Unmarked Passives – Evidence from German let-middles and let-passives

14
Unmarked Passives Evidence from German let-middles and let-passives 1 Marcel Pitteroff University of Stuttgart 1. Introduction Generally, argument alternations such as verbal passives require morphological marking that differentiates them from the corresponding active sentence (e.g., Haspelmath 1990; Kiparski 2013). This is shown in (1b) for German, where the passive contrasts with the active in its use of the auxiliary werden ‘become’ and a past/passive participle. (1) a. Peter las den Roman. (active) Peter read the.ACC novel ‘Peter read the novel.’ b. Der Roman wurde gelesen. (passive) the.NOM novel became read.PARTICIPLE ‘The novel was read.’ Unmarked passives are claimed not to exist. This generalization has lead to numerous theoretical accounts of passives in which “passive morphology” (i.e. passive participial morphology) was attributed some core function in the derivation of passives (see, e.g., Baker, Johnson and Roberts 1989, but also Haider 1984, 1986b, a.o.). In this paper, I provide evidence in support of the existence of unmarked passives in a very well defined domain. Empirically, my argument is based on two constructions in German that involve the causative predicate lassen ‘let’: the German equivalent of the Romance Faire Par construction in (2a) (henceforth let-passive), and the so-called let- middle in (2b) (cf. Fagan 1992, Kunze 1996, Pitteroff & Alexiadou 2012, Pitteroff 2014). 1 This paper has profited from numerous discussions with a number of people. I would like to thank Artemis Alexiadou, David Embick, Terje Lohndal, Heidi Harley, Florian Schäfer, Giorgos Spathas, Jim Wood, Susi Wurmbrand, and the audiences at NELS 45, the University of Leipzig, and NTNU Trondheim where versions of this paper have been presented for helpful and stimulating comments.

Transcript of Unmarked Passives – Evidence from German let-middles and let-passives

Unmarked Passives – Evidence from German let-middles and let-passives1

Marcel Pitteroff

University of Stuttgart

1. Introduction

Generally, argument alternations such as verbal passives require morphological marking

that differentiates them from the corresponding active sentence (e.g., Haspelmath 1990;

Kiparski 2013). This is shown in (1b) for German, where the passive contrasts with the

active in its use of the auxiliary werden ‘become’ and a past/passive participle.

(1) a. Peter las den Roman. (active)

Peter read the.ACC novel

‘Peter read the novel.’

b. Der Roman wurde gelesen. (passive)

the.NOM novel became read.PARTICIPLE

‘The novel was read.’

Unmarked passives are claimed not to exist. This generalization has lead to numerous

theoretical accounts of passives in which “passive morphology” (i.e. passive participial

morphology) was attributed some core function in the derivation of passives (see, e.g.,

Baker, Johnson and Roberts 1989, but also Haider 1984, 1986b, a.o.).

In this paper, I provide evidence in support of the existence of unmarked passives in a

very well defined domain. Empirically, my argument is based on two constructions in

German that involve the causative predicate lassen ‘let’: the German equivalent of the

Romance Faire Par construction in (2a) (henceforth let-passive), and the so-called let-

middle in (2b) (cf. Fagan 1992, Kunze 1996, Pitteroff & Alexiadou 2012, Pitteroff 2014).

1 This paper has profited from numerous discussions with a number of people. I would like to thank

Artemis Alexiadou, David Embick, Terje Lohndal, Heidi Harley, Florian Schäfer, Giorgos Spathas, Jim

Wood, Susi Wurmbrand, and the audiences at NELS 45, the University of Leipzig, and NTNU Trondheim

where versions of this paper have been presented for helpful and stimulating comments.

Marcel Pitteroff

(2) a. Der Lehrer ließ den Roman lesen. (let-passive)

the teacher let the novel read.INF

‘The teacher had the novel be read.’

b. Der Roman lässt sich gut lesen. (let-middle)

the novel lets REFL well read.INF

‘The novel reads well.’

In neither construction is the embedded external argument overtly realized. I will argue,

however, that the external argument introducing projection VoiceP (cf. Kratzer 1996) is

nevertheless present in the infinitival complement, and that this VoiceP is identical to the

one involved in verbal passives. Let-middles and let-passives are thus evidence that

passive morphology in German must be dissociated from Voice, as its absence in (2)

would otherwise be unexpected. This conclusion provides independent support for

analyses of passives à la Bruening (2012), where passive morphology is located in a

functional projection on top of Voice. Such approaches not only have the ability to derive

unmarked passives, but also make the correct predictions concerning where they should

be found: in restructuring infinitives (cf. Wurmbrand 2001, 2004, 2013).

