A regular treatment of
some irregular
constructions
Frank Van Eynde
www.ccl.kuleuven.be/~frank
July 25, 2006
Varna
1
Regularity and irregularity
Regular shapes
July 25, 2006
Irregular shapes
Varna
2
Survey






Irregular combinations in NPs
Lexical type constraints
Phrasal type constraints
Another look at the irregular NPs
Irregular combinations in VPs
Conclusion
July 25, 2006
Varna
3
1. Irregular
combinations in NPs
The German was für (ein) construction
The Dutch van die/van dat construction
Pseudo-partitives in English, French and
Italian
July 25, 2006
Varna
4
The was für (ein) construction
Was für einen Mann willst du eigentlich _?
Wat voor een man wil je eigenlijk _?
what for a man want you actually _?
The preposition für requires an accusative complement
- Das ist für dich/*dir/*du.
- Das ist für einen/*einem/*ein Mann.
Was für einem Mann hast du _ geholfen?
Mit was für einem Mann hat sie _ gesprochen?
Was für ein Mann ist er eigentlich _?
July 25, 2006
Varna
5
The was für (ein) construction
The interrogative was is singular
- Was hat/*haben dich so gestört?
Was für ein Mann hat das gesagt?
Was für Männer haben/*hat das gesagt?
The was für (ein) N’ construction is not a PP, but an NP.
The head is neither the preposition für nor the pronoun
was.
July 25, 2006
Varna
6
The was für (ein) construction
NP
NP
Det
was für ein
July 25, 2006
NOM
Det
NOM
Mann
was für
Unsinn
Männer
Varna
7
The was … für ein construction
Was ist das eigentlich _ für ein
Mann?
Wat is dat eigenlijk _ voor een
man?
what is that actually _ for a man?
NP
Det
NOM
für ein
Mann
Was as a predeterminer
All the boys will come.
The boys will all come.
July 25, 2006
Varna
8
The was … für construction
Was ist das denn _ für Unsinn?
Was habt ihr doch _ für
eigenartige Allergien?
July 25, 2006
NP
Varna
Det ?
NOM
für
Unsinn
9
The van die/van dat construction
Er staan van die dure glazen in de kast.
there stand of those expensive glasses in the cupboard
Er staat van dat vuile water in de sloot.
there stands of that dirty water in the ditch
Ze hebben [van die lange staarten].
they have of those long tails
Een kat met [van die lange haren]
a cat with of those long hairs
July 25, 2006
Varna
10
The van die/van dat construction
Ze heeft de hele dag zitten lezen [in dat boek].
she has the whole day sit read in that book
*Ze heeft de hele dag zitten lezen [Duitse griezelverhalen].
she has the whole day sit read German horrorstories
*Ze heeft de hele dag zitten lezen [van die rare verhalen].
she has the whole day sit read of those weird stories
July 25, 2006
Varna
11
The van die/van dat construction
Ze heeft [gitzwart haar en van die mooie blauwe ogen].
Ze heeft [van die mooie blauwe ogen en gitzwart haar].
*Ze heeft de hele dag [in die Duitse boeken en van die rare
verhalen] zitten lezen.
*Ze heeft de hele dag [van die rare verhalen en in die Duitse
boeken] zitten lezen.
Er staan/*staat van die dure glazen in de kast.
Er staat/*staan van dat vuile water in de sloot.
The van die/van dat N’ construction is an NP. Its head is not
the preposition.
July 25, 2006
Varna
12
The van die/van dat construction
von dem → vom
di la, di gli → della, degli
de le, de les → du, des
NP
Det
van die
van dat
July 25, 2006
NOM
dure glazen
vuile water
no arbitrary gaps
no morpho-phonological
changes
Varna
13
The van die/van dat construction
NP
DetP
P?
Det
van
die
dat
July 25, 2006
NP
NOM
P?
dure glazen
vuile water
van
Varna
NP
Det
NOM
die
dat
dure glazen
vuile water
14
Partitives
A lot of the mice died.
None of his mice died.
All of those mice died.
Some of my mice died.
Few of the mice died.
Five of Kim’s mice died.
NP
NP
none
PP
P
NP
of
the mice
A lot have died of those cute
little mice.
July 25, 2006
Varna
15
Pseudo-partitives
A lot of mice died.
A number of mice died.
A few mice died.
No mice died.
Some mice died.
NP
Det
NOM
a lot of
mice
* A lot have died of cute little
mice.
NP
A lot of mice were killed.
A lot of water was wasted.
July 25, 2006
Varna
Det
NOM
some
mice
16
Pseudo-partitives
Un po’ di macchine sono/*è già
vendute/*o.
