The Expletive Puzzle - Impersonal (let-)middles

Transcription

The Expletive Puzzle - Impersonal (let-)middles
CGSW 28 University of Leipzig October 4‐5, 2013 The Expletive Puzzle ‐ Impersonal (let‐)middles Marcel Pitteroff University of Stuttgart [email protected]‐stuttgart.de 1. Introduction 1.1 The data  Apart from the personal middles in (1) (see Fagan 1992; Ackema and Schoorlemmer 1994, 1995, 2005; Steinbach 2002; Lekakou 2005; a.o. for analysis), German has impersonal middles (IM henceforth; 2). (1) a. The book reads easily. b. Das Buch liest sich gut. German The book reads REFL well (2) a. *It walks nicely in these shoes. b. In diesen Schuhen läuft *(es) sich gut. German in these shoes walks (it) REFL well ‐ (2) shows that the expletive in IM is obligatory.  German has a further construction which qualifies as a dispositional middle: let‐middles (3; see Reis 1976, Hoehle 1978, Grewendorf 1983, Fagan 1992, Ackema and Schoorlemmer 2005 for some comments, and Kunze 1996 for the most elaborate analysis to date): (3) a. Das Buch lässt sich gut lesen. German The book lets REFL well read 'This book reads well.' ‐ In German, let‐middles also have impersonal counterparts (impersonal let‐middles; ILM henceforth, 4): (4) In diesen Schuhen lässt (es) sich gut laufen. German In these shoes lets it REFL nicely walk 'One can walk nicely in these shoes.' - In comparison to IM (e.g. 2b), the expletive in impersonal let‐middles (ILM) is optional.  Ultimately, the data above have to be contrasted with impersonal passives, which render an expletive ungrammatical in German (5). (5) Die ganze Nacht wurde (*es) gelacht. The whole night became it laughed 'There was laughing all night long.' 1
Marcel Pitteroff ‐ The Expletive Puzzle 1.2 The core questions  I focus on the impersonal variants of the middle constructions. The core questions are: ‐ What is the function of the expletive pronoun? ‐ Why is the expletive obligatory in impersonal middles? ‐ Why is the expletive optional in impersonal let‐middles? 1.3 Outline: Section 2: Expletives ‐ a brief overview Section 3: The expletive in IM Section 4: Extension to ILM Section 5: Conclusion Appendix: Discussion of other approaches to impersonal middles 2. Expletives ‐ an overview  In English, two types of expletives are typically distinguished: there‐type expletives and it‐type expletives (see e.g. Bennis 1986, Vikner 1995, Biberauer 2003, Biberauer and Richards 2006 for this split). - there‐type expletives: non‐argumental; signal the absence of a topical(isable) subject as in (8) (see Rosengren 2002, Biberauer 2003, Richards and Biberauer 2005, Biberauer and Richards 2006 for some recent discussion). (8) There was a dog in the garden. - it‐type expletives: occupy an argument position; e.g. weather it (9) (9) It rained.  As is well‐known, the distinction between the two types is obliterated in German SpecCP position. (10a) involves the phonologically identical Vorfeld‐es: (10) a. Es war ein Hund im Garten. It was a.NOM dog in.the garden 'There was a dog in the garden.' b. Es regnete. It rained 'It rained.' - Only argumental es can occur in sentence internal position. I remain silent on whether (11a) involves a covert expletive (see e.g. Grewendorf 1989, Vikner 1995 for this claim) (11) a. weil (*es) ein Hund im Garten war. because it a dog in.the garden was 'because there was a dog in the garden.' 2
CGSW 28 -
University of Leipzig October 4‐5, 2013 b. weil *(es) geregnet hat. because it rained has 'because it rained.' We thus arrive at the following distribution a) Vorfeld‐es; located in SpecCP b) (Quasi‐)argumental es; located VP internally c) ?expletive pro; located in SpecTP? (see Appendix for discussion) Argumental it is not restricted to weather it, but (optionally) turns up with certain predicates that take a clausal complement (Korrelat es in German linguistics): (12) a. Peter hat ??(es) unterlassen, die Eltern zu informieren.. Peter has it omitted the parents to inform 'Peter neglected to inform his parents.' b. Peter hat (es) bedauert, den Kollegen beleidigt zu haben. Peter has it regretted the collegue offended to have 'Peter regretted having offended the collegue.' - Based on extraction possibilities, Bennis (1986) argues that the embedded clause has a different status depending on whether the expletive is present or not (see also Haider 2010; Vikner 1995 claims that the picture is in fact more complicated): (13) The syntactic status of sentential complements with expletive: adjunct status; expletive in object position; wh‐movement blocked (=14a) without expletive: complement status; wh‐movement acceptable (=14b) (14) a. *Wen hat Peter (*es) unterlassen, tWH zu informieren? Whom has Peter it omitted to inform b. Wen hat Peter (*es) bedauert, tWH beleidigt zu haben? Who has Peter it regretted offended to have - The canonical view on the expletive in (12) and (14a) is thus, that the expletive occupies an argument position and takes the sentence adjunct as associate.  With respect to the properties of it/there‐type expletives, I essentially follow Bennis (1986), Vikner (1995), and Ruys (2010) in assuming the following: (17) a. het/it: argumental; requires case (and theta‐role?); full ‐specification [3rd, sg. neut] (see also Cardinaletti 1990) b. er/there: non‐argumental; does not require case; ‐deficient (see e.g. Richards and Biberauer 2005) 3
Marcel Pitteroff ‐ The Expletive Puzzle 3. Expletives in impersonal middles 3.1 The what, why, and where of es - In this section, I am going to argue that - the expletive used in impersonal middles is (quasi‐) argumental. - the need for the expletive is a consequence of the semantics of dispositional statements and the generic operator that is responsible for them. ‐ the es in impersonal middles is base‐generated as the subject of a small clause/Relator Phrase, whose predicate is a locative PP. 3.2 The what of es  es in German IM is argumental in nature (=it‐type expletive): it can occur in sentence internal position. (15) a. Hier tanzt es sich gut. Here dances it REFL easily b. Ich glaube, dass es sich mit 2 Promille nicht mehr so gut singt. I believe that it REFL with 2 per mil not anymore so well sings ‐ A further argument supporting this assumption comes from Dutch. - Dutch, like English, has two expletives that differ in their phonological make‐up, het and er (see e.g. Bennis 1986, Vikner 1995, Ruys 2010). het = it‐type expletive; er = there‐type expletive Dutch impersonal middles require the argumental expletive het and disallow er: (16) a. Het/*er fietst lekker op het fietspad. it/ there cycles nicely on the bikeway 'It is nice to cycle on the bikeway.' b. Het/*er eet niet prettig in een hoog tempo. it/ there eats not pleasantly in a high speed 'It is not pleasant to eat at a high speed.' 3.3 The why of es - Argument: The need for the expletive in IM is a consequence of the dispositional nature of middle constructions. 3.3.1 Dispositional semantics (Lekakou 2005) Lekakou (2005), following Condoravdi (1989), develops a detailed semantics for dispositonal sentences, building on the following three core‐characteristics: 4
CGSW 28 University of Leipzig October 4‐5, 2013 (19) Dispositional generics a. express 'in virtue of' generalizations b. are subject‐oriented c. involve a VP‐level generic operator 
Dispositional generics are 'in virtue of' generalizations: it is generally acknowledged that in middles, some property is ascribed to the structural subject (Roberts 1987, Fagan 1992, Ackema and Schoorlemmer 1994, 1995, 2005, a.o.): (20) These clothes wash easily. = These clothes have the property that it is easy to wash them Dispositionality according to Lekakou (2005): some generalization (e.g. that the property denoted by the VP is generally true of the structural subject) follows in virtue of an implicit, inherent property of the subject‐referent. Consider Fara's (2001) informal characterization of dispositional generics: (21) a. Sugar is disposed to dissolve when put in water. b. 'N is disposed to M when C' is true iff N has an intrinsic property in virtue of which it Ms when C.' Applied to the dispositional middle in (20), this results in the following paraphrase: (20') These clothes wash easily. = In virtue of some inherent property of these clothes (fabric, texture, ...) they can be washed easily. 