The paper proceeds as follows: in Section 2 I provide arguments against a bare vP-

analysis of (2) and in favor of the presence of an embedded VoiceP. In Section 3, I

discuss from Bruening (2012) an approach that is compatible with the conclusion from

Section 2: passive morphology must be dissociated from Voice. Section 4 illustrates that

the infinitival complement in let-passives and let-middles involves a restructuring

infinitive. I then show how this fact interacts with the dissociation of Voice and passive

morphology in order to derive a morphologically unmarked passive argument

constellation. Section 5 concludes.

2. An embedded VoiceP

2.1 The issue

Pylkkänen (2008) proposes a three-way classification of causatives, depending on the

size of the complement of the causative predicate/morpheme: root-embedding, vP-

embedding and phase (=Voice)-embedding causatives.2 Causatives of the type in (2a),

where the embedded external argument (i.e. the causee) is missing, have predominantly

been treated as lacking such an argument altogether. In other words, they are classified as

vP-embedding causatives in Pylkkänen’s typology, lacking the external argument

introducing projection VoiceP (see Taraldsen 1983 for Norwegian; Burzio 1986 for

Italian; Folli and Harley 2007 for Romance Faire Par causatives in general; Enzinger

2010 for German; Tubino Blanco 2011, Harley 2013, 2014 for Hiaki indirect causatives;

2 Pylkkänen works in a Distributed Morphology framework, where the syntax operates with category-

neutral Roots. These Roots are identified as verbs, nouns, adjectives etc. by the immediate environment

they appear in (cf. Marantz 1997; i.e. n0 identifies its complement as nominal, v0 as verbal, etc). I adopt this

perspective in this paper.

Unmarked Passives

Wood 2011 for Icelandic; Key 2013 for Turkish).3 The advantage of such an approach is

obvious: under the assumption that passives require the presence of a Voice projection,

the absence of an embedded VoiceP in let-passives (and let-middles) entails that no

embedded passive is involved, and thus, no passive morphology is expected to surface.

I will now provide a number of arguments in favor of the presence of an embedded

VoiceP in these constructions.

2.2 Arguments for an embedded VoiceP

2.2.1 Voice Adjuncts

One test for the presence of an implicit argument/VoiceP in passives is the possibility to

introduce the suppressed agent as an adjunct by-phrase (3). (4) shows that in cases where

no implicit argument is involved – such as anticausatives (4a) or dispositional middles

(4b) – a by-phrase is infelicitous.

(3) Das Buch wurde von Martin gelesen.

the book became by Martin read

‘The book was read by Martin.’

(4) a. Die Tür öffnete sich (*von Martin). (anticausative)

the door opened REFL by Martin

‘The door opened by Martin.’

b. Das Buch liest sich (*von Kindern) gut. (dispositional middle)

the book reads REFL by children well

‘The book reads well.’

As is well known, let-passives allow the realization of the agent as a by-phrase, patterning

with passives rather than anticausatives or middles (5a). (5b) shows that unlike the

‘canonical’ dispositional middle (4b), let-middles also permit by-phrases (see Fagan

1992, Kunze 1996, Pitteroff & Alexiadou 2012, Pitteroff 2014 for German, Broekhuis

and Corver to appear, Ackema and Schoorlemmer 2006 for Dutch).

(5) a. Hans ließ die Tür (von Martin) öffnen.

Hans let the door by Martin open

‘Hans had Martin open the door.’

b. Das Buch lässt sich (von Kindern) gut lesen.

the book lets REFL by children well read

‘The book can be read well by children.’

3 To be precise, Folli and Harley (2007), following Guasti (1993, 2006), treat Faire Par causatives as

embedding a nominalized infinitive. I list them here as they claim that no VoiceP is present in the

infinitival complement, in agreement with all other authors mentioned.

Marcel Pitteroff

Defendants of a vP-embedding approach to (5) explain the acceptability of by-phrases by

treating them as event modifiers/vP-adjuncts. It seems, however, that such an account

overgenerates: what prevents the attachment of an agentive by-phrase to the middle in

(4b) or the anticausative in (4a), for that matter?4

Under the assumption, then, that by-phrases are indicative of the presence of Voice

(e.g. Alexiadou, Anagnostopoulou & Schäfer 2006, 2015, Lyngfelt and Solstad 2006; see

also Collins 2005, Bruening 2012), the acceptability of (5a,b) strongly suggests that the

infinitival complement in let-passives and let-middles projects a VoiceP.

Another type of adjunct that arguably targets the Voice projection is anti-assistive

selbst ‘self’, illustrated in (6) (Hole 2002, Spathas, Alexiadou & Schäfer to appear).

(6) a. Der Koch hat die Blaubeeren selbst gepflückt.

the cook has the blueberries self picked

‘The cook has picked the blueberries himself.’ (Hole 2002: 136, (6))

b. Die Blaubeeren wurden selbst geflückt.

the blueberries became self picked

‘The blueberries were picked on one’s own.’