Alcune macchine sono già
vendute.
NP
Un peu de maisons seront/*sera
détruites.
Tant de maisons seront/*sera
détruites.
Quelques maisons seront
détruites.
July 25, 2006
Varna
Det
NOM
un po’ di
un peu de
macchine
maisons
17
Pseudo-partitives
A whole lot of mice
An awful lot of trouble
A fat lot of luck
Lots of people
A large number of people
NP
DetP
NOM
Un bel po’ di tempo
Un petit peu de lait
Un tout petit peu de lait
Peu de gens
July 25, 2006
Varna
NP
Det ?
a whole lot
un bel po’
un tout petit peu
of
di
de
mice
tempo
lait
18
Pseudo-partitives
N’’
N’’’
a lot
of
NP
N’
NP
N
a lot
mice
July 25, 2006
NOM
of
NOM
mice
Varna
19
Taking stock
We have observed irregularities in
 the German was für (ein) N’ construction
 the Dutch van die/van dat N’ construction
 the pseudo-partitives in English, Italian and French
In all 3 there is something unusual about the preposition.
July 25, 2006
Varna
20
2. Lexical type
constraints
A lexical type hierarchy
Type constraints
Multiple inheritance
July 25, 2006
Varna
21
A lexical type hierarchy - 1
lexeme
ARGUMENTSELECTION
PART-OF-SPEECH
v-lx
…
p-lx
noarg-lx
argsel-lx
intr-lx
July 25, 2006
Varna
trans-lx
22
Multiple inheritance - 1
noarg-lx → [ARG-ST < >]
argsel-lx → [ARGST nelist ]





n-lx & noarg-lx: proper nouns, pronouns
det-lx & noarg-lx: every, which
adv-lx & noarg-lx: yes, no
adj-lx & noarg-lx: wooden, former
p-lx & noarg-lx: particles
Ze stellen de vergadering uit.
they put the meeting out
Er ruft sie jeden Tag an.
he calls her every day on
July 25, 2006
Varna
23
A lexical type hierarchy - 2
lexeme
PART-OF-SPEECH
v-lx
July 25, 2006
…
HEAD-SELECTION
p-lx
Varna
hdsel-lx
nonhdsel-lx
24
Multiple inheritance - 2
hdsel-lx → [SS | LOC | CAT | HEAD | SELECT synsem]





adj-lx & hdsel-lx: black horses, very long journeys, the person
responsible for this mess
v-lx & hdsel-lx: people speaking five languages, people who speak
five languages
adv-lx & hdsel-lx: never say anything, hardly ever speak, speak
(very) quickly
p-lx & hdsel-lx: the horses in the stable, speak with difficulty, arrive
after midnight
det-lx & hdsel-lx: every black horse, almost every horse
July 25, 2006
Varna
25
Multiple inheritance - 2
nonhdsel-lx → [SS | LOC | CAT | HEAD | SELECT none]



adj-lx & nonhdsel-lx: that horse is black, the journey is very long,
who is responsible for this mess
v-lx & nonhdsel-lx: he was cleaning the stable, they say (that) they
speak five languages
p-lx & nonhdsel-lx: put the horses in the stable, believe in miracles,
look after the children, ruft sie an,stellen de vergadering uit
July 25, 2006
Varna
26
A lexical type hierarchy - 3
lexeme
PART-OF-SPEECH
…
July 25, 2006
ARG-SELECTION
p-lx
noarg-lx
argsel-lx
Varna
HEAD-SELECTION
hdsel-lx
nonhdsel-lx
27
Multiple inheritance - 3
SS | LOC | CAT | HEAD | SELECT synsem
ARG-ST < >




adj-lx & noarg-lx & hdsel-lx: former governor, wooden tables
det-lx & noarg-lx & hdsel-lx: which horse, every door
adv-lx & noarg-lx & hdsel-lx: not knowing the answer, he decided
to keep silent
p-lx & noarg-lx & hdsel-lx: ??
