Case and Passive in Head-Driven Phrase Structure Grammar
Case and Passive
in Head-Driven Phrase Structure Grammar
Stefan M¨uller
Theoretische Linguistik/Computerlinguistik Fachbereich 10 Universit¨at Bremen Institut f¨ur Linguistik Universit¨at Potsdam Stefan.Mueller@cl.uni-bremen.de
December 2, 2005
Case and Passive in Head-Driven Phrase Structure Grammar Case
Case and Case Principles
Case and Case Principles
•What kind of cases are there?
•How does case depend on the syntactic environment?
•Until now case was maximally specified in valency lists, if we now the principles of case assignment, this is unnecessary.
We capture generalizations and reduce the number of lexical entries for verbs likelesen(‘read’) in (1):
(1) a. Er he-nom
m¨ochte wants
das the
Buch book
lesen.
read
‘He wants to read the book.’
b. Ich I
sah saw
ihn him-acc
das the
Buch book
lesen.
read
‘I saw him read the book.’
The case of the subject (and the object) is assigned by a principle.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 1/61
Case Structural vs. Lexical Case
Structural Case
Structural and Lexical Case
•If the case of an argument depends on the syntactic environment, the case is called structural case. Otherwise the argument has lexical case.
•Examples for structural case are:
(2) a. Der the
Installateur plumber-nom
kommt.
comes b. Der
the Mann man-nom
l¨aßt lets den the
Installateur plumber-acc
kommen.
come c. das
the Kommen coming
des of.the
Installateurs plumber-gen
•In (2) we have a subject that changes case, in (3) it is the object:
(3) a. Karl Karl
schl¨agt beats
den the
Hund.
dog-acc b. Der
the Hund dog
wird is
geschlagen.
beaten c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 2/61
Structural vs. Lexical Case Lexical Case
Genitive
Lexical Case
•Genitive that depends on a verb is a lexical case:
The case of the genetive object does not change in passives.
(4) a. Wir we-nom
gedenken remember
der the
Opfer.
victims-gen b. Der
the Opfer victims-gen
wird is
gedacht.
remembered c. * Die
the Opfer victims-nom
wird/werden is/are
gedacht.
remembered (4b) = impersonal passiv, there is no subject
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 3/61
Structural vs. Lexical Case Lexical Case
Dative
Is the Dative a Lexical Case?
•No change with dative objects:
(5) a. Der the
Mann man
hat has
ihm him-dat
geholfen.
helped b. Ihm
him-dat wird was
geholfen.
helped
•But what about (6)?
(6) a. Der the
Mann man-nom
hat has
den the
Ball ball-acc
dem the
Jungen boy-dat
geschenkt.
given
‘The man gave the ball to the boy as a present.’
b. Der the
Junge boy-nom
bekam became
den the
Ball ball-acc
geschenkt.
given
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 4/61
Structural vs. Lexical Case Lexical Case
Dative
Is the Dative a Lexical Case?
•The status of the dative is still a controversial issue.
Three possibilities for dative arguments:
1. All datives are lexical.
2. Some are lexical others structural.
3. All datives are structural.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 5/61
Structural vs. Lexical Case Lexical Case
Dative
Dative as a Lexical Case
•If we treat dative as a lexical case, we have to assume that lexical case can be changed into structural case in the dative passive.
•With lexical dative Haider’s examples in (7) and (8) are explained (Haider, 1986, p. 20):
(7) a. Er he-nom
streichelt strokes
den the
Hund.
dog-acc b. Der
the Hund dog-nom
wurde was
gestreichelt.
stroked c. sein
his Streicheln stroking
des of.the
Hundes dog-gen (8) a. Er
he-nom hilft helps den the
Kindern.
children-dat b. Den
the Kindern children-dat
wurde was
geholfen.
helped
‘The children were helped.’
c. das the
Helfen helping der of.the
Kinder children-gen d. * sein
his Helfen helping der of.the
Kinder children-gen c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 6/61
Structural vs. Lexical Case Lexical Case
Dative
Structural Case and Bivalent Verbs
•If the only difference betweenhelfenandunterst¨utzenis the case, one of the cases has to be lexical.
(9) a. Er he-nom
hilft helps
ihm.
him-dat b. Er
he-nom unterst¨utzt supports
ihn.
him-acc
Von Stechow and Sternefeld (1988) and von Stechow (1990) and authors who see the structural/lexical case issue from a semantic point of view (Kaufmann, 1995; Stiebels, 1996; Olsen, 1997; Rapp, 1997) therefore assume that the dative of bivalent verbs is a lexical dative.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 7/61
Structural vs. Lexical Case Lexical Case
Dative
Dative Passive with Bivalent Verbs
•The case of ditransitive verbs can be derived by principles, but this is impossible with bivalent verbs (unless one has complex semantic strories).