(6a) means (roughly) that the cook picked the blueberries himself, and no one assisted

him in doing this (for a more thorough discussion of the semantics of anti-assistive

modifiers, see the literature mentioned above and the references therein). Crucially, anti-

assistive selbst ‘self’ does not need an overt agent DP, but can occur in passives (6b),

suggesting that the licensing factor is the presence of an (agentive) VoiceP. If anti-

assistive selbst modifies Voice, and the infinitival complement in let-passives and let-

middles lacks this projection, one would expect this modifier to be unacceptable in these

constructions – contrary to fact (7)-(8). (9) shows that anti-assistive selbst cannot occur in

contexts where an (agentive) VoiceP is missing (i.e. anticausatives and middles).

4 Furthermore, what are the restrictions that allow only some languages to have such an agentive vP-

modifier? As Wood (2011) shows, Icelandic causatives without overt causee disallow the addition of a by-

phrase (the same is true for the corresponding Turkish causatives (Jaklin Kornfilt, p.c.)):

(i) a. Þeir létu mig byggja húsið. (Icelandic; Wood 2011)

they let me build house.the

‘They made me build the house.’

b. Þeir létu byggja húsið (*af mér).

they let build house.the by me

INTENDED: ‘They made the house get built by me.’

Arguably, a by-phrase is acceptable in Icelandic let-passives if it denotes a property-like expression. Einar

Freyr Sigurðsson, p.c., finds (ii) acceptable.

(ii) Jón lét opna hurðina af fagaðila

John let open the.door by professional

‘John has a professional open the door.’

I leave it open what this entails for the syntax of Icelandic let-passives. I would like to stress, however, that

the Icelandic counterpart of (5a) is unacceptable. Thus, even if the restriction above is indicative of a vP-

adjunct (as argued in Dotlačil and Šimík 2013 for the by-phrase in Czech retroactive infinitives), the fact

that it does not hold in German supports the analysis of let-passives as embedding a VoiceP.

Unmarked Passives

(7) a. Der Lehrer hat die Prüfungsaufgabe (von seinen Schülern) immer

the teacher has the examination.question by his students always

selbst auswählen lassen.

self pick let

‘The teacher has always made his students pick the exam questions

themselves.’

b. Ich lasse selbst aussuchen, was gegessen wird. 5

I let self pick what eaten becomes

‘I let people choose for themselves what will be eaten.’

(8) a. Schmuckstücke, die nicht nur einzigartig und originell sondern

Pieces.of.jewelry that not only unique and fancy but

auch typgerecht sind, lassen sich am besten selbst herstellen. 6

also type.fitting are let REFL at.the best self produce

‘Pieces of jewelry that are not only unique and fancy, but also suit your

personal type, can be best produced oneself.’

b. Die vielen neuen Änderungen lassen sich am Besten selbst entdecken.7

the many new changes let REFL at.the best self discover

‘The many new changes can be best discovered oneself.’

(9) a. So ein Schmuckstück stellt sich leicht (*selbst) her.

such a piece.of.jewelry produces REFL easily self prt

‘Such a piece of jewelry can easily be produced by oneself.’

b. Diese Tür öffnete sich (*selbst).

this door opened REFL self

Intended: ‘Someone opened the door himself.’

Agent-oriented modifiers like bewusst ‘consciously’ or absichtlich ‘intentionally’ are

also used to test for the presence of an implicit argument/VoiceP (see, e.g., Key 2013 for

the claim that an agent-oriented modifier can target the causee only if the infinitival

complement projects a VoiceP). Such modifiers are hard to find in (let-) middles due to

their semantics (see Lekakou 2005 for discussion), but they are not unacceptable (10).

(10) a. Die ansonsten typische erdig grüne Holznote lässt sich nur

the otherwise characteristic earthy green wood.flavor lets REFL only

unbewusst wahrnehmen.8

unconsciously notice

‘The otherwise characteristic, earthy, green, wooden flavor can only be

noticed unconsciously.’

5<http://www.9monate.de/community/thread/Archiv-Kinderernaehrung-Erfahrungen-und

Tipps/Untergewicht? threadId=13162154> 6 < http://www.katrin-und-frank.de/2011/05/> 7 <http://www.dynasite.de/dynasite.cfm?dsmid=76745&dspaid=587223> 8 < http://www.parfumo.de/Benutzer/Imel/Kommentare/10>

Marcel Pitteroff

b. Die Blase entleert sich reflexartig und lässt sich nicht bewusst

the bladder empties REFL as.a.reflex and lets REFL not consciously

kontrollieren.9

control

‘The bladder empties as a reflex and cannot be controlled consciously.’

c. Aber dieses Gefühl lässt sich nicht absichtlich herbeiführen.10

but this feeling lets REFL not intentionally bring.about

‘But this feeling cannot be brought about intentionally.’

