Ilse Zimmermann

33. Jahrestagung der DGfS 22.-25.2.2011 Georg-August-Universität Göttingen AG 11

Phonological, morphosyntactic and semantic properties of es Handout

1.

Introduction

The contribution deals with the German multifunctional item es and its suppletive forms dessen, dem and das/da(r). Their phonological, morphosyntactic and semantic lexical properties will be considered. The focus of attention will be on the interrelation of the anaphoric and cataphoric functions of es. Sudhoff (2003) differentiates between 5 types of German es: (1) (2) (3) (4) (5)

Positional es: Es kamen viele Gäste. The quasi-argument es: Gestern hat es geregnet.. The pronoun es: Ich lese ein Buch. Es ist interessant/ein Roman. The pro-form es: Mein Nachbar spielt Klavier. Es stört mich nicht. The correlate es: Es stört mich nicht, dass mein Nachbar Klavier spielt.

I will take into account also the occurrence of es in cleft-sentences like (6) and as propredicative as in (7): (6) (7)

Das Telefon läutete. Es waren meine Eltern(, die anriefen). Peter ist feige/ein Feigling. Paul ist es auch.

The suppletive forms mentioned above, es has only as pronoun, pro-form or correlate (see (8)-(10)) and also as quasi-argument as in (2’): (8) (9) (10) (2’)

Auf dem Tisch lag ein Buch. Darunter befand sich das Testament. A: Paul ist faul. – B : Dem kann ich nicht widersprechen. Ich bin mir dessen nicht bewusst, gelogen zu haben. Das HAT gestern geregnet.

In view of these different usages of es and of its suppletive forms, the following questions will be raised: How many different lexical entries do exist for es? By which properties are they related? How are they differentiated semantically? Does information structure play any role in the distribution of es? Is there a difference between the pro-form and the correlate es? What does it mean to be a pro-sentential item, semantically? How can a propositional complement get nominal character? How are matrix-predicates subclassified wrt. the different types of es?

2.

Basic assumptions

-

A conception of minimalism The lexicon as interface of different levels Fully derived and inflected word forms in the lexicon Syntactic representations of clauses (Zimmermann 1999 ff.) and DPs:

(11) (12)

(ForceP) CP (GivenP*) MoodP TenseP PolP (vP*) VP ([DP )α [DP D (XP)] (YP])

Functional domains can be fused. ForceP is absent in embedded clauses. CP characterizes the various sentence types (BRRZ 1992, Zimmermann 2009, 2010). GivenP marks possible positions for given XPs and clitics (cf Schick/Zimmermann 1997). In PolP, the decision between affirmation and negation takes place. Sentence adverbs and modifiers are adjuncts. DPs can be reduced to their functional head D. XPs are predicate expressions. As for transformations, the less syntactic movements are to be assumed the better. Given XPs move to SpecGivenP. Pronouns are located before complex given XPs. Not all overt syntactic movements are visible to LF. Syntactic representations do not contain any logical operators. They are of purely syntactic nature. The semantic amalgamation of meaning parts operates on configurations of LF, a syntactic level. As regards semantic representations, I differentiate between grammatically determined Semantic Form and Conceptual Structure. In the semantic amalgamation of constituents, semantic templates (lambdaabstraction, identification, modification etc.) apply. A relevant equivalence in the semantic amalgamation is (13). (14) is an example, where the argument structure of a functor is applied to argument expressions of type e and t, respectively, or of the corresponding lifted types (cf. Chierchia 1995: 182f., 200). (13)

(14)

3.