Dispositional generics are subject‐oriented: the property in virtue of which the generalization holds has to be a property of the subject (21‐22). This does not hold for non‐dispositional statements (23): 1 (21) This car drives well... a. ... because the suspension is engineered well. b. ??... because we're driving on smooth pavement. (Dowty 2001) 1
It can be shown that it is not the topic of a sentence that is assigned the dispositional property. Consider the following two sentences, were arguably the nominative DP and the locative PP function as topics respectively: (i) Landrovers fahren sich auf schwierigem Gelände besonders gut. Landrovers drive REFL on difficult terrain particularly well (ii) Auf schwierigem Gelände fahren sich Landrovers besonders gut. On difficult terrain drive REFL landrovers particularly well Both (i) and (ii) essentially attribute the property to the nominative DP, independent from whether it is a topic or not. The same can be shown for let‐middles. Both (i) and (ii) are statements about the referent of the nominative (a book), rather than the dative DP (a teenager): (iii) Ein Buch lässt sich einem Teenager heutzutage nur noch schwer verkaufen. A book lets REFL a teenager nowadays only still difficult sell (iv) Einem Teenager lässt sich heutzutage nur noch schwer ein Buch verkaufen. A teenager lets REFL nowadays only still difficult a book sell 'It is difficult to sell a book to a teenager these days.'
5
Marcel Pitteroff ‐ The Expletive Puzzle (22) (23)
-
The clothes wash with no trouble... a. ... because they are machine‐washable. b. *... I have lots of time. (van Oosten 1977) It's no trouble to wash the clothes... a. ... because they are machine‐washable. b. ... I have lots of time. (van Oosten 1977) The subject‐orientedness has been captured differently in the literature (see McConnell‐Ginet 1994 and Tao 2011 for a semantic analysis incorporating causation, or Zwart 1998 for a syntactic approach involving a causative/permissive little v). -
For Lekakou it follows from the VP‐level generic operator. Dispositional Generics involve a VP‐level GEN: Lekakou (2005) extends Brennan's (1993) theory of root modals as VP‐level operators to generic operators: () S‐level GEN: habitual sentences VP‐level GEN: dispositional sentences - Roughly: while the conversational background (see Kratzer XXXX) of S‐level operators involves propositions, the conversational background of VP‐level operators involves bundles of property expressions, which are interpreted relative to the syntactic subject.  inherent properties of the structural subject determine the conversational background  Dispositional generics: the inherent properties of the structural subject restrict the set of possible worlds that GEN/ quantifies over to exactly those worlds in which the subject‐referent has the same set of properties it has in the actual world (see Lekakou 2005 for details). In sum: "The dispositionality component of the middle interpretation requires that what occupies the subject position is the subject of the dispositional predicate." (Lekakou 2005, 95) -
Since in German, movement of the subject to SpecTP is not necessary (see den Besten XXXX, Wurmbrand 2006, a.o. as well as the example in (27)), I implement the subject orientedness formally by assuming that the element to which the dispositional property is assigned to needs to value [u:__] on T, but does not necessarily need to occur in SpecTP itself. (27) weil sich einem Obdachlosen eine Decke gut verkauft. because REFL a.DAT homeless.person a.NOM blanket well sells 'because a blanket can be sold well to a homeless person.' 6
CGSW 28 University of Leipzig October 4‐5, 2013 3.3.2 Disposition ascription in IM #1 - I return to the question of why the expletive is required in IM. - Based on the preceding section, the question arises as to what determines the accessibility relation/the conversational background in the absence of a (referential) subject?  Apart from the obligatory manner adverb, impersonal middles differ from personal ones in requiring a further modifier: a locative PP 2 (see also Abraham 1987, Hoekstra and Roberts 1993, Fagan 1992, Lekakou 2005, Broekhuis and Corver in prep for this observation). (29) a. weil es sich *(auf diesen ruhigen Straßen) schön fährt. because it REFL on these quiet streets nicely drives b. weil es sich *(auf diesem Stuhl) gut sitzt. because it REFL on this chair well sits c. weil es sich *(an ruhigen Nachmittagen) am Besten arbeitet. because it REFL at quit evenings at.the best works - Semantically, it is in virtue of some property of the DP within the locative/temporal PP that the generalization expressed by the IMs in (29) hold: (29') a. In virtue of the quietness of the streets, one can drive well there. b. In virtue of some inherent property of the chair, one can sit well on it. c. In virtue of the quietness of the afternoons, one can work best during such afternoons. - That the DPs within the PPs in (29) can have a similar function as the subject in personal middles is supported by Dutch adjunct middles, where those DPs function as subjects: (30) a. Deze stillen wegen rijden lekker. (Broekhuis and Corver, in prep: 489) These quiet roads drive nicely b. Deze stoel zit gemakkelijk. (Broekhuis and Corver, in prep: 462) This chair sits easily c. Rustige middagen werken het prettigst. (Broekhuis and Corver, in prep: 489)
Quiet afternoons work the most.pleasant - This explains the apparent obligatoriness of the adjunct PP in IM: it provides the property in virtue of which the generalization holds. In formal terms: it determines the conversational background, thus restricting the generic operator. o Question: if the generic operator is subject oriented, how can an adjunct PP have the same effect as a subject or: How can an adjunct PP be associated with the subject position? 2
In fact, temporal or instrumental PPs are possible, as well. I restrict the discussion to locative modifiers.