July 25, 2006
Varna
28
3. Phrasal type
constraints
A phrasal type hierarchy
The head-functor phrase type
July 25, 2006
Varna
29
A phrasal type hierarchy
phrase
CLAUSALITY
clause
HEADEDNESS
non-clause
hd-ph
hd-fun-ph
hd-arg-ph
hd-comp-ph
July 25, 2006
Varna
non-hd-ph
hd-subj-ph
30
Head-functor phrases
phrase: SS | LOC | CAT | COMPS < >
hd-ph: SS | LOC | CAT | HEAD [1] part-of-speech
HD-DTR | SS | LOC | CAT | HEAD [1]
hd-subj-ph: SS | LOC | CAT | SUBJ < >
SUBJ-DTR | SS [1] synsem
HD-DTR | SS | LOC | CAT | SUBJ <[1]>
hd-fun-ph: FUN-DTR | SS | LOC | CAT | HEAD | SELECT [1]
HD-DTR | SS [1] synsem
July 25, 2006
Varna
31
Head-functor phrases
NOM
AdjP [SEL [1]]
July 25, 2006
[1] NOM
ADV
ADJ [SEL [1]]
very
tall
Varna
horses
32
Head-functor phrases
hd-fun-ph: SS | LOC | CAT | MARKING [2] marking
FUN-DTR | SS | LOC | CAT | MARKING [2]
hd-arg-ph: SS | LOC | CAT | MARKING [2] marking
HD-DTR | SS | LOC | CAT | MARKING [2]
Partition of marking: marked, unmarked (bare, incomplete).
NOM [MARKING [2]]
Det [SELECT [1], MARKING [2] marked]
[1] NOM [unmarked]
each
student
July 25, 2006
Varna
33
Prenominal functors
HEAD noun
MARKING unmarked
NOM
[2]
Det
[2] marked
the
HEAD | SEL | … | HEAD noun
| MARKING [1]
MARKING [1] unmarked
NOM
[1]
Adj
[1]
NOM
[1] unmarked
HEAD | SEL | … | HEAD noun
| MARK unmarked
MARKING marked
marking: [FORM form]
dirty
July 25, 2006
tables
water
Partition of form: determinate, indefinite,
…
Varna
34
Partitives
HEAD noun
MARKING marked
NP
[2]
Pronoun
[2] marked
none
HEAD prep
COMPS < NP[determinate] >
MARKING unmarked
PP
COMPS < >
P
COMPS <[1]>
[1] NP
determinate
HEAD | SEL | … | HEAD noun
| MARK unmarked
MARKING | FORM determinate
Partition of form: determinate
(demonstrative, possessive, relative,
interrogative, …), indefinite, …
of
July 25, 2006
his friends
Varna
35
4. Another look at the
irregular NPs
The Dutch van die/van dat construction
The German was für (ein) construction
Pseudo-partitives in English, French and
Italian
July 25, 2006
Varna
36
The van die/van dat construction
p-lx & noarg-lx & hdsel-lx: van
NOM
[2]
P
[2] marked
indefinite
van
July 25, 2006
HD| SEL | …| HEAD noun
| MARKING unmarked
MARKING | FORM demonstrative
NOM
[1]
Det
[1] marked
demonstrative
NOM
unmarked
die
dure glazen
HD|SEL | …| HEAD noun
| MARK | FORM demo
MARKING | FORM indefinite
The result is an indefinite NP, and not
a PP.
Varna
37
The was … für (ein) construction
HD | SEL|…| HD noun
| MARK unmarked
MARK marked [FORM indef ]
NOM
[2]
P
[2] incomplete
für
für
HD | SEL|…| HD noun
|MARK | FORM indef
MARK incomplete [FORM für ]
NOM
[1]
Det
[1] marked
indefinite
NOM
unmarked
Bare nominals are indefinite.
(was) für Männer, für Unsinn
ein
July 25, 2006
Mann
Varna
38
The was für (ein) construction
HD | SEL |…| HD noun
| MARK | FORM für
MARK marked [FORM interrog ]
NOM
[2]
Pro
[2] marked
interrogative
was
July 25, 2006
NOM
[1]
P
[1] incomplete
für
NOM
unmarked
indefinite
für
Unsinn
The result is an interrogative NP
which is headed by a common
noun. Its case is independent of
the requirements of für and its
number is supplied by the common
noun, not by the pronoun.
Varna
39
Pseudo-partitives
p-lx & noarg-lx & hdsel-lx: of
NOM
[2]
Det
[2] marked
plenty
July 25, 2006
HD| SEL | … | HEAD noun
| MARK unmarked
MARK incomplete [FORM of ]
NOM
[1]
P
[1] incomplete
of
NOM
unmarked
of
tables
HD|SEL| …| HEAD noun
| MARK | FORM of
MARK marked [FORM indef ]
Varna
40
Pseudo-partitives
n-lx & noarg-lx & hdsel-lx: of
NOM
[2]
NOM
SEL [1]
[2]
[1] NOM [of]
Det
[2] marked
indefinite
NOM
SEL [1]
unmarked
a
lot
July 25, 2006
of tables
HD|SEL| …| HEAD noun
| MARK | FORM of
MARKING unmarked
The result is a marked indefinite
plural NP.