λxn ... λx1 [ ...xn ... x1 ... ] (a) with xn, a ∈ {e, t} ≡ λΨ ... λx1[Ψ (λxn [ ...xn ... x1 ... ])] (λP [P a]) with Ψ ∈ {, } λxn ... λx1 [ ...xn ... x1 ... ] (a) with a = ||Paul|| or ||dass Paul schnarcht|| ≡ .... λx1 [ ... a ... x1 ... ] ≡ λΨ ... λx1[Ψ (λxn [ ...xn ... x1 ... ])] (λP [P a]) ≡ ... λx1[λP [P a] (λxn [ ...xn ... x1 ... ])] ≡ ... λx1[λxn [ ...xn ... x1 ... ] (a)] ≡ ... λx1[ ... a ... x1 ... ]

Lexical entries

In the sound-meaning correlation of constituents, the lexicon plays a central role. Every lexical entry contains the phonological characterization (except for zero

morphemes), the morphosyntactic categorization and the semantic form of the pertinent lexical item. 3.1

The positional es

(15)

/es/ -accent [DP__]; DP +EF; λX [X]

Phonologically, positional es is a DP and cannot be accented. The edge feature must be checked in SpecCP. 1 Semantically, positional es is empty (an identity function). 3.2

The quasi-argument es

(16)

/es/ -accent [DP__]; DP +ES αEF -αgov

Phonologically, quasi-argument es marked by the feature +ES behaves like positional es. As to case, it occurs in the unmarked nominative or in the governed accusative. Only nominative es can appear in SpecCP (see (2’’) vs. (17’)). Its semantics does not matter. 2 (2’’) Es hat gestern geregnet. (17) Paul hatte es eilig mit der Antwort. (17’) *Es hatte Paul eilig mit der Antwort. 3.3

The pronouns and pro-forms es, dessen, dem, das, da(r) („anaphoric“ items)

(18) (19) (20) (21) (22)

/es/ -accent [DP__]; +D+def+given αEF -αgov; (24) /das/; +D+def+given αgov ; (24) /dessen/; +D+def+given +obl (+masc); (24) /dem/; +D+def+given +obl+gov (+masc); (24) /da/ __P; +D+def+given -anim; (24)

Phonologically, es is a DP and cannot be accented. Syntactically, it cannot occur in SpecCP with the case feature +governed. All pronouns and pro-forms considered here are categorized as +D+def+given. 3 As „anaphoric“ items, they are coreferent with a present or understood „antecedent“. As to the φ-feature +masc in (20) and (21), it can be present whenever the argument x in (24) is of type e.

1

Declarative root clauses and V2-complements have the feature +EF in C. The argument structure of unpersonal verbs has one position marked with the feature +ES. But there is no place where to put the semantics of the correponding quasi-argument (see (i)):

2

(i) 3

λx ... λx1 [ ... x1 ... ] (||DP||) ≡ ... λx1 [ ... x1 ... ]

Of course, the feature +def is redundant, together with +given.

The suppletive form da needs a preposition as right phonological host and cannot refer to animate entities (cf. (8) and (23)). (8) (23)

Auf dem Tisch lag ein Buch. Darunter befand sich das Testament. Das Kind schlief fest. Neben ihm /*Daneben wachte der Hund.

As argument expressions, pronouns and pro-forms are generalized quantifiers: (24)

λP∃!x [x = y] ∧ [P x] ∈ with β ∈ {e, t, }

The free variable y in the restrictor relates to an item given in the cotext or contextually. 3.4

The related item of anaphoric pronouns and pro-forms

In the examples given above for anaphoric pronouns and pro-forms, their related phrase occurs to the left and has the respective semantic type of the identified arguments x and y of the pronouns and pro-forms, namely e, t, or . But the related phrase can also appear to the right of the pronoun or pro-form: (9’) (9’’)

A: Paul ist faul. – B : Dem, dass Paul faul ist, kann ich nicht widersprechen. A: Paul ist faul. – B : Dem kann ich nicht widersprechen, dass Paul faul ist.

