Light Nouns and predicative Infinitives

4 downloads 53841 Views 283KB Size Report
with an overt head or bare VPs, resp. nominalizations. But this ..... projection which is selected by the DP hosting the light noun. ..... I want something good to eat.
Bayer & Brandner – Light nouns and predicative infinitives

Light Nouns and predicative Infinitives Josef Bayer & Ellen Brandner University of Konstanz abstract In this paper, we will discuss an infinitival construction in German and (some of ) its dialects that has a semantically restricted distribution in that the infinitive can combine only with indefinite pronouns and under a certain class of (existential) verbs. Furthermore, the infinitival verb can be only an object-drop verb. The interesting thing about the dialects under discussion (Alemannic and Bavarian) is that in Alemannic the infinitive shows a special suffix ending in a dental stop and that in both dialects only a cliticized form of the infinitival marker zu is allowed (whereas in other contexts the complex form zum is used). We will argue for an analysis where the indefinite pronoun is taken as a light noun which functions as the subject of a small clause whose predicate consists of the infinitival form. The analysis will shed new light on the discussion of light nouns and the relative positioning of the adjective, see Kishimoto (2000), Larson/Marusic (2004). The adjective will be analyzed as a small-clause predicate too which accounts naturally for its post-nominal position.

1. The phenomenon 1.1.

The dental infinitive

Alemannic(A) shows a special infinitival form, called in traditional dialect grammars "gerund", e.g. Staedele (1927), Gabriel (1963): (1)

gib ire ebbes z'essit give her something to-eat "Give her something to eat"

A

Because of the characteristic dental stop occurring as the ending of the infinitive, we will call it from now on "dental infinitive".1 The dental infinitive has the same restricted distribution as a certain kind of Standard German (SG) (morphologically inconspicuous) zu-infinitives, first discussed in Tappe (1984). They occur basically only with the indefinite pronoun etwas (something) and its negative counterpart nichts (nothing): (2)

Im Kühlschrank steht etwas/nichts/*ein Bier/*alles zu trinken In-the fridge stands something/nothing/a beer/everything to drink "There is something to drink in the fridge"

Compare now the following data from Alemannic: (3)

a. *do schtot a Bier/alles z'trinkit there stands a beer/everything to-drink "There is a beer to drink" b. *ich gib dir des Buch z'lesit I give you the book to-read "I give you the book to read"

A A

As can be seen in (3) the dental infinitive in Alemannic never occurs with a restricted NP nor a strong quantifier, just like the Standard German construction. As shown in (4) it is not only the indefinite pronoun (with its negative counterpart) but also certain kinds of so-called degree adverbs which combine with the dental infinitive. This holds also for Standard German, as seen in (5): (4)

S’het / viel / gnueg z’trinkit gea it-has much / enough to-drink existed “There was nothing / much / something to drink”

A

1 The traditional term "gerund" does not seem to be adequate in the light of the fact that there is no agreement inflection which is normally taken to be typical for gerunds. The dental stop originates probably from the “inflected infinitive” used in older stages of German, cf. zi essanne which later developed into zi essande. Commonly it is assumed that this inflection is a dative, governed by the preposition. But note that there is never agreement in number and gender – which sets it apart from the Latin gerund. We will leave the issue open here.

1

Bayer & Brandner – Light nouns and predicative infinitives

(5)

es gab genug / viel zu trinken it gave enough / much to drink

SG

In Bavarian (B), there is no special marking of the infinitive itself but in constructions like in (1) only a clitic form z' is used whereas in other contexts the infinitival marker is zum/zun2, see next section: (6) Host wos z’trinka? have-2sg something to-drink “Do you have something to drink?”

B

(7)

B

Es hot a Bier/ allas *z’trinka geem it has a beer/ all to-drink given “There was a beer/everything to drink”

Alemannic and Bavarian thus overtly signal that the infinitive in this construction is of a different kind than the "normal" infinitive found e.g. in control constructions. This distinction is not seen overtly in Standard German but only via its restricted distribution. Note also that the restriction on weak quantifiers (Milsark (1974/1977) cannot solely come from the matrix verb (Da steht ein Bier – there stands a beer, is perfectly grammatical). So the important thing to note is that it is not definiteness per se which is relevant in this construction but rather DPs that are weak in Milsark's sense and that do not have a restriction. 1.2. Difference in infinitival marker As already hinted at, both in Alemannic and in Bavarian3, there are two variants of the infinitival marker: one is a reduced version, z' and the other one is zum (A), resp. zun (B), consisting of the Standard German version zu plus the contracted dative determiner. A few examples of infinitival forms are given below, where b. gives the form of the bare infinitive. It can be seen that in Alemannic, zum combines with the form of the bare infinitive whereas z' combines with the dental infinitive: (8)

a. b. c.

z’machit, z’lachit, z’essit, z’tend mache, lache, esse, tue... zum trinke, zum lache, zum esse, zum tue make, laugh, eat, do

A

(9)

a. b. c.

z’ macha, z’ lacha, z’ essn, z’ toa macha, lacha, essn, toa zun macha, zun lacha, zun essn, zun toa make, laugh, eat, do

B

The a.-versions occur with the construction from above with its restrictions. Forms like *zum machit or *z’mache are strictly excluded. Thus, the different forms cannot be phonological variants, as has been suggested for Bavarian by Weiß (1998). However, whereas Standard German does not have a grammatical outcome for this infinitive with a restricted DP, this is possible in the dialects under discussion with the infinitival marker zum (and in Alemannic crucially the bare infinitive form). Note also that zum can additionally appear with an indefinite pronoun, cf. (11): (10)

a. b. c.

*Ich möchte ein Bier zu trinken I want a beer to drink I möchte a Bier zum trinke I want a beer to.the-DAT drink I mecht a Bier zun trinke I want a beer to-the-DAT drink

SG A B

2

In certain varieties of Bavarian zum can also be realized with an alveolar nasal as zun. Both phonological variants originate from the preposition zu with a contracted dative marked determiner. 3 Our Alemannic data are from the Island of Reichenau (Southern Baden), the Bavarian data represent Middle Bavarian (Oberpfalz, Lower Bavaria).

2

Bayer & Brandner – Light nouns and predicative infinitives

(11)

a. b.

