Working Papers in Scandinavian Syntax

39

Sten Vikner

MODALS IN DANISH AND EVENT EXPRESSIONS

Department of English

lmiversity of Geneva

22 Blvd des Philosophes

CH-1205 Geneve

Switzerland

June 1988.

Modals In Danish and Event. E.xpressions

Contents l . Introduetion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 . Modal Verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2. 1 Two types of meaning: Epistemic vs. root . . . . . . .... ... . . . . . . . . . . . . . . . . . 4

2. 2 Differences between epistemic and root moelals . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 . 2. l Tenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2. 2. 2 Empty categories as subjects . .. . . . . . . ... . . . . . . . . . . ... . .... .. . . . . 8

2 . 2 . 3 Ordering and combination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2. 2.4 Pronominal complements . . . . . . . .... ... . .. . . . . . ... .. .... . . .. .. .. .. 10

2 . 2 . 5 Pseudo-clefts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3. Analysis . . . .. . . . . .. . . . . . . . . . . .. . . . . . . . .. .. . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . 12

3. 1 9-roles and Additional 8-roles . . . . .. . . . .. . . . .. . .. . . .. .. . . . .. . ... ... . . 12

3. 2 Raising or Control . .. . . . . . .. . . .. . . . . . . .. . . ..... . . .. . . . . . . . .. . . . . . . . . . 13

4. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5. State Expressions Replace Event Expressions: ;!:å + NP, .bl.iYe + AP . . . . . . . . . 19

6. Special Constructions: Motion and Passive . . .. . . . . . . .. .... . ... .. . . . ... . ... 20

6. l K&!mme vs . Bare Adverbials o f Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

6. 2 Illi.Ye-passive vs. 6-passive . . . . .. . . . . . .. . . . . . . . .. .. . . . . . .. . . . . . . . . . . . 23

6.3 Perfect Infinitives . . . . .. . . .. .. .. . . . . . . .. . . . . .. . . . . . . . . . . . . . . . . . . . . . . 24

7 . Some Other Consequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

7 .l Idioms . . . . . .. . . . ... . .. . . ... . . . .. . . .. .. . .. ... .. . .... . . ......... . . . . . . . 26

7. 2 Source phrases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

7. 3 Agent Ambigui ty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

7. 4 Preposed Objects . . . . . . . . . . . . .. .. . . . .. ... . .. . . .. . ... ..... ... . ....... .. 27

7. 5 Adverbials of Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

7. 6 Adverbials of Duration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

l. Introduetion

In this paper I will discuss the effects t.hat a Danish modal verb of deontic

obligation has on the VP with event interpretation that. it may govern. These

VPs with event interpretation differ from VPs with event interpretation

governed by other modals in that they do not. allow the event verbs .få ( "get"),

.bl.iYe ( "become"), and one of the two types of kQrome ( "come"), tut require

e i ther the state verbs haYe ("have") , Yære ( "be" ) o r some special construct.ion

(bare adverbial of direction or 6-passive). l

Section 2 is an introduetion to modals in Danish, especially w. r.t. the dis­

tinetion epistemic/root. Before the central data are set forth in section 4,

section 3 introduces the basic idea of the analys is, that some verbs, al though

they do not take any arguments (i. e. t.hey do not assign any thematic ro les) ,

assign an "additional" thematic role. Section 5 and 6 analyse t.he data and

attempts an account in terms of the analysis suggested in section 3. Section 7

lists a series of further facts that the proposed analysis accounts for

straightforwardly, before the conclusion in section 8.

The different types of data discussed in this paper were first. b:rought

together and discussed in a very interesting paper by Erik Hansen ( 1972), and

in the appendix I comment specifically on a proposal of his, the so-c�lled

"interessee"-construction, in that I suggest that it does not exist as such.

2 Modal Verbs

I will consider the following verbs to be modals in Danish: Yllle , � . Jlåt:!& , lwnne , burde , :tl.l..l::d.e , and �. As for their translations and use, see

(5)-(11) in section 2. 1 below.

Here I will briefly state three (sets of) properties that these verbs have:

i) They may be followed by infinitives without the infinitival marker at

(l) a. Jeg ill (*at) gå hjem nu

"I want to go home now"

b. Jeg r6rul.lrer *Cat) gå hjem nu

"I wish to go home now"

( 2) a. Jeg gider: ikke (at) læse mere

8. Conclusion .. ... ... .. . . . . .. . . . . . . . . . .. . . . . . . .. . . . . .. .. . . . ... . .. .. . ... .. ... 29 "I do not feel like reading anymore"

Appendix: The "Interessee" -Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

A . l 'ii.lll;: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

A. 2 Sw.lle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

A.3 M!:tJ& . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

A. 4 � . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . 32

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

- 2 -

b. Jeg overkomroer ikke *(at) læse mere

"I am too exhausted to read anymore"

l I am very grateful to Carl Vilmer for help and comments. I should also like to thank the following: Christer Platzack, Ian Roberts, Jane Grimshaw, Niels Davidsen-Nielsen, Paul Harrison, Lars-Ake Henningson, Ramona R�isch, and audlences at the University of Lund in March 1988, and the Univerflity of TromsØ in May 1988.

Part of the research for thifl paper was made pollsible by a grant from the "Fonds national suisse de la recherche scientifique", Berne, Swit..erland.

- 3 -

Erik Hansen ( 1977) reports that both :tl.!rdf: ("dan:�") and � ( "feel inclined

to") are beginning to accept infinitives with a.t (hence the lack of a star

inside the bracket in (2a)), whereas � ("need"), which is not treated as a

modal in the present paper, is beginning to be accepted with infinitives

without a.t. ii) Modals also behave differently from (most) other verbs w. r. t. tag ques­

tions, as shown in Lundskær-Nielsen ( 1983: 8) . Most frequently tags are just

� ("not'') for postive statements and :iel for negative ones, but real tag

questions like the ones used in English are also poesible, and here the verb

im ("do" l must be inserted in all cases, except. haY:e ("have"), Yæ.re ( "be"),

and the modals where i t eannot be inserted :

(3) a.

b.

c.

d.

(4) a.

b.

c.

d.

Hun haJ: ikke læst min artikel, hat:!*� hlm vel?

"She has not read my art.icle, has/does Bhe?"

Han el: gået' w*� han ikke?

"He is gone, is/does he not?"

Han kan svømme, k.an/*� han ikke ogsl\?

"He can swim, can/does he not (too)?"

Hun Yil ikke sove godt i nat, Yil!*� hun vel?

"She will not sleep well tonight, will/does she?"

Han fik malet huset, gjQrde/*fik han ikke også?

"He got the house painted, did/got he not (too)?"

Hun bleY alligevel ikke syg, �*bl.eY hun vel?

"She became not ill anyway, did/became she?"

Hun lw.m ind på universitetet, �*&:m� hun ikke?

"She came into the university, did/came she not?"

Hun 1.æa:te min art.ikel, �*læa:te hun ikke?

"She read my art.icle, did/read she not?"

iii) Finally, modals are often said to be morphologically deficient, in that

they do not have e.g. an imperative or a present part.iciple. This is not unique

to modals, as it also goes for e.g. få ("get"), haYe ("have"), beaidde., e.iJa (both "possess"), � ( "owe"). The question iB also whether this is really a

morphological phenomenon, as i t could be argued to be more of a sernantic one, especially as regards the imperative.

2 l Two types of roeaning: Eoistemic ys root

The distinction between epistemic modals and J�t modals is a well-known one. As a brief summary o f this distinction, let me quote Platzack ( 1979: 44) , who says that "the epistemic sense . . . qualifles the truth value of the

sentence containing the modal", and that "the root sense ... expresses neces­

sity, obligation, permission, volition, or ability on behalf of an agent which usually, but not necessarily, is expressed by the . . . subject of the sentence".

- 4 -

Below I list the modale in Danish, giving examples of their epistemic and

their root use. Only when relevant in this art.icle have I made a further dis­

tinetion between more than one epistemic o r bet.ween more than one root use. As shown by Davidsen-Nielsen (1988), many more such distinctions are possible.

(Note that the examples are listed as to their preferred interpretation. Almost

all may be ambiguous given the appropriate context.)

EPISTEMIC

( 5 ) a . vi..lle. Han vil overleve chocket

"He will survive the chock"

( 6) a. skulle ( report.)

Han skal tjene mange penge

"He is said to make a lot

of money"

b. skulle (promise)

Jeg skal nok rydde op efter mig

"I shall clean up after myself"

( 7 ) a . !llåtte.

