Académique Documents
Professionnel Documents
Culture Documents
PRO-DROP PARAMETER
by
Gary Martin Gilligan
December 1987
UMT
Dissertation Publishing
UMI DP71322
Published by ProQuest LLC (2015). Copyright in the Dissertation held by the Author.
Microform Edition ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code
Pro
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, Ml 481 06- 1346
.. Dissertation
D ate
December
1987
DISSERTATION COMMITTEE
co-chairman
co-chairman
jB?.
..
ACKNOWLEDGMENTS
I am pleased to boast of the friends and colleagues
|
i
I
|
i
by alternating between conflict and reassurance, always in |
I
order to extract my best effort.
Foremost among the
|
I
combatants I must thank Osvaldo Jaeggli, who I may yet
j
please.
his cautions.
gentlemen.
I
My gratitude extends also to Mario Saltarelli and
Where would I
get a job.
Kudos also go to my parents, who taught me to dream.
i
i
iii
TABLE OF CONTENTS
PAGE
LIST OF TABLES
xi
ABSTRACT
xiii
j
!
f r a m e w o r k ........
.......................
11
15
19
..........
31
35
................... 40
....
45
46
. .
74
74
78
.................
94
105
113
........
. . 120
125
.................. 127
129
...........................
146
............
147
149
2.3.4 Conclusions
...............................
155
i
i
....
163
172
173
. 179
183
.......................
185
. . . 196
i
j
. . 197
..............
........
200
. . 201
..............
206
208
vi
3.4.1.9 Summary
...........................
209
.........
218
218
219
.....................
228
.....................
. .. .
234
239
identification .........................
. . . . .
242
. 242
vii
279
....................
289
...................................
299
4.3.5 Summary
identified by predication
.......................
301
. . . 329
inSpanish . . . 339
................... 353
374
..........
378
387
i
j
. 387
398
viii
. . 410
5.3.1
......................
419
5.3.2
429
................. 436
451
455
ix
LIST OF TABLES
chart 1:
59
91
92
chart 2:
chart 3:
chart 4:
chart 5:
98
102
131
389
i glosses:
ABL
ABS
ACC
AOR
ASP
AUX
CL
CLAS
CMPR
COND
CONJ
CTR
DAT
DEM
DO
EMP
ERG
FUT
GEN
IMP
INST
10
NEG
NML
NOM
NPST
PART
PL
POSS
PRES
PRET
PRF
PROG
PST
PTCL
PUNC
0
Qprt
REFL
SG
TOP
TRAN
TRNS
ablative
absolutive
accusative
aorist
aspect
auxiliary
clitic
class, classifier
comparative
conditional
conjunction
contrastive
dative
demonstrative
direct object
emphatic
ergative
future
genitive
imperative
instrumental
indirect object
negative
nominalizer
nominative
nonpast
partitive
plural
possessive
present
preterit
perfective
progressive
past
participial
punctual
quantifer
question particle
reflexive
singular
topic
transitive
translative
j Other abbreviations:
j
j
1
i
A
AGR
CP
CPLT
e
EC
INFL
IP
NP
NSL
0
0+A
0-A
*0+A
*0-A
OP
POST
PP
PREP
PRO
SPEC
Std
VP
adjective
agreement
CPLT phrase, aka S
complement
trace of movement
empty category
inflection node
INFL phrase, aka S
noun phrase
null subject language
pronominally interpreted EC
null pronoun with agreement
null pronoun without agreement
obligatorily lexical pronoun with agreement
obligatorily lexical pronoun without agreement
null operator
postposition
adpositional phrase
preposition
the null subject of nonfinite sentences
specifier
standard of comparison
verb phrase
Conventions include:
*
(X)
*(X)
(*X)
/
+
i
i
ungrammatical utterance
X is optional
ungrammatical when X does not appear
ungrammatical when X does appear
morpheme boundary
fused morpheme, e.g., NOM/PL is a single
morpheme designating both nominative Case and
plural
clitic boundary
separates two abbreviations, e.g., DO.CL is a
direct object clitic
xii
ABSTRACT
I
I
i
i
counterexamples.
j
i
;
i
j
i
I
xiii 1
1. 0
But the
The success of
b.
zu-k
a-ri
libruru-a irakurri d-io-zu
you-ERG her-DAT book-ACC read
3sDO-3sIO-2sS
'you have read the book to her'
0 0 0 irakurri diozu
'you have read it to her'
This
core of most
I want 0 to be free
i
i
0 bring me a beer!
1.1.2
The choice,
As
A dis
b.
c.
Hinds'
Furthermore,
Li &
For instance, it is
b.
lai
come
come
(7) a.
muchos estudiantes
piensan que 0
i
son
i
many
students
think
that
are
inteligentes
intelligent
many students think that they are intelligent'
b. *muchos estudiantes
son
i
inteligentes
'many students think that they are intelligent'
A discourse-based account intuitively captures neither the
behavior in these two types of languages nor the similar
ity between the two.