Ich möcht ebbes z'trinkit/zum trinke I want something to-drink/to-the- DAT drink I mecht wos z'trinka/zum trinka I want something to-drink/to-the- DAT drink ' I want something to drink'

A B

(10b,c) with zum are pragmatically odd because of a redundancy that stems from the fact that beer is normally drunk. This effect disappears if a contrastive reading is forced, e.g. I gave you the beer for drinking and not for tipping away! Additionally, if the infinitival part is focused, it can divide from its head noun and occur in Spec-CP (via V/2 movement) and then the construction again is fine: (12)

zum trinke/*z'trinkit het I gern a Bier (und zum esse/*z'essit a Schnitzel) to-the drink had I prt a beer (and to-the eat-inf a schnitzel) ‘To drink, I would like to have a beer (and to eat a schnitzel)

A

Note that focusing does not rescue the dental infinitive in this construction. The same would be true for Bavarian. This goes together with the fact that in Standard German, the only possible grammatical outcome for (10a) is a construction with a purpose clause: (13)

Ich möchte ein Bier (haben), um es zu trinken I want a beer (have), in-order it to drink "I want a beer, in order to drink it"

(13) has the same flavor of redundancy, but this does not affect the grammaticality, in contrast to (10a). Note that the resumptive pronoun in the embedded clause is necessary. This shows that there are two different clauses underlying. In contrast, in Alemannic and Bavarian, an infinitive with the marker zum is possible with a restricted DP. It remains to be shown whether this is a nominalized verb or whether there are also two clauses involved. Note that there is no presumptive pronoun in (10b,c) with the zum-infinitive. We will return to this briefly in section 5. But for the meanwhile, the data give the following picture: zum-infinitives can combine with any kind of DP (including indefinite pronouns) whereas dental infinitives combine only with indefinite pronouns. In this sense, the distribution of the dental infinitive is a subset of the distribution of the zum-infinitive. In sum, Alemannic and Bavarian signal overtly via the morphology of the infinitive (dental infinitive) that (1) deviates in both - structure and interpretation - from an ordinary infinitival clause. Secondly, the zum-infinitive can be used generally as the modifier of a noun, irrespective whether it is an indefinite pronoun or a restricted DP. This form is obviously not (productively4) available in Standard German and thus the zu-infinitive in SG is ambiguous in the sense that it behaves on the one hand like the dental infinitive in combining only with indefinite pronouns – but also as the form occurring in other clausal infinitival constructions. So the questions that have to be answered are the following: a)

what is the structure of the dental infinitive construction and how can it be distinguished from ordinary zuinfinitivals (in SG)?

b)

is there a structural explanation for the demand that only indefinite pronouns (and degree adverbs are allowed?)

c)

what is the structural distinction between z' and zum?

4

There are some occurrences of zum together with a infinitival form in Standard German, e.g. (i) kommt ihr bitte zum Essen come-pl you please to-the eating (dinner) It can be shown easily that this is pure nominalization, because it allows only incorporated objects, cf: (ii)Wir sind zum Fisch-Essen/*Fische essen eingeladen we are to-the fish-eating/fish-pl eat invited 'We are invited for dinner (where they will serve fish)' where only a bare noun (in a type-reading) can occur. However, in colloquial German, the zum-form is much more widespread than in written texts, cf. Pizza zum Mitnehmen (pizza to take away).

3

Bayer & Brandner – Light nouns and predicative infinitives

1.3. Background on Alemannic and Bavarian infinitives Before going on with the further discussion of this special infinitival form, it is useful to provide some background on infinitives in these dialects in general. As mentioned above, there are three different kinds of infinitival forms. The distribution of these infinitives deviates rather sharply from the distribution of zu-infinitives in Standard German. The facts for Bavarian are discussed in Bayer (1993), see also Donhauser (1986), and Weiss (1998) for some recent discussion, so we will concentrate first on Alemannic. First of all, the (phonetic) form zu does not exist in Alemannic, but this may be due to phonological reaons5. More remarkable is the fact that ordinary control infinitives, as they are well known from Standard German and also from English, virtually do not exist but are realized as finite clauses: (14)

Er hat versprochen [das Buch am Montag zu bringen] he has promised the book on Monday to bring "He has promised to bring the book on Monday

SG

(15)

a. Er hot verschproche,[ dass er s'Buch am mäntig bringt] he has promised that he the book on Monday brings "He has promised to bring the book on Monday

A

b. *er hot versproche [am mäntig s'Buch z'bringe]6 he has promised the book to-bring "He has promised to bring the book on Monday"

A

On the other hand, lexical restructuring verbs of the type discussed in Wurmbrand (2000), which are traditionally known as "optional coherent verbs" select for a bare infinitival complement. Note that the bare infinitival complement shows up in "extraposed" position – an ordering which is excluded in contemporary SG, see e.g. Bech (1955/57), also Askedal (1998): (16)

Es hot aag’fange rängle it has started rain “It has started to rain”

A

(17)

ich han em verbote along uff d'Ddilli (uffi)go I have him forbidden alone on the attic (on-to) go "I have him forbidden to go alone on the attic"

A

Purpose clauses are introduced by the complex complementizer zum, as already shown above. This complementizer occurs also with infinitival complements of nouns, as illustrated in (19): (18)

Ich bruuch etz der Bese zum no schnell d’Garage fürbe I need now the broom to-the just quickly the-garage clean “I need the broom to clean the garage”

A

(19)

Ich ha koa Luscht zum mit dir da anigoh I have no desire to-the with you there to-there-go "I don't want to go with you to that place"

A

As can be seen in (18) and (19), these complements are really clausal since they allow fully specified objects and adverbial material. Finally, nominalizations (embedded under various types of prepositions) are another strategy to express Standard German zu-infinitives:

5

Note also that in its historical development, the infinitival marker occuured in OHG and MHG mostly either as "zi" or as "ze", indicating that it was a reduced vowel. Only later the form "zu" consolidated, presumably due to prescriptive norms. 6 There are (younger) native speakers who accept these version. Sentences like this are also often found in written versions of Alemannic (e.g. columns in local newspapers etc). But this can surely be due to interference, since older speakers insist on the finite version, see Brandner (in prep.) for further discussion. Interestingly, a version with an initial zum is much more accepted, even by the older speakers: (i) er hot versproche zum s'Buch bringe he has promised to-the the book bring "He has promised to bring the book" This could be a real development since clause-initial zum, originating from purpose clauses, see below and the examples in e.g. (9), seems to develop into a general (left-peripheral) infinitival complementizer, see also Müller (2000) for the discussion of similar phenomena in Swabian, a closely related dialect.

4

Bayer & Brandner – Light nouns and predicative infinitives

(20)

(21)

a.

Er sollte aufhören zu rauchen he should stop to smoke

SG

b.

Er sott uffhöre mit‘m rauche/s’rauche uffhöre he should stop with-the smoke /the-smoke stop “He should stop smoking”

a.

Er hat Angst davor, nass zu werden Er hat Angst before wet to get

SG

b.

Er hot angscht vorem (*gli) nass werde he has fear before-the (instantly) wet get "He fears to become wet instantly"

A

A

(21b) shows that there is no adverbial material allowed, i.e. no event modification which hints at a nominal structure. In sum - without going further into discussion - it can be observed that in many cases where Standard German shows zu-infinitives, Alemannic uses other syntactic variants. If an infinitival marker is used (e.g. zum) it occurs at a left peripheral position which is not available in SG. Bavarian prefers control constructions also to be expressed with finite clauses, but the complements of optional coherent verbs show the infinitival marker zum/zun. This complementizer does not show up in left peripheral position – as in Alemannic - but has by and large the distribution of the Standard German zu, see especially (25) where zum appears before the verb7. It is argued in Bayer (1993) that the Bavarian infinitives nevertheless preserved the nominal character of the infinitive to a higher degree than the Standard German constructions. (22)

Er hot g’schaut, dass a nauskummt he has looked that he out-gets “He tried to get out”

B

(23)

Etz hot’s aa no s-renga oogfangt now has-it also still the-rain(ing) started “Above everything else, it has now started to rain”

B

(24)

Etz hot’s aa no zum/zun renga oogfangt now has-it also still to-the rain(ing) started " Above everything else, it has now started to rain"

B

(25)

Er hod [de Sei zum fiadn] vagessn he has the pigs to-the feed forgotten "He has forgotten to feed the pigs"

B

We thus can divide basically between purely verbal projections (bare infinitives) and full CP-structures, introduced by zum (in Alemannic). Additionally, zum heads constructions with nominalized verbs, both in ALemannic and in Bavarian. An infinitival TP-projection seems not to exist in the dialects under discussion. Whether this can be traced back to the nonavailability of the infinitival marker zu in the form that it occurs in SG has to remain an open issue for the moment. But interestingly, the dental infinitive construction together with z' is basically the only productive usage of an infinitive particle construing closely with the verb in Alemannic.