7
Marcel Pitteroff ‐ The Expletive Puzzle -
I follow Ackema and Schoorlemmer 1994, and Bennis and Wehrmann 1987 in assuming that PPs are impossible as grammatical subjects. Core Hypothesis The expletive in impersonal middles is required for syntactic reasons. I.e. it links the adjunct PP to an argument position. The adjunct PP itself is required for semantic reasons, as it functions as the subject of the dispositional predication. The expletive thus establishes this relation syntactically (see also Hoekstra and Roberts 1993, Lekakou 2005, and Broekhuis and Corver in prep for suggestions in this direction).  es values [u:__] on T  es is associated with the adjunct PP (see below)  the PP modifier ends up being assigned the dispositional property indirectly via its relation to es  Is there any evidence for the connection between the expletive and the PP? 1) the presence of the PP interacts with the nature of es: if the PP is absent, es has to be a referential pronoun, where possible, otherwise ungrammaticality obtains. Consider (30) in contrast to (29): (30) a. weil es sich gut tanzt. (es = das Sommertänzchen) because it REFL well dances (it = the summer dance) b. *weil es sich gut sitzt. because it REFL well sits c. weil es sich angenehm singt. (es = das Lied) because it REFL pleasantly sings (it = the song) - Clearly, (30) are all instances of personal middles, based on the predicates in (31). The fact that sitzen lacks a transitive counterpart accounts for the ungrammaticality of (30b). (31) a. Mark tanzt es/das Sommertänzchen gerne. ‘Mark likes to dance it/the summer dance.‘ b. *Mark sitzt es/das Bänkchen gerne. ‘Mark likes to sit on it/the bench.’ c. Mark singt es/das Lied gerne. Mark likes to sing it/the song.’  Quasi‐argumental es and the locative PP are interdependent, suggesting that the proposed relation between the two elements exists. 2) The "linking" function of argumental es, e.g. associating adjuncts with argument positions is seen elsewhere in the Grammar.  es as CP‐placeholder (see example XXX and discussion) 8
CGSW 28 
University of Leipzig October 4‐5, 2013 es as subject. Consider the following example involving the psych‐verb please (see Bennis and Wehrmann 1987 for similar constructions in Dutch): (32) weil es mir in meinem Bett gut gefällt. because it me in my bed good pleases 'it pleases me in my bed.' - The sentence in (32) shows the same properties as IM, e.g. the expletive is obligatory (33a); it has quasi‐argumental status; the PP is obligatory unless the expletive is interpreted as a referential pronoun (33b). (33) a. weil (*es) mir in meinem Bett gut gefällt. b. weil es mir gut gefällt. (es = e.g. das Bett) - It is thus plausible to assume that the locative PP in (32) is semantically the subject, and that this relation is syntactically expressed via argumental es.  German es can have the function of "linking adjuncts to argument positions", supporting the view that the expletive in IM has the same function (see also Bennis and Wehrmann 1987, Ruys 2010 who show that Dutch het can have the same function). ‐ Formally, the link between the expletive and its associate has been expressed via co‐
indexing (Bennis and Wehrmann 1987, Hoekstra and Roberts 1990, Broekhuis and Corver in prep) or co‐superscripting (Ruys 2010). The syntax of IM could thus look as follows (adopting Schäfer's 2008 analysis of personal middles as involving expletive Voice): (34) [TP esi [P,N,G] [T' T [uP, uN, uG] [Gen [VoiceP REFL [vP PPi [vP adverb [vP [VP V] v]]] Voice]  The expletive values the phi‐features on T, thereby satisfying the formal requirement for a subject.  Potential problem: the process of coindexing/co‐superscription seems incompatible with certain minimalist assumptions (such as Inclusiveness). Could there be an alternative? 3.4.1 A small clause/Relator Phrase analysis - It could be assumed that the expletive in IM functions as the subject of a small clause or Relator Phrase (see Den Dikken 2006) in the complement position of V, essentially assimilating impersonal middles to secondary predication constructions such as John ran himself to exhaustion, etc. (that unergatives can combine with such a small clause whose predicate is a locative PP has been argued independently in Alexiadou XXXX): 3 (37) [VoiceP sich [vP [VP [VP [SC es dort] tanzt] gut ]v] VoiceEXPL] 3
I remain agnostic as to the precise structure of the predication‐relation involved, e.g. whether it is to be analyzed as a SC, a Relator Phrase, or a PredP (Bowers 1993). In the text, I will make use of SC, but I am not dedicated to this particular implementation. 9
Marcel Pitteroff ‐ The Expletive Puzzle 
The predication relation between the expletive and the PP would replace the co‐
indexation mechanism (37) would also account for some further observations: All of the native speakers I consulted judged the following middles from Abraham (1987) outright ungrammatical, which is expected under (37): the small clause and an internal argument compete for the same position. (38) a. *weil es sich diese Bücher schön liest. because it REFL these.ACC books nicely reads 'because these books can be read nicely.' b. *weil es sich den Löwen nicht so leicht aus der Ruhe bringt. because it REFL the lion not so easily out the calm brings 'because the lion cannot be unsettled that easily.'  In a (very) small questionnaire study I carried out, impersonal middles involving lexically case‐marked datives were judged as degraded by almost all speakers (see also Fagan 1992 for the same finding). This can be clearly seen by comparing them to the corresponding let‐middles, which are perfect. (39) a. ??weil es sich einem Obdachlosen leicht hilft. because it REFL a.DAT homeless.person easily helps b. weil (*es) sich einem Obdachlosen leicht helfen lässt. because it REFL a.DAT homeless.p. easily help lets 'because it is easy to help a homeless person.' (40) a. *weil es sich diesem Argument nicht leicht widerspricht. because it REFL this argument not easily counterargues b. weil (*es) sich diesem Argument nicht leicht widersprechen lässt. because it REFL this argument not easily counterargue lets 'because it is not easy to counterargue this argument.' ‐ Note furthermore that in the let‐middle in (39b, 40b), the expletive is unacceptable, unlike in let‐middles with an adjunct PP, where the expletive is optional. Again, this could be a blocking effect (reminiscent of Keyser and Roeper's 1992 Abstract Clitic Hypothesis). Problems:  Marcel den Dikken points out that it is not clear why under a small clause analysis, a locative interpretation of the PP is forced, while a directional one is blocked. Consider the following observation from Broekhuis and Corver (in prep): (41) a. Jan heeft op de trampoline gesprongen. Jan has on the trampoline jumped 'Jan has jumped on the trampoline.' b. Jan is op de trampoline gesprongen. Jan is onto the trampoline jumped 'Jan has jumped onto the trampoline.' 10
CGSW 28 University of Leipzig October 4‐5, 2013 (42) a. Het springt lekker op de trampoline. it jumps nicely on the trampoline Available reading: 'It is nice to jump on the trampoline.' Impossible reading: 'It is nice to jump onto the trampoline.' - Typically, unergatives combining with small clause complements result in a resultative interpretation (see Hoekstra and Mulder 1990).  An even more severe problem is that this analysis predicts that impersonal middles are incompatible with an additional resultative (Hubert Haider, p.c.). This, however, does not seem to be the case: (43) a. In diesem neuen Stuhl arbeitet es sich leicht in einen Rausch. in this new chair works it REFL easily in a flush 'In this new chair, one can easily work oneself in a flush b. Von dieser Klippe springt es sich viel leichter ins Meer. From this cliff jumps it REFL much easier in.the sea 'From this cliff, one can jump into the sea much easier.' 3.4.2 The PP as stage topic - A different alternative could be based on the assumption that a direct connection between expletive and PP is only apparent. ‐ E.g. the PP functions as a stage topic in the sense of Erteshik‐Shir 1997, 1999 Cohen and Erteshik‐Shir XXXX, Basillico 2003, Lahousse 2007: (44) Stage Topic (Erteshik‐Shir 1999: 124) A stage topic defines the spatio‐temporal parameters of the utterance. Stage topics may be overt (...) or discoursally implied (the here‐and‐now). - In this sense, stage topics are not what the sentence is about (e.g. aboutness topics in the sense of Reinhardt 2001), but restrict the main predication to a certain domain. ‐ Stage topics thus restrict the generalization expressed by impersonal middles to certain times and places.  As not all PPs in IMs need to be locative or temporal, this explanation would fall short of accounting for the whole set of data.  One could imagine that it is not the subject that underlies dispositional predication, but a topic. Since topics are mapped into the restrictor (Krifka et.al 1995) and the restrictor under a modal approach to genericity determines the accessibility relation (REFERENCE), their function would be identical to the one of subjects described above.  Problem: the PP does not have to be the topic and the sentence is still grammatical 11
