Head-Driven Phrase Structure Grammar (HPSG)

Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Motivation
Head-Driven Phrase Structure Grammar (HPSG)
The main mechanism of HPSG for linguistic analysis lies in its
feature architecture
Head-Driven Phrase Structure Grammar (HPSG)
Trees, bar-level distinctions, subcategorization, agreement,
etc. are handled via feature structures
Linguistics 614
Rough comparison to LFG:
With thanks to Detmar Meurers
Similar: use of feature structures, highly lexical
Spring 2015
Introduction
Signs
Different: only one layer of syntax, functions are not
primitives, stronger reliance on structure-sharing
Phrases
Raising/Equi
UDCs
Introduction
Signs
Motivation
Motivation
Preview: lexical entry for letter
Unpacking letter

word
phon








syn









"
#



head 
pernum 3sng 


agr 1


gend
neut 


* "
#+




1
agr
spr

D


count +




D
E
comps PP
<letter>


noun
Phrases
Raising/Equi
UDCs
syn: holds syntactic information (vs. phon or sem)
synsem is also commonly used (e.g., these slides)
head: category & inflectional properties
spr & comps: subcategorization properties
Sometimes 3 lists (subj, spr, comps), sometimes 1 list
(subcat)
Note that detailed properties of selected items can be specified
(This entry more or less follows Carnie (2013) / Sag & Wasow (1999),
which most of the rest of the slides don’t.)
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Components: from a linguistic perspective
Components: from a linguistic perspective
HPSG grammars
Lexical rules
Linguistic perspective
Phrases
Raising/Equi
An HPSG grammar, from a linguistic perspective, consists of
a) a lexicon: licensing basic words
b) lexical rules: licensing derived words
c) immediate dominance (ID) schemata: licensing
constituent structure
d) linear precedence (LP) statements: constraining word
order
e) a set of grammatical principles: expressing generalizations
about linguistic objects
e.g., plural rule (adapted from Carnie 2013):

noun

phon

D1
E
arg-st count +


word

FNPL 1 
syn|head|agr|num pl
⇒ phon
Any feature not specified is unchanged
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Components: from a linguistic perspective
Components: from a formal perspective
Grammatical principles
HPSG grammars
Phrases
Raising/Equi
UDCs
Raising/Equi
UDCs
Raising/Equi
UDCs
Formal perspective
Grammatical principles allow one to generalize over entire classes
of objects
e.g., every complete sentence in English is tensed
An HPSG grammar formally consists of
These can be interpreted as constraints on the acceptable feature
structures
I. the signature as declaration of the domain, and
HPSG is constraint-based: grammars are sets of constraints
which hold simultaneously
This determines the collections of admissible linguistic
structures
II. the theory constraining the domain.
Constraints do not define an order of the derivation or
generation of linguistic signs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Components: from a formal perspective
Components: from a formal perspective
The signature of an HPSG grammar
Signature: some examples
Signs


sign
phon
list(phonstring)
synsem synsem
The signature
defines the ontology (‘declaration of what exists’):
which kind of objects are distinguished, and
which properties of which objects are modeled.
word
consists of
the type hierarchy (or sort hierarchy) and
the appropriateness conditions, defining which type has
which appropriate attributes (or features) with which
appropriate values.
Part of speech
part-of-speech
adj
Introduction
Signs
Phrases
Raising/Equi
phrase
UDCs
Introduction
Signs
adv
det
noun
prep
Phrases
Components: from a formal perspective
Components: from a formal perspective
Models of linguistic objects
An example for a typed feature structure
What do the mathematical structures used as model for HPSG
theories look like?
The objects are modeled by typed feature structures, which
can be notated as directed graphs.
These models represent objects in the world → they are total
with respect to the ontology declared in the signature.
Naturally, knowledge about objects in the world is partial.
Formally speaking, the feature structures used as models are
totally well-typed: Every type has every one of the attributes
and their values which are appropriate for it.
sort-resolved: Every type is maximally specific.
type and sort as well as attribute and feature are used synonymously.
verb
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Components: from a formal perspective
Components: from a formal perspective
The theory of an HPSG grammar
How do we express a theory? Descriptions
Raising/Equi
UDCs
A description language describes sets of objects.
The AVM (attribute value matrix) notation is used to
compactly write down these descriptions.
A theory is a set of description language statements, often
referred to as the constraints.
Descriptions consists of three building blocks:
Type descriptions single out all objects of a given type
The theory singles out a subset of the objects declared in the
signature, namely those which are grammatical.
Example:
word
Attribute-value pairs describe objects that have a particular
property.
A linguistic object is admissible with respect to a theory iff it
satisfies each of the descriptions in the theory and so does
each of its substructures.
The attribute must be appropriate for the particular type of
object, and the value can be any kind of description.
Example:
h
i
spouse name mary
Tags (structure sharing) to specify token identity
Example: 1
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Components: from a formal perspective
Components: from a formal perspective
Descriptions (cont.)
An example
AVM - The pronoun she