2. Meaning Before turning to further syntactic properties of this construction, let us first have a closer look at its meaning. There are two important aspects: despite its appearance as an infinitival clause in Standard German, the construction is nominal in character,

7

See Bayer (1993) for arguments that zum indeed introduces a nominalized verb form in Bavarian.

5

Bayer & Brandner – Light nouns and predicative infinitives

see 2.1. Secondly, although combining with the auxiliaries to have and to be, there is no modal reading evolving in the dialects, see 2.2. 2.1. A nominal with a "verbal" restriction If one tries to paraphrase the construction, it turns out that the most adequate paraphrase is either a relative clause headed by the indefinite as in (26) or an –able adjective modifying the indefinite pronoun (27): (26)

I want something that one can eat

(27)

I want something edible

There is no event of eating implied in the strict sense (one could do something else with this object, e.g. give it to somebody else), specifically there is no control relation between the subject of the matrix and an implied agent of the embedded construction (e.g. give me something to eat for the children). The whole infinitival construction is rather interpreted as a nominal which has the property P, where P is expressed by the lexical content of the verb. The noun's denotation corresponds to a property-reading of a verb. The question is how this interpretation is achieved and what the syntactic structure is that allows this "transfer". More specifically: Do we have a verbal/clausal projection of the embedded verb? How can the lexical content of a verb become the content of a noun? Can we derive the semantic restrictions on a syntactic basis? And finally: how come that only something/nothing and degree adverbs are allowed? In the following, we would like to suggest that this construction is the mirror image of those well-studied light verb constructions in which an incorporated nominal delivers the specific content of a verb denoting merely "activity", i.e. to do, as discussed in detail in Hale/Keyser (1993,2002). In these cases, the incorporation (or conflation) starts from a position where movement is allowed according to the general restrictions on movement, specifically that the moved element c-commands its trace. Nominals that are generated in higher (specifier) positions of the verbal projection, i.e. subjects, therefore cannot conflate with the verb. This is the reason according to Hale/Keyser why we do not get verbal derivations like *to book the shelf etc. In the case under discussion – although the basic configuration is the same – movement is not possible since it is the verb which should incorporate into the nominal. This is not possible since this would involve an illicit operation of lowering, i.e. the moved element cannot c-command its trace. (28) light verb:

light noun: VP

VP

NP

V0

NP

V0

N0

-do

N0

eat

dance

something

Therefore, another mechanism has to be invoked to bring the two elements into a formal relationship and we would like to suggest that this a small clause configuration of the depictive kind. Consider: (29) I met John hungry This encodes an event where I met John with the additional information that John had the property of being hungry (during that time). Basically the same holds in the construction: (30) John wants something to eat

6

Bayer & Brandner – Light nouns and predicative infinitives

Something to eat can also be divided into two parts: one is that John wants something and the second is that this something should have the property of being edible. This parallelism in interpretation should be captured in parallelism of structure, and this is what we would like to argue for, i.e. there is no projection of the verb in these cases, the verb rather acts like the adjective in a small clause configuration. Tappe (1984) suggested that there is a passive-like operation at stake, capturing the intuition that it is always and only the direct object (DO) of the embedded verb which is involved in this operation. Extending and modifying this idea, we would like to claim that the “original” DO of the verbs acts as the subject of the small clause such that we get roughly the structure in (31): (31) NP NP

PredP

something verb with nom/adj. features

Such an analysis is confirmed by the paraphrases in (26,27). (26) could also be paraphrased as "something that can be eaten", i.e. with a passive form. For (27), it is standard to consider -able as a function-changing morpheme that yields an unaccusative construction, see (xx). However, it will be shown later, that the parallelism to passive is not that straightforward, especially concerning the role of the direct object, but (31) captures the basic intuition. 2.2. The lack of modal interpretation Another important aspect that should be mentioned here is the fact that a construction with a dental infinitive construed with to have never allows a modal reading as it is familiar from other Germanic languages (including Standard German) where the construction have + to-infinitive gives rise to an obligation modal interpretation, see e.g. Gronemeyer (2001): (32)

Ich ha no ebbes z'lesit I have still something to read a. # There is something that I must still read b. There is still something that I can read (= reading stuff)

A

(33)

Ich habe noch etwas zu lessen I have still something to read a. There is something that I must still read b. There is still something that I can read (= reading stuff)

SG

As can be seen from (33) the SG version is ambiguous between the two readings. But the ambiguity disappears if a restricted NP is involved. Then, only the obligation modal reading is possible: (34)

Ich habe noch vier Bücher zu lesen SG I have still four books to read a. There are still four books that I must read b. #There are still four books that I can read (=reading stuff)

In Alemannic, (34) is ungrammatical with a dental infinitive, as already demonstrated in section 1., but it is possible with a zum-infinitive, cf. also above (8b). The same is true for Bavarian z'/zum infinitives: (35)

a. *Ich ha no vier Bücher z'lesit I have still four books to read b. Ich ha no vier Bücher zum lese I have still four books to-the read

7

A

Bayer & Brandner – Light nouns and predicative infinitives

However, the modal reading still does not arise in (35b), at least in the Alemannic and Bavarian variants under discussion. If we assume with Bhatt (1999) that the modal interpretation of an infinitive involves crucially a C0-head, then the interpretation of the a.-version gives further support for our proposed analysis, namely that in this case, there is no clausal projection of the infinitive but rather a small clause (there is no C0-head in a small clause). The general unavailability of the modal reading in Alemannic and Bavarian must probably be treated separately8. Finally, the apparent modality that can be observed in the paraphrase in (12) is not of the above discussed obligation or deontic modality, but rather of the availability reading, see e.g. Izworski (1998) for the same facts in other languages. Especially, there is no range of possible modal interpretations, as discussed in Bhatt (1999, ch. 2) – only the could-reading is available. This restricted interpretation shows again, that there is no C0-head involved in these constructions. To conclude, the interpretation of the dental infinitive as the restriction of a (semantically) empty noun as well as the lack of the modal reading shows that the projection of the verb in this case is of a different kind than the usual event-reading. Additionally, the clitic version of z’ in Alemannic and Bavarian does not allow for a clausal projection(i.e. involving a C-head). Together with the facts presented in section 2., we can conclude that there is a parametric difference between Standard German and the South German dialects under discussion w.r.t. the clausal structures of infinitives.

3. Further restrictions In the following, we will take a closer look at the occurrence restrictions of the dental infinitive. We will first discuss the properties of the embedded verb. It will turn out that only a restricted class of transitive verbs may enter the construction. Then we will show that also the matrix verb underlies certain restrictions.