Marcel Pitteroff ‐ The Expletive Puzzle (45) Q: Wo kann man gut tanzen? Where can one well dance? 'Where can one go dancing?' A: In dieser Diskothek tanzt es sich gut. In this disco dances it REFL easily  In (45), the locative PP is focus, and thus mapped into the nuclear scope. It cannot restrict the set of possible worlds, and (45) should either be ruled out, or mean something different. This is not the case.  I conclude that both a small clause, as well as a (stage) topic analysis come with certain problems. As it seems that the co‐indexation/superscription mechanism is needed independently, I will make use of it for the rest of this presentation, pending a treatment of this mechanism which is more in line with minimalist assumptions. 3.5 Summary o The expletive in IM is quasi‐argumental, values [u:__ ] on T, and is associated with the adjunct PP, thereby syntactically establishing the relation between the adjunct and the disposition ascription. o The expletive in IM is required as a consequence of the dispositional semantics of middles. 4. Extension to let‐middles 4.1 The syntax of let‐middles - In Pitteroff and Alexiadou (2012), it was argued that let‐middles have a more complex syntax than canonical middles: they involve a reflexively marked anticausative (sich lassen) embedding a passive VoiceP. 4 Additionally, let‐middles qualify as restructuring constructions, and are thus monoclausal. (43) a. Das Buch lässt sich gut lesen. The book lets REFL well read b. [VoiceP sich [vP [VoiceP [vP gut [vP [VP das Buch lesen] v]] VoicePASSIVE] lässt] VoiceEXPL]  In personal let‐middles, the embedded theme can neither be licensed by embedded v (passive Voice), nor by matrix v (expletive Voice). It thus values [u:__] on T, receives nominative case, and functions as the target for the disposition ascription. 4.2 ILM and expletive es  First, even though the expletive in ILM is optional, it is argumental in nature (e.g. an it‐
type expletive), as shown by the fact that it is not restricted to sentence‐initial position: 4
The embedded passive VoiceP ensures the ungrammaticality of let‐middles like (i): (i) Er lässt sich ihr helfen He lets RELF her.DAT help
12
CGSW 28 University of Leipzig October 4‐5, 2013 (46) a. weil (es) sich dort gut tanzen lässt. because it REFL there well dance lets ‘because one can dance well there.’ b. Auf dieser Matratze lässt (es) sich angenehm schlafen. on this mattress lets it REFL comfortably sleep ‘One sleeps comfortably on this mattress.’  Furthermore, let‐middles appear to be subject‐oriented: (47) a. Der Stein lässt sich leicht hochheben, weil er nicht schwer ist. The stone lets REFL easily up.lift because it not heavy is b.*Der Stein lässt sich leicht hochheben, weil Menschen so stark sind. The stone lets REFL easily up.lift because humans so strong are (48) a. Die Decke lässt sich leicht berühren, weil sie nicht sehr hoch ist. The ceiling lets REFL easily touch because it not very high is b.?Die Decke lässt sich leicht berühren, da ich groß bin. The ceiling lets REFL easily touch because I big am - This is confirmed by the observation that a locative PP is not easily added to a personal (let)‐middle. Clearly, the properties in virtue of which reading the book is easy depend on properties of the subject, and not some subject‐external factor: 5 (49) ??Dieses Buch lässt sich in einer Bücherei gut lesen. This book lets REFL in a library well read  In ILM, the addition of a PP‐modifier is obligatory, as shown in (50): (50) a. weil (es) sich *(dort) gut tanzen lässt. (es ≠ referential pronoun) because it REFL there well dance lets b. weil (es) sich *(auf der Terrasse) angenehm sitzen lässt. because it REFL on the terrace comfortably sit lets  Due to these similarities to IM: In the contexts where the expletive in let‐middles is present, the same analysis as in IM could be applied: the expletive functions as the formal subject and is co‐indexed with the PP adjunct. The expletive mediates the disposition ascription to the modifier PP. - The fact that modification is obligatory in ILM is particularly noteworthy, as personal let‐
middles (unlike their canonical counterparts) do not need any kind of modification: (51) weil sich dieses Fenster einschlagen lässt. because REFL this window break.in lets 5
In that regard, (let‐)middles behave like individual‐level‐predicates (see e.g. Kratzer 1995, Cinque 1995, Diesing 1992 for discussion). Despite these parallelisms, I cannot adopt a Kratzer‐Diesing style analysis of ILP for middles (as, e.g. involving a subject which is base‐generated externally). This is because of the evidence provided in Schäfer (2008) and Pitteroff and Schäfer (to appear) that the structural subject in (German) personal middles behaves like an internal argument. 13
Marcel Pitteroff ‐ The Expletive Puzzle 
In the face of (51), the question arises why while the modification is obligatory, the expletive is not. Note that the structure in (43b) allows for a second way to introduce the locative PP (Pos 2 in (52)). (52) [VoiceP sich [vP Pos 2 [vP [VoiceP [vP [VP [SC Pos 1] V] v] VoicePASSIVE] lassen]] VoiceEXPL] - Let us assume that in let‐middles, no covert Gen is involved as in canonical middles, but that (sich)‐lassen, due to its lexical semantics (involving causation/permission) leads to a similar interpretation as VP‐level GEN (see McGonnell‐Ginet 1994, Zwart 1998, Tao 2011 for the claim that a causative component is involved in dispositionality). 6 - I argue that if the expletive is present, the PP is introduced as a modifier of the lower VP. - If the PP is merged high, it adjoins to the vP containing lassen. It thereby directly modifies the causative component and thus, the component which arguably leads to the 'in‐virtue‐of'‐reading in let‐middles. In other words, it specifies where/when/how the (implicit) causing event takes place. (53) a. In dieser Disko lässt es sich gut tanzen. = in virtue of some property of this disco, dancing there is fun. b. In dieser Disko lässt sich gut tanzen. = something in this disco causes dancing to be fun.  Is there any evidence for such a structural difference? i) Topicalization of embedded verb and locative PP is only possible in let‐middles including es. (53) a. weil (es) sich am See wohl gut feiern lassen muss,... because it REFL at.the lake apparently well celebrate let must b. Am See feiern muss *(es) sich wohl gut lassen,... at.the lake celebrate must it REFL apparently well let (...warum sonst sollten all die Jugendlichen dorthin wollen?) 'because apparently the lake must be a good place for celebration, why else should all the teenagers want to go there?' (54) a. weil (es) sich im Meer schon immer gut hat schwimmen lassen. because it REFL in.the sea already always well has swim let 'because the sea was always a good place for swimming.' b. Im Meer schwimmen hat *(es) sich schon immer gut lassen. in.the sea swim has it REFL already always well let 6
In the philosophical literature on dispositionality, for example, it is frequently assumed that dispositionals are best treated as conditionals (Hans Kamp, p.c.): (i) X is soluble in water  if X is put in water, then it dissolves. Interestingly, Lewis (1973) has also argued for a conditional analysis of causation, suggesting that an assimilation as suggested in the text might not be too far‐fetched.