Varna
41
Pseudo-partitives
HD | SEL | … | HEAD noun
| MARK unmarked
MARK marked [FORM poss ]
NOM
[3]
Det
[3] marked
possessive
his
July 25, 2006
NOM
[2]
NOM
SEL [1]
[2] unmarked
[1] NOM [of]
incomplete
large number
of questions
Varna
42
Summing up
Making use of a lexical type which inherits from
the independently needed p-lx, noarg-lx and
headsel-lx, and of a phrasal type which subsumes
the phrases of type head-adjunct, head-specifier and
head-marker, we can account for the irregular NPs
without any further stipulations.
July 25, 2006
Varna
43
5. Irregular
combinations in VPs
The English infinitival to
The English for-to clauses
July 25, 2006
Varna
44
The English infinitival to
PP[to]
COMPS < >
P[to]
COMPS <[1]>
[1] VP[base]
to
go to Varna
July 25, 2006
Why not P? Pullum 1982
[2]
One can relate it to his lack of vision.
*One can relate it to see better.
[4]
He went right to the captain
I wish (*right) to see the captain
[7]
To the transporter room with him!
*To get out of my sight with him!
[10]
Mudd, who did you give it TO?
*Proceed I intend TO, Mr. Sulu.
Varna
45
The English infinitival to
[3]
I want to go home
*I want to my home planet
[5]
I agree to boldly go where no man
has gone before.
*I agree to boldly transportation to
strange new worlds.
[6]
To survive an ion storm will be
tough even for the Enterprise.
*To within a thousand miles of a
black hole will be tough.
July 25, 2006
[9]
I’m gonna explore strange new
worlds.
*I’m gonna Epsilon Delta IV.
[1]
No other prepositions take
uninflected VP complements.
[8]
Problem with for-to clauses:
S → NP PP or S → PP PP
Varna
46
The English infinitival to
I can’t come, but he can _.
He didn’t come, but she did _.
*I can’t come but I’d like _.
*I couldn’t come, but I wanted
_.
I can’t come but I would like to
_.
[1]
COMPS < >
[1] V[inf, + AUX]
COMPS <[2]>
[2] VP[base]
to
go to Varna
Why infinitival?
*toes, *toing, *toed, *to to
July 25, 2006
Varna
47
The English infinitival to
No verbal morphology: *toes, *tid, *ton’t, *toesn’t, *tidn’t,
*tone, *toing, *to to
Special VFORM value which only applies to to. In this way
every morphologically invariant word can be argued to be a
verb (with a defective paradigm).
No conjunction with other (subject raising) verbs:
She [is and remains] our best goalkeeper.
We will [invade and conquer] the valley.
They were all [eating and drinking].
*It may [seem and to] to amuse her.
July 25, 2006
Varna
48
The English infinitival to
'Only auxiliaries and to license VP ellipsis'
BUT not all of them
* He shouldn’t be punished for a trifle like that, but he is
being _.
? They haven’t (got) to leave, but I’ve had _.
AND they are not the only ones
John is more reliable than Kim _.
Pete has more sheep than all of us together _.
Nobody is as fast as my friend from Illinois _.
July 25, 2006
Varna
49
The English infinitival to
p-lx & noarg-lx & hdsel-lx: to
[1] VP[inf]
[2] marked [to]
P, SELECT [1]
[2] marked [to]
[1] VP[inf]
unmarked
HD|SEL|…|HEAD verb [VF inf]
|SUBJ <NP>
|COMPS < >
|MARKING unmarked
MARKING marked [FORM to]
Why Pullum’s arguments against P
status do not apply.
to
July 25, 2006
go to Varna
Varna
50
The English infinitival to
Partition of synsem: canonical-synsem, noncanonical-synsem (prosynsem, gap-synsem).
The prepositions can require their head to have a canonical
synsem value (German was für, Dutch van die/van dat, French
de, Italian di, English of ), but they can also allow it to be of
a noncanonical type.
LOC|CAT|HEAD|SELECT gap-synsem [LOC [1]]
SLASH { [1] }
... and proceed I intend to _, Mr. Sulu.
July 25, 2006
Varna
51
The English for-to clause
VP[inf], to
SUBJ < >
[3] NP
[2]
VP[inf], to
SUBJ <[3]>
P, SELECT [1]
[2] for
[1] NP[acc]
for
us
July 25, 2006
Varna
to go to Varna
52
6. Conclusion
Exploiting the possibilities of HPSG’s
constraint-based lexicalism it is possible to
provide a regular treatment of a number of
irregular constructions
July 25, 2006
Varna
53
Descargar

A regular treatment of some irregular constructions