Hun måtte sove tungt

"She had to be sound asleep"

( 8 ) a . ltu.nn.e

ROOT

b. vi..lle.

Han vil læse lingvistik

"He wants to read linguistics"

c. skulle Han skal gå i skole hver dag

"He must go to school every day"

b. !llåtte. (obligation)

Han må se l v tage ansvaret

"He must take the responsibili ty

himself"

c. !llåtte. (permission)

Hun må låne min bil "She may borrow my car"

b. kunne (ability)

Det kan let gå gal t Han kan ikke svØmme

"Things may easily go wrong" "He eannot swim"

(9)a. �

Det burde slå til

"That ought to be enough"

- 5 -

c. ltu.nn.e (permission)

Du kan låne min c')'kel "You may borrow my bicycle"

b. �

Han burde opfØre sig pænt

"He ought to behave"

(l O ) a . :tJ..u:de 2

Det turde være en overdrivelse

"It is probably an exaggeration"

( 11 ) a . *ltide.

b. :tw::de Han tØr ikke gå ud alene

"He dare not go out alone"

b. t.tilie. Han gider kun at se TV "He only feels like watching TV"

The root modals may be divided into two groups, deontic (Yil,k, akuJ.le,

o:åt:l&l , irurme (permjssion), b.mie) and non-deontic (kurme labilHyl, :tl.l.rde, gide_). The deontic modals may be further divided into those =ncerned with

obligation (Yil,k, akuJ.le, måtte (obligation), b.u::deJ and permission (o:åt:l&l ( reil!lissionl, kurme (permission)).

However, the main two groups to be distinguished below are one consisting of

deontic obligation modal s ( y:jj,k, l3.kl.ll.l.fl, måtte (obligation l , b.mie) and

another one consisting of all the others (måtte (ooil!lissionl, lrunne (oormis­ciQnl , kurme (abilityl' :tl.l.rde , gide_).

2 2 Differences between eoistemic and root moctals �

The five subsections below diseuse different distinctions between epistemic

and root modale . This is done mainly so that the :reader may see on what I base

the classification of a particular use of a modal as epistemic or root. This is

particularly relevant in the appendix.

In other words, all that sections 2 . 2 . l - 2. 2 . !5 amount to is a diseuseion o f

various tests for epistemic vs. root. Although it obviously is desirable, I

will not try to explain the data, but content myself with grouping them for

classificational purposes.

2 2 l Tenses Whereas epistemic modals eannot occur in perfect o r pas t perfect, root ones

may:

(12) a. *Han bar villet tjene mange penge

"He has been going to earn a lot of money"

b. Han hal:: yjllet tjene mange penge

"He has wanted to earn a lot of money"

However, Davidsen-Nielsen (1988:ch. 2, p. 26) lists the two following exam­

ples of eplstemles in the perfect ( 13) , which admi ttedly are not completely

unacceptable (my underlining and juclgments):

2 I agree with Skyum-Nielsen (1971:213) that the epistemic use of :tJ..u:de is archaic, and below I will simply aseurne that :tw::de has no epistemic use in modem Danish (as does Davidsen-Nielsen (1988:ch. 2, pp. 26-28)).

- 6 -

(13) a. ?Han bar skullet bo i Arhus

"He has been said to live in Arhus = He is said to have lived in

Arhus"

b. ?Der hal:: måske nok � være tale om en fejl

"There has maybe probably could be talk about a mistake

= It might have been a mistake"

He is therefore lead to say that epistemic modal s "do not normallv occur in the

past participle" (my emphasis).

In my view the perfect in (13), although clearly realised on the epistemic

modal s, really is the perfect o f the main verbs. One argument is that (13a, b)

mean exactly the same as ( 14a, b) , where i t is the main verb that cx:ccurs in the

perfect tense:

(14) a. Han skal have boet i Arhus

"He is said to have li ved in Arhus"

b. Der kan måske nok have været tale om en fej l

"There can maybe probably have been talk about a mistaks

= It might have been a mistake"

Another argument is based on the interpretation of (13a) when an adverbial like

"since 1983" is added:

(15) ?Han har skullet bo i Arhus siden 1983

"He has been said to li ve in Arhus since 1983"

The interpretation o f ( 15) clear ly is that his li ving in Arhus reportedly

started in 1983, and not that the reports of his living in Arhus started in

1983. In other words, the adverbial modifies the main verb, not the modal. As

an adverbial like "since 1983" can only modify a perfect tense, as c.an be seen

in (16) , I again find that, in spite of appearances, i t is the main verb that

is in the perfect rather than the modal in ( 13) and ( 15) .

(16) a. Han skal haye boet i Arhus siden 1983

"He is said to have lived in Arhus since 1983"

b. *Han skal w i Arhus siden 1983 "He is said to live in Arhus since 1983"

The two views make two different predictions w.r.t. the possibility of both the

modal and the main verb occurring in the perfect: If the modal is poesible in

the perfect, both verbe should be possible in the perfect, whereas i f the per­

fect somehow is misplaced from the main verb to the modal in the examples

a bove, then both verbe should not be possible in the perfect. To my ear two

perfects are not possible, and this is compatible only with the latter view:

- 7 -

(17) a. *Han har skullet have boet i Arhus

"He has been said to have li ved in Arhus."

b. *Der har måske nok kl.lno.e:t have været tale om en fej l

"There has maybe probably could have been talk about a mistake"

This analysis might suggest that the occurrenc:e of epistemic modals in per­

fect o r past perfec:t is not prevented by a syntactic condit ion as such, as ( 13)

does occur, but rather a restrietion of a different kind, perhaps a sernantic

one. 3

2 2 2 Empty categories as subiects Epistemic modals eannot have PRO as subjec:t (�orhether it is arbitrary PRO, as

in ( 18) , or not, as in ( 19)) , whereas root modalE; can:

(18) a. *Ilet er behageligt PRO at ville tjene mange penge hurtigt

"It is pleasant to be going to earn a lot of money quickly"

b. Det er moderne PRO at ville tjene mange penge hurtigt

"It is fashionable to want to earn a lot of money quickly"

(19) a. *Han drØmte om PRO at kunne være reJst tH London

"He dreamt that he might have gone to London"

b. Han drØmte om PRO at kunne svØmme

"He dreamt about being able to swim"

Both kinds may have a !ID-trace as subject,

( 20) a. Hvem t kan have sendt de 50 rØde roser?

"Who might have sent the 50 red roses?"

b. Hvem t kan svØmme over Kanalen?

"Who is able to swim across the Channel?"

whereas the pieture is less clear when i t comes to havi.ng an NP-trace as sub­

ject (i.e. being embedded under a raising verb). Here the epistemic seems to

vary somewhat, (21a,b):

3 In the framework of Vikner (1985:93-96), such a restrietion could be said to rule out interpretations of sentences involvir�g epistemic modals where the event point precedes the second point of reference. Of course the question of why these representations should be rul ed out is still unanswered. Also unans­wered is the question of why the perfec:t of the epistemic in (13) is better than the one in ( 12a) .

- 8 -

(21) a. Det ser ud til t at ville blive godt vejr i eftermiddag

"It seems to will become nice weather this afternoon =

This afternoon seems to be going to be nice"

b. *Hun forekommer mig t at måtte have sovet under forelæsningen

"She appears to me to must have slept during the lecture"

c. Han ser ud til t at ville gØre alting på en gang

"He seems to want to do everything sirnul taneously"

2 2 3 Ordering and combination Epistemic modals may not be embedded under root ones, (22b), whereas root

modale may be embedded under epietemic ones, (22a):

( 22) a . De flkal Yill.e bygge et hus

"They are said to want to build a house"

b. *De ill gerne� have tjent en milion"

"They would like to be said to have made a million"

A modal eannot be combined with iteelf, not even in ite two different read­

inge:

( 23) a. *Han flkal � gå til lægen hver uge

"He is said to have to see a cloetor every week"

b. *Hun kan � svØmme længere end du tror

"She might be able to swim further than you think"

As for the combination of two epistemic, (24a-d), or two root modals,

(24e-h), i t seems that both are possible, but only if the sec:ond of the two

modale is � (Cf. also Østkjær Jense.n ( 1987: 28) and the ex. in Thråinsson

(1986:262, n13)):4

4 Though this kl.Jl:lne , as well as the one in the following example is tmdoub­tedly epistemic, i t seems that i t may appear in the perfec:t tense , cf. sec:tion 2.2.1. I have no satisfactory explanation to offer:

(i) a. Det måtte have kunnet stå på en side "It must have been poesible to fit it onto one page"

b. ? ?Det måtte have kunnet have stået på en side "It must have been poesible to have fitted it onto one page

( ii) a. Der ville let have kunnet gå noget gal t "It would easily have been poesible that something went wrong"

b. ??Der ville let bave kunnet være gået noget galt "It would easily have been poesible that something had gone wrong"

(iii) a. Det må have kunnet regne i dagevis i det gamle Rom, at dØmme efter ... "It must have been poesible for it to rain for days in ancient Rome,

judging from ... " b. ??Det må have kunnet have regnet i dagevis i det gamle Rom,

"It must have been poesible for it to have rained for days in ancient Rome, ... "

- 9 -

(24) a.

b.

c.

d.

e.

f.

g.

h.