Complex descriptions are obtained through
conjunction (∧)
disjunction (∨)
negation (¬)
In the AVM notation, conjunction is implicit.
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction



























































word
phon
<she>

local




cat










synsem|loc 
cont










context


Signs
Aspects of the signature
Aspects of the signature
Signs
Motivating subcat
Aspects of the signature


sign
phon
list(phonstring)
synsem synsem
word
phrase
dtrs constituent-structure



 local
category
synsem



head

local
category categoryhead
local
content content subcat list(synsem)
non-local non-local
context context marking marking

(1) a.
b.
c.
d.
I
I
I
I
Raising/Equi









cat





noun




head

case nom 



subcat hi





ppro






ref




per third





index 1




num sing 





gend fem 


restr {}




context




 




 psoa



backgr reln female 



 inst 1
 
Phrases
laugh.
saw him.
give her the book.
said that she left.
Raising/Equi
(<NP>)
(<NP NP>)
(<NP NP NP>)
(<NP S[that]>)
Cannot always be derived from semantics:
(2) a.
b.
Paul ate a steak.
Paul ate.
(3) a. Paul devoured a steak.
b. * Paul devoured
UDCs
(<NP>)
(<NP NP>)
(<NP>)
(<NP NP>)
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Aspects of the signature
Aspects of the signature
Syntactic category information
Properties of particular categories
Raising/Equi
UDCs
head
functional
spec synsem


substantive

prd boolean
mod mod-synsem
vform
finite infinitive base
case
gerund present-part. past-part. passive-part. nominative
pform
accusative of
to
...
marker determiner


noun
preposition
...
adjective verb
vform vform  case case pform pform


aux
boolean
inv
boolean
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Aspects of the signature
Aspects of the signature
Motivating VFORM
Motivating CASE
(4) a. Peter will win the race.
b. * Peter will won the race.
c. * Peter will to win the race.
(base form)
(5) a. Peter has won the race.
b. * Peter has win the race.
c. Peter has to win the race.
(→ different verb)
(past participle)
Signs
(7) a. He left.
b. * Him left.
(acc)
(to-infinitive)
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Aspects of the signature
Indices
Semantic representations


index
person person 


number number
gender gender
there
UDCs
(nom)
Aspects of the signature
referential
Raising/Equi
(8) a. She sees him.
b. * She sees he.
(6) a. Peter seems to win the race.
b. * Peter seems win the race.
c. * Peter seems won the race.
Introduction
Phrases
Raising/Equi
content
quant
it
person
number
gender
first second third
singular plural
masculine feminine neuter
psoa


nom-obj
index

index
restriction set(psoa)



laugh’
give’
think’


laugher ref 
giver ref thinker ref
given ref thought psoa
gift ref
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Aspects of the signature
Aspects of the signature
Auxiliary data structures
Abbreviations for describing lists
⊤
boolean
true false
list
...


empty-list non-empty-list
first ⊤

rest list
Signs
Phrases
Raising/Equi
is abbreviated as
e-list, <>
non-empty-list
is abbreviated as
ne-list
is abbreviated as
first 1
rest 2
. . . 1 | hi
is abbreviated as

Alternative names for the attributes first (ft) and rest (rt) of
non-empty-list are head (hd) and tail (tl).
Introduction
empty-list
UDCs

first 1

first 2 
rest
rest 3
Introduction
is abbreviated as
Signs
Theory: Lexicon
Abbreviations for common AVMs
Theory: The Lexicon
S: 1
VP: 1
abbreviated AVM


synsem



head noun 


local category

subcat hi
content|index 1


synsem



head verb 


local category subcat hi

content

synsem



local category
content
Introduction
Signs
1 , 2 | 3

word
phon






synsem|loc






head verb

subcat synsem 
1
Raising/Equi
UDCs
Introduction
Signs

Phrases
An example lexicon

phon <drinks>






verb


head
 



 cat 
vform fin


 


 
subcat NP[nom] 1 [3rd,sing], NP[acc] 2 
s|l 




 

drink’
 