→Dative withhelfenis said to be lexical.
Prediction: dative passive is impossible with such verbs.
•Wegener (1985; 1990) provides the examples in (10):
(10) a. Er he
kriegte got
von by
vielen many
geholfen helped
/ gratuliert congratulated
/ applaudiert.
applauded
‘Many helped / congratulated / applauded him.’
b. Man one
kriegt gets
t¨aglich daily
gedankt.
thanked
‘One is thanked on a daily basis.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 8/61
Structural vs. Lexical Case Lexical Case
Dative
Dative Passive with Bivalent Verbs (Corpus Examples)
(11) a.”Da there
kriege get
ich I
geholfen.“1 helped
‘Somebody helps me there.’
b.”Kl¨arle“
Kl¨arle h¨atte had
es it
wirklich really
mehr more als than
verdient, deserved
auch also
mal once
zu to
einem a
”unrunden“
insignificant Geburtstag birthday
gratuliert congratulated
zu to
bekommen.2 get
‘Kl¨arle would have more than deserved to be wished a happy birthday, even an insignificant birthday.’
c. Mit with
dem the
alten old
Titel song
von by
Elvis Elvis
Presley Presley
[. . . ] bekam got
Kassier cashier
Markus Markus
Reiß Riss zum
to.the Geburtstag birthday
gratuliert, congratulated
[. . . ]3
‘The cashier Markus Riss was wished a happy birthday with the old Elvis Presley song [. . . ].’
1Frankfurter Rundschau, 26.06.1998, p. 7.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 9/61
Structural vs. Lexical Case Lexical Case
Accusative
Accusative
We already saw structural accusatives, but there is also lexical accusative:
(12) a. Ihn him-acc
d¨urstet.
is.thirsty b. Die
the Mutter mother
lehrte taught
ihre her
Tochter daughter-acc
ein a
neues new
Lied.
song-acc
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 10/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
Adjectival Environments
Lexical Case in Adjectival Environments (I)
The case of objects that depend n adjectives does not change.Adjectives may assign genitive and dative:
(13) a. Ich I
war was
mir me-dat
dessen that-gen
sicher.
sure
‘I was sure of this.’
b. Sie she ist is
ihm him-dat
treu.
faithful
‘She is faithful to him.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 11/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
Adjectival Environments
Lexical Case in Adjectival Environments (II)
Accusative is also possible, but not so frequent (Haider, 1985):(14) a. Das this ist is
diesen this
Preis price
nicht not
wert.
worth
‘This is not worth this price.’
b. Der the
Student student ist is
das the
Leben live
im in.the
Wohnheim dormitory
nicht not
gewohnt.4 used
‘The student is not used to the live in the dormitory.’
c. Du you
bist are
mir me
eine an
Erkl¨arung explanation
schuldig.5 owe
‘You owe me an explanation.’
4(Helbig and Buscha, 1972) 5(Heidolph et al., 1981) c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 12/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
Adjectival Environments
Structural Case in Adjectival Environments
The case of the subject of an adjective depneds on the syntactic environment (Wunderlich, 1984):(15) a. Der the
Mond moon-nom
wurde got
kleiner.
smaller b. Er
he sah saw
den the
Mond moon-acc
kleiner smaller
werden.
get
‘He saw how the moon got smaller.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 13/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
The Case of Unrealized Subjects
The Case of Unrealized Subjects (I)
H¨ohle (1983, Chapter 6):The case of elements that do not surface can be determined.
ein- nach d- ander-(‘one after the other’) may refere to constituents with plural reference.
Case and gender has to agree with the antecedent phrase.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 14/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
The Case of Unrealized Subjects
The Case of Unrealized Subjects (II)
We have reference to subjects and objects in (16):
(16) a. [Die the
T¨uren]i doors-nom-pl-fem
sind are
[eine one-nom-fem
nach after der the-dat-fem
anderen]i other
kaputt broke
gegangen.
went
‘The doors broke one after another.’
b. [Einer one-nom-mas
nach after dem the-dat-mas
anderen]i other
haben have
wiri we-nom
die the
Burschen lads-acc
runtergeputzt.
down.cleaned
‘We took turns in bringing the lads down a peg or two.’
c. [Einen one-acc-mas
nach after
dem the-dat-mas
anderen]i other
haben have
wir we-nom
[die the
Burschen]i lads-acc-pl-mas
runtergeputzt.