Det må � stå på en side

"It must be possible to fit it onto one page"

Der Yil let kunne gå noget galt

"It will easily be possible that something goes wrong"

*Han Yil 6lmll.e have læst bogen

"He will be said to have read the book"

*Han 5kal ri.lle opfØre sig pænt

"He is said to be going to behave"

Han 5kal kunne svØJmJe for at få jobbet

"He must be able to swim to get the job"

?Hun Yil � forstå fransk

"She wants to be able to understand French"

*Han Yil turde gå op i Eiffeltårnet

"He wants to dare to go up in the Eiffel Tower"

*Hun 5kal gide at gå på indkØb

"She must feel like doing the shopping"

2 2 4 Pronominal comolemente This is actually not a difference between epistemic and root modals, as I

disagree with Platzack (1979:46), who claims that only root modals may have det

("it") as complement (in Swedish). I do not find this to be the case for

Danish, both kinds may have this complement. The examples show that epistemic

modals do indeed have "it" as a (substitute for t.heir) complement:

(25) a. Han vil være hjemme hele dagen. M vil hun desuden også.

"He will be at home all day. So will she. "

b. Han skal være rejst til London. - Nå, dm< skal han? Jeg har ellers lige

set ham.

"He is supposed to have gone to London. - Oh, he is? I just saw him."

c. Jeg lover Dem, aviserne skal afslØre denne valgsvindel. Og � skal de

elektroniske medier også.

"I promise you that the papers will revectl this election fraud. And so

will the electronic media"

d. Klokken må have været g<Xl.t halv ti. - Ja, det må den vel.

"It must have been araund 9.30 - Yes, I guess it must have."

e. De må have været te�m�elig berusede. Og det må vi egentlig også.

"They must have been rather drunk. And so must we . "

f. Kan brevene være kOI!Illet allerede i mandags? - Ja, de!. kan de g<Xl.t.

"M:ight the letters have arrived already �!onday? - Yes, they might."

g. Pengene burde række. - Nå, så de!. burde de?

"The money should be enough. - Oh, they should?"

- 10 -

(see also Hansen (1972:24) and Østkjær Jensen (1987:53)).5

2 2 5 Pseucio-clefts Root modals may occur in this construction, (26), epistemic ones eannot,

(27) (cf. Thrainsson (1986:255), Davidsen-Nielsen (1988: ch. 2, p. 13)):

(26) a.

b.

c.

d.

e.

f.

g.

h.

i.

Det eneste han Yil er at svare på sPØrgsmålet

"The only thing he wants to is answer the question"

Det eneste han 5kal er at svare på sPØrgsmålet

"The only thing he has to is answer the question"

Det eneste han absolut må er at svare på sPØrgsmålet

"The only thing he absolutely must to is answer the question"

Det eneste han g<Xl.t må er at låne min cykel

"The only thing he may is borrow my bicycle"

Det eneste han godt kan er at låne min cykel

"The only thing he can is borrow my bicycle"

En af de ting han ikke kan er at svØmme over Kanalen

"One of the things he eannot is swim across the Channel"

En af de ting han absolut w er at svare på sPØrgsmålet

"One of the things he absolutely ought to is answer the question"

En af de ting han ikke � er at svØmme over Kanalen

"One of the things he dare not is swim across the Channe l"

En af de ting han ikke � er at svØmme over Kanalen

"One of the things he doesn't feel like is swim across the Channel"

5 For some unknown reason, det has to be in topicalised position (Specifier of CP), as in the (a) examples below, and eannot occur in normal object posi­tion, cf. the (b) examples below. This is different from main verbs, (iv), and al so from the main verb use of � ( "can") , ( iii) :

(i) a. Det skal han ikke b. *Han skal det ikke "It shall/is-said he not" "He shall/is-said it not"

(ii) a. Det kan han ikke b. *Han kan det ikke "It may/can he not" "He may/can it not"

(iii) a. Det kan han ikke b. Han kan det ikke "It knows he not" "He knows it not"

(i v) a. Det hØrte han ikke b. Han hØrte det ikke "It heard he not" "He heard it not"

- 11 -

(27) a. *Det han vil i 100rgen er at tabe kampen om mesterskabet "What he will tomorrow is lose the fight about the charopionship"

b. *Det han efter sigende skal er at stemme på de Konservative "What he is said to is vote for the Coru3ervatives"

c. *Det han nok skal er at rydde op efter s.i.g en anden gang "What he shall is to clean up after himself another time"

d. *Det hun måtte var at have sovet under foredraget "What she must was have slept during the talk"

e. *Det hun kan er at have sovet over sig "What she might is to have overslept"

f. *Det hun l::urde var at have vidst det "What she ought to was have known it"

3 Analysis

3 l 9-roles and Additional B-roJes

Thematic roles, or G-roles are roles that are assigned by verbs, preposi­tions, or adjectives, to argument NPs. Examples of G-roles are agent, theme, goal, source, experiencer. The assignment of 9-roles is constrained by the G-criterion, which says a) that each 8-role must be aBelgned to one and only one argument, and b) each argument must receive one and only one G-role.

This has often been IOOdified by the addition of other kinds of roles to the ones constrained by the G-criterion, e.g. Zubiza:rreta (1982) or Grimshaw (1986), to mention but a few.

Here I want to suggest the existence of what I will call an "additional G-role". The idea is that this kind of 8-role is aBelgned by obligation root IOOdale and also by få ( "get"), � ( "become"), and one of the two types of lwmme ( "come").

This additional G-role may be assigned to an argument that alr�eady has a G-role, cf. the idea in Zubizarreta (1982:41,123) that 8-roles exist that are invisible for the G-criterion. However, as opposed to Zubizarreta, I want to suggest that one additional G-role may be assigned to euch argument, BUT NCYI'

TWO . The intuition is that two additional 9-roles would give the argument too

much "8-burden", much like an argument eannot receive 100re than one normal G-role, which again suggests that additional G-n)les are not completely invisible to the G-criterion.

Thus one half of the G-eri terion see.ms to hold completely for addi tional 8-roles as they IWBt be assigned to one and only one argument. The other half of the G-criterion only holds halfway, as an argument may not receive 100re than one additional G-role (in addition to a normal 8--role), but i t may receive less than one additional G-role. Furthermore, as will be discussed in connection wi th (57) , there is a case where an argument would seem to be grammatical �dthout any normal G-role, surviving on an addit:Lonal El-role only.

The three verbe that assign an additional G-n)le (apart from the obligation root modale) , få ( "get" ) , � ( "become" ) , and one type of ko.mw.e ( "come" ) , are

- 12 -

all event verbe that are the counterparts of the state verbe l:laYe ("have"), � ( "be"). The difference between the two sets is thus that the event verbe ass:ign an addi tional 9-role to their subject, and the stat.e ones do not. This corresponds to the intuition that the event expressions have all the irnplica­tions of the state ones plus some 100re.