 

 cont drinker 1 

drunken 2


phon <drink>


 
verb


head
 

vform fin
 cat 



W 

 

subcat NP[nom] 1 [plur], NP[acc] 2 
s|l 



 


drink’
 

 



 cont drinker 1

drunken 2
W

phon <give>



 
verb


 
head




 
vform
fin



 cat 

 

NP[nom] 1 [plur], NP[acc] 2 , 
 

subcat
 

PP[to] 3
 



s|l 

 
give’

 



 
giver

1

 cont 



 

gift 2 


given 3





verb

head


cat 


vform
fin





subcat NP[nom] 1 [3rd,sing] 






laugh’
content
laugher 1
<laughs>

Theory: Lexicon

UDCs
... 1
with each of the lexical entries being descriptions, e.g.:
Theory: Lexicon
word →
Raising/Equi
word → lexical-entry 1 ∨ lexical-entry 2 ∨ . . .
1
Phrases
UDCs
The basic lexicon is defined by the Word Principle as part of the
theory. It is an implicational statement defining which of the
ontologically possible words are grammatical:
Pollard and Sag (1994) make use of the following abbreviations for
describing synsem objects:
NP 1
1 | 2
Phrases
Aspects of the signature
Abbrev.
Raising/Equi
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Theory: Lexicon
W

phon




s|l


W
Introduction
Signs
Phrases
<to>
W



preposition 
head

cat 

pform to 





NP[acc]
subcat
1 


cont index 1


phon




synsem|loc



phon <think>






verb


head






vform fin
cat 








subcat NP[nom] 1 [plur], S[fin]: 2 
s|l








think’







cont thinker 1 

thought 2
W

Introduction
Signs
Phrases

phon
<poets>
UDCs
Raising/Equi
UDCs


head noun

cat


subcat hi


"
#



per third 

cont index
num plur

<wine>



head noun


cat





subcat hi


synsem|loc 
"
#


per third 


cont index
num sing
Raising/Equi
UDCs

Introduction
Signs
Phrases
Introduction
Introduction
Different kinds of phrasal constructions in HPSG
The structure of a simple phrase

phrase
phon




synsem













dtrs







For phrases, the dtrs feature has a constituent-structure type:
constituent-structure


head-struc
head-dtr sign

comp-dtrs list(phrase)
head-comps-struc
Introduction
Raising/Equi
Theory: Lexicon


coordinate-structure
conj-dtrs
set(sign)
conjunction-dtr word




head-filler-struc
head-adjunct-struc
head-marker-struc



head-dtr
phrase
phrase
 head-dtr phrase
 head-dtr

marker-dtr word  adjunct-dtr phrase filler-dtr phrase
comp-dtrs elist
comp-dtrs
elist
comp-dtrs elist
Signs
Phrases
Raising/Equi
UDCs
Introduction




verb



head


local|cat
vform
fin





subcat elist



head-comp-struc






word



phon



<walks>




 
head-dtr 
verb
 





synsem|loc|cat|head


vform
fin


 





+
* phrase


 
phon
<Kim>




comp-dtrs 
 

noun
 

synsem|loc|cat|head
case nom
<Kim,walks>



Signs
Phrases
Raising/Equi
UDCs
Introduction
Introduction
Sketch of an example for head-complement structures
Sketch of an example for head-complement structures with
a ditransitive verb
c



phon <Kim>
"


verb
head


vform fin 
synsem|loc|cat 
subcat hi

c
h


noun

synsem|loc|cat|head
case nom



phon <walks>
synsem|loc|cat


verb

synsem|loc|cat|head
vform fin
#
head 4
subcat hi
h
phon <John>
synsem 1
"
synsem|loc|cat
h
"
##
head 4
subcat 1
c
c
 phon <the book>
phon <her>



