretation restricted, and not similar to the behavior of prenominal possessors? •

by-phrases in the nominal domain: (Fox & Grodzinsky 1998):

(34) a. a book by John b. CK1 by Calvin Klein The nominal by-phrase is thematically restricted, it can only be interpreted as an Agent or Creator. It does so, in the absence of passive morphology/argument structure. •

English passives: external arguments are licensed by Voice.

•

English nominalizations: Voice is absent. In the absence of Voice, the by-phrase assigns the thematic role of Actor (Fox & Grodzinsky 1998), if this is compatible with the root semantics.

•

This predicts a) that English nominalizations should be, like adjectival passives, compatible with reflexive action (Rossdeutscher 2007, cf. Kratzer 1994)

5

It follows from this that they cannot assign accusative Case to their internal argument, as opposed to the gerund, which contains Voice, has an external argument and can assign accusative.

8

(35) the enrollment students •

of

the

(agent

theme) 6

=

/

≠

and b) that other prepositions not similarly restricted could in principle introduce NonActor external arguments. This prediction is confirmed. The preposition ‘of’ is typically used to introduce internal arguments. But in (36)-(37) ‘of’ can introduce experiencers.

(36) the fear of John

(experiencer)

(37) the love of Mary towards her son

(experiencer)

3.2

Greek and Hebrew verbal passives

Different features of Voice are involved in the formation of passives across languages. We concentrate here on the difference [± agentive] (Agent vs. Causer): Structure-feature combinations of Voice

1 2 3 4

agentivity syntactic realisation specifier + specifier implicit + implicit 7 -

For some reason, 4 is the combination that Greek and Hebrew lack, the result being that passive in these languages is strictly agentive. 3.3

Infinitival -en nominalizations in German

Infinitival -en nominalizations in German behave like the Greek and Hebrew verbal passive. While the anticausative version licenses only Causer PPs, as in the verbal domain, the passive version licenses only Agent PPs, unlike the verbal domain. •

This suggests that combination 4 is available in the verbal domain, but not in the nominal domain.

•

Recall that there is a second type of nominalizations, -ung nominalizations, which never combine with a reflexive but which allow both, Agent and Causer PPs.

(38) a. die the b. die the

Öffnung open-ung Öffnung open-ung

der the-gen der the-gen

Tür door Tür door

durch through durch through

Peter Peter den Wind the wind

(agent) (causer)

Question: Does this suggest that we should assume two types of nominal passive Voice?

6 7

•

No. -ung nominalizations are much like -ation nominals in English. They lack Voice.

•

Infinitival -en nominals like verbal passives are incompatible with reflexive action; -ung nominals, like adjectival passives are compatible with reflexive action (Rossdeutscher 2007, cf. Kratzer 1994):

p.c. Thomas McFadden. We use the term implicit argument here without making any claims concerning its representation in the syntax.

9

(39) a. Das the b. Das the

Kind child Kind child

wurde gekämmt was combed war gekämmt was combed

(40) a. das the b. die the

Anmelden der Gäste announcing the.gen guests Ameldung der Gäste announcement the.gen guests

(agent ≠ theme) (verbal passive) (agent = / ≠ theme) (adjectival passive) (agent ≠ theme) (nominalized infinitive) (agent = / ≠ theme) (-ung nominalization)

Conclusion: in the absence of Voice conceptual licensing jumps in as in English -ation nominals. However, the German preposition ‘durch’ seems not to be restricted to actors but is compatible with other theta-roles, too (cf. (16/17) above). 3.4 •

Greek nominalizations Recall that the Greek passive is restricted, while the nominalization is not.

No:

While Greek passive cannot select the feature combination 4, Greek nominals can select 4. This explanation is ad hoc and unrestricted.

Proposal 2: No:

Greek nominalization is like the English nominalization. There is evidence that it contains Voice.

Proposal 1:

Evidence: I. Greek nominalizations license adverbs that are Voice related. Initiator manner adverbs such as carefully are licit in the DP domain (Alexiadou 2001, Alexiadou & Anagnostopoulou 2007): (41) i katastrofi ton egrafon toso prosektika the destruction the documents-gen that carefully II.

Nominalizations of de-adjectival verbs can license Agent PPs. Since in the case of these roots, the Agent must be structurally licensed, as nothing from the semantics of the root could license it, it must be licensed by Voice. The preposition apo differs from by in that it is not strictly associated with the Actor role:

(42) to adiasma tu kutiu apo tin katharistria the emptying the box by the cleaner III.