14
CGSW 28 University of Leipzig October 4‐5, 2013  (53) and (54) show that the locative only forms a constituent with the embedded predicate if the expletive is present. In the absence of the expletive, no such constituent is built. 4.3 A missing piece  The above said leaves the question of how to explain the argument‐adjunct asymmetry, illustrated in (57) vs (58) (taken from Hoehle 1978, 63): (57) a. Einem so verbohrten Kerl lässt (*es) sich schwer helfen. a so stubborn guy lets it REFL with.difficulty help 'It is difficult to help such a stubborn guy.' b. Diesem Argument lässt (*es) sich nicht widersprechen. this argument lets it REFL not counterargue 'This argument cannot be contradicted.' c. Von den Folterungen lässt (*es) sich nicht einfach absehen. from the torture lets it REFL not easily ignore 'One cannot easily ignore the torturing.' (58) a. In diesem See lässt (es) sich sehr gut schwimmen.. in this lake lets it REFL very good swim 'One can swim very well in this lake.' b. Mit solchen Gedanken lässt (es) sich schlecht einschlafen. With such thoughts lets it REFL badly in.sleep 'It is difficult to fall asleep with such thoughts in mind.' c. Auf glattem Boden lässt (es) sich gut tanzen. on slick ground lets it REFL well dance 'It is easy to dance on a slick floor.' - In other words: why is the expletive unacceptable with lexical datives or PP complements? How is the dispositional property assigned in these cases?  The descriptive generalization seems to be the following: (59) ‘canonical’ middles: target of disposition ascription = subject let‐middles: target of disposition ascription = argument  This might be a further hint towards the idea that the semantics in let‐middles are construed slightly differently from how they are achieved in canonical middles. E.g. the behavior of lassen differs in certain ways from the one of GEN. - Maybe let‐middles require the target of the disposition ascription to somehow associate with CAUS. There are two possibilities of how this can be achieved: 1) via direct modification (comp. above) 2) indirectly via the effects of restructuring 15
Marcel Pitteroff ‐ The Expletive Puzzle -
-
Under a complex predicate approach, 2) makes immediate sense: oblique arguments would not be arguments of the embedded predicate alone, but of the lassen(CAUS)‐
Infinitive complex. Yet, I rejected the complex predicate approach. Can the matrix relatedness be captured in a XP‐complementation approach to restructuring?  Bobaljik and Wurmbrand (2005, 2007) argued, based on antireconstruction effects, that in restructuring contexts, movement of the embedded theme into the root clause has to take place: (60) a. weil Peter alle Fenster zu schließen vergessen hat. because Peter all windows to close forget has Reading 1:  < NEG (forget); for all windows, Peter forgot to close them. Reading 2: NEG (forget) < ; Peter forgot to close all windows. b. weil alle Fenster zu schließen vergessen wurden. ( < NEG; *NEG < ) because all.pl.Nom windows to close forget was.pl c. weil alle Fenster zu schließen vergessen wurde. ( < NEG; NEG < ) because all.pl windows to close forget was.sg - (60b) is a restructuring construction, and disallows narrow scope of the theme argument, suggesting that movement into the matrix clause took place, from where the theme outscopes NEG (which is part of the lexical semantics of vergessen ‘to forget’). Reconstruction is blocked as a consequence of the following generalization: (61) Agreement‐Scope‐Correlation (Bobaljik and Wurmbrand 2005, 812) A DP may not be interpreted (for scope and binding) in a position lower than in the domain in which it undergoes Case/agreement‐checking. - Yet, there is no evidence that such movement takes place in the case of datives: (62) a. weil Peter allen Schülern zu helfen vergessen hat. ( < NEG; NEG < ) because Peter all.DAT students to help forget has b. weil allen Schülern zu helfen vergessen wurde. ( < NEG; NEG < ) because all students to help forget was  A new puzzle: There is a clear difference wrt to the acceptability of oblique arguments in middles ‘canonical’ middles: oblique arguments are degraded or even judged ungrammatical let‐middles: oblique arguments are perfectly acceptable - In impersonal let‐middles involving oblique arguments, the dispositional property is clearly attributed to the oblique argument itself. - This cannot even be captured under an approach which analyzes the structural subject in personal let‐middles as an argument of lassen (e.g. à la Kratzer‐Diesing‐style analyses of Individual‐level‐predicates; or Zwart’s (1998) analysis of Dutch middles), as the oblique argument is clearly an argument of the embedded predicate. 16
CGSW 28 University of Leipzig October 4‐5, 2013 - This gives rise to the following question: (63) How can an argument of a verb X be interpreted as if it was an argument of a c‐
commanding verb Y? 5. Conclusion - I argued that  impersonal canonical middles require es, because the dispositional property needs to be assigned. As this cannot be done to the (obligatory) locative adjunct PP, the argumental expletive es is inserted, taking the PP as associate and transmitting the disposition ascription.  let‐middles render the expletive optional because of their structural complexity, providing two positions for the modifier PP to adjoin to.  As a consequence, impersonal passives in German are not expected to involve expletive es: since German lacks an (obligatory) EPP‐feature on T (see e.g. Haider 1993, 2010, Wurmbrand 2006, a.o.; see also the Appendix), SpecTP needs not be occupied. Furthermore, impersonal passives lack the semantic requirement of middles that a dispositional property needs to be assigned, and thus, allow the subject position to remain empty. References Abraham, W. 1987. Was hat sich in 'Damit hat sich's'? In Passive im Deutschen, Centre de recherche en linguistique germanique Nice (ed), 51‐71. Tübingen: Niemeyer. Ackema, P., and Neeleman, A. 1998. Conflict Resolution in Passive Formation. Lingua 104: 13 –29. Ackema, P., and Schoorlemmer, M. 1994. The middle construction and the syntax‐semantics interface. Lingua 93: 59‐90. Ackema, P., and Schoorlemmer, M. 1995. Middles and nonmovement. LI 26:173‐197. Ackema, P., and Schoorlemmer, M. 2005. Middles. In The Blackwell Companion to Syntax, Vol. III, Martin Everaert and Hank van Riemsdijk (eds.), 131‐203. Oxford: Blackwell. Bayer, J., and Kornfilt, J. 1990. Restructuring Effects in German. In: Parametric Variation in Germanic and Romance: Proceedings for a DYANA Workshop. E. Engdahl, M. Reape, M. Mellor, and R.P.Cooper (eds). Edinburgh Working Papers in Cognitive Science, 21‐42. Bech, G. 1955. Studien über das deutsche Verbum infinitum. Linguistische Arbeiten 139. Niemeyer, Tübingen. Bennis, H. 1986. Gaps and Dummies. Dordrecht:Foris. Bennis, H., and Wehrmann, P. 1987. Adverbial Arguments. In Linguistics in the Netherlands F. Beukema & P. Coopmans (eds), 1‐11. Dordrecht: Foris Publications. Biberauer, T. 2003. Reconsidering the EPP and SpecTP in Germanic. Cambridge Occasional Papers in Linguistics (COPiL) 1:15 ‐ 40. Biberauer, T., and Richards, M. 2006. True optionality: When the Grammar doesn't mind. In: Boeckx, C. (ed.). Minimalist Essays. Amsterdam: John Benjamins, 35 ‐ 67. Bobaljik, J.D., and Wurmbrand, S. 2005. The domain of Agreement. Natural Language and Linguistic Theory 23(4): 809‐865. Bobaljik, J.D., and Wurmbrand, S. 2007. Complex predicates, aspect, and antireconstruction. Journal of East Asian Linguistics 16: 27‐42. 17