3.1. Restrictions on the embedded verb As observed in Tappe (1984), only transitive verbs with an accusative object are possible with this kind of zu-infinitives in Standard German. (And the same facts hold for Alemannic and Bavarian as will be demonstrated below). He attributes this to the passive-like operation that seems to be at stake in this construction, i.e. only structurally accusative-marked DPs can be “absorbed”. However, we will show in the following that the restriction goes even further in that (i) no additional argument may be involved and (ii) that the accusative objects of these verbs can be left unprojected, as it is typical for verbs like to eat and to drink. This points into the direction already hinted at above – namely that there is no verbal projection involved in the dental infinitive construction. Rather the verb is transformed into a predicative projection that acts as the predicate of a small clause. The restrictions on the embedded verb (the dental infinitive) are illustrated with the following examples. We also show whether zum would be possible or not. (i) no ditransitives: (36)

Hosch em Pfarrer ebbes zum beichte / *z’beichtit have-2sg the priest-dat something to-the confessing / to-confess

A

(37)

Ich ha ihm nünt zum gea / *z’gebit I have him-dat nothing to-the giving / to-give

A

(38)

Host du am Pfaarer no wos zun beichtn / *z’beichtn? have you the priest-dat still something to-the confessing / to-confess “Do you still have to confess something to the priest?”

B

8

One possible solution could be that it is not only the presence of a C0 –head which is necessary to get a modal interpretation but that additionally this C0 head must be empty. This would fall together with the observations in section 2, namely that there are either C-projections with an overt head or bare VPs, resp. nominalizations. But this remains a topic for future research.

8

Bayer & Brandner – Light nouns and predicative infinitives

(39)

Mia ham uns nix mehr zun song / *z’song we have us-dat nothing more to-the saying / to-say “We are finished with each other”

B

(ii) no verbs selecting for a dative argument, as e.g. German helfen - to help9: (40)

*er hot ebber z'helfit/zum helfe gfunde he has somebody to-help found

A

(41)

*er hot oan z'häifa g'funda B he has somebody to-help found "He has found somebody that he could help" (intended meaning, not available)

This case is interesting because it does not allow the zum-version with the intended reading. But there is a grammatical construction with an infinitival zum, however this has a completely different reading in that the agent of to help is targeted: (42)

ich such ebber zum helfe I look for somebody to-the help

A

(43)

I suoch oan zun häifa I look-for somebody to-the help "I'm looking for somebody who can help me"

B

The indefinite in this case is interpreted as the subject of the embedded verb. This is in fact the only possible interpretation. It shows again, that the zum-infinitive is of a different nature in that it can modify any type of argument, irrespective of its grammatical function. (iii) no verbs with an (additional) oblique argument: (44)

Ich ha nünt uff de Tisch * z’stellit/zum uff de Tisch stelle I have nothing on the table to-put /to-the on the table put "I have nothing to put on the table"

A

(45)

Ich suech ebbes * z'schriibit/zum schriibe look-for something to-write/to-the write "I'm looking for something to write" with zum: “to write with”

A

(46)

I suoch was aaf'n Diisch *z'stäin/zun stäin I search something on the Table to-put "I'm looking for something to put on the table"

B

As indicated in (45), the only possible interpretation of the zum-infinitive is an instrumental one, i.e. if the indefinite is not interpreted as the object/result of the writing-action. This should be seen in line with the to help example, i.e. the possible interpretations do not directly depend on syntactically overt arguments, but rather it seems as if a syntactically legitimate zumphrase can target additional arguments that are only conceptually present. Concerning the main point of this section, the data above indicate again that it is not merely the absence/presence of an accusative marked DP which matters but rather the absence of a full verbal projection. As shown, the zum-versions are possible (with the restrictions just mentioned).

(iv) no particle verbs (47)

Do schtoht ebbes z'wegkaiet/*zum wegkaie there stands something to throw away/to-the throw away "There is something that can be thrown away"

9

A

The other lexically assigned verbal case in German, genitive cannot be tested in Alemannic, because the verbal genitive is not present in this variant.

9

Bayer & Brandner – Light nouns and predicative infinitives

(48)

*Do host was z'austrinka there have-you something to-updrink "Here you have something to drink up" XXXjosef

B

It is characteristic for particle verbs that the argument must always be realized, in contrast to simple transitive verbs like to eat and to drink. If we follow Zeller (2001) in the analysis that particles (which are actually prepositions) head their own projection within a complex VP, then we see again that it is not the absence/presence of the direct object that is relevant to identify the correct class of verbs allowed in this construction. Rather, it is the non-projection of additional V-layers which plays the crucial role. In sum, constructions with an oblique marked argument are not possible as well as those which necessarily require some additional VP-structure. However, if a verb does not select an argument at all, i.e. intransitives, then the construction is not possible either. This pertains unaccusatives and unergatives alike. Note also that in this case, a zum-infinitival is equally not possible: (49)

*Mir hond gnueg z'danzit/zum danze we have enough to-dance

A

(50)

*Do het's gnueg z'stinkit/zum stinke there has-it enough to-stink

A

(51)

*Mir ham gnuag z'danzn/zun danze we have enough to-dance

B

(52)

*Do hot's gnuag z'stinka/zun stinke there has-it enough to-stink

B

translationxxxx?? An exception seems to be the intransitive verb lachen (“laugh”): (53)

Etz hond er nünt mee z’lachit/zum lache now have you-pl nothing more to laugh/to-the laugh “It is no more fun for you”

A

(54)

Der hot dahoam nix z’lacha/ zun lacha he has at-home nothing to laugh/to-the laugh “He doesn’t have much fun at home”

B

However, the indefinite quantifier can be interpreted as a nominal adverb of reason, as in SG Was lachst du denn so blöd? (“Why do you laugh so stupidly?”) where we see that the NP was adopts an adverbial interpretation. Exactly this reading is relevant in (53) and (54). To sum up: The argument structure of the embedded verb plays a crucial role whether a construction with a dental infinitive is possible or not. The data above lead to the following generalization: (55)

Only transitive verbs that may leave their internal argument unprojected syntactically (i.e. object-drop verbs) may enter the dental infinitive construction

It is the class of verbs that may be used both transitively and intransitively but which always have the internal argument represented as a variable in their argument structure. This aspect will turn out to be central for our account.

3.2. Restrictions on the matrix verb Dental infinitives are only possible under verbs that do not affect their nominal complement in the sense of "change of state", but rather foreground the EXIST-Component (Szabolsci 1986) and thus induce also the “Definiteness Effect”: •

verbs of existence:

10

Bayer & Brandner – Light nouns and predicative infinitives

geben - there exists/is, sein - there is, haben – have, stehen - there is •

verbs of transfer: geben - give, nehmen - take, kriegen - get, holen - fetch/bring



verbs of desire: mögen - like/want, wollen - want



intensional verbs: suchen - look-for, finden - find, sehen - see

These are the verbs used in the examples above, so we will not give further illustrations. Matrix predicates which induce a change of state, i.e. accomplishment verbs, cannot be used with the dental infinitive but are again possible with a zuminfinitive: (56)

D’Muetter het ebbes zum esse / *z'essit kocht the-mother has something to-the eat / to-eat cooked “Mother made something to put on”

A

(57)

B’Muata hot wos zun / *z’essn kocht the-mother has something to-the / to-eat cooked “Mother cooked something to eat”