While in (9’) the coreferent CP appears adjoined to the proform, in (9’’) it is an adjunct to GivenP. In both cases, it functions as facultative explication of the variable y in the SF (24) of the pro-form. Now the question is whether in sentences like (5) es can also be regarded as pro-form and how (5’) and (5’’) are to be analysed: . (5) (5’) (5’’)

Es stört mich nicht, dass mein Nachbar Klavier spielt. Mich stört es nicht, dass mein Nachbar Klavier spielt. Mich stört nicht, dass mein Nachbar Klavier spielt.

In constructions like (5’’), where es or its suppletive forms are absent, the sentential complement represents new information. In connection with es or its suppletive forms as in (5) and (5’), it can express given information if an explicit pro-form or a correlate is not mandatory (cf. Zimmermann 1993). Furthermore, with which parts of the antecedent is es coreferent in (25)-(27)? (25) (26) (27)

A: Ob viele Besucher kommen werden? – B: Ich bezweifle es. Wenn viele Besucher kämen, würde es mich freuen. Ihre Freundin muß operiert werden. Es wird drei Stunden dauern.

In general, anaphoric expressions need not fully correspond to their antecedent. Their explicitness varies from mere or partial repetitions of the related expression to reduced forms as in nominalizations, with nominal classifiers like diese Tatsache, dieser Umstand, dieses Problem etc. and with mere pronominal expressions. In this connection, it seems alarming to ask which semantic conditions determine

coreference. It seems important to take into account of which semantic type the antecedent and the argument y in (24) are in the respective cotext. While the pro-forms in the examples (4), (9), (9’), (10) and also in (25) and (26) relate to a unit of type t, this is not the case in (27), in the cotext of the matrixverb dauern. Here, the pronominal item es is not a pro-form, but a pronoun referring to an event of type e. 4 In cases like (5) and (5’), es refers to units of type t, irrespective of its status as pro-form or as correlate. Both the pro-form and the correlate relate to propositional units of the semantic type t. But while the pro-form es and its suppletive forms by themselves can constitute a DP, with a facultative adjoined CP (see (9’)), the correlate has the CP as obligatory explicative co-constituent. 3.5

The correlate es and its suppletive forms

I assume that the correlate and its propositional complement basically are coconstituents of a DP (for Russian cf. Zimmermann 1983). As in Zimmermann (1993), I propose the basic syntactic configuration (28): (28)

[DP [DP es/dessen/dem/das/da] CP]

In contrast to DPs like [DP [D das][NP Haus]], in (28) there is no NP-complement of D. The correlate cooccurs with an explicative CP adjoined to its right. Both constituents can be governed by predicate expressions or prepositions from the outside. This guarantees that the DP as argument expression gets case and the propositional complement can be selected for its sentence type, namely dass- or V2-clauses, oband/or w-, wie-clauses or infinite constructions with the formative zu (+2.status of the infinite verb form), which are all of type t. This implies that these clauses cannot constitute the lexical XP of the DP (see (12)). The matrix-predicates indicate these conditions of cooccurrence in their argument structure (see below and Zimmermann 1991, 1993). Furthermore, the proposed constellation does not allow any extractions out of CP (see below). The correlate expressions differ from their corresponding pro-forms in (18)(22) only in two respects: They have an obligatory propositional co-constituent and they are not necessarily +given. (29)

/(es/)γdessen/dem/das/da/ (-accent)γ [DP__]; +D+def δgiven αgov; λyλP∃!x [x = y] ∧ [P x] ∈

Like the pro-form es in (18), the correlate cannot be accented and is a DP, phonologically. This implies that its explicative CP must leave the DP by extraposition in PF. Unlike the pronoun, the accusative correlate es can occur in SpecCP, as in (30). (30)

[DP Es tCP ]i hörtenj alle Gäste ti tj, [CP dass mein Nachbar Klavier spielt].

But the semantic integration of CP takes place within the DP:

4

The same is true wrt. the pronominal items dessen and da in the cotext of temporal prepositions as in währenddessen, dabei, davor, danach, dazwischen.