synsem
synsem
2
3
verb



head 4 vform fin 

synsem|loc|cat 

subcat 1 , 2 , 3

phon <gives>
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Introduction
Introduction
Head-Feature Principle:
Subcategorization Principle:
Raising/Equi
UDCs
The head value of any headed phrase is structure-shared with the
head value of the head daughter.
In a headed phrase (i.e. a phrasal sign whose dtrs value is of sort
head-struc), the subcat value of the head daughter is the
concatenation of the phrase’s subcat list with the list (in order of
increasing obliqueness) of synsem values of the complement
daughters.
phrase
→
dtrs headed-structure
synsem|loc|cat|head
1
dtrs|head-dtr|synsem|loc|cat|head 1
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
Introduction
Immediate Dominance Schemata
Subcat Principle: (cont.)
Immediate Dominance Principle (for English):
dtrs headed-structure →


synsem|loc|cat|subcat 1
#
"

head-dtr|synsem|loc|cat|subcat 1 ⊕ 2 


dtrs
comp-dtrs synsem2sign 2
rest
rest
2
2 ⊕ 3
:= "e-list.
#
first synsem 1
first 1
.
:=
synsem2sign
synsem2sign e-list
rest 2
Introduction
Signs
rest synsem2sign 2
Phrases
Raising/Equi


verb
head
∨
¬
verb


synsem|loc|cat 
inv −




subcat hi








head-comp-struc


dtrs

head-dtr phrase
comp-dtrs sign



verb
∨ ¬ verb 
synsem|loc|cat head
inv −




∨
subcat synsem






head-comp-struc
dtrs
head-dtr word
and synsem2sign encoding conversion of synsem to sign lists defined
as follows
phrase
→
dtrs headed-struc

with ⊕ standing for list concatenation, i.e., append, defined as
follows
e-list
⊕
:= 1 .
1
first 1
first 1
⊕ 3 :=
.
UDCs
Introduction
Signs
Phrases
Immediate Dominance Schemata
Immediate Dominance Principle (for English):
Head-Subject Phrases (Schema 1)

"
synsem|loc|cat
c
synsem 2
(Head-Subject)
(Head-Complement)
Raising/Equi
Immediate Dominance Schemata


verb
synsem|loc|cat head
inv + 



∨
subcat
hi
(Head-Subject-Complement)





head-comp-struc 
dtrs
head-dtr word
head-marker-struc
∨ dtrs
(Head-Marker)
marker-dtr|synsem|loc|cat|head marker



head-adjunct-struc


(Head-Adjunct)
∨ dtrs adj-dtr|synsem|loc|cat|head|mod 1 
head-dtr|synsem
1
UDCs
head
1
subcat hi
#
h
##
"
head
1
synsem|loc|cat
subcat 2
"
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
Immediate Dominance Schemata
Immediate Dominance Schemata
Head-Complement Phrases (Schema 2)
Head-Subject-Comp Phrases (Schema 3)
"
"
"
"
##
head 1
synsem|loc|cat
subcat 2
c
c
"h
## head 1
synsem 3 .. . synsem n
synsem|loc|cat
subcat 2 , 3 ,..., n
Introduction
Signs
Phrases
Raising/Equi
UDCs
"
synsem|loc|cat
#
head 1
subcat hi
c
c
## "h
head 1
synsem
synsem
2
n
..
.
synsem|loc|cat
subcat 2 ,..., n
Introduction
Signs
Phrases
Raising/Equi
Immediate Dominance Schemata
Immediate Dominance Schemata
Sketch of an example with an auxiliary
Sketch of an example with an inverted auxiliary
c
phon <John>
synsem 1



phon <John, can, go>


head 3 

synsem|loc|cat
subcat hi
phon




#
head 3

synsem|loc|cat
subcat 1
<can, go>
"
h
<can>




verb


vform fin



head 3




synsem|loc|cat 
+
aux




inv
−


D
subcat 1 NP nom , 2 VP bse
Introduction
Signs
Phrases
UDCs


phon <can, John, go>


head
3


synsem|loc|cat
subcat hi
h
phon
UDCs
h
c
 phon <go>

synsem 2







E

Raising/Equi
c
c
 
phon <go>
phon <can>
phon <John>





synsem
synsem
1
2
verb









head 3vform fin





synsem|loc|cat 
+
aux







inv
+


D
E

subcat 1 NP nom , 2 VP bse
UDCs
Introduction
Signs
Immediate Dominance Schemata
Immediate Dominance Schemata
Complementizer as heads
Markers
Phrases
Raising/Equi
head
Some verbs select a complementized sentential complement
headed by a verb with a specific verb form:
functional
spec synsem
marker determiner
(9) I demand that he leave/*leaves immediately.
substantive
adjective verb noun preposition . . .
marking
marked
unmarked
comp
that
...
for
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
Immediate Dominance Schemata
Immediate Dominance Schemata
Marking Principle:
Lexical entry of the marker that
In a headed structure, the marking value coincides with that of
the marker daughter if there is one, and with that of the head
daughter otherwise.