Marcel Pitteroff ‐ The Expletive Puzzle Bowers, J. 1993. The syntax of predication. Linguistic Inquiry 24: 591–656. Brennan, V. 1993. Root and Epistemic Modal Auxiliary Verbs. PhD dissertation, UMass. Broekhuis, H., and Corver, N. in prep. Syntax of Dutch. Verbs and Verb phrases, vol. 1 (chapter 3). Amsterdam: Amsterdam University Press. Burzio, L. 1991. The morphological basis of anaphora. Journal of Linguistics 27: 81‐105. Cabredo Hofherr, P. 1999. Two German impersonal passives and expletive pro. Catalan Working Papers in Linguistics 7: 47‐57. Cardinaletti, A. 1990. Impersonal Constructions and Sentential Arguments in German. Padova: Unipress. Carlson, G. 1989. On the semantic composition of English generic sentences. In Properties, Types and Meaning II, ed. Gennaro Chierchia, Barbara Partee, and Raymond Turner, Dordrecht: Kluwer, 167–192. Chierchia, G. 1995. Individual level predicates as Inherent Generics. In The Generic Book, ed. Gregory Carlson and Francis Jeffry Pelletier. Chicago: Chicago University Press, 176‐223. Condoravdi, Cleo. 1989. The middle: where semantics and morphology meet. MIT Working Papers in Linguistics 11:18–30. Dikken, M. den. 2006. Relators and Linkers. The Syntax of Predication, Predicate Inversion, and Copulas. Linguistic Inquiry Monograph 47. Cambridge: MIT Press. Diesing, M. 1992. Indefinites. Cambridge, MA: MIT Press. Dobrovie‐Sorin, C. 1998. Impersonal se‐constructions in Romance and the Passivization of Unergatives. LI 29: 399‐439. Dowty, D. 2001. The semantic asymmetry of `argument alternations' (and why it matters). In: Making Sense: From Lexeme to Discourse. (Groninger Arbeiten zur germanistischen Linguistik, 44). Geart van der Meer and Alice G. B. ter Meulen (eds), Center for Language and Cognition Groningen, Groningen. Fagan, S.M.B. 1992. The syntax and semantics of middle constructions: A study with special reference toGerman. Cambridge:CUP. Grewendorf, G. 1989. Ergativity in German. Dordrecht: Foris. Grewendorf, G. 1990. Small pro in German. In Scrambling and barriers, G.Grewendorf & W. Sternefeld (eds), 295‐315. Amsterdam. Haider, H. 1985. Von sein oder nicht sein: Zur Grammatik des Pronomens 'sich'. In Erklärende Syntax des Deutschen, W.Abraham (ed), 223‐254. Tübingen: Narr. Haider, H. 1987. Nicht‐sententiale Infinitive. Groninger Arbeiten zur Germanistischen Linguistik 28:73‐
114. Haider, H. 1993. Deutsche Syntax ‐ Generativ. Tübingen: Gunter Narr Verlag. Haider, H. 2003. V‐clustering and clause union: Causes and effects. In Verb constructions in German and Dutch (Current issues in Linguistic Theory 242), ed. P.Seuren & G. Kempen, 91‐126. Amsterdam, John Benjamins. Haider, H. 2010. The Syntax of German. Cambridge: Cambridge University Press. Haider, H. & Rindler‐Schjerve, R. 1987. The parameter of auxiliary selection: Italian German contrasts. Linguistics 25: 1029–1055. Höhle, T. N. 1978. Lexikalistische Syntax. Die Aktiv‐Passiv‐Relation und andere Infinitiv‐konstruktionen im Deutschen. Tübingen: Niemeyer. Hoekstra, T., and Mulder, R. 1990. Unergatives as Copular Verbs: Locational and Existential Predication. The Linguistic Review 7:1, 1‐79. Hoekstra, T, and Roberts, I. 1993. Middle constructions in Dutch and English. In Knowledge and Language II. Lexical and Conceptual Structure, ed. Eric Reuland and Werner Abraham, 183–220. Dordrecht: Kluwer. Huber, W. 1980. Infinitivkomplemente im Deutschen – Transformationsgrammatische Untersuchungen zum Verb lassen. Ph.D. dissertation, Freie Universität Berlin. Keyser, S.J., and Roeper, T. 1992. Re: The Abstract Clitic Hypothesis. Linguistic Inquiry 23:1. 89‐125. 18
CGSW 28 University of Leipzig October 4‐5, 2013 Kornfilt, J. 1994. On some unusual passives in Turkish and in German. In: Proceedings of the 1994 Annual Conference of the Canadian Linguistic Association. P. Koskinen (ed.). Toronto Working Papers in Linguistics, 289‐300. Kratzer, A. 1995. Stage‐Level and Individual‐Level Predicates. In The Generic Book, ed. Gregory Carslon and Francis Jeffry Pelletier, 125–175. Chicago: Chicago University Press. Kunze, J. 1996. Plain middles and lassen middles in German: Reflexive constructions and sentence perspective. Linguistics 34: 645‐695. Lekakou, M. 2005. In the middle, somewhat elevated. The semantics of middles and its cross‐linguistic realization. Ph.D. dissertation, University of London. Lewis, D. 1973. Causation. Journal of Philosophy 70, 556–567. McConnell‐Ginet, S. 1994. On the Non‐Optionality of Certain Modifiers. In Proceedings of SALT 4 , ed. Mandy Harvey and Lynn Santelmann, 230–250. Ithaca: Cornell University. Pitteroff, M., and Alexiadou, A. 2012. The Properties of German sl‐middles. In Proceedings of WCCFL 29, 214‐222. Cascadilla Press. Pitteroff, M., and Schäfer, F. to appear. The argument structure of reflexively marked anticausatives and middles ‐ Evidence from datives. In Proceedings of NELS 43. Reis, M. 1976. Reflexivierung in deutschen A.c.I.‐Konstruktionen. Ein transformations‐grammatisches Dilemma. In: Papiere zur Linguistik 9: 5‐82. Richards, M., and Biberauer, T. 2005. Explaining Expl. In: den Dikken, M. & C. Tortora (eds.). The function of function words and functional categories. Amsterdam: Benjamins, 115 ‐ 154. Roberts, Ian. 1987. The Representation of Implicit and Dethematized Subjects. Linguistic Models Series. Dordrecht: Foris. Rosengren, I. 2002. EPP: A syntactic devise in the service of semantics. Studia Linguistica 56:2, 145‐
190. Ruys, E. 2010. Expletive Selection and CP arguments in Dutch. Journal of Comparative Germanic Linguistics 13, 141‐178. Schäfer, F. 2008. The syntax of (anti‐)causatives. External arguments in change‐of‐state contexts. Linguistics Today 126. Amsterdam: John Benjamins. Schäfer, F. 2012. The passive of reflexive verbs and its implications for theories of binding and case. Journal of Comparative Germanic Linguistics, 15.3, 213‐268. Schäfer, F. 2013. Passives of Reflexive Verbs: The Repair of a Principle A Violation. In Repairs: The added value of being wrong. P. Brandt & E. Fuss (eds.), Berlin: Mouton de Gruyter. Steinbach, M. 2002. Middle voice. Amsterdam: John Benjamins. Szatmári, P. 2004. Das heterogene ‘sich lassen’. Beiträge zur germanistischen Sprachwissenschaft 17. Hamburg: Buske. Tao, Y. 2011. Chinese Middle Constructions: A Case of Disposition Ascription. PhD dissertation, The Hong Kong Polytechnic University. Van Oosten, J. 1977. Subjects and agenthood in English. Chicago Linguistic Society 13:451–471. Vikner, S. 1995. Verb movement and Expletive Subjects in the Germanic Languages. Oxford: Oxford University Press. Wurmbrand, S. 2001. Infinitives: Restructuring and clause structure. Berlin: de Gruyter. Wurmbrand, S. 2004. Two types of restructuring: Lexical vs. functional. Lingua 114:8. 991‐1014. Wurmbrand, S. 2006. Licensing Case. Journal of Germanic Linguistics 18:3. 175‐234 Wurmbrand, S. 2007. How complex are complex predicates? Syntax 10:3. 243‐288. Zwart, J. W. 1998. Nonargument Middles in Dutch. Groninger Arbeiten zur germanistischen Linguistik 42: 109‐128. Groningen: University of Groningen. Appendix A Further approaches A.1 Satisfying the EPP 19