Even the discourse analysis claim that lexicaliza
tion, where there is a null alternative, implies emphasis
turns out to be false.
b.
mvulana a-li-ni-piga
(mimi) na msichana
boy
3sSU-PST-lsOB-hit me
and girl
'the boy hit me and the girl'
(10)
(el ) quier-e
PRO venir
i
i
he
want-PRES/3s
come
'he wants to come'
(11)
(el ) se
compr-o
0 un helado
i
i
i
he
REFL buy-PRET/3s
a ice crea
'he bought himself an ice cream'
In
On a general
very
representation,
this
theory.
accounts
pronouns
heightened
reflect
sense
greater
of
This
null
principles
likely
to
underlying
null pronouns.
I turn now to a general consideration of the GB
framework.
1.2
1.2.1
Levels of derivation
A basic tenet of every version of transformational
A deri
This insight
d-structure
s-structure
PF
LF
10
by
LF is
, and is
The
D-structure modules
X-bar theory minimally is responsible for the build
ing of structure.
11
The
^ ^ X P ^
SPEC
X
head
CPLT
X
(15)
--> NP INFL VP
12
(16)
^CP
SPEC
Nsx IP
NP
INFL
VP
When put
Theta-marking of the
predication
analysis
of Williams
(19800,
which
accom
An early
Positions
At subsequent levels,
More generally, it is
effectively
exists
ruled
out,
since
the
complement
position
and movement
in
(16).
Move
16
Note that the EPP predicts that these subjects are present
even in languages where there are no lexical counterparts
to there and it in (21).
Not to be overlooked is the adjunction type of move
ment.
XP ?
XP
adjunct
everyone
thinks that he
i
is intelligent
i
(24).
18
As noted above,
Furthermore, while
S-structure/LF modules
The two major modules constraining the output of move
and the rule of free indexing, and thus operating at sstructure and LF, are government theory and binding
theory.
Government theory plays a very large role in GB
theory.
19
an NP
b.
an NP
verb
c.
an NP
is oblique if governed by P
(27)
^lexical
[+AGR]
NP without Case
20
(29)
^IP^
NP
^ I ' v
INFL
Cal
^-VP.
SPEC
v"
V ?*
NP
PRES
eat
bugs
(30)
NP
I f.
INFL
VP.
V\
SPEC
V"
IP
SPEC
I .
INFL
I
believe him to
PRES
VP
I
be a fool
(32) a.
m-commands
1)
does not dominate fi , and
2) every maximal projection that dominates
P dominates oC
21
(32) b.
^ excludes ft if no segment of
dominates ^
(34)
22
(35) a.
*who
As bounding
In particular, it
b.
c.
23
the guitar
John
chapter four.
24
1.2.5
In these
These parameters,
But there is
Yet the
unwelcome results.
Second, it creates
28
The word
30
parameters, there is independent evidence, i.e., crosslinguistic verification, that a parametric difference
among languages has been sorted out: the two parameters
predict four types of languages and only those four are
found.
As we shall see in chapter two, no such crosslinguistic evidence justifies any of the commonly assumed
versions of the Pro-drop Parameter.
1.3
1.3.1
In a number of cases, he
32
b.
These
Note
TENSE
prefix
suffix
25%
25%
25%
25%
33
(43)
VO
OV
TENSE
suffix
prefix
24
18
60
2
To extend markedness
For instance,
Because
34
Typological methods
The methods employed in typological investigations of
The best
35
To
Unfortunately, biblio
However, it is
He
The
Further
In
kolme
poika-a on tull-ut
three/NOM boy-PART be come-PTCL
'three boys [lit., three of boys] have come'
For instance,
But in
Postpositions in
38
talo-n
kohda-lta
house-GEN point-ABL
'in the neighborhood [lit., from the point] of
the house'
39
Even if Greenbergs
For
Typology and GB
The strongest GB responses to typology begin and end
Coopmans, a particularly
40
science.
One might
facts about word order co-occurrences that GB syntacticians are only beginning to offer explanations and
analyses for.
It is unthinkable that
42
I simply want to
Deep typology
To summarize the situation thus far: on the one hand,
The benefits of
43
Typology,
I therefore propose to
Given a typology of
Though surveys of an
1.4.1
The theta-directionality
46
X-bar theory
In
47
is Mandarin (cf.
1985).
Each of these analysts has proposed that Mandarin is at dstructure an OV language, despite the fact that this order
is not reflected in all surface forms.
b.
ta nian shu le
he read book ASP
'he read the book'
ta ba shu nian le sange xiaoshi
he BA book read ASP three hour
'he read the book for three hours'
b.
c.
b.
cEo
wEE ji
mi
bo
na
man/DEF INFL water drink cabin in
'the man drank water in the cabin'
mari ye
(a) fO
seku yE ko a a
na
Mary INFL it tell Sekou to
he INFL come
lu
ma
house to
'Mary is telling Sekou to come home'
49
Since this
50
Typologists
It is
To be fair,
theta-assignment.