(28) a. Han har tre biler "He has three cars"

(29) a. Han er professor "He is a professor"

(30) a. Han er i London "He is in London"

3.2 Raising or Control

b. Han får tre biler "He gets three cars"

b. Han bliver professor "He becomes a professor"

b. Han k ollliller til London "He arrives in London"

In accordance with the thematic properties set out above, i.e. that IOOdals do not assign normal G-roles, I will follow Zubizarreta's (1982:133--34) sugges­tion (given for French) that IOOdals are like raising verbs: The subject of a IOOdal is base-generated somewhere else, and moved into the subject position.s (31a) is an example of a raising verb and (3lb) of a modal (in this case ambiguous w.r.t. epistemic/deontic)

r-0---, (31) a. Hani ser ud til ti at more sig

"He seems to enjoy himself"

,o-, b. Hani skal ti 100re sig

"He is said to/must enjoy himself"

An alternative would be to analyse modale as parallel to control verbs:

,-G� ,-G---, (32) Hani lovede hende PROi at more sig

"He promised her to enjoy himself"

It should be noted that raisins verbs and modale are different in important ways, e.g. in that raisins verbs allow exPletive subJects, and modalB do not:

( i) Det ser ud til at han morer sig "It seems that he amuses himself"

( ii) *Det skal at han morer sig "It is said/must that he amuses himself"

- 13 -

AB my analysis in the following sections is built on a prohibition against assigning more than one addi tional G-ro le to an argument, I have to adopt the raising analysis, where there is only one argument, the chain in (33a), whereas a control analysis would have three arguments, as. jn (33b), the chains headed by hun, by the firs t PRO, and by the eecond PRO:

,-CGJ1 ,ceJ1 ,-e----, (33) a. *Huni vil ti blive ti arresteret ti

1CGJ1 ,-(9)1 ,-e----, b. *Huni vil ti blive ti arresteret ti

"She wants to become arrested"

(AB discussed in section 4, this is an ungrammatical sentence, and I a=ount for the ungrammaticality by assuming that (33a) is the right analysis, and that at most one !l-role (as assigned by arresteret) and one additional 1;1-role (as assigned by bliYe_, and also by ill in its root sense) may be assigned to one chain. (33a) violates this as two additional !l-roles are assigned to the chain, but if the analysis of (33b) was on the right track, I could not rule out the sentence.)

I have no independent evidence to offer at present. The characteristic dif­ference between raising and control is that in raising constructions like (31a) "he" has one role, the one of "amuser of himself"' which is assigned to the trace by "amuse", whereas in (32) "he" has two roles, one of "promiser of some­thing to her", which is assigned to "he" by "promise", as well as one of "arnuser of himself" which is assigned to PRO by '"amuse". Thanks to the coreference (the so-called "control") between "he" and PRO in (32), the referent of "he" ends up with both these two roles. When comparing this to the modal construction, (31b), it is not so clear whether the subject of the modal ende up with one role, that of "amuser of himself", or two, "amuser of himself" as well as "someone whom something is reported of"/"someone who must some­thing". Indeed, the analysis that obligation root modale assign additional G­roles is based on the intuition that there is such a role as "someone who must BOIIlething", as this is exactly what is represented by the additional 9-role assigned by flklJ.lle in its root use in (31b) = (34,a).

To sum up, my suggestion ( which admi ttedly is not based on strong indepe dent evidence) is that a main verb, like IIIQre ( "amuse") assigns a real G-role, whereas an obligation root modal assigns an additional G-role, as in (34a) (and so does få ( "get"), bliYe. ( "become"), and one of the two types of l:wll:lr.e ( "come" ) ) , and a modal which is not an obligation root one, does not assign any 9-role whatsoever, as in (34b):

- 14 -

,-(O), ,�;�, (34) a. Han! skal ti more sig

"He must enjoy himself"

,-G, b. Han! skal ti more sig

"He is said to enjoy himself"

4 Data

(Obligation root)

(Epistemic)

The hypothesis presented in section 3. l, that no more than one "addi tional 9-role" rnay be assigned to an argument, makes the prediction that none of the different kinds of verbs that assign such an additional G-role role rnay be com­bined. AB for combinations of (obligation) root modals with each other this has

already been mentioned above, (24g,h), and will not be discussed further.7 Below I will concentrate on combinations of obligation root modals with the three other verbs mentioned in section 3, U ( "get"), bliYe. ( "become"), and koiJ:Ime ( "come"). These data were originally brought together and discussed in

Hansen (1972), though some of them, especially the bare adverbial facts, also had been notedin earlier works (cf. the references in Hansen (1972)).

The constructions in the left column in ( 35) eannot occur embedded under an obligation root modal, and they are replaced by the ones on the right:

(35) a. få NP ("get NP") b. blive AP ( "become AP")

c. komme Adv ("come Adv") d. blive V-t ( "become V-en")

Examples of this are

(36) epistemic Yille e vent

a. Hun vil få tre biler i 1990

"She will get three cars in 1990"

b. Hun vil blive klog "She will become wise"

c. Hun vil komme hjem "She will come home"

d . Hun vil b li ve arresteret "She will become arrested"

by by by by

have NP ( "have NP" ) være AP ( "be AP")

Ad v V-e.

state Hun vil have tre biler i 1990

"She will have three cars in 1990"

Hun vil være klog "She will be wise" Hun vil være hjemme "She will be home" Hun vil være arresteret "She will be arrested"

7 As pointed out by Jane Grimshaw, this means that my analys is provides a reason for the impossibility of combining an obligation root modal with another one, but not for the impossibility of combining an epistemic modal with another one. As the restrietions seem to be the same in the two cases, cf. section 2.2.3, this is not completely satisfactory.

- 15 -

(37) root Yille (obligation)

e vent

a. Hun vil have tre biler i 1990

"She wants to get three cars in 1990''

b. Hun vil være klog

''She wants to become wise··

c. Hun vil hjem

"She wants to (go) home··

d. Hun vil arresteres

''She wants to become arrested''

stateB

Hun vil have tre biler i 1990

"She wants to have 3 cars in 1990"

Hun vil være klog

"She: wants to be wise"

Hun vil være hjemme

"She wants to be home"

Hun vil være arresteret

"She: wants to be arrested"

This means that for the first two canatruetions there will be ambigui ty

statejevent in the obligation root construction, (37a, b) & (39a, b), as opposed

to the epistemic canatruetion w i th the same verbs, ( 36a, b) & ( 38a, b) , and the

non-obligation root construction, (40a,b), in all of which there is a dif­

ference between the state and the event. For the (c,d) constructions, the state

and the event are distinct throughout.

(38) epistemic måtte

e vent state

a. Hun må få tre biler i 1990

"She must get three cars in 1990"

b. Hun må blive klog

"She must become wise"

c. Hun må kODJDe hjem

"She must come home"

d. Hun må blive arresteret

"She must become arrested"

Hun må have tre biler

"She must have three cars in 1990"

Hun må være klog

"She must be wise"

Hun må være hjemme

"She: must be home"

Hun må være arresteret

"She must be arrested"

s Thoush the first reaction of many native speakers may be that this is not so, obligation root modale are actually perfectly· compatible with state inter­pretations as shown by e . g.

(i) Jeg vil/skal være i London den 21. marts "I want to/must be in London on March 21st"

- 16 -

(39) root � (obligation)

e vent

a. Hun må have tre biler i 1990

"She must get three cars in 1990"

b. Hun må være klog

"She must become wise"

c. Hun må hjem

"She must (go) home"

d . Hun må arresteres

"She must become arrested"

state

Hun må have tre biler i 1990

"She must have three cars in 1990"

Hun må være klog

"She must be wise"

Hun må være hjemme

"She must be home"

Hun må være arresteret

"She must be arrested"

( 40) root måtte (gerne) (non-obligation,

e vent

i.e. permission)

a. Hun må (gerne) få tre biler i 1990

b.

"She is allowed to get 3 cars in Hun må (gerne) blive klog

"She is allowed to become wise"

c. Hun må (gerne) komme hjem

"She is allowed to come home"

1990"

d. ?Hun må (gerne) blive arresteret

"She is allowed to become arrested"

state

Hun må (gerne) have tre biler ...

"She is allowed to have 3 . . . " Hun må (gerne) være klog

"She is allowed to be wise"

Hun må (gerne) være hjemme

"She is allowed to be home" Hun må (gerne) være arresteret

"She is allowed to be arrested"

The rest of this section is just going to be a list of the same canatrue­

tions with the other modals. In other words, you may wish to skip to p. 19.