Marcel Pitteroff ‐ The Expletive Puzzle 
Under the assumption that the EPP is a universal principle, one way of analyzing the expletive in impersonal middles is as an element that is inserted to satisfy the EPP: (1) a. Es tanzt sich dort gut. It dances REFL there well 'One can dance well there.' b. [TP es T[EPP] [VP tanzt sich dort gut]. - This analysis, however, does not straightforwardly extend to impersonal passives, or expletiveless impersonal let‐middles (2). How is the EPP satisfied in these constructions? (2) a. Dort wurde getanzt. 'There was dancing.' b. Dort lässt sich gut tanzen. 'One can dance well there.' A.1.1. Semi‐pro‐drop (Grewendorf 1989, 1990; Cardinaletti 1990)  German has sometimes been treated as a semi‐pro‐drop language (Grewendorf 1989, 1990; Cardinaletti 1990; Vikner 1995): while referential pronouns must be overtly realized, expletive pronouns are dropped. SpecTP in (3a), as well as in contexts where DPnom stays in VP‐internal position, is filled by the expletive pronoun pro (which then, in fact is the true counterpart of English there). (3) a. Dort wurde getanzt. [CP dort C [TP pro T [VP wurde getanzt]]] b. weil dem Mann ein Buch geschenkt wurde. because the.DAT man a.NOM book given became 'because a book was given to the man.' [CP weil [TP pro T [VP dem Man ein Buch gegeben wurde]]] - such an account immediately raises the question of why pro cannot occur in impersonal middles, as the ungrammaticality of the following example without the expletive shows: (4) *In diesem Bett schläft sich gut. in this bed sleeps REFL well 'One can sleep well in this bed.' - The solution Grewendorf (1989) proposes is that in middle constructions, es is in fact a quasi‐argument base generated in SpecvP, moving to TP to satisfy the EPP. Evidence for this assumption comes from his claim that in contrast to personal middles, the external argument in impersonal middles is in fact syntactically active (see Cardinaletti 1990 for the same argument): (5) a. Hier arbeitet es sich schwer, ohne PRO gut zu verdienen. Here works it REFL difficult without PRO well to earn 'One works hard here without being paid well.' b. Hier diniert es sich hervorragend, ohne PRO allzu viel bezahlen zu 20
CGSW 28 
i)
ii)
University of Leipzig October 4‐5, 2013 Here dines it REFL excellently without PRO too much pay to müssen. have 'One dines here excellently without having to pay too much.' (Grewendorf 1989: 156) Problems: (5a) is supposedly not a middle. Note that is is paraphrased as "Hier wird schwer gearbeitet" (People work hard here), lacking the modal component characteristic of dispositionality. The adjunct clause is acceptable in personal middle clauses as well, in which no external argument is projected: (6) a. Ihr neuestes Buch liest sich gut auch ohne PRO die anderen Romane her newest book reads REFL well also without PRO the other novels zu kennen. to know 'Her latest book reads well even if one does not know the other novels.' b. Auch ohne PRO ihn groß zu bewerben verkauft sich der Film gut. Also without PRO him big to promote sells REFL the movie well 'This movie sells well even without promoting it too much.' - Furthermore, the adjunct clause is licensed in contexts where there is no potential controller in the matrix clause, suggesting a case of non‐obligatory control: (7) a. Das U‐Boot wurde zerstört PRO ohne Japan vorzuwarnen. The submarine was destroyed PRO without Japan to warn 'The submarine got destroyed without warning Japan in advance.' b. Ohne PRO stark auf die Bremse drücken zu müssen kommt dieses Auto without PRO hard on the brake press to must comes this car schnell zum stehen. fast to.the stand 'Without having to push too hard on the break pedal, the car comes to a halt very quickly.' - In (7a), whoever destroyed the submarine must not be the one to warn Japan. And in (7b), the adjunct clause is added to a generic unaccusative, which does not involve a potential controller at all. This suggests that ohne‐zu‐clauses cannot be taken as a test for the syntactic presence of controller.  Cardinaletti (1990) provides further evidence for the claim that the external argument in impersonal middles is syntactically active: (8) a. Hier tanzt es sich gut miteinander. Here dances it REFL well together. 21
Marcel Pitteroff ‐ The Expletive Puzzle b. Hier redet es sich gut miteinander. Here talks it REFL well together - Cardinaletti's data are problematic since if it was the quasi‐argumental it that licenses the anaphor, one would expect that it does so in all cases, contrary to fact: (9) a. *An diesem Tisch sitzt es sich gut miteinander. At this table sits it REFL well with each other b. *Über andere lacht es sich gut miteinander. over others laughs it REFL well with each other iii)
In general, Grewendorf's and Cardinaletti's approach seem to suffer from the same defect: why should impersonal middles differ from personal ones in the proclaimed way? The only reason seems to be to save the EPP, resulting in some kind of circular logic: "personal middles lack an external argument because they contain a different element that can satisfy the EPP; impersonal middles project a (quasi‐argumental) external argument because they need to satisfy the EPP." iv)
Neither Grewendorf’s analysis nor Cardinalleti’s can explain why in impersonal let‐
middles, the expletive is optional. - The most straightforward extension to let‐middles would be to assume that in the absence of the quasi‐argument, pro is involved just as in passives: (10) a. [CP dort lässt [TP es T [sich gut tanzen]]] there lets it REFL well dance b. [CP dort lässt [TP pro T [sich gut tanzen]]] there lets pro/there REFL well dance - It is completely unclear why these two possibilities should exist: 1) with let‐middles, but not with canonical middles 2) at all; consider the Avoid Pronoun Principle, which both Grewendorf and Cardinaletti need ot assume in order to explain why pro in impersonal passives need to remain covert: (11) Avoid Pronoun Principle Lexical pronouns are blocked by empty pronouns where possible. - According to (11), due to the fact that (10b) is possible, (10a) should not arise. v)
Finally, the evidence for the existence of pro in German is flimsy at best. Haider (1993, 2010), Cabredo Hofherr (1999), as well as Biberauer (2003) have argued convincingly against German as a semi‐pro‐drop language. A.2 Cognitive Object approach (Cabredo Hofherr 1999) 22