However, in the
In many
(exemplified below).
Furthermore, OV
(48) a.
b.
54
Typology provides
It is
Furthermore, the GB
55
(Greenberg 1966).
languages is reported.
AStd
vso
SVO
sov
Post
Prep
StdA
5
9
0
0
1
9
1
13
10
0
both
0
1
0
1
0
Each is
b.
56
38
14
8
6
3
1
70
StdA
Both
VO
OV
23
0
2
12
1
1
Post
Prep
Both
2
22
0
10
2
0
1
0
1
58
Bambara
Basque*
Bini
Burmese*
Burushaski*
Cebuano
Chibcha*
Cocoma
Daga
Dani
Dutch
Eskimo
Ewe
Fij ian
Finnish*
Fore
Fulani*
Georgian
Goaj iro
Greek*
Guarani*
Guay mi*
Hakka Chinese
Hawaiian
Hebrew*
Hindi*
Hixkaryana
Hua
Idoma
Ilokano
Jacaltec
Japanese*
Javanese
Kannada*
Khassi
Kpelle
Lingala
Malagasy
Malay*
Mandarin
Mandinka
Maori*
Margi
Mundari
Navaho
Nenets
Ngandi
Norwegian*
chart
SVO/SOV
Prep/Post
SOV
SOV
Post
Post
Prep
SVO
SOV
SOV
VSO
SVO
SOV
VSO
SVO
SVO
SOV
SVO
SVO
SVO
SOV
ovs
SOV
SVO
VSO
VSO
SOV
SVO
SOV
SVO
SOV
SVO
VOS
SVO
SVO
SOV
VSO
SVO
SOV
SOV
SOV
SOV
SVO
verbal
StdA
Q
Q
StdA
StdA
Prep
Case
StdA
Q
Q
Q
AStd
Post
Post
Post
Post
SVO
SOV
AStd
AStd
Prep
SVO
VSO
SVO
type
Post
Post
SOV
SOV
SOV
SOV
SOV
AStd/StdA
Prep
AStd
ptxs
ptxs
AStd
Post
Post
Prep
Prep/Post
Post
Prep
Post
Prep
Prep
Post
Post
Prep
Prep
Prep
Post
Post
Post
Prep
Prep
Prep
Post
Prep
Post
Prep
Post
Prep
Prep
Prep
Prep
Post
Prep
Prep
Post
Post
Post
Post
Prep
StdA
AStd
AStd
AStd/stda
Q
ptxs
verbal
AStd
StdA
AStd
AStd
AStd
AStd
Q
ptxs
Q
Q
Q
StdA
verbal
Q
Q
AStd
AStd
StdA
StdA
Case
Q
ptxs
verbal
AStd
AStd
AStd
StdA
AStd
StdA
Q
Q
Q
Q
Post
AStd
AStd
AStd
AStd
AStd
verbal
verbal
Q
Q
StdA
AStd
AStd
AStd
verbal
verbal
Q
verbal
StdA
StdA
StdA
AStd
Q+Case
verbal
Prep
Q
9
Case
ptxs
59
Nubian*
Northern Paiute
Papago
Papiamentu
Persian
Quechua*
Serbo-Croatian*
Somali
Songhai*
Sre
Swahili*
Tagalog
Thai*
Tibetan
Tiv
Turkish*
Welsh*
Xhosa
Yapese
Yidin
Yoruba*
Zapotec*
SVO/SOV
Prep/Post
AStd/StdA
type
SOV
SOV
Post
Post
StdA
StdA
Post
VSO
SVO
SOV
SOV
SVO
SOV
SOV
SVO
SVO
VSO
SVO
Prep
Prep
Prep
Post
Prep
Prep
Post
Prep
Prep
Prep
Prep
Post
SOV
SVO
Prep
SOV
VSO
SVO
VSO
Post
Prep
Prep
Prep
SOV
SVO
VSO
AStd
AStd
AStd/StdA
StdA
AStd
StdA
AStd
AStd
AStd
AStd
AStd
StdA
AStd
StdA
AStd
AStd
AStd
Post
Prep
Prep
AStd
AStd
Q
Q
Q
Q
Q
Q
Q
verbal
Q
verbal
Q
Q
Post
verbal
Q
Q
Prep
Prep
ptxs
verbal
b.
ituja-ve xehe-gui
old-CMPR I-'than'
'older than I'
hata-ve
ita-gui
hard-CMPR rock-'than'
'harder than a rock'
(Guasch 1956)
(Pederson 1977)
60
For
Further
b.
(57)
a. in
pesar az an
doxtar boland-tar ast
thisboy
than that girl
tall-CMPR is
this boys is taller than that girl
b.
u bistar asb
darad taman (daram)
he more
horse has
than I
have
he has more horses than I d o
62
b.
goft ke farda
miayam
said that
tomorrow I am coming
?he said that he was coming tomorrow'
geza'i
ke mixorim varede xun
misavad
food that we eat arrive blood come
'the food that we eat enters the blood'
A parallel
Again,
The
non callid-ior
es quam hie
NEG cunning-CMPR are than he
'you are not more cunning than he'
non nascitur ex
malo bonum, non magis quam
NEG born
from bad good
NEG more than
ficus ex
olea
fig from olive
'Good is not born out of bad, no more than a
fig tree from an olive'
b.
c.