down.cleaned
‘One after the other, we brought the lads down a peg or two.’
d. Ich I
ließ let [die
the Burschen]i lads-acc-pl-mas
[einen one-acc-mas
nach after dem the-dat-mas
anderen]i other
einsteigen.
enter
‘I let the lads get in (get started) one after the other.’
e. [Uns]i us-dat
wurde was
[einer one-dat-fem
nach after der the-dat-fem
anderen]i other
der the
Stuhl chair vor before
die the T¨ur door
gesetzt.
set
‘We were given the sack one after the other.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 15/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
The Case of Unrealized Subjects
The Case of Unrealized Subjects (III)
In (17) we have reference to dative or accusative objects of embedded infinitives:
(17) a. Er he hat has uns us
gedroht, threatened
[die the
Burschen]i lads-acc-pl-mas
demn¨achst soon
[einen one-acc-mas
nach after dem
the-dat-mas anderen]i other
wegzuschicken.
away.to.send
‘He threatened us that soon he would send the lads away one after the other.’
b. Er he hat has
angek¨undigt, announced
[uns]i us-dat
dann then
[einer one-dat-fem
nach after der the-dat-fem
anderen]i other
den the Stuhl
chair vor before
die the T¨ur door zu to
setzen.
set
‘He announced that he would then sack us one after the other.’
c. Es it
ist is
n¨otig, necessary
[die the
Fenster]i, windows-acc-pl-neu
sobald as.soon es it
geht, goes
[eins one-acc-neu
nach after dem
the-dat-neu anderen]i other
auszutauschen.
to.exchange
‘It is necessary to exchange the windows one after the other, as soon as possible.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 16/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
The Case of Unrealized Subjects
The Case of Unrealized Subjects (IV)
Reference to the subject of the infinitival VP:
(18) a. Ich I
habe have
[den the
Burschen]i lads-dat-pl-mas
geraten, advised im in.the
Abstand distance von of
wenigen few
Tagen days [einer
one-nom-mas nach after dem the-dat-mas
anderen]i other
zu to
k¨undigen.
hand.in.their.notice
‘I advised the lads to hand in their notice one after the other, at intervals of a few days.’
b. [Die the
T¨uren]i doors-nom-pl-fem
sind are
viel much
zu too
wertvoll, precious um compl
[eine one-nom-fem
nach after der
the-dat-fem anderen]i other
verheizt burnt
zu to
werden.
be
‘The doors are much too precious to be burnt one after the other.’
c. [Wir]i we-nom-pl
sind are
es itextra
leid, tired
[eine one-nom-fem
nach after der the-dat-fem
anderen]i other
den the
Stuhl chair vor
before die the T¨ur door
gesetzt set
zu to
kriegen.
get
‘We are tired of being given the sack one after the other.’
ein- nach d- ander-is nominative→Subjects are nominative as well.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 17/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
The Case of Unrealized Subjects
The Case of Unrealized Subjects (V)
We have to make sure that non-realized subjects get case. If the case of the subject would be left unspecified, sentences like (19) would get a wrong reading:
(19) # Ich I
habe have
den the
Burschen lads-dat-mas
geraten, advised im in.the
Abstand distance
von of
wenigen few Tagen
days einen
one-acc-mas nach after
dem the-dat-mas
anderen other
zu to
k¨undigen.
fire
‘I advised the lads to fire (them) one after the other, at intervals of a few days.’
einen nach dem anderenis the object ofk¨undigenand cannot refer to the subject of the infinitive, which is coreferential withden Burschen.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 18/61 Case
The Analysis The Case Principle
The Case Principle (I)
•Dative is treated as a lexical case.
•ditransitive verbs likegeben(‘give’) have the followingsubcatvalue:
(20) hNP[str], NP[str], NP[ldat]i str= structural case,ldat= lexical dative.
•The assignment of structural case is done via the following principle (Przepi´orkowski, 1999b; Meurers, 1999):
Case Principle:
•In a list that contains both the subject and the complements of a verbal head, the least oblique element with structural case gets nominative, unless it is raised by a higher head.
•All other elements that have structural case and are not raised get accusative.
•In nominal environments, elements with structural case get genitive.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 19/61
Case The Analysis
The Case Principle
The Case Principle (II)
•This principle is similar to the one by Yip, Maling and Jackendoff (1987) and therefore can explain the case facts of the languages that were discussed by these authors, in particular the complicated case system of Icelandic.