(41) epistemic ekulle e vent

a. Hun skal få tre biler i 1990

"She is said to get 3 cars in 1990"

b. Hun skal blive klog

"She is said to become wise"

c. Hun skal komme hjem

"She is said to come home"

d . Hun skal b li ve arresteret

"She is said to become arrested"

(42) root ekulle (obligation)

e vent

a. Hun skal have tre biler i 1990

"She must get three cars in 1990"

b. Hun skal være klog

"She must become wise"

c. Hun skal hjem

"She must (go) home"

d. Hun skal arresteres

"She must become arrested"

- 17 -

state

Hun skal have tre biler i 1990

"She is said to have 3 cars

Hun skal være klog

"She is said to be wise" Hun skal være hjemme

"She is said to be home"

Hun skal være arresteret

"She is said to be arrested"

state

Hun skal have tre biler i 1990

"She must have three cars in 1990"

Hun skal være klog

"She must be wise"

Hun skal være hjemme

"She must be home"

Hun skal være arresteret

"She must be arrested"

(43) epistemic �

event

a. Hun kan få tre biler i 1990

"She may get three cars in 1990''

b. Hun kan blive klog

"She may become wise"

c. Hun kan komme hjem

"She may come home"

d. Hun kan bl i ve arresteret

"She may become arrested"

state

Hun kan have tre biler

"She may have three cars in 1990"

Hun kan være klog

"She may be wise"

Hun kan være hjemme

"She may be home"

Hun kan være arresteret

"She may be arrested"

(44) root � (non-obligation, i.e. permission and/or ability)

event

a. Hun kan få tre biler i 1990

"She can get three cars in 1990"

b. Hun kan blive klog

"She can become wise"

c. Hun kan kOIIlllle hjem

"She can come home"

d. ?Hun kan b li ve arresteret

"She can become arrested"

(45) epistemic burde

e vent

a. Hun bØr få tre biler i 1990

"She ought to get three cars in 1990"

b. Hun b;Sr blive klog

"She ought to become wise"

c. Hun bØr komme hjem

"She ought to come home"

d. Hun bØr blive arresteret

"She ought to become arrested"

(46) root burde (obligation)

event

a. Hun bØr have tre biler i 1990

"She ought to get three cars in 1990"

b. Hun bØr være klog

"She ought to become wise"

c. Hun bØr hjem

"She ought to (go) home"

d. Hun bØr arresteres

"She ought to become arrested"

- 18 -

state

Hun kan have tre biler i 1990

"She can have three cars in 1990"

Hun kan være klog

"She can be wise"

Hun kan være hjemme

"She can be home"

Hun kan være arresteret

"She can be arrested"

statE,

Hun bØr have tre biler i 1990

"She ought to have 3 cars in 1990"

Hun bØr være klog

"She ought to be wise"

Hun bØr være hjemme

"She ought to be home"

Hun bØr være arresteret

"She ought to be arrested"

statE,

Hun bØr have tre biler i 1990

"She ought to have 3 cars in 1990"

Hun bØr være klog

"She ought to be wise"

Hun bØr være hjemme

"She ought to be home"

Hun bØr være arresteret

"She' ought to be arrested"

(47) root turde (non-obligation)

e vent

a. Hun tØr få tre biler i 1990

"She dare get three cars in 1990"

b. Hun tØr blive klog

"She dare become wise"

c. Hun tØr komme hjem

"She dare come home"

d . Hun tØr b li ve arresteret

"She dare be=me arrested"

(48) root gide (non-obligation)

e vent

a. Hun gider (godt) få tre biler i 1990

"She does not mind getting 3 ... "

b. Hun gider (godt) blive klog

"She does not mind becoming wise"

c. Hun gider (godt) komme hjem

"She does not mind coming home"

d. Hun gider (godt) blive arresteret

"She does not mind becoming arrested"

state

Hun tØr have tre biler 1 1990

"She dare have three cars in 1990"

Hun tØr være klog

"She dare be wise"

Hun tØr være hjemme

"She dare be home"

Hun tØr være arresteret

"She dare be arrested"

state

Hun gider (godt) have tre biler ..

"She does not mind having 3 ... "

Hun gider (godt) være klog

"She does not mind being wiae"

Hun gider (godt) være hjemme "She does not mind being home"

Hun gider (godt) være an-es te ret

"She does not mind being arrested"

5. State Exoressions Reolace Eyent Expressions: få + NP, blive t .AE.� Eå and bliY.e are ungran:matical embedded under obligation root modals as

would be predicted:S

(49) a. *Hvis du Yil få din lØn udbetalt kontant, så ...

"If you want to get your salary paid out in cash, then ... "

b. *Han 6kal bliY.e nummer et hver gang, ellers bliver han sur

"He must become number one every time, otherwise he gets cross"

c. *Han Yil bliY.e fotograferet hele tiden (OK: a-passive, cf. 6.2)

"He wants to be photographed all the time"

Instead, it is poesible to have � and � embedded under an obligation

root modal:

(50) a. Hvis du Yil � din lØn udbetalt kontant, så ...

"If you want to have your salary paid out in cash, then ... "

b. Han 6kal � nummer et hver gang, ellers bliver han sur

"He must be number one every time, otherwise he gets cross"

s Cf. also the appendix on p. 29 about Hansen's (1972) so-called "interessee" -constructions .

- 19 -

This presents a serious conceptual problem: Not only is it poesible to sub­

stitute state expressions for event expressions, tut the state expressions also

take on event meaning, as can be seen when comparing (50) with (49a,b). If

event meaning follows from the additional G-role suggested in section 3.1, then

why is it not the event meaning that is ilDpoesible with obligation root modale,

tut only the event expressions? I have no satisfatctory answer to this ques­

tion. 10

Note that it is not the case that the state expressions are the "unmarked"

expressions that may be used both for events and continui ties and that the

event expressions are somehow "marked": This would predict that the state

expressions could have an event interpretation in all the constructions, not

only when embedded under an obligation root modal. This is not the case

((51a,b) are epistemic modals, (51c) is a deantic non-obligation one, and (51d)

is a non-deantic (and therefore also non-obligation) one) :

(51) a. *Hvis ikke du skriver til forlaget, Yil tx:>gen ikke Yæi:e trykt

"If you do not write to the p..�blisher, the book will not be printed"

b. *Hun lOkal have � millionær på bare ft:>.m år

"She is said to have become a millionah-e in only five years"

c. *Hvis du er interesseret, kan du haJle din lØn udbetalt kontant

"If you are interested, you may have your salary paid out in cash" d. *De gidei: ikke Yæi:e hjem

"They do not feel like coming (or going) home"

6 Soocial Constructions: Motion and Passive

6 l Ko!mne ys Bare Adverbials of Direction

The situation w. r. t. lmmme is more complicated than the one w. r. t. få and

blive deseribed above. It is poesible to find kQmwe embedded under obligation

root modale :

10 From a comparative point of view, it is in��resting to note the following in Swedish: Although the event oonatruetions themselves are not impossible in the way they are in Danish (cf. ( 49) ) , the eta te expressions may take on event interpretation in exactly the same cases as in Danish: if and only if embedded under obligation root modale. The same goes for the construction with the bare adverbial of direction, it is also poesible in �{actly the same con�s as in Danish. (As for the f!-passive, it is much more widespread in Swedish anyWay and is therefore poesible not only in the same contexts as in Danish but also e.g. after epistemic modale) . This seerne rather difficul t to account for, tut per­haps a poesible analysis could be to simply say t.hat the event expressions are not analysed as assigning an addi tional G-role in Swedish. The question of why the state expressions may take on event interpretation in certain cases and not in others is left unanswered as before, but is now even more o f a mystery.

- 20 -

( 52) a. Han 6kal lromm.e

"He must come"

b. Hun Yil kQmme hjem hver dag

"She wants to come home every day"

c. Han må lmmme. til London ØJeblikkeligt

"He has to come to London right away"

This may be a different kQmme, though, as it is ungrammatical with an adverbial

of direction that has a figurative reading:

(53) a. *Han 6kal kQmme igang

"He has to get started"

b. *Hun Yil kQmme ind i Folketinget

"She wants to become a member of parliament"

c . *Han må lromm.e op på tæerne

"He must get to be on his toes"

I shall therefore suggest the following: There are two types of komme, which we

might call an auxiliary lromm.e and a main verb kQmme.

The auxiliary kQmme does not occur on its own (as in (52a)), i t only occurs

with adverbials of direction, both "literal", as in (52b,c) and figurative

ones, as in (53). It assigns an additional Q-role to its subject. It may not

occur embedded under obligation root modals, cf. that ( 52b, c) and (53) are

grammatical without kQmme (though with a slightly different interpretation for

(52b,c), which may be due to deixis, cf. the diseuseion of (56) below).

The main verb kQmme does not occur with figurative adverbials of direction,

tut only on its own or with "literal" adverbials, as in (52). It assigns an

ordinary G-ro le to i ts surface subject. 11 The main verb lmmme. may oc--cur

embedded under obligation root modal s , cf. (52) .

That lmmme. also exists as an auxiliary may be seen as supported by two sets o f facts of a more morphological nature. One is that i t turne up as a .. real"

auxiliary in some languages: Italian has a passive formed with venire , French

has a pas t tense formed w i th .lleniJ:: , and Swedish forms future w i th kQill!!lfl att :

11 The surface subject of lmmme (which is an ergative verb) is basegenerated as an object, whereas the surface subject of an intransitive verb like � ("sleep") is basegenerated as a subject (cf. Vikller & Sprouse (1988) and the references cited there). This distinction is not crucial for the present analy­sis.