*amicitia,
quam quae nihil
melius habemus
friendship, than whom nothing better have
'friendship, which we have nothing better than'
(cf. *amicitia, quae nihil melius quam habemus)
64
b.
accidit ut
esset luna plena
happen that was
moon full
'it happened that it was a full moon'
c.
mitto
quod possessa per vim
disregard that possess by violence
'I disregard the fact that they were possessed
by violence'
comparatives is explained in both Persian and Latin: at dstructure, these complements are generated in the pre-head
position in accord with the general rule that assigns
theta roles in VPs and comparative phrases in the same
direction; then the complements extrapose at s-structure,
for whatever reason, thus yielding the surface forms.
The remaining two counterexamples, Dutch (exemplified
in (64)) and Guaymi, pose a related problem.
Each can be
65
b.
de jongen
is groter dan ik
the boy-NOM is taller than I-NOM
'the boy is taller than I'
ik vind hemaardiger dan
I find him-ACC nicer
than
'I find him nicer than you'
jou
you-ACC
In the latter
66
1.5
Summary
In the preceding sections, I have set the groundwork
This
approach differs from other generative analyses of crosslinguistic phenomena, or parameters, in as much as the
range of language types is not simply restricted to con
firming the findings of a particular analysis, but per
forms an important function in the development of that
analysis.
67
The
Instead, a
Though
It is demonstrated
I first propose a
Where
This hypo
lexical pronoun,
71
2.0
Each of
The
Given
A large
73
2.1
2.1.1
possibility of null subjects cooccurred with the grammaticality of extracting subjects from subordinate clauses
headed by an overt complementizer.
(nosotros) he-mos
trabaja-do todo el dia
we
have-PRES/lp work-PTCL all the day
we have worked all day'
(2) a.
qui^n
diji-ste
i
que e
sal-i-cT
temprano
who
say-PRET/2s that
leave-PRET-3s early
'who did you say that left early'
b.
las cosas
que dij-i-ste
que e
i
i
the things that say-PRET-2s that
pasa-ron
happen-PRET/3p
'the things that you said that happened'
74
must
can
(6) a. *who
S'
(8) d.
(9) a. *qui
s 'est evanoui
i
b. *les ^venements
se sont
i
drouls
Perlmutter subsequently formulated a single principle (cf.
section 2.1.2) to generalize over the data in (1) through
(9).
In Italian, a language
76
77
b.
lo
dio" 0 0 ayer
i
J
i j
Juan lsDO.CL 3smI0.CL give
yesterday
Juan gave it to me yesterday
present,
In 2.1.3, I
78
This
Dutch. Some
to
wie
vertelde je dat e
i
(Dutch A)
gekomen was
i
who said
you that
come
was
'who did you say that had come?'
(15) a. *(hun) var farinn til Island
she was gone
to Iceland
'she went to Iceland today1
b.
sag^ir \r a^S
hver
i
i dag(Icelandic)
today
LA
who
say
you that
was
come to
'who did you say that had come to LA?'
LA
79
(16)
(17)
df gaer
var (*y6ai) mikill snjor a jfir^inni
yesterday was there more
snow
on ground
'yesterday, there was more snow on the ground.'
null thematic
subjects
null nonthematic
subjects
(-FILTER)
+DELETION
free
subordinate subj ect
extractions
Perlmutter's generalization was later readdressed in
Chomsky & Lasnik (1977).
80
*[
CP
This
null thematic
subjects
null nonthematic
^ subjects
+DELETI0N
+FILTER
*[
x y]
COMP
A that-t
82
(The
null
thematic<------ >rich
subjects
AGR
null
4
that-t
nonthematic <----- >AGR<---- >f ilter
subjects
violations
In Rizzi (1982), Taraldsen's correlations are altered
in two ways.
b.
The scope
The key
for Rizzi was the observation that the wide scope reading
84
It
null
thematic<----- >INFL
subjects
[+ref]
null
nonthematic<--- >INFL<
subjects
[+pnl]
subject
that-t
>inversion<---- >f ilter
violations
Rizzi's
85
(28)
null
null
thematic nonthematic subject
subjects subjects
inversion
+
+
+
+
+
that-t
filter
violations
+
+
Just as Taraldsen
86
For
(eles) saira-m
they leave-3p
'they left'
b. *sa:ram eles
Mandarin also displays the same set of characteristics
(Huang 1982,1984).
87
(30)
null
nonthematic<
subjects
>N0M-drop
parameter
that-t filter
violations
Null nonthematic subjects, he suggested, are possible
whenever a language may optionally realize nominative Case
(NOM-drop parameter).