•An important difference is that the principle above is monotonic, i.e. case that was assigned cannot be changed by a higher predicate.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 20/61 Case
The Analysis Active
Active
prototypical valency lists:
(21) a.schl¨aft(‘sleeps’): subcathNP[str]ji b.unterst¨utzt(‘supports’):subcathNP[str]j, NP[str]ki c.hilft(‘helps’): subcathNP[str]j, NP[ldat]ki d.schenkt(‘gives as . . . ’):subcathNP[str]j, NP[str]k, NP[ldat]li The first element in thesubcat-Liste gets nominative.
All other elements with structural case get accusative.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 21/61
The Analysis Passive
Agentive Passive
Passive
(22) a.schl¨aft(‘sleeps’): subcathNP[str]ji b.unterst¨utzt(‘supports’):subcathNP[str]j, NP[str]ki c.hilft(‘helps’): subcathNP[str]j, NP[ldat]ki d.schenkt(‘gives as . . . ’):subcathNP[str]j, NP[str]k, NP[ldat]li When these verbs get passivized, we get the followingsubcatlists:
(23) a.geschlafen wird:subcath i b.unterst¨utzt wird:subcathNP[str]ki c.geholfen wird: subcathNP[ldat]ki d.geschenkt wird: subcathNP[str]k, NP[ldat]li The first position is occupied by a different NP in (23).
If this NP has structural case, it gets nominative,
if it has not (as in (23c)) the case remains the way it is, namely lexically specified.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 22/61 The Analysis
Passive Dative Passive
Dative Passive
(24) c.hilft(‘helps’): subcathNP[str]j,NP[ldat]ki d.schenkt(‘gives as . . . ’):subcathNP[str]j, NP[str]k,NP[ldat]li The dative argument becomes the first argument and
the lexical dative of the embedded verb is turned into a structural case:
(25) a.geholfen bekommt: subcathNP[str]ki b.geschenkt bekommt:subcathNP[str]l, NP[str]ki The former dative argument is now in first position.
Since it has structural case, it gets assigned nominative.
In (25b), the second element (the direct object) gets accusative.
This change of lexical case into structural case is not nice, but there seems to be no better way.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 23/61
The Analysis Passive
The AcI Construction
The AcI Construction (I)
(26) a.schl¨aft(‘sleeps’): subcathNP[str]ji b.unterst¨utzt(‘supports’):subcathNP[str]j, NP[str]ki c.hilft(‘helps’): subcathNP[str]j, NP[ldat]ki d.schenkt(‘gives as . . . ’):subcathNP[str]j, NP[str]k, NP[ldat]li The AcI Construction is analyzed as argument composition:
the arguments of the embedded verb become arguments of the AcI verb:
(27) a.schlafen l¨aßt: subcathNP[str]i, NP[str]ji b.unterst¨utzen l¨aßt:subcathNP[str]i, NP[str]j, NP[str]ki c.helfen l¨aßt: subcathNP[str]i, NP[str]j, NP[ldat]ki d.schenken l¨aßt: subcathNP[str]i, NP[str]j, NP[str]k, NP[ldat]li NP[str]istands for the subject of the AcI verb.
NP[str]j, NP[str]kand NP[ldat]lare the arguments of the embedded verbs.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 24/61 The Analysis
Passive The AcI Construction
The AcI Construction (II)
(28) a.schlafen l¨aßt: subcathNP[str]i, NP[str]ji b.unterst¨utzen l¨aßt:subcathNP[str]i, NP[str]j, NP[str]ki c.helfen l¨aßt: subcathNP[str]i, NP[str]j, NP[ldat]ki d.schenken l¨aßt: subcathNP[str]i, NP[str]j, NP[str]k, NP[ldat]li Only the valency lists in (28) are relevant for case assignment.
The elements in the valnecy representations of the base verbs are irrelevant, since the case principle does not assign case to elements that are raised.
The first element in the lists in (28) gets nominative, the remaining elements with structural case get accusative.
Hence, the logical subjects of the embedded Vs get realized in accusative.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 25/61
The Analysis Passive
Subjects of Adjectives
Subjects of Adjectives
The case assignment to the subject of adjectives work analoguosly.
The copula is combined with the adjective and we get a valency list that contains the arguments of the adjective (29a).
If such a complex is embedded under an AcI verb we get (29b):
(29) a.kleiner werden: subcathNP[str]ji
‘smaller become’
b.kleiner werden sah: subcathNP[str]i, NP[str]ji
‘smaller become saw’
The first NP gets nominative, the second one accusative.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 26/61
Case and Passive in Head-Driven Phrase Structure Grammar Case
Summary
Intermediate Summary
•Case is assigned on a valence representation (heresubcat, in other HPSG publicationsarg-st).