The genitive DP is interpreted only as an internal argument (Alexiadou 2001). Thus (43) is not compatible with reflexive interpretation:

(43) i anagelia ton kalesmenon the announcing the guests-gen Proposal 3:

(agent≠theme)

Greek nominals derived from alternating verbs are structurally ambiguous between the passive and anticausative form.

10

Evidence: I. We saw above that nominals can have Voice. However, they do not need to contain Voice. Nominalizations of internally caused verbs never license an Agent PP/VoiceP: (44) to sapisma ton filon apo tin igrasia / *apo ton kipuro the rotting the leaves-gen from the humidity / by the gardener (45) a. Ta The b. *I the II.

fila sapisan leaves rotted igrasia / *o humidity / the

apo tin igrasia / *apo ton kipuro from the humidity / by the gardener kipuros skuriase ta fila gardener rotted the leaves

Modification by adjectives/adverbs like spontaneous/suddenly:

(46) a. to the b. to the

afthormito anigma tis portas spontaneous opening the door-gen ksafniko anigma tis portas sudden opening the door-gen

(47) a. to the b. to the

anigma opening anigma opening

(48) a. *to the b. to the

afthormito sapisma ton filon spontaneous rotting the leaves-gen ksafniko sapisma ton filon sudden rotting the leaves-gen

•

tis the tis the

portas door-gen portas door-gen

→ anticausative interpretation → passive interpretation → anticausative interpretation → *passive interpretation → anticausative interpretation

The Greek (46a) & (46b) correspond to two structures.

(49) a. Passive:

b. Anticausative:

[ n … [ Voice [ V …]] •

afthormita spontaneously ksafnika suddenly

→ passive interpretation

[ n … [ V …]]

In German, these patterns are morphologically distinct in the nominalized infinitive, as shown by the data in (50). In the absence of ‘sich’ the adjective ‘spontan’ (spontaneous) modifies an implicit agent's action; in the presence of ‘sich’ it modifies the unfolding of the inchoative event. 8

(50) a. das the b. das the

spontane spontaneous spontane spontaneous

Öffnen der Türen open-en the-gen doors Türen sich Öffnen der REFL open-en the-gen doors

A further problem: Greek nominals derived from verbs that do not form a passive are passive, i.e. contain Voice:

8

The corresponding –ung nominalization is ambiguous, as expected.

11

•

Recall that there are Greek verbs that do not form a nominalizations do, e.g. break and burn: (51) a. to afthormito spasimo ton piaton apo to the spontaneous breaking the dishes-gen by the b. to afthormito kapsimo ton vivlion apo to the spontaneous burning the books-gen by the

passive but the corresponding Jani John Jani John

•

At present we do not have a clear understanding of the gaps (lack of productivity) in the verbal passive. Nevertheless, in the nominal domain the formation of a passive is possible. All we can conclude for now is that the gap is dependent on the morphosyntactic context in which the root appears.

•

It seems that what is special is the verbal passive, not the nominalization.

Finally: The contrast between English and Greek nominalizations concerning goal/recipient PPs. (52)

*the receipt of the package by John

(53)

i paralavi tu paketu apo to Jani the receipt the package-gen by the John

•

This contrast is expected if Greek nominalizations have a Voice projection introducing external theta roles that are sentient in Dowty's term (besides the Causer).

•

English nominalizations do not contain Voice, and as we saw the English by phrase is semantically restricted.

4. Conclusions We argued for two different ways of introducing PPs relating to external arguments in the nominal domain: 1)

Presence of functional structure.

2)

In the absence of functional structure, PPs can be licensed via an interplay of the encyclopedic meaning of the root involved and the properties of the preposition itself.

Two questions: I) (54) •

Why is 2) only possible in the nominal domain? In other words, why is the anticausative in (54) ungrammatical? *The vase broke by John The domain of middles shows some interesting parallels: middles, on the one hand, involve an implicit Agent, which, however, is not structurally represented. This conflict was resolved by proposing that in this case the implicit Agent is introduced via an interplay of the generic semantics and the encyclopedic meaning of the root involved (Schäfer 2007). 12

II)

Why do we have Agent exclusivity in the passives of some languages (Greek/Hebrew) and in some constructions (German -en nominalizations, also Hebrew nominalizations, Sichel 2007)?