- 21 -

(54) a. I t. Il libro viene letto

''The book comes read = The book is read"

b. Fr. Il vient de mourir

"He comes from dyi.ng = He just died"

c. Sw. Han kommer att lasa boken

"He comes to read the book = He will read the book"

The other (perhaps more sp;trious) argument supporti.ng that auxiliary ls.o.DJr.le is

somehow related to the ( other) two event expressions få and � is the fol­

lowi.ng:

(55) a. få is

b. bliYe is

""bekQmmen''

''beQQme''

in German

in English and "deYenir" in French

The above analysis with its suggestion that two kinds of ls.o.D:m:le exist, an

auxiliary and a main verb, makes the prediction that (main V) kOl!llle should be

possible embedded under an obligation root modal whenever a bare place adver­

bial is poesible there. There are many exceptions from this, but they may

presumably all be accounted for by the well-known deictic requirements on

k.OJ!Ille, i.e. the movement deseribed by kOJ!Ille must be towards a place where

either the speaker or the hearer is (or may be) situated, cf. e.g. Fillmore

(1975:50ff). This means that either the hearer or the speaker must be in London

for (56a) to be a=eptable, whereas there is no such requirements on (56b):

(56) a. Han vil geme komme til London

"He would like to come to London"

b. Han vil geme til London

"He would like (to go) to London"

Thus, with the arnendment that there are two kinds of ls.Qmme , the analysis is

able to account for the ungrammaticality of kQwme under obligation root modale

in a way parallel to the account concemi.ng få, and bliYe . As for the bare adverbial of direction embedded under an obligation root

modal, it is grammatical as predicted, as it presumably does not assign any

addi tional Q-ro le. '!'hus i t is not poesible if embedded under a modal which is

not an obligation root one, (57a),l2 presumably because it assigns no Q-role at

12 This is different from German, where the bare adverbials occur after all root modals, inelucting the ones meaning permission and not obligation:

(i) , dal3 Peter nur ein Mal im Jahr nach Frankfurt k&ln/da.d/� ", that Peter canjmayjwould like tojneed (go) to Frankfurt only once a year"

As for the other constructions, Østkjær Jenser, (1987:116-19,157,187) reports that læiu ( "be") may never replace � ( "becorne"), whereas llaben ("have") often is an acceptable al temstive to bekoomen or � (both "set"), whether it is embedded under an obligation root modal like mLl.l3en ("must") or under non­obligation ones like .dl1!::ff:n ( "may") or brauchen ( "need").

- 22 -

all. Cf. that it never occurs without a modal either, (57b):

(57) a. *fu kan hjem med det samme

"You may home right away"

b. *Jeg så dem hjem

"I saw (=watched) them home"

(OK:gå hjemlkO!Wle hjem) "go hornejeorne home"

(OK:gå hjem/kOJ!Ille hiem) "go hornejeorne home"

The intuition is that as the bare adverbial assigns no 9-role, the subject

of (57a) receives no 9-role at all, hence the ungrammaticality. This implies

however, that in the cases of obligation root modals with bare adverbials (e.g.

(56b)), it is poesible for an argument to survive even though it is only bei.ng

assigned an additional Q-role and no normal one.

(57b) is ungrammatical for a parallel reason, dem receives no G·-role.

Although e! assigns a 9-role to its complement, the role of "what is seen",

thi s ro le is assigned to the whole consti tuen t dem h Jem, cf. that in

(58) Jeg så dem danse

"I saw them dance"

dem has the role of dancer(s), and dem danse the role of what is seen. In (58)

dem thus receives a G-role from d.!JJ:l.f!e, but dem in (57b) can get no 8-role from hjem, as the bare adverbial assigns no tl-role.

6 2 Blive-passive vs s-passive As far as embeddi.ng under root obligation modals are concemed, the

a-passive is parallel to the bare adverbial case: it is grammatical as

predicted, as it does not assign any additional 9-role.

The potential embeddi.ng of the a-passive under non-obligation modals may

present more of a problem. These facts are rather different from the bare

adverbial case in that the a-passive may occur on i ts own, and therefore eannot

be assumed to be "Q -deficient" in the way assumed for the bare adverbial of

dirRetion in the previous section:

(59) a. Nye opdagelser � hver dag

"New discoverles are made every day"

b. I England� grØntsager altid for længe

"In England vegetables are always boiled for too long"

Nonetheless, the a-passive is ungrammatical with epistemic modals:

(60) a. *Han vil udnævnes til professor, men han ved det ikke endnu

"He will be appointed professor, but he does not know yet"

b. *De skal forsendes hver for sig selvom det ikke er nØdvendigt

"They are said to be sent separately even if it is not necessary"

- 23 -

(cf . that Skyum-Nielsen ( 197 1 : 73 ) in his corpus has found two examples which he

qualifles as peculiar ( "ejendommelige" ) ) .

Furthei1DOre the a-passive breaks the pattern completely, as it may occur

embedded under two non-obligation root mcxials , lwnne. and måtte (oeuniseion) : 1 3

( 6 1 ) a . Pakken kan � efter klokken tre

"The parcel can be pieked up after three o 'clock"

b . Pakken må ikke � fØr juleaften

"The present may not be opened before Ch.ristmas eve"

It may be interesting that these two non-obligation root modale are the two

deontic non-obligation ones , whereas the two non-obligation root modale which

do not seem to allow the a-passive , turde and gide , are not deontic but what

has been called "dynamic" ( cf . e . g . Davidsen-Nielsen ( 1988 : ch . 3 , p . 2 ) :

( 62 ) a . ? ?Hun tØr ikke hejses op i flagstangen

"She dare not be hoisted up on the flagpole"

b. ? ?Han gider ikke klippes hver måned

"He doesn ' t feel like having his hair cut every month"

Both of these are fine with the bliYe-passive .

Summing up this section , we might say that the questions of why the embed­

dinge that were poesible under obligation root modale are not poesible under

other modale remain less than satisfactorily answered as regards the a-passive

facts , which do not follow the lines that have emerged so far , i . e . the parti­

tion between the obligation root modale on one side , and all the other modale

on the other .

6 3 Perfect Infinitlyes

When placed in a context that requires a perfect infinitive of the other

embedded verbs , the bare adverbial of direction and the a-passive are ungram­

matical :

( 63) a . *Han .alrulJ.e have jgang for længe siden

"He should have got started long ago"

b . *Hun Yille haYe ind i Folketinget "She wanted to have become a member of parliament"

c . *Han .ekulle haYe arreateredes "He should have been arrested"

1 3 In (61 ) the bliYe-passive is not very good at all , cf . also ( 40d) and ( 44d) . With blive hentet and blive åbnet the examples seem to only really have an epistemic interpretation .

- 24 -

r · • J ' · ,, " ' J � �

Thi& may be because the bare adverbial and the a-passive have no past

participle . Cf . the present tense version of ( 63 ) , where a past participle is

not necessary :

( 64) a . Han llkal jgang nu

"He must get started now"

b . Hun Yil ind i Folketinget "She wants to become a member of parliament"

c . Han llkal arresteres "He must be arrested"

In the cases where a past participle is necessary , the bare adverbial of

direction and the a-passive are then replaced with constructions with været ( "been " ) , as appears when comparing ( 63) w i th the following ( I believe that

this observation was firet made by Jespersen ( 1932 : 206 ) ) :

(65) a . Han .ekulle have været igang for længe siden

"He should have got started long ago"

b . Hun Yille have været ind i Folketinget

"She wanted to have become a member of parliament"

c . Han .alrulJ.e have været arresteret

"He ehould have been arrested"

Even if the ungrammaticality of (63) is accounted for by the (morphological )

unavailability of a past participle , another queetion remains : Given that være may occur in the cases where a perfect infinitive (of the bare adverbial or of

the a-passive ) is called for , why can være not also appear in the present

tenee : 1 4

( 66) a . *Han llkal være jgang nu

"He must get started now"

b . *Hun Yil være ind i Folketinget

"She wants to become a member of parliament"

c . *Han llkal være arresteret

"He must be arreeted"

I have no answer to this , but i t is interesting to note the following : The

event interpretation o f the eta te expression ( i . e . the one w i th haYe ( "have" )

or være ( "be" ) ) is poesible in all cases bt the two where there is an al tema­

ti ve way (of expreseing an event) which is not also interpretable as a etate :

the movement construction in present tense , as in ( 66a, b) where the unambiguoue

1 4 ( 66a , c ) are fine with state interpretation , but it is the event inter­pretation that we are concerned w i th here . ( 66b) eannot be interpreted as a eta te because o f the directional adverbial , .l..nd.__Ul ( " in to" ) , cf . section 7 . 5 , on p . 28 .