(31)

... (DP λyλP∃!x [x = y] ∧ [P x] (CP λX [X] (MoodP∃s [[mein Nachbar Klavier spielt’] (s)]))) ... ≡ ... λP∃!x [x = ∃s [[mein Nachbar Klavier spielt’] (s)]] ∧ [P x] ...

(31) shows that the complementizer dass is semantically empty. 5 The identification of x and of the CP-interpretation is the basis of the semantic equivalence of CPs of type t embedded with or without the help of an correlate (cf. Zimmermann 1993). 6 It depends on the matrix-predicate whether the propositional argument can be embedded directly - as with empfehlen - or with the help of a DP or PP - as with widersprechen or abhängen von. Sudhoff (2003) assumes that the DP support is present for all matrixpredicates which belong to the bedauern-class (see below), irrespective of the visibility of the correlate. In structures like (5’’’), it is phonologically not realized. The same is true in the example (5’’), because the verb stören belongs to the bedauern-class. (5’’’) [DP [DP ∅] daß mein Nachbar Klavier spielt], stört mich nicht. Extraction phenomena as in (32) and (33) are decisive for Sudhoff’s analysis. (32) (33)

[CP Wann{i, j} empfahl dir der Hausarzt ti, [CP ti dass du tj einen Spezialisten aufsuchen sollst?]] [CP Wann{i,* j} stört [DP es/∅ tCP] dich ti, [CP tj dass dein Nachbar tj Klavier spielt?]]

Thus, besides their semantic function as cataphoric items, the correlates can be looked at as a means to supply embedded propositions with a nominal support and to create a barrier for extractions. Of course, the matter needs a separate study. According to the distribution of phonologically realized correlates and the zero-correlate, I assume that ∅ appears only if the corresponding case feature in the argument structure of the matrix-predicate (for example stören) can be neutralized and the DP is not given. Given DPs move to SpecGivenP and thereby take scope over MoodP. The semantic amalgamation results in an SF like (35) (see below). 5

For the development of the complementizer dass from the pronoun thaz, see Axel (2009). Her analysis amounts to the following stages (with ’=’ as implicit identity relation): (i) (ii) (iii)

[... thazi ... ][thazi = [ ... ]i] [[... thazi ... ][thazrel [ti = [ ... ]i]]] [... thazi = [thazc ...]i ... ]

6

Of course, wrt. the semantic amalgamation of a predicate expression with a sentential argument there is a difference depending on whether the embedding is direct or with the help of a DP. ||CP|| is of type t, while ||[DP CP]|| is a generalized quantifier of type , where ||CP|| appears in the restrictor of the correlate (cf.31). Consequently, one gets (i) vs. (ii) together with behaupten and bedauern, respectively: (i) (ii)

λxλs [s INST [ASSERT ||CP|| x]] λxλs∃!y [y = ||CP||] ∧ [s INST [REGRET y x]]

4.

Right dislocation vs. extraposition

According to Sudhoff (2003), clauses at the right edge of complex sentences are rightdislocated phrases base-generated as adjuncts of CP or transformationally extraposed complements of a correlate. These configurations differ wrt. information structure and intonation. Right-dislocated phrases represent given information or an after-thought and are separated by a relatively long pause. Extraposed clauses as in (5) leave their DP-shell for phonological reasons, because es in PF must be directly dominated by DP. Another reason for leaving the host-DP is the complexity of this argument expression. It is easier to compute clause structures one after the other. Therefore it is by far preferred to have the complementclause of the suppletive forms of es extraposed instead in situ (cf. (9) vs. (9’) and (34) vs. (34’)). (34) Viel wird davon abhängen, WER dirigiert. (34’) Viel wird davon, WER dirigiert, abhängen. As to the exact position of an extraposed phrase, there are very different proposals, which I cannot consider here. In any case, the position is less high in the syntactic structure than the place for right-dislocated phrases. Thus, the example (5) is ambiguous. Its interpretation depends on the status of es as correlate or as pro-form and, correspondingly, whether the embedded clause is extraposed or right-dislocated. Extraposed clauses are not visible in LF. But rightdislocated phrases must be semantically integrated. (35) illustrates this for (5), in case es is a pro-form (see (18)). (35)