•There is no zero case for non-realized subjects, these elements get case according to the normal principles.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 27/61
Passive The Phenomenon
Agentive Passive
Agentive Passive
“personal” passive:
(30) a. weil because
Karl Karl-nom
den the
Schrank cupboard-acc
¨offnet opens b. weil
because der the
Schrank cupboard-nom
ge¨offnet opened
wird.
is
‘because the cupboard is opened.’
“impersonal” passive:
(31) a. weil because
Karl Karl-nom
arbeitet works b. weil
because gearbeitet worked
wird.
is
‘because it is worked.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 28/61 Passive
The Phenomenon Dative Passive
Dative Passive
(32) a. weilbecause Peter Peter-nom
dem the
Jungen boy-dat
den the
Ball ball-acc
wegnimmt away.take
‘Peter takes away the ball from the boy.’
b. weil because
der the
Junge boy-nom
den the
Ball ball-acc
weggenommen away.taken
bekommt gets
‘The ball is taken away from the boy.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 29/61
Passive The Phenomenon
lassen Passive
lassen
Passive
(33) a. weil because
er he-nom
einen an
Fachmann expert-acc
den the
Wagen car-acc
reparieren repair
l¨aßt lets
‘because he has an expert repair the car.’
b. weil because
er he-nom
den the
Wagen car-acc
(von by
einem an
Fachmann) expert
reparieren repair
l¨aßt lets
‘because he has an expert repair the car.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 30/61 Passive
The Phenomenon Modal Infinitives
Modal Infinitives
(34) a. weilbecause ihr you-nom
den the
Aufsatz paper-acc
zu to
schreiben write
habt have
‘because you have to write the paper.’
b. weil because
der the
Aufsatz paper-nom
(von by
euch) you
zu to
schreiben write
ist is
‘because the paper is to be written by you.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 31/61
Passive The Phenomenon
Morphological Identity of Forms (Participle)
Morphological Identity of Forms (Participle)
The form of the participle is the same in the perfect, the agentive passive, the dative passive:
(35) a. Der the
Mann man-nom
hat has
den the
Ball ball-acc
dem the
Jungen boy-dat
geschenkt.
given
‘The man gave the ball to the boy.’
b. Der the
Ball ball-nom
wurde was
dem the
Jungen boy-dat
geschenkt.
given
‘The ball was given to the boy.’
c. Der the
Junge boy-nom
bekam got
den the
Ball ball-acc
geschenkt.
given
‘The boy got the ball as a present.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 32/61 Passive
The Phenomenon
Morphological Identity of Forms (Bare Infinitive)
Morphological Identity of Forms (Bare Infinitive)
The form of the bare infinitive is the same in the future, AcI,lassenpassive, and middle constructions:
(36) a. weil because
ein a
Mechaniker mechanic-nom
den the
Wagen car-acc
reparieren repair
wird.
will
‘because the mechanic will repair the car.’
b. weil because
Karl Karl-nom
einen a
Mechaniker mechanic-acc
den the
Wagen car-acc
reparieren repair
l¨aßt.
lets
‘because Karl has the mechanic repair the car.’
c. weil because
Karl Karl-nom
den the
Wagen car-acc
(von by
einem a
Mechaniker) mechanic
reparieren repair
l¨aßt.
lets
‘because Karl has somebody / the mechanic repair the car.’
d. weil because
sich self der the
Wagen car-nom
nicht not
reparieren repair
l¨aßt.
lets
‘because it is impossible to repair the car.’
Forzuinfinitives, we also have two modal infinitive constructions with different argument realizations.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 33/61
The Phenomenon Unaccusativity Passivizability
Unaccusativity: Passivizability
Although certain intransitive verbs allow passivization, there are others that do not allow the passive.
(37) a. daß that
der the
Zug train
ankam arrived b. * Dort
there wurde was
angekommen.
arrived c. daß
the der the
Mann man
ihr her
auffiel noticed
‘She noticed the man.’
d. * Ihr her-dat
wurde was
aufgefallen.
noticed Caution:
Not all verbs that cannot be passivized belong to the same class!
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 34/61 The Phenomenon
Unaccusativity Adjectival Participles (I)
Adjectival Participles
Such verbs can be used as prenominal adjectival participles:
(38) a. der the
angekommene arrived
Zug train b. dem
the Regime regime-dat
aufgefallene
part.fell ”Vaterlandsverr¨ater“6 traitors.to.their.country
‘the “traitors to their country” noticed by the regime’
The subject role of the particple is filled be the modified noun.