B

Each of the predicates which are possible with a dental infinitive shares a semantic component that has been characterized by Szabolcsi (1986) as an EXIST component. Some have nothing but this; others have further components such as CAUSE TO BECOME AVAILABLE, ... IN A PARTICULAR FASHION etc., which however, as Szabolcsi claims, must be backgrounded. As Grosu (2002) points out, the matrix predicates which license what he calls modal existential constructions (MEC) may vary from one language to the next. For Alemannic and Bavarian it seems clear that the z’-construction is placed rather centrally in the group of languages with MECs10. Given that the predicates in question require (and therefore highlight rather than presuppose) the existence of their direct object, it becomes plausible why z’-infinitives are only construed with weak quantifiers. Notice that weak quantifiers are the ones which involve an existential (∃) operator; (cf. Szabolcsi, 1986; Izvorski, 1998, Grosu & Landman, 1998, Grosu, 2002). Definite DPs can be ruled out because they presuppose rather than state the required existence part. Strong quantifiers can be ruled out accordingly. However, what does not follow from the type of matrix verbs is the fact that the NPs in question may not even have a restrictor. The basic contrast is repeated under (59) for convenience: (58)

a. *Do schtot a Bier/alles z'trinkit there stands a beer/everything to-drink "There is a beer to drink"

A

b. S’het ebbes / nünt/ viel / gnueg z’trinkit gea it-has something/ nothing/ much / enough to-drink existed “There was nothing / much / something to drink” So there must be an additional factor that rules out the combination of a restricted NP and the dental infinitive. We will assume that it is a (lexical) property of these verbs that their selectional requirements are fulfilled if an existential quantifier heads the nominal projection which is its complement. This means that their c-selectional requirements are fulfilled already by a simple D0, filled with an existential quantifier. (59) EXIST verb [D0 ∃x [ ...NP…]

10

The restrictions in the German dialects go, however, farther than in the Albanian, Arabic, Bulgarian, French, Greek, Hebrew, Hungarian, Macedonian, Polish, Romanian, Russian, Serbo-Croatian and Spanish examples presented in Grosu (2002). In these languages, the complements are full-blown CPs with various choices for operators in SpecCP. So the constructions are only provisory comparable. But that the class of matrix verbs coincide is surely not an accident. We have to leave open an in-depth analysis of these predicates.

11

Bayer & Brandner – Light nouns and predicative infinitives

The idea is then that the dental infinitive is acting as the restrictor of the DP, i.e. that the existential quantifier is introduced via selection by the matrix verb and that it then combines with its restrictor made up of the dental infinitive. This entails that we assume that indefinites are made up of two parts (such that the two can be "targeted" by different operations) and thus in the following section we will have closer look at the properties of the nominal elements allowed in this construction. 4. The concept of light noun 4.1. Previous analysis The concept of “light noun” must of course be seen in the context of light verbs. These are characterized as verbs with a limited semantic content, basically expressing only ‘activity’. They can combine with other predicates to build complex predicates, or – as already hinted at above – they are part of the lexical structure in that they combine with a noun, originating as their complement. The conflation of the two yields a verb with the lexical content of the noun. We would like to emphasize that we see the parallelism between light nouns and light verbs not so much in their syntactic behavior, as it has been claimed by Kishimoto (2000), see below, but rather in their similar semantic characterization. The important thing is that light nouns – like light verbs – are devoid of lexical content and they must derive their semantic license by combining with a contentful element. But as already discussed in section 2., there is no "simple" conflation possible, due to structural reasons. Before turning to our own analysis, let us have a brief look at recent analysis of light nouns: Kishimoto (2000) presents an analysis of light nouns in purely syntactic terms. He suggests that they behave syntactically like light verbs11 in that they (must) move to a higher functional projection. Light verbs move to an I-projection and light nouns move to a Number head within the functional domain of DP. This is meant to explain the fact that adjectives occur in these constructions obligatorily in post-nominal position, (60), whereas DPs with a restriction have the adjective to the left: (60)

a. I heard something new b. * I heard new something

(61)

a. I have a new book b. * I have a book new

Larson/Marusic (2004) show convincingly that such an analysis is not able to explain all the occurrences of post-nominal adjectives, especially if one also considers the different interpretative possibilities, cf. the visible stars vs. the stars visible, (see section 4.4 for further discussion). They suggest that the two orders are base generated and the defectiveness of the Nprojection with the indefinite, cf. (60b) prevents the adjective to adjoin to pre-nominal position. However, their structural analysis is equally not completely satisfying since it leaves unexplained what is special about indefinites, especially if – as they claim – all cases of post-nominal adjectives should be treated alike. So, the unavailability of (61b) remains unclear. Leu (2004) on the other hand suggests a solution in terms of an empty noun that is preceded by the adjective and therefore the post-nominal position is only apparent. The noun which is modified by the adjective is the complement of a functional projection which is selected by the DP hosting the light noun. Evidence for this comes from the fact that in German (and also Swiss German), the adjective appears in an inflected form in the light noun construction. This is typical for a pre-nominal occurrence whereas in predicative function, agreement is not possible: (62)

a. b. c.

etwas schön-es something nice-agr es ist schön/*es it is nice ein schön-es Buch a nice-agr book

Leu brings this pattern together with the French light noun construction. In French, adjectives modifying a light noun are introduced by the preposition de which is the (partitive) genitive marker in French:

11

He understands under light verb basically auxiliaries and modals, a rather unusual categorization.

12

Bayer & Brandner – Light nouns and predicative infinitives

(63)

a. quelque chose de magnifique some thing of fantastic b. quelque chose de beau some thing of beautiful

The structure is like in (64), (details omitted), where NP2 refers to the empty noun which is modified by the pre-nominal adjective. As such the adjective shows normal behavior in terms of inflection and additionally the "post-nominal" appearance finds an explanation. (64) DP

deP de0

something quelque chose

-s de

NP Adj

NP2

schönbeau

ec

Crucially, the light noun itself is not further decomposed as in the proposals, e.g. by Abney (1986) and Kishimoto (2000). Instead it is seen as an un-analyzable unit. Although the composition of indefinite pronouns in English points to a complex analysis, Leu claims that (i) the non-productivity of the pattern in English (e.g. *nohow, *somewhen etc) and (ii) the morphological opaque forms in Swiss German (öppis, corresponding to ebbes in Alemannic < etwas SG) do not allow a decomposed analysis. Nevertheless, we would like to argue that the two components are indeed decomposed and that they head different positions within the DP. 4.2. Analysis 4.2. 1. Something and nothing Let us first discuss the properties of something and nothing. In German, they consist of two components, although these are not easily detectable in Swiss German and Alemannic - due to regular phonological processes (assimilitation): (65) et – was

existential prefix + pronoun

(66) ne –wiht (= nicht)

negation prefix + 'thing'

ete- is used in MHG as an existential prefix, see Paul et al (199824) and combines with a pronoun12. Ne-wiht is the exact counterpart to Modern English no-thing. We will take et- and ne- as the spell-out of the existential quantifier, resp. its negative counterpart and as observed above, see section 3.2., these can act as the sole argument of the EXIST-verbs. They occupy the D0 position and the nominals –was and –wiht fill out the NP slot. (67)

DP

D0 etne-

NP was wiht

4.2.2. Degree adverbs

12 Until the 16th century, the form et-wer (Alm: ebber) was used also in the Standard language. It was given up in favor of ieman = generalizing particle + 'man'.