... (GivenP (GivenP λpλy [p] (DP λP∃!x [x = y] ∧ [P x] (λpλx [p] (MoodP ∃s ... [s INST [DISTURB sp x]]))))(CP ∃s’ [[mein Nachbar Klavier spielt’] (s’)])) ... ≡ ... ∃!x [x = ∃s’ [[mein Nachbar Klavier spielt’] (s’)]] ∧ ∃s ... [s INST [DISTURB sp x]] ...

In addition to the meaning of the constituents, lamda-abstraction (36) is applied two times her (36)

TSLA: λpλz [p] ∈ of a theta-role λx includes morphosyntactic conditions on the pertinent argument expression, which concern cases of DPs and/or sentence types of embedded clauses. ’/’ separates alternative conditions. 9 The idiosyncratic feature +nc determines the neutralization of the case feature for a dependent DP of a correlate construction. 8

(38)

/weigern/; +V-N; λPλyλxλs [s INST [REFUSE [P x] x]] ∈

(39)

/wähnen/; +V-N; λPλyλxλs [s INST [WRONGLY BELIEVE [P y] x]]] ∈

(40)

/scheinen/; +V-N; λP()αλy()-α(λx)(λz)-αλs [s INST [BELIEVE [P y] x]] (TRUE)-α ∈ with β∈ {e, t}

The verb meinen belongs to the behaupten-class allowing embedded dass- or V2clauses or a pro-form of type t (cf. (18) or (19)) with the case feature +gov(erned). The verb stören is an instance of the bedauern-class and is compatible with a pro-form, the correlate together with a dass-clause, with anaphoric or cataphoric classifying DPs as in (41), with nominalizations as in (42) or with DPs not corresponding to clauses like der Nachbar. Furthermore, this matrix-verb can combine with infinitival clauses in the 2. status, instead of dass-clauses. These clauses have a PRO-subject (cf. (43)). (41) (42) (43)

Den Professor stört es/das/die(se) Tatsache, dass er ständig unterbrochen wird. Den Professor stört die ständige Unterbrechung. Den Professor stört es, ständig unterbrochen zu werden.

Sich weigern in (38) is a reflexivum tantum with strong inherent control. 10 This implies that this verb does not embed propositional units, but a predicate expression in the 2.status. Consequently, it does not combine with pro-forms, correlate constructions or nominalizations. The same is true for the matrix-verb wähnen in (39) embedding a so-called small clause and with accusative case-marking of the DP in the specifier position of the small clause. Only in the semantic representation there is an embedded proposition, which does not have a syntactic correspondent. Like the copula, this verb has a predicate and its argument as separate theta-roles. And again, there is no possibility to combine verbs like wähnen with a pro-form or a correlate construction or a nominalization. In contrast to sich weigern, wähnen is not restricted to a verbal predicate expression, but can be used with a predicative DP, AP or PP. The most complicated matrix-predicate is the verb scheinen. It takes a predicate and its argument as theta-roles. Belonging to the behaupten-class, it allows a dass- or V2-complement. In addition, it takes the quasi-argument es and can have a dative experiencer. It is important that the dass-clause is associated with the predicate TRUE, lexically. As a result of this analysis of scheinen, there is no room for a nominalization or complex anaphoric DPs or pro-sentential items like the pro-form or the correlate es. There is no Was/das/es scheint (mir). In sentences like Mir scheint es/das nur s., so is the predicate expression and es or das are pro-form arguments of so specifying y in (40). It is an open question how many different matrix-predicate classes exist. With respect to the pro-form and the correlate es and its suppletive forms, the classes illustrated in (36)-(40) seem illuminating. In order to use the pro-sentential item es or its suppletive forms, the corresponding argument must be of type t. And in order to supply a sentential item of type t with nominal character it must be embedded in a DP. 10

On matrix-predicates with strong inherent control see Stiebels (2010: 403ff.).