phon







synsem—loc—cat





phrase
→
dtrs headed-structure

synsem|loc|cat|marking 1


head-mark-struc
dtrs
marker-dtr|synsem|loc|cat|marking 1


synsem|loc|cat|marking 1

∨
¬head-mark-struc
dtrs
head-dtr|synsem|loc|cat|marking 1

Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Determiners
Sketch of an example for a head-marker structure
SPEC Principle:
Introduction
Signs
Phrases
Raising/Equi
UDCs
#
"
phrase
dtrs (marker-dtr ∨ comp-dtrs|first) |synsem|loc|cat|head functional
h



mark

head spec 3 




subcat hi

marking 1 that






verb


head




vform
fin
∨
bse



loc—cat 



marking unmarked
 


subcat hi



If a nonhead daughter in a headed structure bears a spec value, it
is token-identical to the synsem value of the head daughter.


phon <that,John,laughs>



head 2




s|loc|cat subcat hi 
marking 1
m

mark




spec




head






subcat hi
marking that
Signs
Immediate Dominance Schemata

phon <that>




synsem|loc|cat

<that>

UDCs


phon <John laughs>


 
verb


 
head 2 vform fin 
 


s 3loc|cat 

 
marking unmarked 
subcat hi
Phrases
Raising/Equi
→
"
dtrs
UDCs
Introduction
#
(marker-dtr ∨ comp-dtrs|first) |synsem|loc|cat|head|spec 1
head-dtr|synsem
1
Signs
Determiners
Determiners
Determiners
Determiners (cont.)
Phrases
Raising/Equi
UDCs
Determiners select their nominal sister:
Common nouns subcategorize for their determiners:


phon <book>





head
noun

*"
#+






cat 
head
det 



subcat
loc|cat




subcat
hi



synsem|loc 






index 1





cont 

reln
book





restr
inst 1

phon <every>




det














cat head spec









synsem|loc 




subcat hi





quant




cont 5det


restind

qstore 5



"
#



head noun


cat

subcat DetP 
loc 


cont 2









forall


2

Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Determiners
Possessives
Sketch of the head-comp phrase every book
Lexical entry for the possessive pronoun my

phon <every,book>



head
5




cat
subcat hi 
s|loc
cont 2
c
h


phon <every>





det





head

cat 



spec 3 








synsem 4loc

subcat
hi










det
forall 


cont
restind 2
Introduction
Signs

phon <book>

 

 
cat
s 3loc


cont
Phrases






det










index
head
1








spec
N’:
cat






restr 2 










subcat hi





synsem|loc 





ref












per
1st
cont|index
3








num sing






context|c-indc|speaker 3



 


 det

the














index
1












qstore 
 reln

poss 


 







restind








restr possessor 3  ∪ 2 











 possessed 1





"

#

head 5 noun

subcat 4


2
Raising/Equi
UDCs
phon
Introduction
<my>


Signs
Phrases
Possessives
Lexical entry for possessive ’s
Sketch of an example for a determiner phrase
Signs
Phrases

phon <Mary’s>







det


#
"









1
index
head










spec
N’:
cat






restr 2 











subcat
hi




"
#






synsem|loc 
npro

cont






3
index


















reln
naming











context|backgr bearer 3
 






 name

mary








det
the















index
1















qstore 


poss 
 

 reln


restind 











restr possessor 3  ∪ 2 












 possessed 1


Raising/Equi
UDCs
Introduction
Signs
Adjuncts
Adjuncts
Lexical entry of an attributive adjective
Semantics Principle

word
phon












synsem|loc













adj



prd

−






#
"





head
noun


D
cat|head 

E 

cat






loc|cat|head
det
subcat






mod|loc 








index 1 index

content


restr 2








index 1


(
)