- 25 -

alternative is a bare adverbial of direction , ( 64a , b) ; and the passive con­atruetion in present tense , as in ( 66c) where the unamb:iguous al ternat i ve is

the a-passive , ( 64c ) . The same oonatruetions with a perfect infinitive , (65)

and ( 63 ) , are fine with the state expression having event interpretation , and

so are the two other oonatruetions (with haYe + NP replacing få + NP , and with Ylll.I:e + AP replacing b1.iYe + AP) wi th a present as well as w i th a perfect

infinitive .

7 Sqme Other Conseauences Below will be listed six properties that haYe (or Yænl) show when embedded

under an obligation root modal , the ( c ) examples , oot not elsewhere , the ( a )

examples . As the same properties are regular properties o f få ( o r bliYe ,

�) , cf . the ( b ) examples , they are accounted. for by the analysis preseted so far . It is also correctly predicted that the baYe ( or Ylll.I:e) examples with

these properties only have event interpretation .

7 l Idioms (Hansen ( 1972 : 26 ) )

� may not normall y replace få in idioms ( 67a, b ) , oot embedded under an

obligation root modal it not only may , ( 67c ) , it actually must , ( 67d ) . The

analysis predicts that få is out in combination Hi th obligation root modale , and IlllSt be replaced by haYe .

(67) a . *Han har altid sin vilje

''He always has his way"

b. Han få!.:: altid sin vilje

"He always gets his way"

c . Han Yil a l tid haYe sin vilje

"He always wants to have his way''

d. *Han Yil altid få sin vilje

"He always wants to get his way"

Other expressions : få.....taly ( "get beaten up" ) , �.n:nJ.yd ( "obtain Bilence so

that one may be heard" ) , få en overraskelse ( "have a surprise" ) , etc .

7 2 Source phraaes � does not normally allow a phrase ( outside the object NP) that gives the

source of the object, ( 68a ) , whereas få does ( 68b ) . Embedded under an obliga­

tion root modal , however , haYe does allow a source phrase, ( 68c) . This is as

predicted by the analysis , få is out in cernbination wi th obligation root

modale , ( 68d ) , and must be replaced by ha.Ye, and only in these circumstances

does haYe behave exactly as få , e . g . i t allows source phrases .

- 26 -' ,l

( 68) a . *Hun haYde ingen penge af sine forældre "She had no money of her parents "

b . Hun fik ingen penge af sine forældre " She got no money of her parents"

c . Hun Y.ille ingen penge haYe af aine forældre "She wanted to have no money of her parents"

d. *Hun Y.ille ingen penge få af sine forældre " She wanted to get no money of her parents"

( OK without source phrase )

7 3 Agent Amblguity (Hansen ( 1972 : 27 ) , Wiwel ( 1901 : 180) )

(69) a . Vi havde aflivet hunden

"We had killed the dog" b . V i fik aflivet hunden

"We got killed the dog" c . Vi oorde have aflivet hunden

"We ought (to) have killed the dog"

( 69b) can either have the interpretation "We put the dog to sleep" or "We had the dog put to sleep (by someone else ) " , i . e . the agent of the participle

that follows få may or may not be the same as the subject of få itself. This is not the case with ha.Ye, the agent of the following participle must be that same

as the subject of have , and therefore ( 69a) only has the interpretation "We had

put the dog to sleep" , and eannot have the interpretation "We had the dog put

to sleep (by someone else ) " . However , in just those cases where haYe is embedded under an obligation root modal , it is nevertheless poesible to have

both interpretations , and therefore ( 69c ) may mean either "We ought to have put

the dog to sleep" or "We ought to have the dog put to sleep (by someone else) " .

7.4 Preposed Objects. (Hansen ( 1972 : 28 ) )

� does not normally allow the object NP to occur before the participle , ( 70a) , whereas få does ( 70b ) . Embedded under an obligation root modal , however ,

have does allow a preposed object , ( 70c ) . This i s a s predicted by the analysis ,

få is out in cernbination with obligation root modale , ( 70d ) , and must be

replaced by have , and only in these circumstances does haYe behave exactly as få, e . g . it allows preposed objects .

- 27 -

( 7 0 ) a . *Vi ll.allde probleroet ordnet

"'We had the problem fixed""

b. Vi fik problemet ordnet ""We got the problem fixed""

c . Vi må baY.e problemet ordnet

"'We must have the problem fixed' '

d . *Vi må få problemet ordnet

"'We must get the problem fixed"'

7.5 Adyerbials of Direction (Hansen ( 1972 : 29 ) )

HaYe and Yære do not normally allow an adverbial of direction , ( 71a) and

( 72a ) , whereas få and � do , ( 7 1b) and ( 72b) . Embedded under an obligation

root modal , however, baY.e and Yære do allow an adverbial of direction , ( 71c )

and ( 72c) . This is as predicted by the analys is, få and lwmme are out in com­

bination with obligation root modale , ( 71d) and ( 72d) , and must be replaced by

baY.e and Y<ere , and only in these circumstances do baY.e and Yære behave exactly

as få and � . e . g . they allow adverbials of direction .

( 7 1 ) a. *Htm ll.allde alle brikkerne ned i æsken igen

"'She had all the pieces back into the box again"'

b . Hrm fik alle brikkerne ned i æsken igen

" 'She got all the pieces back into the box again"'

c . Htm Yi..lle baY.e alle brikkerne ned i æsken igen

"'She wanted to have all the pieces back into the box again"'

d. *Hun Yi..lle få alle brikkerne ned i æsken igen

" 'She wanted to get all the pieces back into the box again"'

(72) a. *Hun havde � mrez: til naboen

"'She had been (over) to the neighbour"'

b . Htm var kQmme1 mrez: til naboen

"'She had come (over) to the neighbour"'

c . Htm Wmlle. have � mrez: til naboen

"'She should have been ( over) to the neighbour"'

d. *Htm Wmlle. være kQmme1 mrez: til naboen

" 'She should have come (over) to the neighbour"'

7 6 Adverbials of Duratioo HaYe and Yære do not normally allow an adverbial of duration, ( 73a ) and

( 74a ) , whereas få and bliYe do ( 73b) and (74b) . Embedded rmder an obligation

root modal , however , baY.e and Yære do allow an adverbial o f duration , ( 73c) and

(74c ) . This is as predicted by the analysis , få and � are out in combina­

tion with obligation root modale , ( 73d) and ( 74d ) , and !DUBt be replaced by baY.e

and Y<ere, and only in these circumstances do haYe and Y<ere behave exactly as få

and �. e . g . they allow adverbials of duration .

- 28 -

(73) a .

b.

c .

d .

(74) a.

b.

c.

d .

*Hun ll.allde en ny sekretær i lØbet af en uge " 'She had a new secretary in a week .. Hun fik en ny sekretær i lØbet af en uge " 'She got a new secretary in a week"'

Hun ill baY.e en ny sekretær i lØbet af en uge " 'She wanted to have a new secretary in a week"'

*Hun ill få en ny sekretær i lØbet af en uge " 'She wanted to have a new secretary in a week "'

*Htm fll: cand . mag . på fem år " 'She is an M . A . in five years"'

Hun bli.Ye.!:: cand . mag . på fem år " 'She becomes an M . A . in five years "'

Hrm ekal Yære cand . mag . på fem år " 'She has to be an M . A . in five years" '

?Hrm ekal � cand . mag . på fem år " 'She has to be an M . A . in five years" '

8 Conclusion I have tried to show that the analysis of so-called ' "additional" " 0-roles

makes an account poesible of the intriguing facts conceming certain VPs w i th

event interpretation embedded tmder obligation root modals .

I have also been lead to aseurne that there are two verbs lronlne ( .. come"' ) , an

auxiliary and a main verb . Certain questions have been left rmanswered , notably

ones concerning the distribution of the two different passives . On the other

hand a series of properties of the state expressions which are normally associ­

ated exclusively with event expressions have been accormted for .

AP.oend i x : The " 'Interessee"' -Construotion

Hansen ( 1972 ) suggests what he calls an "' interessee" '-construction ( Da .