88
(31)
a.
b.
c.
null
null
thematic nonthematic subject
subjects subjects
inversion
+
+
+
+
+
d .
that-t
filter
violations
+
e.
Among the types in (31), three (31ace) are predicted
by Rizzi's typology and one (31b) corresponds to the type
defined by Brazilian Portuguese, assuming that apparent
that-t filter violations are indeed something else (cf.
section 2.2.2.3).
It is true
There
89
Huangs analysis.
The possibility that the other counterexamples are
only apparent is an obvious ploy, one which is aided
immeasurably by the limited sampling of languages for
which there have been generative analyses of null sub
jects.
In each in
90
null
thematic
subjects
Arabic
Levantine
Bani-Hassan
Chamorro
Dutch:A
B
English
Flemish:west
French:modern
old
German:standard
Bavarian
Hausa
Hebrew
Icelandic
Irish
Italian
Japanese
Mandarin
Portuguese
European
Brazilian
Quechua: Imbabura
Spanish
Swedish
ND:
(1):
(2):
(3):
(4):
null
nonthematic subj ect
inversion
subjects
thatfilt<
violat
yes(l)
yes
yes
no
no
no
yes
no
yes
no
yes
yes
yes
no
yes
yes
no( 3)
no( 3)
yes(1)
yes
ND
no
yes
no
yes
no
yes
yes
yes
yes
yes
yes(2)
yes
yes
ND
yes
yes(1)
yes
yes
no
no
no
no
no
no
no
no
yes
yes
yes
no
yes
no
no
no
yes
yes
no
yes
no
yes
no
ND
no
yes
yes
yes
yes
yes
yes
ND
ND
yes
yes
yes(l)
yes
no
yes
yes
ND
yes
no
yes(4)
no
yes(1)
yes
no
no
yes
no
yes
yes
This
91
Indo European
Dutch A, B (Maling & Zaenen 1978)
English (Perlmutter 1971)
west Flemish (Bennis & Haegemann 1984)
modern French (Perlmutter 1971)
Old French (Adams 1987)
standard German, Bavarian German (Bayer 1983)
Icelandic (Maling & Zaenen 1978; Platzack 1987)
Irish (McCloskey & Hale 1984)
Italian (Rizzi 1982)
European Portuguese (Zubizarreta 1983)
Brazilian Portuguese (Chao 1981)
Spanish (Perlmutter 1971, Jaeggli 1982)
Swedish (Platzack 1987)
Afro-Asiatic
Levantine Arabic, Bani-Hassan Arabic (Kenstowicz 1984)
Hausa (Tuller 1982)
Hebrew (Borer 1984)
Chamorro (Chung 1982ab,1984,1987)
Japanese (Hasegawa 1985, Hoji, p.c.)
Mandarin (Huang 1984)
Imbabura Quechua (Cole 1982,1987)
chart 2a: genetic and bibliographic listing of chart 2
languages
approach to troublesome counterexamples clearly parallels
the explanation given by Pesetsky (1982) for the lack of a
direct correlation between null thematic and nonthematic
subjects.
Cer
Nevertheless, if the
92
Choosing
Summary
Before leaving this section, it is useful to provide
94
null
null
thematic nonthematic subject
subjects subjects
inversion
+
+
+
+
+
that-t
filter
violations
+
+
95
2.2
null
null
thematic nonthematic subject
subjects subjects
inversion
+
+
+
+
+
-
that-t
filter
violations
+
Of the
I have
language family
# of lgs
Khoisan
31
Niger-Kordofanian
1064
Nilo-Saharan
138
Afro-Asiatic
241
Caucasian
38
144
Indo-Hittite
Uralic-Yukaghir
24
Altaic
63
Chukchi-Kamchatkan
5
Eskimo-Aleut
9
Elamo-Dravidian
28
Sino-Tibetan
258
Indo-Pacific
731
Austric
1175
Miao-Yao
4
Austro-Asiatic
155
Daic
57
Austronesian
959
Australian
170
Na-Dene
34
Amerind
583
Algic
16
Mosan
27
Siouan-Yuchi
11
Caddoan
4
Kutenai
1
Keresan
2
Iroquoian
7
Penutian
68
Hokan
28
Uto-Aztecan
25
Tanoan
7
Oto-Manguean
17
Chibchan-Paezan
43
macro Ge
21
macro Pano
49
macro Carib
47
Andean
18
Equatorial
145
Tucanoan
47
isolates,unclassified
21
creoles
37
total
roughly 5000
% of total lgs
# in survey
1
18
3
5
1
3
1
1
1
1
1
5
13
19
0.6
21.3
2.7
4.8
.8
2.9
.5
1.3
.1
.2
.6
5.2
14.6
23.5
.1
3.1
1.1
19.2
3.4
.7
11.7
0
3
1
15
3
1
17
.3
.5
.2
.1
.0
.0
.1
1.4
.6
.5
.1
.3
.9
.4
1.0
.9
.4
2.9
.9
.4
.7
100.0
1
1
0
0
0
0
1
1
1
1
1
1
2
1
1
1
1
2
1
2
4
100
98
percent.