6Die Zeit, 26.04.1985, p. 3.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 35/61
The Phenomenon Unaccusativity
Adjectival Participles (I)
Adjectival Participles (II)
Transitive verbs are different:The object role of the participle is filled by the modified noun:
(39) a. die the
geliebte loved
Frau woman b. der
the geschlagene beaten
Hund dog
Verbs that do not have an accusative object usually do not allow for adjectival particples:
(40) a. * der the
getanzte danced
Mann man b. * der
the
(ihm) geholfene him
Mann helped man
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 36/61 The Phenomenon
Unaccusativity Nominatives with Object Properties
Nominatives with Object Properties (I)
•Arguments of certain verbs that are nominative in active sentences have object properties.
•Such verbs are called unaccusative (Perlmutter, 1978) or ergative (see for instance Grewendorf, 1989).
•Grewendorf (1989): fourteen tests to differenciate between unaccusative and unergative/transitive verbs.
•Fanselow (1992) six additional tests
•Despite this big number of test researchers do not agree which verbs should be treated as unaccusative.
•Some tests produce contradicting results (M¨uller, 2002, In Preparation).
•Kaufmann (1995): Many differences btween unacc and trans/unerg verbs can be explained differently.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 37/61
The Phenomenon Unaccusativity
Nominatives with Object Properties
Nominatives with Object Properties (II)
•Data is explained if one assumes, that the subject of unaccusative verbs is an underlying object:
Passive = Suppression of the subject. No subject present→ passivization ofankommenandauffallenimpossible
•The formation of adjectival participles is possible, if there is an element with accusative object properties.
•Since the subjects ofankommena.auffallenare underlying objects, the well-formedness of the phrases in (41) is explained.
(41) a. der the
angekommene arrived
Zug train b. dem
the Regime regime-dat
aufgefallene
part.fell ”Vaterlandsverr¨ater“7 traitors.to.their.country
‘the “traitors to their country” noticed by the regime’
7Die Zeit, 26.04.1985, p. 3.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 38/61 The Phenomenon
Unaccusativity Resultative Constructions
Resultative Constructions
Resultative Constructions: Verb + Accusative + Predicate (Wunderlich, 1997; M¨uller, 2002, Chapter 5):
(42) weil because
niemand nobody-nom
den the
Teich pond-acc
leer empty
fischt fishes
‘because nobody fishes the pond empty’
If the verbs are unergative,
the resultative predicate predicates over the accusative.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 39/61
The Phenomenon Unaccusativity
Resultative Constructions
Resultative Constructions with Unaccusative Verbs
Resultative predicate predicates over the subject of the verb:
(43) a. [. . . ] und and im in.the
Winter winter
fror froze
sein its
Wasser water
zu to
Eis.8 ice
‘and in the winter its water froze to ice.’
b. den the
Tonb¨andern tapes
im in.the
Archiv, archive die which
in in der the
tropischen tropical
Hitze heat
zu to
einer a schwarzen
black Masse mass
schmolzen.9 melted Data are explained,
if the result predicate predicates over the element with object properties.
Subjects in (43) are not normal subjects but underlying objects.
8Frankfurter Rundschau, 16.09.1999, p. 3.
9Frankfurter Rundschau, 05.08.1997, p. 3.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 40/61
Case and Passive in Head-Driven Phrase Structure Grammar Passive
The Analysis
Two Strategies
•In principle, there are two strategies for tackling the passive problem:
1. One (or more) lexical entries for the perfect participle and the passive participle(s). (Bresnan, 1978, 1982; Pollard and Sag, 1987, p. 214–218;
Bierwisch, 1990, p. 189; Kunze, 1996, p. 656; Manning and Sag, 1998;
Michaelis and Ruppenhofer, 2001, Chapter 4; Vierhuff, Hildebrandt and Eikmeyer, 2003, p. 231)
2. One lexical entry for the participle that can be used in different environments. (Bech, 1955, p. 37; H¨ohle, 1978; Haider, 1986; Toman, 1986;
Fanselow, 1987, p. 165; Hoekstra, 1987, p. 283; von Stechow, 1990, p. 171)
•Solution 1: Lexical rules that map active to passive forms or stems to various participles.
•Solution 2: Argument Attraction: The auxiliary determines which arguments of the embedded verb are realized and determines the form.
•Argument for 2: No variation in morphological form.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 41/61
The Analysis Preliminaries
The Subject of Non-Finite Verbs
The Subject of Non-Finite Verbs
•Subject of non-finite verbs is not represented in thesubcat-Liste, but undersubj
(Borsley, 1989; Pollard, 1996; Kiss, 1992, 1995)
•VPs are maximal projections (uniform treatment of extraposition): A neccessary condition for extraposition is maximality.