•

Although, we do not have an answer to this question, we note that the existence of such cases argues against an analysis of the passive as involving suppression of the external argument (because the corresponding transitives do not show this restriction). See Doron (2003) for an alternative approach.

References Alexiadou, A. 2001. Functional structure in nominals: nominalization and ergativity. Amsterdam: John Benjamins. Alexiadou, A. & E. Anagnostopoulou. 2007. Structuring participles. Proceedings of WCCFL 26. Alexiadou, A, E. Anagnostopoulou & F. Schäfer. 2006. The properties of anticausatives crosslinguistically. In Phases of Interpretation, ed. M. Frascarelli, Berlin: Mouton de Gruyter. Alexiadou, A. & E. Doron. 2007. The syntactic construction of two non-active voices: passive and middle. Paper presented at the Workshop on Global selective comparison, GLOW XXX, Tromsø. Baker, M., K. Johnson & I. Roberts. 1989. Passive Arguments raised. Linguistic Inquiry 20: 219-251. Borer, H. 1993. Parallel Morphology. Ms., University of Massachusetts at Amherst. Borer, H. 2003. The forming, the formation and the form of nominals. Handout USC. Borsley, R. & J. Kornfilt. 2000. Mixed Extended Projections. In Syntax and Semantics, Volume 32, pp. 101-131, edited by R. Borsley Academic Press, New York, NY Collins, C. 2005. A Smuggling Approach to the Passive in English. Syntax, 8.2: 81-120. Doron, E. 2003. Agency and Voice: the Semantics of the Semitic templates. Natural Language Semantics 11 (1): 1-67. Fox, D. & Y. Grodzinsky. 1998. Children's Passive: a View from the by-phrase. Linguistic Inquiry 29: 311-332. Fu, J., T. Roeper & H. Borer. 2001. The VP within Nominalizations: evidence from Adverbs and the VP Anaphor do-so. Natural Language and Linguistic Theory 19: 549-582. Grimshaw, J. 1990. Argument Structure. Cambridge, Mass.: MIT Press. Harley, H. & R. Noyer. 2000. Formal vs. Encyclopedic properties of vocabulary: evidence from nominalization. In B. Peters (ed.) The Lexicon-Encyclopedia Interface. Elsevier Press. Hornstein, N. 1997. S and X' Convention. Linguistic Analysis 3: 137-176. van Hout, A. & T. Roeper. 1998. 'Events and Aspectual Structure in Derivational Morphology'. MIT Working Papers in Linguistics 32: 175-220. Jaeggli, O. 1986. Passive. Linguistic Inquiry 17: 587-663. Kratzer, A. 1994. The Event Argument and the Semantics of Voice. Ms., University of Massachusetts, Amherst. Marantz, A. 1997. No escape from syntax: Don’t try morphological analysis in the privacy of your own lexicon. In University of Pennsylvania Working Papers in Linguistics, Vol. 4.2, eds. A. Dimitriadis, L. Siegel, C. Surek-Clark, and Alexander Williams, 201-225. Philadelphia. Marantz, A. 2005. Objects out of the lexicon: objects as event. Ms., NYU. Markantonatou, S. 1992. The Syntax of Modern Greek Noun Phrases with a derived nominal head. Ph.D. Diss. University of Essex. 13

Pesetsky, D. 1995. Zero Syntax. MIT Press. Ramchand, G. 2006. First phase syntax. Ms., University of Tromsø. Rappaport, M. 1983. 'On the Nature of Derived Nominals'. In B. Levin, M. Rappaport, and A. Zaenen (eds) Papers in Lexical-Functional Grammar. Indiana University Linguistics Club, pp. 113-142. Rossdeutscher, A. 2007. Resultatslesarten und modulare Repräsentation. Ms. University of Stuttgart. Schäfer, F. 2007. Middles as Voiced Anticausatives. In E. Efner and M. Walkow (eds.), Proceedings of NELS 37, Amherst, MA: GLSA. Sichel, I. 2007. Agent exclusivity in nominalization. Paper presented at the Workshop on Bare nouns and nominalizations, Stuttgart, June 2007. Williams, E. 1985. 'PRO and subject of NP'. Natural Language and Linguistic Theory 3: 297315.

14