13

Bayer & Brandner – Light nouns and predicative infinitives

It was argued above that the light nouns something and nothing have a complex structure consisting of the quantifier and the noun. As we observed earlier, it is not only these light nouns which can enter into the dental infinitive construction but also so-called degree adverbs like much, enough etc. How can these be integrated into the proposed structure? The semantic and syntactic properties of these items are discussed extensively in Doetjes (2002). We cannot go into a detailed discussion but simply draw from her work the conclusion that the degree words under discussion can always function as a determiner, Doetjes (2002:7). This sets them apart from adverbs introducing iterativity. Compare the following contrast13: (68)

a. he has bought a lot b. he buys often (useless thing)

(69)

a. he has often bought 3 kilos of olives b. * he has bought a lot 3 kilos of olives

The important observation is that a lot can be used as quantifying the event in (68) but it does do not imply iterativity of the event (in contrast to often). It rather refers to the quality/intensity of the event. Observe that in (68a) an NP can be added which realizes the (understood) object of buying (a lot of things). In (70b) it seems as if the degree adverb and the determiner 3 kilos compete for the same position. This can be explained if we assume that a lot in fact occupies the D0 position14. Now, in the dental infinitive construction only degree adverbs like enough, a lot, and few are possible and never iterativity implying adverbs like often. We therefore assume that the analysis for degree adverbs is such that these adverbs occupy the D0 position and that the structure is thus the same as in (68) with the sole difference that the N-part is not spelled out, i.e. there is no –was overtly but the NP is nevertheless present15. This has probably to do with the fact that et- and ne- are bound morphemes and have to attach to an overt morpheme whereas this is not the case with the degree adverbs. 4.3. The structure of the dental infinitive Now as said above, the important thing about light nouns is not so much their syntactic distribution (this should fall out from the analysis) but rather their special semantic properties, namely that they are devoid of content. And as already hinted at above we will assume that the “transfer of content” takes place in a small clause configuration. This means that the VP headed by the dental infinitive must be prevented from projecting an event-like type. Rather it must be turned into a property, i.e. a purely predicative element. Now a VP is per se predicative and in the semantic tradition there is no difference (in type) whether the VP is finite (and eventually turned into a proposition) or whether it is infinite. But what seems to be happening in the dental infinitive construction is that the V-projection is of a syntactic and semantic type that hinders this projection to extend, i.e. combine with further functional heads such that the type and category of a proposition is created. The distinction we are looking for is discussed in Carlson (2002). In the context of Diesing's (1992) Mapping hypothesis, Carlson explores the borderline between VPs and IPs and how it can be captured in semantic terms. Diesing's central claim is that (unbound) material inside the VP is existentially bound (mapped to the nuclear scope) whereas material outside the VP can be bound by various operators (mapped into the restrictive clause). This is meant to explain the fact that e.g. an indefinite located outside of the VPs always receives a specific interpretation and never an existential one.

13

Doetjes discusses French examples but the contrasts can be reproduced in English as well as in German. That these two kinds of adverbs must occupy different positions can also be seen by the fact that two different kinds of adverbs can be coordinated, (a). This is not possible if they belong to the same class (b): a. er hat oft viel Bier getrunken he has often a lot beer drunk b. *er hat viel genug Bier getrunken he has a lot enough beer drunk 15 This must also be assumed for the (rare) cases where etwas can be left out. This is especially found in older styles and special registers (e.g. bible translations): (i) und dann gab er ihnen [zu essen] and then gave he them to-eat 14

14

Bayer & Brandner – Light nouns and predicative infinitives

Carlson suggests that this distinction can find a natural explanation if one assumes that VPs denote only eventualities in the sense of Bach (1986). A VP per se denotes only kinds of events16. So VPs are instances of types and not of tokens. And this is what we need for the interpretation of the dental infinitive. It is merely the type of event described by the verb that is referred to in this construction. We can conclude from this that the category of the dental infinitive is a bare VP. The question is now what happens to the arguments that are selected by the verb in question. Recall that we identified as the relevant class transitive, active verbs which allow their DO to remain unprojected. In this case they get the interpretation of an activity verb although the DO, i.e. the theme of “eating” is conceptually present. The same holds for the agent. Recall that there is no control relation between the subject of the matrix and the agent of the embedded verb, rather we get either a PROarb interpretation or the agent is contextually identifiable. It seems plausible then that only a V0 projects to a maximal level and that this Vprojection acts as a predicate. So the claim is that the projection of e.g. essit consists of the V0-head and its maximal projection level without any arguments structurally represented and as such it denotes only an eventuality. We can assume that the special morphology in Alemannic is an overt signal for this kind of maximal VP and that it does not combine with further functional heads (i.e. Tense or Aspect). Such a VP does not have the potential of a proposition, i.e. of being either true or false w.r.t. the actual or a possible world. Let us now cometo the role z´. The preposition zu has in its core a directional meaning. If we assume that this directionality is transferred from space to time we can attribute to it a future or prospective interpretation17. Merger with the dental infinitive, e.g. essit, yields the property of ‘being edible’. The fact that there may be a future event of eating can only be inferred. It is clearly not expressed. In sum, the compositional analysis of (i) the eventuality reading of the verb together with (ii) prospective meaning of zu yield an object that can be used as a predicative. Syntactically we would like to suggest that z´ is the head of the small clause, i.e. the overt head of a PredPhrase in the sense of Bowers (1993), or as Moro (2000) terms it: it is a “rich small clause”. The next question is where the nominal –was originates, and why only this least marked form can arise in the dental infinitive construction. Recall that there is no DO projected within VP. If the construction were equivalent to a passive or raising construction, any kind of structurally Case-marked DO would be expected. But this is not what we find. The constraint cannot be attributed to a purely semantic incompatibility restriction since the versions with zum are possible with a restricted noun, see section 2. So there must be a syntactic reason. Additionally, if –was would originate in the VP, it would be part of its own predication – a rather awkward situation. The idea then is that –was is “late inserted” as a default in order to avoid the violation of the ban on empty predication. It combines with the ∃-operator, introduced by the matrix verb and thus we get the following complete structure: (70) V0 suchen

DP D0 et-

NP (= PredP) N0 -was

Pred´ Pred0 z´

VP V0 essit

There is no formal relationship between –was and the DO of the embedded verb expressed in (70), although – at first sight etwas is interpreted as the object of the matrix as well as the object of to-eat. But recall that a simple raising analysis would not give us any clue how to account for the restriction that only indefinite pronouns are allowed. The identification of these 16 At this (semantic) level, the arguments of the verb are not referring expressions but rather build together with the verb a more specific kind of event (i.e. cake-eating is more specific than eating but both are eventualities). Important for us is the fact that the non-projection of arguments will lead unavoidably to an eventuality reading.