CGSW 28 
University of Leipzig October 4‐5, 2013 Cabredo Hofherr, following an analysis by Dobrovie‐Sorin (1998) for Romanian, develops an analysis of the distribution of expletives in impersonal middles and passives, based on the assumption that impersonal passives are in fact personal passives, involving a null‐
cognate object, which is assigned nominative case in‐situ. (12) a. weil gelacht wurde. b. [TP T [VoiceP wurde [VP eCOGN gelacht]]]] - Since the null object needs case, merging the lexical expletive in subject position is excluded for case reasons. - In middle constructions, Cabredo Hofherr argues that the surface subject is base‐
generated as an external argument in the personal variants. By extension, the cognate object in impersonal middles is merged into the external argument position as well: - German does not license null external subjects and uses es instead. (13) a. weil es sich leicht lacht. b. [TP T [VoiceP es [VP sich leicht lacht]]]  Problems i) This account does not extend easily to let‐middles, as it would require an analysis as the one in Kunze (1996): no expletive in impersonal SLM: construction involves a passive, with the cognate object in theme‐position, therefore remaining covert expletive in impersonal SLM: construction involves an embedded middle, with the cognate object in external argument position, therefore necessarily being overt. ‐ We have seen above that there is no evidence for such a split‐analysis. ii) The argument that the structural subject in personal middles is base‐generated in SpecVoice, even though frequently proposed (see e.g. Ackema and Schoorlemmer 1994, 1995, 2005; Lekakou 2005; Steinbach 2002) has been called into question. - Schäfer (2008) shows that the tests applied to determine the argument structure of middle constructions are flawed, not showing what they are intended to show. - Pitteroff and Schäfer (to appear) provide a battery of tests which uniformly show that DPnom in middles (and let‐middles) patterns with internal, rather than external arguments, and in that regard fundamentally differ from thematic reflexive constructions as in (14). (14) Er fotografiert sich. 'He takes a picture of himself.' 23
Marcel Pitteroff ‐ The Expletive Puzzle -
If the structural subject in personal (let‐)middles is merged internally, it is rather unlikely that the cognate object in the impersonal variants is base‐generated externally. iii) Similarly, Cabredo Hofherr seems to predict impersonal passives based on predicates that combine with a lexically case marked DP as in (14a) or with a prepositional complement (14b) to be ungrammatical, as no covert object could possibly be contained in these examples. As (15) shows, this prediction is incorrect: (15) a. Dem Mann wurde geholfen. The.Dat man was helped 'The man was helped.' b. Auf Peter wurde vertraut. On Peter was trusted 'People took trust in Peter.' A.3 Binding  Haider (1985), Abraham (1987), and Steinbach (2002): the expletive in impersonal middles functions as the formal antecedent of the reflexive pronoun (=> need to prevent violation of Principle A of the Binding Theory) - Problems i) Cardinaletti (1990): since the nominative DP in German may stay in VP‐internal position (see e.g. Bobalijk and Wurmbrand 2005, Wurmbrand 2006), REFL in (16) does not have a c‐commanding antecendent, which, under Haider's analysis, should trigger insertion of an expletive, contrary to fact: (16) weil (*es) sich [VP trockenes Holz leicht entzündet] generic unaccusative because it REFL dry wood easily inflames ‘because dry wood catches fire easily.’ ii) Does not account for impersonal let‐middles in which the expletive is absent. Those should either be ungrammatical, or come without the reflexive pronoun, contrary to fact (see Kunze 1996 for the same point of criticism). (17) Unter dieser Brücke lässt (*sich) gut schlafen. Under this bridge lets REFL well sleep 'One can sleep well under this bridge.' iii) Furthermore, with passives of reflexive verbs (see Schäfer 2012, 2013) there exists a further construction in which there is no overt antecedent for the reflexive pronoun. Insertion of es in (18) is, as is typically the case in impersonal passives, ungrammatical: (18) a. Zuerst wird (*es) sich geküsst, at.first becomes it REFL kissed 'At first people kiss each other,…' 24
CGSW 28 University of Leipzig October 4‐5, 2013 b. und dann wird (*es) sich geschämt. and then becomes it REFL shamed 'and then people feel ashamed.'  Under the proposed account, the question arises as to why the reflexive pronoun in (18) and in impersonal let‐middles does not need an antecedent, while the one in impersonal canonical middles requires one obligatorily. A.4 Alternative structure approach (Kunze 1996)  In order to explain the distribution of expletives in let‐middles, Kunze argues that impersonal let‐middles are not structurally uniform. (19) Kunze's alternative structure approach to let‐middles The impersonal let‐middle of a verb V is the inherently reflexive (reflexive/ergative) form of lassen combined with the infinitive of a) the impersonal passive of V or b) the impersonal plain middle of V - Evidence for this split‐approach: impersonal let‐middles based on verbs that disallow impersonal passives (Kunze's data and judgments): (20) a. Ein Restaurant, in dem (es) sich recht gut sitzen lässt. a restaurant in which it REFL quite nicely seat lets b. *Ein Restaurant, in dem recht gut gesessen wurde. a restaurant in which quite nicely sat was (21) Ein Restaurant, in dem es sich recht gut sitzt. a restaurant in which it REFL quite nicely sits - While the impersonal passive is unacceptable (20b), the impersonal middle (21) is well‐
formed. Kunze thus argues that (20a) involves an embedded middle rather than a passive, which correlates with the rather obligatory character of the expletive. - Problems: i) Kunze does provide no explanation for why the expletive in IM is obligatory. ii) Verbal passives including the same manner adverb as the corresponding middle are always ill‐formed (22). (20b) improves if the manner adverb is left out (23): (22) a. Das Buch lässt sich gut lesen. the book lets REFL well read b. Das Buch wurde (*gut) gelesen. the book was well read (23) a. In dem Restaurant wurde gesessen (...getanzt und gelacht). 25
Marcel Pitteroff ‐ The Expletive Puzzle iii) iv) in the restaurant was sat ... danced and laughed b. Der Laden an sich ist sehr schön, auch draußen kann gesessen werden. the place at REFL is very nice also outside can sat be 'The place itself is very nice, one can even sit outside.' If the ILM in (20) involves an embedded IM, it is unclear why the expletive is optional, unlike in the corresponding IM (21). In general, Kunze needs to resort to conclusions like "a let‐middle based on verb X is more plain‐middle‐based" (679). This indeterminacy is difficult to make sense of. 26