6.

Complex anaphoric and cataphoric DPs

In this section, I would like to have a look on complex anaphoric and cataphoric DPs which also relate to clauses of type t. Among them, factivity expressions (see Zimmermann 2010) play a special role. Cf.: (44)

A: Der Professor ist schwerhörig. – B: Wir werden es/die(se) Tatsache berücksichtigen(, dass der Professor schwerhörig ist). (44’) A: Der Professor ist schwerhörig. – B: Wir werden die(se) Tatsache(, dass der Professor schwerhörig ist,) berücksichtigen.

Wir werden ∅DP/es/die Tatsache berücksichtigen, dass der Professor schwerhörig ist. (45’) Wir werden ∅DP/*es/die Tatsache, dass der Professor schwerhörig ist, berücksichtigen. (45)

(46) (47)

Wir werden die (Tatsache der ) Schwerhörigkeit des Professors berücksichtigen. Wir werden diese Tatsache, die Schwerhörigkeit des Professors , berücksichtigen.

The matrix verb berücksichtigen is a factive predicate. It belongs to the bedauernclass of Sudhoff (2003). As illustrated in (45) and (45’), this verb does not need an explicit nominal shell. Consequently, in these examples there must be a silent DP embedding the complement clause, as proposed by Kiparsky & Kiparsky (1970). I assume the following basic syntactic configurations of the complex anaphoric anaphoric and cataphoric DPs in (44)-(47): (48) (49)

([DP)α [DP [D’ die(se) [NP Tatsache]]] ([CP dass ...]/DP]])α [DP [D’ die [NP [NP Tatsache] [CP dass ...]/DP]]]

Whereas in configurations like (48) the embedded clause or a corresponding nominalization serves as facultative explicative apposition, (49) is an instance of a nominal head with a restrictive sentential or nominalized modifier. As adjunct to the NP Tatsache as in (49) the explicative DP appears in the genetive, whereas it agrees in case with the head DP as apposition as in (49) (cf.(46) and (47)). The relevant lexical entries for these factive constructions are given in (50)(52). (50) (51) (52)

/berücksichtigen/; +V-N; λy()λxλs [s INST [TAKE INTO ACCOUNT y x]] ∈ with α ∈ {e, t} /tatsache/; +N-V +fem; λx() [FACT x] ∈ with α∈ {t, e} /die(se)/; +D+def βgiven +fem; λP1λP2∃!x [P1 x] ∧ [P2 x] ∈ with α ∈ {t.e}

. The verb berücksichtigen demands a complement of type t or e in the accusative, which can be neutralized in the context of a declarative clause. The noun Tatsache, too, has an argument of type t or e and ,too, can neutralize its case feature.

Now it is important to note that the noun Tatsache by itself does not combine directly with a complement, but only mediated by the determiner. And in order to integrate the complement, one needs identification between two arguments. This is the case in constructions like (53) and (54), where the DP eine Tatsache is used predicatively, and like (44)-(47), where die(se) Tatsache is used anaphorically or cataphorically. (53) (54)

Daß der Professor schwerhörig ist,/Die Schwerhörigkeit des Professors ist eine Tatsache. Es ist eine Tatsache, dass der Professor schwerhörig ist.