" ' interessent-konstruktion"' ) to accormt for cases where obligation root modals

occur with the eYent expressions having event interpretations ( i . e . with få ,

�. and Jmmme.) . According to him , this construction has a different inter­

pretation from the event interpretations of the state expressions , not w . r . t .

event/state ( in both cases the interpretation is one o f event ) , h.tt w . r . t .

where the obligation comes from , in that the special feature of these interessee-constructions is that the obligation does not come from the subject

o f the sentence but from somewhere else ( Hansen ( 1972 : 12) ) .

I disagree with this suggestion, as I do not think this kind of construction

really exists . There are four potential sets of cases , as there are four

obligation root modals , and below I will go through them one by one .

- 29 -

A l Ville.

Hansen ( 1972 : 1 1 ) gi ves the following examples , all o f which he finels gram­

matical (every time I cite an example from Hansen, the juclgments and the under­

linings are mine , not his) :

(75) a . *Jeg ill gerne få det at vide

"I would like to get it to know = I would like to be told"

b . *Jeg ill gerne bJ.iE formand for Studienævnet

" I would like to become chairman o f the Board o f Studies"

c . Jeg ill gerne kQli.IIIle hjem

"I would like to come home "

d . *Jeg ill gerne bJ.iE undersØgt

" I would like to be examined"

As indicated by the stars, I find all these examples rather unacceptable

(Hansen admits that they are "realised very rarely" ) , with the exception of the

one involving lromme , ( 75c ) . This was discussed in connection with ( 52 ) , where I

argued that there are two types of kæ,me , and only the auxiliary one is ungram­

matical under obligation root modale . As ( 75c) is interpretable as a main V

occurrence of kæ.me , it is thus not ruled out . If replaced by a kæ.me which

JWSt be an auxiliary occurrence , such as one with a figurative adverbial of

direction, cf . ( 53 ) , the example becomes just as una=eptable as ( 75a, b, d) :

( 78) Han 6kal lromme i Studienævnet

"He will to get onto the Board of Studies"

However, in none o f these examples does clwl.k have an interpretation o f

obligation, oot instead one of promise (by the speaker) . I think that here we

have a variant of epistemic clwl.k ( as indicated by my listing in ( 23d) ) , and

this is supported by the following syntactic properties :

i ) They are unacceptable in perleet (o r pas t perfect) (cf . section 2 . 2 . l ) :

(79) a . *Han har s&1llet få det at vide

"He has been going to be told"

b . *Han bar skullet blive student til somroer

"He has been going to graduate this summer"

c . *Han har 5&1llet komme i Studienævnet

"He has been going to get onto the Board of Studies "

d . *Han har skullet blive hentet

"He has been going to be pieked up"

ii) They eannot have PRO as a subject ( cf . section 2 . 2 . 2 ) :

(80) a . *Det er ubehageligt at PRO skulle få det at vide

( 76 ) *Jeg ill gerne kQI!Ille. i Studienævnet " It is unpleasant to be going to be told"

"I would like to get onto the Board of Studies" b . *Det er ubehageligt at PRO skulle blive student til sommer

Gerne locks :ti.l,le in to i ts root interpretation, as i t is an adverbial meaning

roughly "voluntarily" or "with pleasure" . I thus see adverbials as requiring a

particular thematic s trueture rather than adding to i t ( as opposed to Zubizar­

reta ( 1982) .

A 2 Skulle. Hansen ( 1972 : 16-1 7 ) gi ves the following examples :

( 77 ) a . Han 6kal få det at vide

"He will get it to know = He will be told"

b . Han 6kal bJ.iE student til sommer

"He will become a student this summer = He will graduate this summer"

c . Han 6kal ko.wme ud

"He will come out"

d . Han 6kal bJ.iE hentet

"He will be pieked up"

and in this case I agree with him, all the examples are acceptable, and this

would also go for one with ko.wme and a figurative adverbial of direction :

- 30 -

" It is unpleasant to be going to graduate this summer"

c. *Det er ubehageligt at PRO skulle komme i Studienævnet

" It is unpleasant to be going to get onto the Board of Studies" d . *Det er ubehageligt at PRO skulle b li ve hentet

" It is unpleasant to be going to be pieked up"

The PRO has arbi trary reference here, oot the juclgments are the same if the

subject is a PRO that does not have arbitrary reference, i . e . if d!ll......eJ:.:

ubehageligt is replaced by han drØmte om cf . ( 18 ) and ( 19 ) ) .

iii) They do not allow the pseudo-cleft construction (cf . section 2 . 2 . 5 ) :

( 8 1 ) a . *Hvad han akal er at få det at vide

"What he is going is to be told"

b . *Hvad han akal er at blive student til sommer

"What he is going is to graduate this summer

c . *Hvad han akal er at komme i Studienævnet

"What he is going is to get onto the Board o f Studies"

d. *Hvad han akal er at blive hentet

"What he is going is to be pieked up"

- 31 -

Cf. also that Davidsen-Nielsen (1988 : ch . 2 , p. 15-16) says that this kind of

� behaves synta�tically like an epistemic modal (though he finds it

"semantically non-epistemic").

Taking this � (of promise) te be epistemic also accounts for its dif­

ference in interpretation from the � of obligation : There is no additional

G-role assigned te the subject , and therefore the subject eannot be the origin

of an obligation.

A 3 Måtte

Hansen (1972 : 20-21) gives the following examples , all of which he finde

granJDatical , but ag a in I disagree :

( 82) a . *Han må få en reprimande

"He must be reprimanded"

b. *Han må b li ve medlem snarest

"He must become (a) member very soon"

c . *Han må komme af med mindst 10 kg

"He must get rid o f at least 22 pounds"

d . *Han må blive opstillet til formandsposten

"He must become a candidate for the chair"

Hansen (1972 : 21) himself calls them "very rare" . According te ØstkJær Jensen

( 1987 : 23) such examples are rare , and also gramna.tical but unacceptable

(" . . . sind grammatikalisch korrekt aber scheinen vom Sprachusus abzuweichen" ) .

A 4 llirde Hansen (1972 : 23) gives no examples , but says that "the facts . . . seem te be

rather unclear" . Skyum-Nielsen ( 197 1 : 200) has in his corpus found no examples

at all of � being followed by a passive with bl.m .

References Davidsen-Nielsen , Niels (1988) : "On Tense and Mood in English and Danish" . Ms ,

Copenhagen School of Economics , llisiness Adlministration , and Modern

Languages.

Fillmore , Charles (1975) : Santa Cruz Leetures on Lle.ixifi , Bloomington , IN :

Indiana University Linguistics Club.

Grimshaw , Jane (1986) : "Nouns , Arguments and Adjulllcts". Ms , Brandeis llniv­

ersity.

Hansen, Erik (1972) : "Modal interessens. Nu bØr det komme frem" in �

Studie.x:, vol . 67 , pp . 5-36.

Hansen , Erik (1977) : "BehØver vi at?" in Nyt fra Spro.gnæynet , vol 19 , pp, 1-4 .

Jespersen, Otte (1932) : Tanker og studier . Copenhagen: Gyldendal/Nordisk For­

lag .

- 32 -

' i , ,

Lundskær-Nielsen, Tom (1983) : "The Grammar of Auxiliaries in Danish" . Ms ,

lmiversity of Cambridge.

Østkjær Jensen , Helle (1987) : Eine kontrastive Analyse der dl.inischen Moctalver­ben "behqlve" und "måtte" und ihrer deut8cher1 Aquivalente "brayc.hen" . "d.iirfen" , "!llOgen" und "milssen" . Odens e : llniversity Press .

Platzack , Christer (1979) : The Sernantic Interpretation of Aspect and Aktions­arten : A Study of Internal Time Reference in Swedish . furdrecht : Foris .

Skyum-Nielsen , Peder (1971) : "Modalverberne i Nudansk". Ms , lmiversity of

Copenhagen .

Thrainsson , Hoskuldur (1986) : "On Auxiliaries , AllX and VPs in Icelandic" in

Lars Hellan & Kirsti Koch Christensen (eds . ) : Topjes in Scandinavian � . pp. 235-265 . Dordrecht : Heidel.

Vikner , Sten (1985) : "Reichenbach Revisited : One , Two , or Three Temporal Rela­

tions?" in Acta Linguistica Rafniensia , vol. 19(2) , pp. 81-95 .

Vikner, Sten & Rex A. Sprouse (1988) : "HavWEe Selection as an A-Chain

Membership Requirement" . Horking Papers on Scandinavian Syntax , no. 38.

Wiwel , H. G . ( 1901) : Synspunkter for dansk sproglære . Copenhagen : Nordisk For­

lag .

Zubiaarreta, Maria Luisa ( 1982) : "On the Relationship of the Lexicon te Syntax". Ph.D. , M . I. T.

- 33 -