The situation for let-passives is more complex. To understand why this is the case, one

has to keep in mind that adjuncts in let-passives are ambiguous: they can be interpreted as

modifying the matrix or the embedded event (11).11

(11) Der Lehrer ließ den Text zweimal lesen.

the teacher let the text twice read

i. ‘Twice did the teacher make someone read the text’

(2 causing events)

ii.‘The teacher made someone read the text twice.’

(1 causing, 2 reading events)

As a consequence, an agent-oriented modifier in a let-causative should in principle be

able to modify the matrix or the embedded external argument. Yet, even in let-causatives

where the causee is overtly realized, an agent-oriented modifier has the very strong

tendency to modify the matrix agent ((12a); see Siloni (2011), who shows the same for

French causatives). A “low-scope” interpretation is only possible if the matrix subject

cannot be construed as the target of the modifier. This is exemplified in (12b), where the

matrix subject is inanimate.

(12) a. Der Psychologe ließ Martin das Auto absichtlich zerkratzen.

the psychologist let Martin the car intentionally scratch

i. ‘The psychologist intentionally made his patient scratch the car.’

ii.??‘The psychologist made his patient scratch the car intentionally.’

b. Der Neid ließ die Schüler absichtlich teure Autos zerkratzen.

the jealousy let the students intentionally expensive cars scratch

i.*‘Jealousy intentionally made the students scratch expensive cars.’

ii. ‘Jealousy made the students scratch expensive cars intentionally.’

The contrast between (12a) and (12b) suggests a prominence effect. Where possible, an

agent oriented modifier picks the more prominent agent, i.e. the matrix subject, rather

than the causee. Modification of the embedded agent is only possible where modification

9<http://www.special-harninkontinenz.de/service/glossar/blasenerkrankungen-blase-harnwege

id74180.html> 10 <http://www.spiegel.de/spiegel/print/d-9286027.html> 11 This property is evidence against let-causatives as root-selecting. Furthermore, the ambiguity in the

interpretation of adjuncts is often attributed to the restructuring status of this construction.

Unmarked Passives

of the matrix subject is ruled out for independent reasons. Two predictions follow from

this: First, the prominence effect should be even stronger if the embedded agent is covert.

Second, modification of the implicit agent should be possible if the matrix subject is

inanimate. While the first prediction is borne out (13a), the second one is not (13b).

(13) a. Der Psychologe ließ das Auto absichtlich zerkratzen.

the psychologist let the car intentionally scratch

i. ‘The psychologist intentionally made someone scratch the car.’

ii.*?‘The psychologist made someone scratch the car intentionally.’

b. *Der Neid ließ absichtlich teure Autos zerkratzen.

the jealousy let intentionally expensive cars scratch

‘Jealousy made someone/people scratch expensive cars intentionally.’

(13b), however, does not falsify the prominence argument: the example is ungrammatical

even without the modifier. This is because let-passives, unlike the let-causatives in (12),

are incompatible with an inanimate matrix subject (see Huber 1980 for German; Folli and

Harley 2007 for Romance Faire Par causatives).

What is needed is thus a let-passive with an animate matrix subject which cannot be

or is at least unlikely to be the target of an agent-oriented modifier. (14) provides such a

scenario, and a low interpretation for the modifier is acceptable.

(14) Gestern ließ der Hypnotiseur von seinen Gästen unbewusst ein Auto

yesterday let the hypnotist by his guests unconsciously a car

demolieren.

demolish

i.??‘Yesterday, the hypnotist unconsciously made his guests demolish a car.’

ii. ‘Yesterday, the hypnotist made his guests demolish a car unconsciously.’

(14) shows that once once the prominence effect is neutralized, an agent-oriented

modifier can target the implicit causee. This therefore suggests that the infinitival

complement of let-passives does indeed involve Voice.12

2.2.2 Stem Allomorphy

This argument builds on the observation that certain predicates mark the causative

alternation via stem allomorphy. An example of this is provided in (15).

(15) a. Peter muss das Schiff versenken.

Peter must the ship sink.TRANSITIVE

‘Peter has to sink the ship.’

12 Pitteroff (2014) argues that the failure of the implicit causee to control into purpose clauses is also a

consequence of the prominence of the matrix subject. Vecchiato (2011) proposes the same for Italian Faire

Par causatives. In support, she advances the English example in (i), where control by the causee is

impossible, even though the infinitival complement clearly involves a passive.

(i) The major had the enemies attacked by General Custer, without informing the government.

Marcel Pitteroff

b. Das Schiff muss versinken.

the ship must sink.INTRANSITIVE

‘The ship must sink.’