While in (53) and (54) the identification is connected with the copula, in anaphoric or cataphoric DPs as in (44)-(47), the embedded clause or its nominalization are adjuncts and need the additional templates (55) and (56) for their semantic integration. (55) (56)

TSIDENT: λyλx [x = y] ∈ TSMOD: λQ2λQ1λx [Q1 x] ∧ [Q2 x] ∈

The type shift (55) delivers the identification of two arguments. (56) combines a modificator with its modificandum (see Zimmermann 1992). (58) exemplifies the application of these two templates for the complex cataphoric DP (57). (57) (58)

[DP die [NP [NP Tatsache] [CP dass der Professor schwerhörig ist]]] λP1λP2∃!x [P1 x] ∧ [P2 x] (λQ2λQ1λx [Q1 x] ∧ [Q2 x] (λyλx [x = y] (∃s ...∃!z [PROFESSOR z] ∧ [s INST [HARD OF HEARING z]]))(λx [FACT x])) ≡ λP2∃!x [[FACT x] ∧ [x = ∃s ...∃!z [PROFESSOR z] ∧ [s INST [HARD OF HEARING z]]]] ∧ [P2 x]

While here the restrictive clause [P1 x] of the determiner is specified as [FACT x] and the identification of x with the interpretation of CP is brought about by a template, in corresponding constructions with the correlate es or with its zero substitute one gets (59). (59)

λyλP∃!x [x = y] ∧ [P x] (∃s ...∃!z [PROFESSOR z] ∧ [s INST [HARD OF HEARING z]]) ≡ λP∃!x [x = ∃s ...∃!z [PROFESSOR z] ∧ [s INST [HARD OF HEARING z]]] ∧ [P x]

Evidently, (59) would be the simpler complement of the factive verb berücksichtigen, because with factive matrix-predicates the DP die Tatsache is redundant.

7.

Concluding remarks

This contribution stresses the role of lexical entries in the sound-meaning relation of linguistic expressions and in the division of labour between phonology, morphology, syntax and semantics.

As for the pro-form and the correlate es and its suppletive forms , I followed the distinctions of two matrix-predicate classes made by Sudhoff (2003) and tried to add semantic considerations. It was shown how the related CPs of the correlate and of the pro-form are integrated into the respective constructions, syntactically and semantically. Whereas pro-forms are always +given and move to the left periphery, correlates can be -given and move only for phonological reasons. Therefore, the associated CPs get different scope wrt. to MoodP, where the referential argument s of the matrix-construction is bound (cf. (60) and (61)). (60) (61)

... (MoodP ∃s.[ ... ∃!x [x = ∃s’ [ ... s’ ... ] ∧ [ ... s ... x ... ]]]) ... ... (MoodP ∃!x [x = ∃s’ [ ... s’ ... ] ∧ (MoodP ∃s [ ... s ... x ... ]])) ...

The structure in (61) results from two possible positions of the embedded clause, as right-dislocated adjunct of GivenP or as right adjunct of the definite DP (cf. (9’)). In both cases it is an facultative explicative apposition to the pro-form, which is +given. In (60), we have to do with a correlate construction, where the embedded clause is an obligatory constituent of DP. And the DP is not necessarily given. In cases where the correlate is +given, correlate constructions also result in configurations like (61). A comparison of es or its suppletive forms with complex anaphoric and cataphoric constructions wrt. their relation to sentences or their nominalizations revealed that in both cases an identification of two arguments is involved, though in different ways. In any case, this recourse to another unit is connected with the embedding via a DP. It is a kind of objectivization (in German „Vergegenständlichung“), which makes the embedding of a sentence less direct as is the case in connection with matrix-predicates of the behaupten-class. All these pecularities of es and of its suppletive forms do speakers of German know or acquire. Our linguistic theory must reflect this fact and offer the appropriate tools and restrictions of sound-meaning relation in the lexicon and in the division of labour between phonology, morphology, syntax and semantics.