Where
All four
In IndoEuropean, fa
single representative was selected, e.g., in Niger-Kordofanian, I was careful to include languages from each major
grouping within the family.
These manipulations proved to be quite simple, espe
cially when compared to the difficulties encountered in
finding source materials for each of the language
99
families.
Thus, it was
101
Khoisan
Niger-Kordofanian
Nilo-Saharan
Afro-Asiatic
Caucasian
Indo-European
Uralic
Altaic
Paleosiberian
Eskimo-Aleut
Dravidian
Sino-Tibetan
Indo-Pacific
Indo-Pacific
Austric
Austro-Asiatic
Daic
Austronesian
Australian
Na-Dene
Amerind
Algic
Mosan
Iroquoian
Penutian
Hokan
Uto-Aztecan
Tanoan
Oto-Manguean
Chibchan-Paezan
macro Ge
macro Pano
macro Carib
Andean
Equatorial
Tucanoan
103
isolates
creoles
That is,
Definitions,
104
analysis.
analyses to tackle the question head on: for him, the null
subject was PRO by default, since the only alternative ECs
within his model were the traces of movement and a move
ment analysis for null, subjects is out of the question.
105
+pronoun
-pronoun
+anaphor
PRO
NP-trace
-anaphor
pro
WH-trace
Chomskys
An EC
role.
The
Again, this
Though this
108
At first glance,
yo
tu
el,ella
nosotros
vosotros
ellos,ellas
cant-o
cant-as
cant-a
cant-amos .
cant-Zis
cant-an
I
you
he,she
we
you
they
sing-ls
sing-2s
sing-3s
sing-lp
sing-2p
sing-3p
Sometimes
a
ft
al
la
a
a
i
al
magn
magn
magna
magna
magnam
magne
magnen
magnen
I
you
he
she
we
you
they
they
Is eat
2s eat
3sm eat
3sf eat
lp eat
2p eat
3pm eat
3pf eat
109
(Cf. the
What seems
In
For example, a
110
[w^dy] .
these: where the subject is marked twice in a nontopicalized sentence, one of the forms is sure to be AGR.
Yet
short form
mo, ng
o
o
a
e
nwon
Is
2s
3s
lp
2p
3p
long form
&mi
1WQ
6un
awa
enyin
awon
tensed sentences.
At any
I return to
Behind this
This, not to
nouns.
113
In other
In fact,
(43)
First, in running
Less
b.
0 yabduu
?anna ?al-?awlaad-a saafaruu
seem/PRES/3sm that DEF-boys-ACC traveled
'it seems that the boys traveled'
...?anna-hu
yabduu
?anna ?al?awlaada
that-CL^sm seem/3sm that boys
saafaruu
traveled
'...that it seems that the boys traveled'
nation of lexical and null nonthematic subjects in verbsecond languages like German or Icelandic, discussed in
greater detail in chapter five.
115
116
(47) a.
b.
0 i-lithibitishwa
na Ahmed kwamba
nyoka
CLAS5-confirm-PASS by Ahmed that
dog
alimwuma
bit
'it was confirmed by A. that the dog bit him'
a. it
b.
it
control PRO.
to rain all day
i
rained all day without PRO
snowing once
i
Borer
ze e
hu kolkax satum margiz oti
that he so
dumb annoy me
'it annoys me that he is so dumb'
b.
es ekelt
mir
vor
dir
it disgust me/DAT before you
'you make me sick'
mir ekelt ('s) vor dir
118
(53) a.
b.
0 contest-o
la pregunta Juan
answer-PRET/3s the question J
'Juan answered the question'
A list of
sentence-types involving these elements in the IndoEuropean languages has no significance for other languages, of
course.
2.2.2.3
As with the
contest-o
a la pregunta Juan
answer-PRET/3s
the question J
'Juan answered the question'
In these con
121
b.
c.
d.
Goat Gruff
figure
(Cf. Belletti
In
(Old French
Focus is not
Tanja ubi-l-a
Mas-u
T
kill-PST-3s M-ACC
'Tanya killed Masha'
b.
c.
d.
Right dislocations
124
This
que
piens-as
i
que
vend-ra
i
125
fainted?
i
b. *que
who
i
b.
que
i
These rules are
not possible
under
anycondi
fainted?
i
b. *who do you regret e fainted?
i
i
^
(65) a. *qui regrettes-tu que e s'est evanoui?
i
i
b. *qui regrettes-tu qui e s'est evanoui?
i
i
Any language which modifies the complementizer and re
i
nonfactive verbs,
(66b),
que
rapazes
acredita-s
que
i
what child
believe-PRES/2s that
tenha-m
gasto
esse dinheiro
have-PRES/3p spend/PTCL that money
which children do you believe that spent
that money?
b. *quem
preocup-a+te
i
ter-em
gasto
who
worry-PRES/3s+2sCL
have/INF-3p spend
esse dinheiro
that money
who does it worry you that have spent the
money
Summary
127
c)
c)
128
2.3
129
2.3.1
The three
|
For ease of exposition, I have adopted the following list
;of abbreviations, which will be used throughout the re
mainder of this chapter.
i
!