In Embick (2010, 2012) stem allomorphy is treated via phonological readjustment rules

(morpheme-morpheme rules in Embick 2012), which are triggered by a morpheme in a

certain local relationship to the Root undergoing the change. Based on Alexiadou,

Anagnostopoulou & Schäfer’s (2006, 2015) analysis of the causative alternation as a

Voice alternation, the sole element to trigger stem allomorphy in the cases above is

Voice.13 (17) provides the syntactic decomposition of causatives and anticausatives, and

(18) the rule that derives the stem allomorphs (I use a simplified readjustment rule here

for the sake of illustration; ‘ ̑ ’ indicates concatenation; see Embick 2010, 2012).

(17) The syntactic decomposition of causatives and inchoatives

a. [VoiceP Voice [vP v [√P √ROOT ]]] (causative)

b. [vP v [√P √ROOT ]] (anticausative)

(18) Readjustment rule

sink → senk /__ ̑ Voice

(18) entails that the causative stem allomorph is only found in the context of Voice. This

therefore suggests that stem allomorphy in the causative alternation is a Voice diagnostic.

Returning to let-middles and -passives, Pitteroff and Alexiadou (2012) and Pitteroff

(2014) note that in let-middles only a causative stem allomorph is acceptable (19). Based

on the discussion above, this means that the infinitival complement involves Voice.

(19) Das Schiff lässt sich leicht versenken / *versinken.

the ship lets REFL easily sink.TRANSITIVE / sink.INTRANSITIVE

‘The ship can be sunk easily.’

For let-passives, the argument is again more involved, since the anticausative variant

is acceptable below lassen ‘let’.

(20) a. Peter lässt das Schiff versenken.

Peter lets the ship sink.TRANSITIVE

‘Peter makes someone sink the ship.’

13 Note that this account does not make use of different flavors of v (as in Harley 1995 and much

subsequent work). The vP involved in the causative is identical to the one in the anticausative – both

introduce an event. Causative semantics are not encoded on a specific syntactic head but read off from the

structure at the CI-interface: i.e. causation equals the combination of a process and a resultant state.

However, Harley (2014) provides a compelling analysis of Hiaki indirect causatives, which also involve an

implicit causee and no passive marking. Her analysis is based on the assumption that agentivity can be split

across vP and VoiceP, such that the former introduces agentive semantics while the latter hosts the

argument. Note, however, that this account requires different flavors of v and is thus not available here. I

leave a detailed comparison of the two approaches for future research.

Unmarked Passives

b. Peter lässt das Schiff versinken.

Peter lets the ship sink.INTRANSITIVE

‘Peter lets/makes the ship sink.’

However, it can still be shown that let-passives, like let-middles, necessarily embed the

causative variant, supporting the claim that let-passives involve an embedded VoiceP.

Recall that let-passives are incompatible with an inanimate matrix subject (cf. (14)).

Changing the animacy of the matrix subject in (20) leads to ungrammaticality only in the

case of the embedded causative variant (21).

(21) Der Sturm lässt das Schiff versinken /*versenken.

the storm lets the ship sink.INTRANSITIVE / sink.TRANSITIVE

‘The storm sinks the ship/makes someone sink the ship.’

(21) thus shows that (20b) is not a let-passive. More specifically, (21) and (22) indicate

that let-passives necessarily embed the causative stem allomorph, thus supporting the

claim that let-passives involve an embedded VoiceP.14

Having shown that the infinitival complement in let-middles and let-passives involves

a VoiceP, it has to be stressed that this VoiceP – although agentive – must not introduce

an argument in its specifier. In other words, the VoiceP in the infinitival complement of

lassen in let-passives and let-middles is identical to the one involved in verbal passives

(see Embick 2004, Alexiadou, Anagnostopoulou & Schäfer 2006, 2015, Schäfer 2008,

Bruening 2012, Harley 2013, a.o. for the claim that verbal passives involve an agentive

but specifierless VoiceP). This then leads us to the morphological issue of this paper.15

3. Separating passive morphology from Voice

3.1 Passive morphology and Pass

The conclusion from the previous section that let-middles and let-passives embed the

same type of VoiceP as is involved in verbal passives, combined with the observation

that no passive morphology must surface (25), leads to the null hypothesis that passive

morphology in German is not the spell-out of Voice0.

(25) a. Peter lässt das Buch lesen / *gelesen (werden).

Peter lets the book read.INF / read.PARTICIPLE become

‘Peter makes someone read the book.’

14 This conclusion is confirmed by the observation that only the causative stem allomorph allows the

addition of a by-phrase typical for let-passives (i). Thanks to Florian Schäfer for pointing this out.

(i) Peter lässt das Schiff von Piraten versenken / *versinken.

Peter lets the ship by pirates sink.TRANSTIVE / sink.INTRANSITIVE

‘Peter makes pirates sink the ship.’ 15 See also Wilder (1990) for the claim that let-passives embed a passive. Following Baker, Johnson and

Roberts’ (1989) account of passive morphology, he treats the infinitival suffix –en as an argument.