•The lexicon contains stems that have all their arguments onsubcat.
The stem is mapped onto inflected forms and the subject of infinitives with and withoutzuis not represented in thesubcatof the output, but undersubj.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 42/61 The Analysis
Preliminaries The Subject of Non-Finite Verbs
The Verbal Complex
•I assume that auxiliaries and embedded verb form a complex.
•The embedding verb takes over all arguments from the embedded verb.
•More motivation tomorrow.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 43/61
The Analysis Preliminaries
Lexical Entry for the Future Auxiliary
Lexical Entry for the Future Auxiliary
•werdenselects a verb inbseform, that is an infinitive withoutzu.
(44) wird(Futur Auxiliary):
head verb
subcat1⊕2⊕ hV[bse, subj1, subcat2]i cat
•wirdtakes over the description of the arguments ofhelfen(Karl,mir):
(45) daß that
Karl Karl mir me
helfen help
wird will
‘that Karl will help me’
Thesubcat-Liste ofhelfen wirdhas the same form as the one ofhilft.
•The auxiliary attracts the arguments of the embedded verb.
(argument attraction, argument composition).
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 44/61 The Analysis
Preliminaries
Lexical Entry for the Future Auxiliary
Analysis of
helfen wird(‘help will’)
"
head 1 subcat2⊕3
#
CL H
4
loc
head
subj 2hNP[nom]i vformbse verb
subcat3hNP[dat]i
head 1
subj h i vformfin verb
subcat2⊕3⊕ h4i
helfen wird
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 45/61
Passive The Analysis
The Representation of Unaccusativity
Unaccusative, Unergative, and Transitive Verbs
•Haider (1986): Argument with subject properties (the designated argument) is marked in the argument structure of the verb.
•The subject of unergative and transitive verbs is the designated argument.
Unaccusative verbs do not have a designated argument.
•Heinz and Matiasek (1994) and Lebeth (1994) use a list-valued featuredafor the representation of the designated argument.
If there is a designated argument, it is a member of both thedaand thesubcat-Liste:
da subcat
ankommen (unaccusative): hi hNP[str]i
tanzen (unergative): h1NP[str]i h1i auffallen (unaccusative): hi hNP[str], NP[ldat]i lieben (transitive): h1NP[str]i h1, NP[str]i schenken (ditransitive): h1NP[str]i h1, NP[str], NP[ldat]i helfen (unergative): h1NP[str]i h1, NP[ldat]i c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 46/61 Passive
The Analysis Agentive Passive
Agentive Passive
•Haider: The designated argument of the participle is blocked.
•If the participle is used in the passive, the designated argument remains blocked. If the participle is used in the perfect, the perfect auxiliary deblocks the blocked element.
subj subcat
angekommen (unaccusative): hi hNP[str]i
getanzt (unergative): hNP[str]i hi
aufgefallen (unaccusative): hi hNP[str], NP[ldat]i geliebt (transitive): hNP[str]i hNP[str]i geschenkt (ditransitive): hNP[str]i hNP[str], NP[ldat]i geholfen (unergative): hNP[str]i hNP[ldat]i
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 47/61
Passive The Analysis
Agentive Passive
Argument Blocking Lexical Rule for Participles
•Argument blocking lexical rule for participles:
. . .
head
"
da1
verb
#
subcat1⊕2
stem
7→
. . .
head
vformppp subj 1
verb
subcat2
word
•The LR brackes thesubcat-Liste of the input into two parts:
The part that corresponds to thedalist and a rest.
•Only the rest is represented as thesubcatvalue of the output.
Thedalist is identified with thesubj-Liste of the output verb.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 48/61 Passive
The Analysis Agentive Passive
The Passive Auxiliary (I)
•werden(Passive Auxiliary):
"
da hi
subcat1⊕ hV[ppp, dahNPi, subcat1]i
#
The passive auxilary selects a particple and attracts its arguments.
(The logical subject of the participle is blocked)
•Auxiliary requires the particple to have a designated argument (an element in theda-Liste).
This excludes passivization of unaccusative verbs, since these do not have anything inda.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 49/61
Passive The Analysis
Agentive Passive
The Passive Auxiliary (II)
•The entry explains both the personal and the impersonal passive:
"
da hi
subcat1⊕ hV[ppp, dahNPi, subcat1]i
#
The followingsubcatlists result after verb complex formation:
geschlafen wird: subcath i unterst¨utzt wird:subcathNP[str]ki geholfen wird: subcathNP[ldat]ki geschenkt wird: subcathNP[str]k, NP[ldat]li
•The case principle takes care of the correct case assignment.