rel red
cont




restr
∪ 2
arg1 1
<red>

Raising/Equi
UDCs
Raising/Equi
UDCs


phon <’s>






det






head



index
1








spec
N’:








restr
2


cat 



+
*

synsem|loc 





npro






subcat
NP:
5




index 3








cont
5



 




det
the
















index 
1












qstore 
 reln
poss 


 








restind








restr possessor 3  ∪ 2 











 possessed 1



Introduction
UDCs

Possessives

Raising/Equi
Phrases
In a headed phrase, the content value is token-identical to that
of the adjunct daughter if the dtrs value is of sort head-adj-struc,
and with that of the head daughter otherwise.
phrase
→
dtrs headed-structure

synsem|loc|content 1

head-adj-struc
dtrs
adjunct-dtr|synsem|loc|content 1


synsem|loc|content 1

∨
¬head-adj-struc
dtrs
head-dtr|synsem|loc|content 1


Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Adjuncts
Raising/Equi
Sketch of an example for a head-adjunct structure
Review of ID Schemata


verb
∨ ¬ verb 
synsem|loc|cat head
inv −




subcat hi
phrase




→
(Hd-Sbj)

dtrs headed-struc


head-comps-struc


head-dtr phrase

dtrs
comp-dtrs sign



verb
∨ ¬ verb 
synsem|loc|cat head
inv −




∨
(Hd-Cmp)
subcat synsem





head-comps-struc
dtrs
head-dtr word



verb

synsem|loc|cat head
inv
+



subcat hi
(Hd-Sbj-Cmp)
∨



head-comps-struc 
dtrs
head-dtr word



phon <red, book>
"
#



head 2




cat
subcat 1 
s|loc
cont 4
a
h


phon <red>




adj



cat|head prd − 



s|loc


mod 3 

cont 4


phon <book>
"
"
##

head 2 noun


D s 3 loc|cat
E 
subcat 1 loc|cat|head det
Introduction
Signs
Phrases
Raising/Equi
UDCs
Towards an analysis: Unsaturated Complements

word
phon










synsem







In English, many verbs and adjectives subcategorize for an
unsaturated complement.
a complement can be specified as
hIn other words,
i
subcat NP , rather than subcat hi
The Head-Subject Schema allows for this.
And this will give consider access to the lower subject, as well
as its own subject.
Signs
Phrases
Raising/Equi
UDCs
Subject oriented equi verbs

word
phon











synsem








Introduction


cat

verb

head
vform

D

subcat NP 1 ,


try
tryer
1 ref
soa-arg 2
Phrases
Raising/Equi
UDCs
<seem>

synsem


local










cat


local 










cont
Signs










cat






verb




head



vform bse



h
D

i E

subcat 1 , VP inf, subcat 1 : 2 





seem
soa-arg 2
Phrases
Raising/Equi
UDCs
Object oriented raising verbs
<try>

synsem


local










cat




local 











cont
Signs
Subject oriented raising verbs
Where does it say in these ID schemata that every subcategorized
item must be realized?
Introduction
Introduction






bse

h
i E

VP inf, subcat NP 1 : 2 





word
phon











synsem








<believe>

synsem


local










cat




local 











cont

cat

verb

head
vform

D

subcat NP 1 ,


believe
believer 1 ref
soa-arg 3












bse

h
i E

2 , VP inf, subcat 2 : 3 





Introduction
Signs
Phrases
Raising/Equi
UDCs
Object oriented equi verbs

word
phon












synsem









Introduction
<persuade>

synsem


local










cat





local 












cont



Introduction
Phrases
Phrases
Raising/Equi

phrase
p <It seems to rain> 

#
"




head 9 

cat

s|l 
subc hi 



cont 10
c


word
p <it>


"
#

 



head noun 


cat
s 5
subc hi

l 


cont index it
h

phrase
p <seems





cat
s|l 


cont
h
"