Marcel Pitteroff

b. Das Buch lässt sich leicht lesen / *gelesen (werden).

the book lets REFL easily read.INF / read.PARTICIPLE become

‘The book reads easily.’

Such a dissociation of passive morphology from Voice has been independently proposed

in analyses of verbal passives such as Embick (2004) or Bruening (2012). For the sake of

concreteness I will follow the latter, although nothing hinges on this choice.

Bruening (2012) proposes a syntactic decomposition of verbal passives as in (26) (see

also Wurmbrand 2013; Alexiadou, Anagnostopoulou & Schäfer 2015; Spathas,

Alexiadou & Schäfer to appear).

(26) Decomposition of passives

[PassP Pass [VoiceP Voice [vP v [√P √ROOT]]]]

Pass0 selects for an unsaturated predicate – a VoiceP that introduces an argument variable

with no DP in its specifier to saturate it. This is exactly the type of VoiceP that is

involved in the infinitival complement of let-middles and let-passives (see section 2).

Pass0 further imposes existential closure over the open external argument variable.

Finally, for Bruening, the Pass-Voice-v sequence is realized as passive morphology.

I propose that (26) can account for a situation where an agentive, yet specifierless

VoiceP does not simultaneously lead to passive morphology: if the latter depends on the

presence of Pass0, it is expected to be absent in contexts where PassP is not projected.16

Of course, where an infinitival complement projects up to TP or CP – as is the case in

most infinitival complements – PassP will necessarily be part of the embedded structure

and passive morphology will surface. In order to find the right context for a potential

unmarked passive, we thus have to look for structurally reduced complements.

Wurmbrand (2001, 2004, 2013) has argued for just such an analysis of restructuring

infinitives (RIs). The proposed explanation for the absence of passive morphology in let-

middles and let-passives would thus find support if it could be shown that their infinitival

complement is a restructuring infinitive. This will be done in the next section.

4. A restructuring infinitive

Numerous tests have been developed to identify restructuring infinitives (see a.o. Haider

1986a,b, 2003, Fanselow 1989, Wurmbrand 2001, 2004 for German). Due to space

limitations, I only employ two such tests to show that let-middles and let-passives involve

restructuring: remnant topicalization and extraposition (see Wurmbrand 2001 for a

critical discussion of extraposition as a restructuring diagnostic).

Remnant topicalization (also known as verbal complex topicalization; see Haider

1986a,b) involves the topicalization of the matrix and the embedded verb to the exclusion

of any other material belonging to the embedded clause – which, as a consequence, must

16 I assume that the infinitival morpheme –en is inserted as a default to ensure morphological acceptability,

since German (unlike English) renders bare stems unacceptable.

Unmarked Passives

have been scrambled into the matrix clause prior to topicalization. Given that scrambling

is clause-bound, remnant topicalization is only possible in monoclausal, i.e. restructuring

configurations. While non-restructuring infinitives disallow this process (27a), the result

is legitimate in the context of let-passives (27b) and let-middles (27c).

(27) a. *[zu versenken geplant] hat Peter das Schiff gestern Nachmittag.

to sink planned has Peter the ship yesterday evening

‘Yesterday evening, Peter planned to sink the ship.’

b. [versenken lassen] hat Peter das Schiff gestern Nachmittag.

sink let has Peter the ship yesterday evening

‘Yesterday evening, Peter made someone sink the ship.’

c. [versenken lassen] hat sich dieses Schiff ganz leicht.

sink let has REFL this ship very easily

‘It was very easy to sink this ship.’

RIs are generally assumed to block extraposition of the infinitival complement. This

is shown in (28b), where long object movement (another restructuring diagnostic

indicated by nominative case on the embedded theme argument; see Wurmbrand 2001) is

incompatible with extraposition.

(28) a. Gestern wurde versucht, [den Traktor zu reparieren].

yesterday became tried the.ACC tractor to repair

‘People tried to repair the tractor yesterday.’

b. *Gestern wurde versucht, [der Traktor zu reparieren].

yesterday became tried the.NOM tractor to repair

‘People tried to repair the tractor yesterday.’

Let-passives and -middles also behave like restructuring configurations in disallowing

extraposition of the infinitival complement (29).

(29) a. *Gestern hat Peter lassen [den Traktor reparieren].

yesterday has Peter let the tractor repair

‘Yesterday, Peter had the tractor repaired.’

b. *Am Anfang hat sich der Traktor lassen [leicht reparieren].

at.the beginning has REFL the tractor let easily repair

‘At the beginning, it was easy to repair the tractor.’