15

Bayer & Brandner – Light nouns and predicative infinitives

(apparent) two occurrences of -was cannot be a matter of narrow syntax, a view which is compatible with the fact that only "object-drop" verbs are possible: the non-projection of the argument (which is crucial for the predicative usage of the verb) does not lead to a violation of the Projection Principle. Nevertheless, in the conceptual-lexical structure, a(n) unspecified object is present (there is no eating-activity without something being eaten!). We would like to suggest therefore that the identification happens in the conceptual component18 and that there is no formal syntactic relationship between the subject of the small clause – consisting of –was and the non-projected direct object of the infinitival verb. But there is nevertheless an thetaidentification via the argument structure of the infinitive verb. What about the complement of the matrix verb? Is the light noun etwas indeed the direct complement of it? We would rather suggest that the matrix verb thematically selects for a small clause, i.e. it introduces categorically the D0 (realized by the existential quantifier) but semantically it calls for a property. 19 So –was is not the direct complement of the matrix verb20 and therefore not theta-marked by it; rather its theta-role is identified via the inferred relation to the (only conceptually present) DO of the embedded verb. The whole constituent then receives a theta-role from the matrix verb, such that we get a picture like the following: (71)

[∃x[ -was z'essit]] θ1 θ2

V

Note that any other structural analysis (e.g. in terms of N-raising within one single N-projection) would then lead to a thetacriterion violation. So the syntactic analysis we suggest is corroborated further by semantic and thematic considerations. Two things remain to be clarified: (i) the category of the SC and (ii) how the nominal checks its case. As can be seen in (70), we assume that it is the late inserted N0 which determines the categorial value to the whole projection. We cannot go into further details but note that this is in line with a suggestion made by Iatridou et al (1999) in the context of relative clauses. They propose that it is the moved NP that projects in this construction, thus exploiting the variability given in the Bare Phrase Structure concept (Chomsky 1995). And since the comparability with relative clauses is given, we think that this assumption is justified. W.r.t. case marking, we assume that –was checks its case with the matrix verb in an ECM-like way. Case cannot be checked in the domain of the dental infinitive as there is no functional structure that could make this possible. This is corroborated by the fact that the outer syntax of the light noun construction is like that of any other ordinary DP, i.e. it can act as the subject in passives, it can occur as the complement of prepositional phrases etc. However, it cannot occur as an argument marked with an oblique case, see (72c). But this can be explained in the theory of oblique case that is advocated in Bayer et al (2001). There it is shown that elements that cannot overtly express case-values (like nothing etc) can never show in oblique case positions. So the ungrammaticality of (72c) is independent evidence for the claim that case is assigned in an ECMconfiguration and this implies structural case marking. (72)

a. es wurde nichts zu essen gefunden EXPL was nothing to eat found “There was not found anything to eat” b. wir haben mit etwas zu essen gerechnet we have with something to eat reckoned “We reckoned with something to eat” c. * dieser Klumpen ähnelt nichts zu essen this lump resembles nothing to eat “This lump doesn’t resemble anything to eat”

17

as it is traditionally done in the context of zu-infinitives, see e.g. Paul et al (199824) among many others. This is reminiscent of adjectival modification discussed in Larson (1999). In a construction like (i) a quick cup of coffee "quick" of course does not modify the cup but rather the implied event of "drinking" where the cup is involved.. In our cases above, the object of "eating" is present already in the lexical entry of the verbs in question, but the mechanism is basically the same. 19 This is typical for intensional verbs, as discussed in Zimmermann (1993). Thanks to H. de Swart for pointing this out. 18

16

Bayer & Brandner – Light nouns and predicative infinitives

4.4. Extension to other light noun constructions As has been pointed out in section 4.1., light nouns are generally discussed in the context of pre- and postnominal adjectives. But as just seen, the properties of the dental infinitive construction find a natural solution in terms of a small clause analysis according to which the light noun is the subject of the small clause and its predicate is a (reduced) VP. Thus, this configuration is expected to occur also with other predicative elements. And this is what we would like to suggest for the post-nominal adjectives in light noun constructions: there is no movement of the head noun nor different base positions of the adjective within a straight projection of N. Instead of a reduced VP it is the adjective which acts as the predicate of the small clause. So the structure for light nouns with adjectives is the same as in (70). Since adjectives are predicative per se and since they do not have a theta-grid like verbs, there are no lexical restrictions expected – in contrast to the dental infinitive. Concerning restricted DPs with a postnominal adjective as in (73) and discussed by Larson & Marusic (2004) in the context of light nouns, we would suggest that these cases should be set apart from the genuine light noun construction. (73) a. the stars visible b. the visible stars The reason for assuming that these are different constructions is that (i) it is only a subclass of adjectives, namely derived ones, that can occur post-nominally with a restricted DP. This does not hold for the light noun construction (something white) and so there is a difference in that the light noun construction is fully productive whereas restricted DPs with post-nominal adjectives have only a limited range of application. (ii) As pointed out in Larson/Marusic, (73b) has only a stage level interpretation (the observation is going back to Bolinger (1967)). Although it is true for the adjectives possible with restricted nouns, it does obviously not hold for something white etc. Thus, although post-nominal position may be a necessary condition for stage-level interpretation, it is by no means the only factor, i.e. it does not automatically lead to that interpretation and this shows again that restrictive DPs with a post-nominal adjective (of a certain kind) is – despite so many similarities of a different kind than the light noun construction. In our view it is the semantics of the light nouns which lead to an analysis in terms of a small clause and thus the postnominal occurrence of adjectives in this construction falls out without further assumptions. The post-nominal occurrence of adjectives with restricted DPs should find an explanation in different terms, probably best in terms of reduced relative clauses, as e.g. in Ross (1981), see also Svenonius (1994) but we cannot go further into that. 4.5. The inflection of the adjectives The question that has now to be answered is why there is inflection on the adjective in the light noun construction. Recall that predicatively used adjectives in German are normally non-inflected, see the examples in 4.1. surrounding the discussion of Leu's proposal. Now – although it is true that the inflection is homophonous to the singular neutral form (-es) – which corresponds to the values of etwas – it is claimed by traditional grammarians that the –es inflection is an old genitivus partitivus form, see e.g. Schirmunski (1962:435). This fits very well with Leu's findings about French. Recall that in the French light noun construction, the genitive particle de must be used. So we would have in both languages a genitive construction and then we can assume that in both languges, the small clause is headed by an exponent of the genitive:

20

This can also be seen the wrong implication: There is nothing to eat ≠> there is nothing

17

Bayer & Brandner – Light nouns and predicative infinitives

(74)

DP D0

etquelque

NP (= PredP)

N0

-was chose

Pred'

Pred0

de -es

AP

A0

schönbeau The difference between French and German would then be – just like in Leu's proposal – that the adjective schön moves to the Pred0 position in order to combine with the genitive inflection. Since adjectives do not build complete functional phrases like e.g. zum-phrases, they never can occur with restricted DPs, i.e. they are either the predicate of a construction like in (74) or they adjoin to the NP in the familiar pre-nominal position. This can also be illustrated with the English example in (75).21 (75)

a. some white cat b. * some cat white c. something white

(75b) is out because it is not the right input structure for (75) since cat projects an NP of its own and thus we would have a double restriction – which is – in the approach here – not only ruled out for semantic reasons but also for syntactic reasons. 5. Z' and zum In this final sub-section, we would like to briefly address the analysis of zum-complements. Recall from section 3. that in many contexts where the dental infinitive is not possible the alternative form with zum yields a grammatical sentence. Some examples are repeated here for convenience: (76)

a. Ich ha nünt zum uf de Tisch stelle/*z'stellit I have nothing to-the on the table put 'I have nothing which could posit on the table'

(77)

Ich hätt gern a Bier zum trinke I had prt a beer to-the drink 'I would like to have a beer to drink'

A

The contracted form of the infinitival marker already hints at a possible solution: in this case we have a nominalized form of the infinitive which is functionally complete in that it has a DP as its maximal projection which in turn is selected by the preposition zu, yielding zum: (78)

PP P0 zu-

DP D0 -m

NP N0 trinke

The interpretation is a purpose interpretation and we would like to suggest that a PP like in (78) can attach to any kind of constituent, i.e. there are no syntactic restrictions as it is the case with the predicative VP-projection of the dental infinitive. In Alemannic, as already hinted at in section 2., zum has even a more wide-spread use in that it can function as a left-peripheral infinitival complementizer. This is probably the reason why it occurs in (76) before the complement of the verb, i.e. the table.