#


head 9

subc 5 

to rain>
10
c



phrase
word
 p <to rain>
p <seems>










#
"
"
#










verb
verb






9


head
6
head











vform
fin
vform
inf
cat




 s 8l 

cat 

D
i E
h




 


s|l 
subc 5 NP it
 

subc 5 , 8 VP inf, subc 5 : 7 





"
#


cont 7 rain





cont 10 seem
soa-arg 7
UDCs
Analysis of: *It wants to rain
Introduction
Signs
Phrases
Raising/Equi

phrase
p

<wants to rain>


"
#






head 9

cat


s |l 

3
subc



con 10
To account for UDCs, we will use the feature slash.
Named after the categorial grammar categories (S/NP as an S
missing an NP).
c


phrase
word
p <wants>
 p <to rain>



 
"
#






verb



 

head 9




 
vform
fin


cat

 s 8l cat


h
D



i E
 





 
5
11
NP
: 7 
subc 3 NP 11 , 8 VP inf, subc
s|l 





con




want






con 10wanter 11 ref





soa-arg 7

This is a non-local feature which:


#
"

verb

head 6 vform inf 





subc 5 NP it

7 rain

originates with a trace,
gets passed up the tree,
and is finally bound by a filler
Introduction
An example for a strong UDC
Signs
Phrases
Raising/Equi
The bottom of a UDC: Traces
S
f
NP
h
S
c
NP
we

word
phon




synsem

h
VP
h
V
know
UDCs
Unbounded dependency constructions (UDCs)

Johni
UDCs


Signs
Raising/Equi
Analysis of: It seems to rain










cat



verb



head
vform bse



h
D

i E


subcat NP 1 , NP 2 , VP inf, subcat NP 2 : 3 





persuade




persuader 1 ref









persuadee 2 ref

soa-arg
3
h
Signs
c
1 NP
she
c
S
loc|cat|subcat hi
h
VP
h
i
loc|cat|subcat 1
h
V
c
2 NP
likes
i
h
i
loc|cat|subcat 1 , 2


hi



local 1

"
#

inherited|slash 1 


nonloc
to-bind|slash {}
phonologically null, but structure-shares local and slash
we’ll talk about to-bind later
UDCs
Introduction
Signs
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
The middle of a UDC
Traces
The Nonlocal Feature Principle (NFP)
Because the local value of a trace is structure-shared with the
slash value, constraints on the trace will be constraints on the
filler.
For each nonlocal feature, the inherited value on the mother is
the union of the inherited values on the daughter minus the
to-bind value on the head daughter.
For example, hates specifies that its object be accusative, and
this case information is local
So, the trace has synsem|local|cat|head|case acc as part
of its entry, and thus the filler will also have to be accusative
(10) *Hei /Him i , John likes
Introduction
Signs
In other words, the slash information (which is part of
inherited) percolates “up” the tree.
This allows the all the local information of a trace to “move
up” to the filler.
i
Phrases
Raising/Equi
UDCs
Introduction
Signs
Phrases
Raising/Equi
The top of a UDC: Filler-head structures
The top of a UDC: Filler-head structures
Filler-head schema
Explanation of the schema
UDCs
phrase
→
dtrs head-filler-struc


loc|cat





head-dtr|synsem 






nonloc


filler-dtr|synsem|local 1
Introduction
Signs


Filler and trace are identified as the exact same thing (as far
as their local structure is concerned)
The trace is “bound” by the to-bind feature; this prevents
the slash value from going any higher in the tree
Only saturated finite verbs (i.e., sentences) license such
structures

verb
head


vform fin 




subcat hi

"
#

inherited|slash element 1 



to-bind|slash
1
Phrases
Raising/Equi
The top of a UDC: Filler-head structures
Example for a structure licensed by the filler-head schema
UDCs
Introduction
Signs
Phrases
NP
f
local 1
h
"
"
##
inherited|slash . . . , 1 ,. . .
nloc
to-bind|slash
1
Raising/Equi
The analysis of the strong UDC example
f
nloc|inherited|slash {}
UDCs
local
Johni
3
S
nloc|inherited|slash {}
h
"
nloc
c
NP
we
"
h
"
S
##
inherited|slash 3
3
to-bind|slash
h
VP
#
inherited|slash 3
nloc
to-bind|slash {}
V
c
S
know
c
1
NP
she


loc|cat|subcat hi

inherited|slash 3 
nloc
to-bind|slash {}
h
VP


loc|cat|subcat 1

inherited|slash 3 
nloc
to-bind|slash {}
h
V
loc|cat|subcat 1 , 2
nonloc|to-bind|slash {}
likes
c
NP
loc 3
2
nloc|inher|slash 3
i
UDCs