(➚Case Principle)
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 50/61 Passive
The Analysis Agentive Passive
Exclusion of Double Passives
Without restrictions one could form a passive from (46b), which would be the impersonal passive in (46c).
(46) a. weil because
sie she
den the
Mann movie
liebt loves b. weil
because der the
Mann man
geliebt loved
wurde was c. * weil
because geliebt loved
worden been
wurde was Thedavalue of the passive auxiliary is the empty list.
Therefore the combination of the auxiliary and the participle is parallel to unaccusative simplex verbs.
Since the passive auxiliary does not allow for the embedding of unaccusatives,geliebt wordencannot be embedded underwurdein (46c).
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 51/61
Passive The Analysis
Perfect
Perfect
•The perfect auxiliary deblocks the designated argument.
haben(Perfect Auxiliary):
subcat1⊕2⊕ hV[ppp, subj1, subcat2]i
•It attracts the concatenation of thesubjvalue and thesubcat-Liste of the embedded verb.
•The blocked designated argument is reintroduced into thesubcat-Liste by the auxiliary.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 52/61 Passive
The Analysis Perfect
The Perfect Auxiliary
sein•Similar tohaben, but does not deblock the designated argument of the embedded participle:
(47) sein(Perfect Auxiliary):
subcat1⊕ hV[ppp, subcat1]i
Since nothing was blocked during the formation of particples of verbs likeangekommenandaufgefallen,
nothing has to be deblocked for the perfect.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 53/61
Passive The Analysis
The Dative Passive
The Dative Passive
bekommen(Dative Passive Auxiliary):
da h i subcatD
1NP[str]2E
⊕3⊕4⊕
*V[ppp,dahNPi, subcat3⊕D
NP[ldat]2E
⊕4] +
•The embedded verb has to have an element inda.
•All non-blocked arguments except the dative are directly raised.
•The dative element is coindexed with an NP with structural case.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 54/61 Passive
The Analysis Modal Infinitives
Modal Infinitives
I follow Haider in assuming that infinitives differ from participles in that the logical subject rather then the designated argument is blocked.
da subcat subj
a. anzukommen (unaccusative):hi hi hNP[str]i
b. zu tanzen (unergative): h1NP[str]i hi h1i c. aufzufallen (unaccusative): hi hNP[ldat]i hNP[str]i d. zu lieben (transitive): h1NP[str]i hNP[str]i h1i e. zu schenken (ditransitive): h1NP[str]i hNP[str], NP[ldat]i h1i f. zu helfen (unergative): h1NP[str]i hNP[ldat]i h1i
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 55/61
Passive The Analysis
Modal Infinitives
The Argument Blocking Lexical Rule
(48) Argument blocking lexical rule for infinitives with and withoutzu:
. . .
"
head verb subcat1
#
stem
7→
. . .
head
vforminf-or-bse subj 2
verb
subcat3
word
∧first-np-str(1,2,3) The subject is not directly marked in the lexicon by a feature.
The element that is represented undersubj
is the first NP in thesubcat-Liste that has structural case.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 56/61 Passive
The Analysis Modal Infinitives
The Auxiliaries
The auxiliaries are parallel to what we saw for the perfect:habendeblocks the element insubj,seinleaves blocked elements untouched.
(49) a. weil because
ihr you-nom
den the
Aufsatz paper-acc
zu to
schreiben write
habt have
‘because you have to write the paper.’
b. weil because
der the
Aufsatz paper-nom
(von by
euch) you
zu to
schreiben write
ist is
‘because the paper is to be written by you.’
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 57/61
Passive The Analysis
The lassen Passive
The
lassenPassive
lass-(Passive Version):
head|dah1i
subcat h1NP[str]i ⊕2⊕D
V[bse, dahNPi, subcat2]E
•lassenrequires the embedded verb to have ada.
•It raises all arguments that are not blocked.
•It introduces its own argument.
•The case principle does the rest.
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 58/61 Passive
The Analysis Adjectival Participles
Adjectival Participles
Adjective Derivation Rule for Participles:
. . .
head
"
vformppp verb
#
subcath1NP[str]i ⊕2
word
7→
. . .
head
"
subjh1i adj
#
subcat2
stem
•The lexical rule maps a participle onto an adjective stem.
•The designated argument of the participle is blocked, if there is any.
•Therefore the element at the first position of thesubcat-Liste is the element with object properties (the direct object of transitive verbs and the subject of unaccusative verbs).
c
Stefan M¨uller 2005, CL, FB 10, Universit¨at Bremen & CL, Uni Potsdam 59/61