Wurmbrand (2001, 2004) argues that the mono-clausal properties of RIs derive from

the truncated nature of the infinitival complement. Although in Wurmbrand (2001, 2004)

RIs are argued to be bare vPs, it has recently been shown that RIs in fact project up to

Voice (Wurmbrand 2013, Chung 2004).

Marcel Pitteroff

Thus, the fact that the infinitival complement in let-passives and let-middles involves

Voice (Section 2) and qualifies as an RI lends support to the account of unmarked

passives sketched in section 3: since passive morphology in German is not located in

Voice, but in a higher projection (PassP; or AspP as in Embick 2004), and lassen ‘let’ in

let-middles and let-passives combines with an RI that projects up to Voice but no higher,

the infinitival complement lacks the projection that introduces passive morphology – thus

resulting in a passive argument constellation without morphological marking.

5. Conclusion

In this paper, I provided evidence for a passive argument constellation in the infinitival

complement of let-passives and let-middles which is, however, morphologically

unmarked. I argued that this strongly suggests that passive morphology in German must

be dissociated from Voice. Support for this claim came from the observation that the

infinitival complement of the two constructions in question qualifies as a restructuring

infinitive – a type of infinitive which has been independently shown to project up to

Voice, but not higher. Thus, although the infinitival complement in let-passives and let-

middles contains the projection that regulates the presence/absence of the external

argument (Voice), it lacks the one that gives rise to passive morphology (Pass).

References

Ackema, Peter & Maaike Schoorlemmer. 2006. Middles. In The Blackwell Companion to

Syntax, Vol. III, eds. M. Everaert and H. van Riemsdijk, 131‐203. Oxford: Blackwell.

Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. 2006. The properties of

anticausatives cross-linguistically. In Phases of Interpretation, ed. M. Frascarelli, 187-

211. Berlin: de Gruyter.

Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. 2015. External

Arguments in Transitivity Alternations: a Layering Approach. Oxford: Oxford

University Press.

Baker, Mark, Kyle Johnson & Ian Roberts. 1989. Passive Arguments Raised. Linguistic

Inquiry 20: 219-251.

Broekhuis, Hans & Norbert Corver. in prep. Syntax of Dutch. Verbs and Verb phrases,

vol. 1. Amsterdam: Amsterdam University Press.

Bruening, Benjamin. 2012. By-Phrases in Passives and Nominals. Syntax 16: 1-41.

Burzio, Luigi. 1986. Italian syntax: A government and binding approach. Dordrecht:

Reidel.

Chung, Sandra. 2004. Restructuring and verb-initial order in Chamorro. Syntax 7.3: 199-

233.

Collins, Chris. 2005. A Smuggling Approach to the Passive in English. Syntax 8.2: 81-

120.

Dotlačil Jakub, and Radek Šimík. 2013. Peeling, structural Case, and Czech retroactive

infinitives. In Proceedings of FDSL 9, eds. Uwe Junghans et al., 105-124. Frankfurt:

Lang.

Unmarked Passives

Embick, David. 2004. On the structure of resultative participles in English. Linguistic

Inquiry 35.3: 355-392.

Embick, David. 2010. Localism versus Globalism in Morphology and Phonology.

Cambridge: MIT Press.

Embick, David. 2012. Contextual conditions on stem alternations: Illustrations from the

Spanish conjugation. In Romance Languages and Linguistic Theory 2010, eds. I.

Franco, S. Lusini and A. Saab, 21-40. Amsterdam: John Benjamins.

Enzinger, Stefan. 2010. Kausative und perzeptive Infinitivkonstruktionen. Studia

grammatica 70. Berlin: Akademie Verlag.

Fagan, Sarah. 1992. The Syntax and Semantics of Middle Constructions: A Study with

Special Reference to German. Cambridge: Cambridge University Press.

Fanselow, Gisbert. 1989. Coherent infinitives in German. Restructuring vs. IP-

Complementation. In Syntactic phrase structure phenomena in noun phrases and

sentences, eds. C. Bhatt, E. Löbel and C. M. Schmidt. Linguistik Aktuell/Linguistics

Today 6, 1-16. Amsterdam: Benjamins.

Folli, Raffaella & Heidi Harley. 2007. Causation, Obligation, and Argument Structure:

On the Nature of Little v. Linguistic Inquiry 38: 197-238.

Guasti, Maria Teresa. 1993. Causative and perception verbs. Turin: Rosenberg and

Sellier.

Guasti, Maria Teresa. 2006. Analytic Causatives. In The Blackwell Companion to Syntax,

Vol. I, eds. M. Everaert and H. van Riemsdijk, 142-172. Oxford: Blackwell.

Haider, Haider. 1984. Was zu haben ist und was zu sein hat – Bemerkungen zum

Infinitiv. Papiere zur Linguistik 30: 23-36.