References Axel, Katrin (2009). Die Entstehung des dass-Satzes – ein neues Szenario. In: Linguistische Berichte, Sonderheft 16, 21-41. Brandt, Margareta, Marga Reis, Inger Rosengren & Ilse Zimmermann (1992). Satzty.p, Satzmodus und Illokution. In: Rosengren, Inger (ed.)(1992/1993). Satz und Illokution (= Linguistische Arbeiten 278). Tübingen: Niemeyer, Band 1, 1-90. Chierchia, Gennaro (1995). Individual level predicates as inherent generics. In: Carlson, Gregory & Francis Jeffry Pelletier (eds.). The generic book. Chicago, London: University of Chicago Press, 176-223. Kiparsky, Paul & Carol Kiparsky (1970). Fact. In: Bierwisch, Manfred & Karl Erich Heidolph (eds.). Progress in Linguistics. The Hague, Paris: Mouton, 143-173. Pütz, Herbert (1986). Über die Syntax der Pronominalform es im modernen Deutsch. 2., durchgesehene Auflage (= Studien zur deutschen Grammatik 3).Tübingen: Gunter Narr. Schick, Ivanka P. & Ilse Zimmermann (1997). Das dativische pronominale Klitikum in der Substantivgruppe des Bulgarischen. In:. Junghanns, Uwe & Gerhild Zybatow (eds.). Formale Slavistik (= Leipziger Schriften zur Kultur-, Literatur-, Sprach- und Übersetzungswissenschaft 7). Frankfurt am Main: Vervuert Verlag, 49-61. Schwabe, Kerstin (2007). Old and new propositions. In: Späth, Andreas (ed.). Interfaces and interface conditions (= Language, Context, and Cognition 6). Berlin, New York: de Gruyter, 97-114.

Stiebels, Barbara (2010). Inhärente Kontrollprädikate im Deutschen. In: Linguistische Berichte 224, 391-440. Sudhoff, Stefan (2003). Argumentsätze und es-Korrelate. Zur syntaktischen Struktur von Nebensatzeinbettungen im Deutschen. Berlin: Wissenschaftlicher Verlag Berlin. Zimmermann, Ilse (1983). Untersuchungen zum Verhältnis von Substantivgruppe und Nebensatz. In: Růžička, Rudolf & Wolfgang Motsch (eds.). Untersuchungen zur Semantik (= Studia grammatica 22). Berlin: Akademie-Verlag, 201- 242. Zimmermann, Ilse (1991). Die subordinierende Konjunktion wie. In. Reis, Marga & Inger Rosengren (Hrsg.). Fragesätze und Fragen. Referate anlässlich der 12. Tagung der Deutschen Gesellschaft für Sprachwissenschaft, Saarbrücken 1990 (= Linguistische Arbeiten 257). Tübingen: Niemeyer, 113-122. Zimmermann, Ilse (1993). Zur Syntax und Semantik der Satzeinbettung. In: Rosengren, Inger (ed.)(1992/1993). Satz und Illokution (= Linguistische Arbeiten 278). Tübingen: Niemeyer, Band 2, 231-251. Zimmermann, Ilse (1999). Die Integration topikalischer DPs in die syntaktische und semantische Struktur von Sätzen. In: Doherty, Monika (ed.). Sprachspezifische Aspekte der Informationsverteilung (= Studia grammatica 47). Berlin: Akademie Verlag, 41-60. Zimmermann, Ilse (2007). German w-clauses at the left and right periphery of copular sentences. In: Späth, Andreas (ed.). Interfaces and interface conditions (= Language, Context, and Cognition 6). Berlin, New York: de Gruyter, 141-155. Zimmermann, Ilse (2009). Satzmodus. In: Kempgen, Sebastian, Peter Kosta, Tilman Berger & Karl Gutschmidt (eds.). Die slavischen Sprachen. Ein internationales Handbuch zu ihrer Struktur, ihrer Geschichte und ihrer Erforschung. Berlin, New York: de Gruyter, Band 1, 484-505. Zimmermann, Ilse (2010). Where are the worlds? Paper presented at the International Conference to Honor Manfred Bierwisch Sentence Types, Sentence Moods, and Illocutionary Forces, November 4-6, 2010, Berlin, ZAS.