18

Bayer & Brandner – Light nouns and predicative infinitives

i.e. the table. This indicates that there is a full purpose clause projected in this case. This is different in Bavarian, cf. (47) where probably only a structure like in (79) is available. We cannot go further into the difference between these two variants, but note that in both cases a purpose reading evolves. But what is important is the fact that the contracted form already indicates that in this case, the verb is in fact turned into a noun (probably already in the lexical component) and that this noun projects to a full DP and thus it cannot act as a restrictor in the sense made precise above, i.e. it is no more predicative in the syntactic sense and thus the complete phrase is modifier that does not underlie further syntactic restrictions.

6. Conclusion There are still many open problems in this area but we hope to have shown that the very restricted distribution of the dental infinitive can find a natural explanation in the syntactic analysis proposed above. But there remains on open question which we would like to address briefly: As discussed in Larson/Marusic (2004), post-nominal adjectives cannot be iterated that easily as pre-nominal adjectives can. This would find a natural explanation in terms of the analysis above since this would also be a double predication. However, as Susanne Trissler (p.c.) and I. Marusic (p.c.) pointed out to me, it is possible to have a combination with a post-nominal adjective and a dental infinitive: (79) I möcht ebbes guets z’essit I want something good to eat

A

The only possible solution that comes to mind is that –was is first merged with the predicate guets and that in this case it projects only to N’. As such it can be re-merged in the specifier position of z’essit, project to the maximal level and then combine with et-, being selected by the matrix verb. Note that only “quality”-adjectives can occur together with a dental infinitive. Let us assume that “quality” itself is (semantically) not complete and thus requires an additional nominal that can deliver something on which it can operate then the solution just hinted at is a viable solution. But this has to be left open for the moment. References (muss noch ergänzt werden) Abney, S. (1987): The English noun phrase in its sentential aspect. Doctoral dissertation, MIT Cambridge, Mass. Askedal, J. O. (1998). „Zur Syntax infiniter Verbalformen in den Berthold von Regensburg zugeschriebenen deutschen Predigten. Vorstufe der topologischen Kohärenz-Inkohärenz-Opposition?. in: Askedal, J.O. (ed). Historische germanische und deutsche Syntax. Frankfurt a.M., Peter Lang. Bach, E. (1986). “The Algebra of Events”. Linguistics and Philosophy 9: 5-16. Bayer, J. (1993): „Zum in Bavarian and scrambling.” in: Abraham, W. & J. Bayer (eds.) Dialektsyntax. Opladen. Westdeutscher Verlag. Bayer, J., M. Bader & M. Meng, (2001): “Morphological underspecification meets oblique case: syntactic and processing effects in German”. Lingua 111. 465-514. Bech, G. (1955/57). Studien über das deutsche verbum infinitum. Kopenhagen, Munksgaard. (appeared also 1983.Tübingen, Niemeyer) Bhatt, R. (1999): Covert Modality in NOn-Finite Contexts. PhD dissertation, Pennsylvania. Bowers, J. (1993). "The syntax of predication". Linguistic Inquiry 24:591-656. Carlson, G. (2000). Weak indefinites. Ms. (Talk given at the NP-DP Conference, Antwerp.) Doetjes, J. (2002). “Comparing adverbs of quantity”. Ms. University of Utrecht. Donhauser,. K. (1986): „Die Infinitivkonstruktionen mit z' und zum im Bairischen.“ in: Koller, E. et al.(eds.) Bayerischösterreichische Dialektforschung. Würzburg. Gallmann, P. (1990): Kategoriell komplexe Wortformen. Das Zusammenwirken von Morphologie und Syntax bei der Flexion von Nomen und Adjektiv. Tübingen. Niemeyer. Gronemeyer, C. (2001). “Modal readings of light verbs with to-infinitivals. Working papers in Scandinavian Syntax 48: 55.65. 21 Maybe the derived adjectives in English that can occur post-nominally, see the discussion in section 4.4., have some additional functional shell which enables them to adjoin to the DP, just like the zum-phrases. But this remains speculation at this point.

19

Bayer & Brandner – Light nouns and predicative infinitives

Grosu, A. (2002): „The syntax-semantics interface of Modal Existential wh Constructions.“ Ms. Tel Aviv University. Iatrodou, S. & E. Anagnostopoulou & R. Izworski (1999): "Some observations about the form and meaning of the perfect". in: M. Kenstowicz, ed. Festschrift for Ken Hale. MIT Press. Izworski, R. (1998): ”Non-Indicative WH-Complements of Possessive and Existential Predicates.” in: Tamanji, P. & K. Kusumoto (eds.) NELS 28. 159-173. Kishimoto, H. (2999). "Indefinite pronouns and overt N-raising". Linguistic Inquiry 31.3:557-566. Larson, R. (1999). Semantics of adjectival modification. Lectures presented at the LOT winterschool. Amsterdam (http://semlab5.sbs.sunysb.edu/~rlarson/. Larson, R. & Marusic, I. (2004): "On indefinite pronoun structures with APs: Reply to Kishimoto". Linguistic Inquiry 35.2: 268-287. Leu, Th. 2004. "Something invisible in English". Proceedings of the Penn Linguistics Club PLC 28. Merkle, L. (1976): Bairische Grammatik. München. Hugendubel. Milsark, G. (1974): Existential Sentences in English. PhD thesis. MIT, Cambridge, Massachusetts. Milsark, G. (1977): “Toward an explanation of certain peculiarities of the existential construction in English.” Linguistic Analysis 3. 1-30. Moro, A. (2000). Dynamic Asymmetry. LI Mongraph 38. Cambridge, Mass. MIT Press. Paul, H. (1920): Deutsche Grammatik, vol.IV Syntax. (Zweite Hälfte). Halle: Niemeyer. Schiepek, J. (189): Der Satzbau der Egerländer Mundart. Prague. Verlag des Vereins für Geschichte der Deutschen in Böhmen. Schirmunski, V. M. (1962): Deutsche Mundartkunde. Akademieverlag, Berlin Staedele, A. (1927): Syntax der Mundart von Stahringen. Dissertation, Freiburg im Breisgau. Szabolcsi, A. (1986): „From the Definiteness Effect to Lexical Integrity“. in: Abraham, W. & S. de Meij (eds..) Topic, Focus, and Configurationality. Amsterdam.Benjamins Weiß, H. (1998. Syntax des Bairischen. Tübingen. Niemeyer. Zeller, J. (2001). Particle verbs and local domains. Amsterdam. Benjamins

20