Haider, Hubert. 1986a. Nicht‐sententiale Infinitive. Groninger Arbeiten zur

Germanistischen Linguistik 28:73‐114.

Haider, Hubert. 1986b. Fehlende Argumente: vom Passiv zu kohärenten Infinitiven.

Linguistische Berichte 101: 3-33.

Haider, Hubert. 2003. V-Clustering and Clause Union-Causes and Effects. In Verb

Constructions in German and Dutch, eds. P. Seuren and G. Kempen, 91-126.

Amsterdam: Benjamins.

Harley, Heidi. 1995. Subjects, Events and Licensing. Ph.D. dissertation, MIT.

Harley, Heidi. 2013. External arguments and the Mirror Principle: On the distinctness of

Voice and v. Lingua 125: 34-57.

Harley, Heidi. 2014. Aggressive Passives – Voice and Implicit Arguments in Hiaki. Talk

at SynSalon, Nov. 17 2014.

Haspelmath, Martin. 1990. The grammaticization of passive morphology. In Studies in

Language 14: 25-72.

Hole, Daniel. 2002. Agentive selbst in German. In Proceedings of Sinn und Bedeutung

VI, eds. Graham Katz, Sabine Reinhard & Philip Reuter, 133-150.

Huber, Walter. 1980. Infinitivkomplemente im Deutschen – Transformations-

grammatische Untersuchungen zum Verb lassen. Ph.D. dissertation, Freie Universität

Berlin.

Key, Gregory. 2013. The Morphosyntax of the Turkish Causative Construction. Ph.D.

dissertation, University of Arizona.

Marcel Pitteroff

Kiparski, Paul. 2013. Towards a null theory of the passive. Lingua 125: 7-32.

Kratzer, Angelika. 1996. Severing the External Argument from its Verb. In Phrase

Structure and the Lexicon, eds J. Rooryck and L. Zaring, 109-137. Dordrecht: Kluwer.

Kunze, Jürgen. 1996. Plain middles and ‘lassen’ middles in German: Reflexive

constructions and sentence perspective. Linguistics 34: 645-695.

Lekakou, Marika. 2005. In the middle, somewhat elevated. The semantics of middles and

its cross-linguistic realization. Ph.D. dissertation, University of London.

Lyngfelt, Benjamin. & Torgrim Solstad. 2006. Perspectives on demotion: Introduction to

the volume. In Demoting the Agent – passive, middle and other voice phenomena, eds.

B. Lyngfelt and T. Solstad, 1-20. Amsterdam: John Benjamins.

Marantz, Alec. 1997. No escape from syntax: Don't try morphological analysis in the

privacy of your own Lexicon. In Proceedings of the 21st Annual Penn Linguistics

Colloquium: Penn Working Papers in Linguistics 4: 2, eds. A. Dimiatridis et al., 201-

225.

Pitteroff, Marcel & Artemis Alexiadou. 2012. The Properties of German sl-middles. In

Proceedings of WCCFL 29, eds. J. Choi et al., 214-222.

Pitteroff, Marcel. 2014. Non-canonical let-middles. Ph.D. dissertation, University of

Stuttgart.

Pylkkänen, Liina. 2008. Introducing Arguments. Cambridge: MIT Press.

Schäfer, Florian. 2008. The Syntax of (Anti‐)Causatives. External arguments in

change‐of‐state contexts. Linguistics Today 126. Amsterdam: John Benjamins.

Siloni, Tal. 2011. Reciprocal verbs and symmetry. NLLT 30: 261-320.

Spathas, Giorgos, Artemis Alexiadou & Florian Schäfer. To appear. Middle Voice and

reflexivization: afto-prefixation in Greek. NLLT.

Taraldsen, Knut Tarald. 1983. Parametric Variation in Phrase Stucture: A Case Study.

Ph.D. dissertation, University of Tromsø.

Tubino Blanco, Mercedes. 2011. Causatives in Minimalism. Linguistics Today 179.

Amsterdam: John Benjamins.

Vecchiato, Antonella. 2011. Events in the Grammar of Direct and Indirect Causation.

Ph.D. dissertation, University of Southern California.

Wilder, Chris. 1990. Passive and the German Infinitive. Frankfurter Linguistische

Forschungen 9: 1-31.

Wood, Jim. 2011. Icelandic let-causatives and case. Working Papers in Scandinavian

Syntax 87, 1-52.

Wurmbrand, Susi. 2001. Infinitives: Restructuring and clause structure. Berlin: de

Gruyter.

Wurmbrand, Susi. 2004. Two types of restructuring: Lexical vs. functional. Lingua

114:8. 991‐1014.

Wurmbrand, Susi. 2013. Complex predicate formation via voice incorporation.

manuscript, University of Connecticut. <http://ling.auf.net/lingbuzz/001968>.

Marcel Pitteroff

[email protected]