I
pro:
EXE:
SI:
subject inversion
'
130
Alyawarra
American Sign Lg.
Angami
Arabic (Classical)
Babungo
Basque
Bhojpuri
Big Nambas
Blackfoot
Bobangi
Burmese
Canela-Kraho
Cape York Creole
Caviteno
Chamorro
Cherokee
Chorote
Chukchi
Cocama
Daga
Dani
Duka
Egyptian (Middle)
Engenni
Ewondo
Finnish
Fore
Fula
Garo
Georgian
Golin
Gonj a
Grebo
Guarani
Guaymi
Hausa
Heiltsuk
Hixkaryana
Hua
Iai
Icelandic
Indonesian (Betawi)
Italian
Izi
Kamba
Kewa
pro
EXE
SI
no( 1)
yes
no( 1)
yes(2)
no
yes
yes
yes
yes
yes
no (1)
yes
no
no( 1)
yes
yes
yes
yes
yes
yes
yes
no
yes
no
yes
yes
yes
yes
no( 1)
yes
yes
yes
yes
yes
no
yes
yes
yes
yes
yes
no
no( 1)
yes
yes
yes
yes
ND
yes
ND
yes(2)
yes
yes
ND
ND
yes
yes
yes
ND
yes
yes
yes
ND
yes
ND
ND
ND
ND
yes
yes
ND
ND
yes
ND
yes
yes
yes
ND
ND
ND
ND
yes
yes
ND
ND
yes
ND
yes(3)
yes
yes
ND
ND
ND
yes
no
no
yes(2)
yes
no
no
ND
yes
no
no
no
no
yes
yes
yes
no
yes
yes
no
no
no
no
no
no
no
no
no
yes
no
yes
ND
no
yes
no
yes
no
yes
no
no
yes(4)
no
yes
no
ND
no
THAT
ND
ND
ND
no
ND
yes
ND
ND
ND
ND
ND
ND
ND
ND
yes
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
no
ND
ND
ND
no
ND
ND
ND
ND
ND
yes
ND
ND
ND
ND
yes
ND
yes
ND
ND
ND
131
Eh
<
CQ
( D O P P P P P P P P P P P P P P P P P P P
P P P P P P P P P P P P P P P P P P P P P P P P P
Eh
CQ
CQ CQ CQ
CQ CQ CQ
CQ
O O O O O 0 O 0 0 0 O O O O 0 0 0 O
QOOOO
CQ
CQ
CQ
CQ CQ
CQ
CQ
( l ) 0 0 0 0 0 0 ( P 0 0 ( D 0 f l ) ( D 0 0 0 0 ( P 0 0 ( D
CQ
C
o
H
-P
(0
rH
0
P
p
0
o
0
0
6
O
G
w
X
w
p p p p
CQ (Q
(D C D P P P P P
CQ CQ
CQ
( D P (D (D P P P P P P P P P P
CQ
C D P P P P P P P
CQ
(DPP
0
CQ CQ CQ CQ
CQ CQ
(D(D<D(DPPPP(D<D
Xl
-p
o
0
X)
CQ CQ CQ CQ CQ CQ CQ w w CQ CQ w CQ CQ w
(D(D<D(D(D(D(DOO(D(DO(D(DO
>1 > 1
CQCQCQCQCQCQCQCQCQCQ
(D(D(D(D(D<D(D(D(D(D
CQ CQ CQ CQ CQ CQ CQ
(D (D (D (D (D (D (D
CQ
ii
CQ w CQ CQ CQ CQ CQ CQ
0 0 0 O 0 0 0 0 0 0
>1 G >i G >1 >1 >1 >1 >1 >1
CQ
r 1
pH
G
0
X!
(0
JD
m
o
g
(0
H 0
0
0 T3
0 G rH O X G
15 O rH ' H 0 0 H
0 X 0 0 G D>,0
>i 0
CQ r1
0 0
& >1
0 0
H H
(D
rI
(0
O
0
3
u
-p w
G
H O "H
P O P
g 0 0 0 -P
0 T3 -P *0 3
C C X C P
g
h o a 0 g 0 0 0 0 0 0 -H 3 3
X X X X P P 2
(0
(D 0
0-H P
X ^3 0
(0 (0 C -H (D
g > (o e c a o
(0 (0 OVH 0
OE3
CQ (0
(0-h o
a a a)
>i
>
>i
U !
(D
P (0
3
cq
-P
& &
\0 co
g cq
O2
0 Eh I
(0
(0
0 -H
rI N
(0
>1
c
(0
00
a c
(0 (0
+> X
(0 0
(D 0 CQ g (0 X 0> H G P
P (0 (0 X! -H -H -H 0 0 3
H 0 0 -H ^ TJ^
B (0 D 0 0 KO 0 0 ,G
2OPUPUPUOiC0QC/)0Q0QC/30Q0Q0QC/) Eh Eh Eh Eh Eh Eh Eh Eh Eh
in
-p
p
0
X
o
CM
CD
Turkish
Vietnamese
Warao
Waskia
West Greenlandic
Yebamasa
Yessan-Mayo
Yoruba
(1):
(2):
(3):
(4):
pro
EXE
SI
yes
no( 1)
no
yes
yes
no
no( 1)
yes
yes
yes
ND
ND
ND
ND
ND
yes
no
no
no
no
no
yes
no
no
THAT
ND
ND
ND
ND
ND
ND
ND
yes
Fortunate
ru?u ni-zee-ri
I
PRF-eat-ls
'I ate'
b. *ni-zee ru?u
.
s v
c. * n - m - z e e
Since subject arguments never occur twice in a single
sentence, one of the forms in (67) must be agreement.
It
134
Discussion
The facts of chart 5 are restated in the following
table.
yes-yes yes-no
pro-EXE
pro-SI
pro-THAT
EXE-SI
EXE-THAT
SI-THAT
24
22
5
14
7
4
0
49
3
25
2
0
no-yes
15
11
2
1
0
3
no-no
ND
2
15
1
1
1
4
61
4
89
61
90
89
For in
135
resolved in (68).
2.3.2.1
EXE
yes
yes
24
no
0
[15
pro
no
61 combinations involving ND
2 doublets: Classical Arabic
Icelandic
102 total
The large number in two of the cells makes them
reasonably secure.
b.
2.3.2.2
At first glance,
SI
yes
yes
14
no
25
EXE
no
1
61 combinations involving ND
2 doublets: Classical Arabic
Icelandic
102 total
137
f>a^
munu kaupa ^essa bok margir studentar
there will buy
this
book many
students
'many students will buy this book'
b.
138
(73)
Note
b.
however.
139
Yet
In
b.
2.3.2.3
(76)
THAT
yes
no
4
0
yes
SI
no
4
89 combinations involving ND
100 total
ten 1983:283.)
(77)
a. Maria a
disparse
M
PRF disappear
'Mary disappeared'
b. *a disparse Maria
(78)
a. ken
bo
kere
ku
e a
bini
i
who you think that
PRF come
'who do you think that had come?'
i
b.
homber
ku
i
taevidente ku
bini
i
the man
that PROG evident that
PRF come
'the man that it is evident that had come'
141
b.
SI > THAT
Rizzi's
THAT
no
yes
2
7
0
1
90 combinations involving ND
100 total
142
But
tion predicts that this cell (along with the yes-yes cell)
should be filled.
The yes-no type, exemplified by Finnish and Georgian
in my survey, must therefore be the focus of any attempted
reanalysis.
Although
c.
(82) b.
c.
(83) a.
b.
2.3.2.5
SI
yes
no
yes \ 22
49
\
no
11 15 \
4 combination involving ND
1 doublet: Classical Arabic
101 total
144
correct.
SI
yes
no
28
58
pro
no
Though the numbers in both the no-yes and no-no cells are
rather small, the types can be well-documented.
Among the
(Garcia, et al 1975:39,103).
(86) a. *(yu) bas-aja
I
sing-PRES
'I sing'
b.
mu-re
cenaroti-aja yu
you-DAT greet-PRES
I
'I greet y o u '
filter violations.
figures.
(87)
yes
THAT
yes
no
5
3
no
pro
1
89 combinations involving ND
100 total
The yes-yes and no-no types are well-known from Perlmutter's original study and require no discussion, despite
the small number from my sample in the latter cell.
The
Summary
146
(88) a.
pro
--> EXE
b.
SI
--> EXE
c.
SI
> THAT
d.
THAT
--> EXE
(statistical)
In tabular
yes-yes
13
9
10
9
12
7
yes-no
0
5
1
9
2
1
no-yes
5
1
2
1
0
5
no-no
5
9
6
4
4
6
ND
1
0
3
1
4
3
147
Arabic:Levantine
Bani-Hassan
Chamorro
Dutch:A
B
English
Flemish:west
French:modern
old
German:standard
Bavarian
Hausa
Hebrew
Icelandic
Irish
Italian
Japanese
Mandarin
Portuguese:European
Brazilian
Quechua: Imbabura
Spanish
Swedish
ND:
(1):
(2):
(3):
(4):
pro
EXE
SI
yes(1)
yes
yes
no
no
no
yes
no
yes
no
yes
yes
yes
no
yes
yes
no( 3)
no( 3)
yes
yes
yes(1)
yes
no
yes(1)
yes
yes
no
yes
no
yes
no
yes
yes
yes
yes
yes
yes(2)
yes
yes
yes
yes
yes
yes
ND
yes
no
yes(1)
yes
yes
no
no
no
no
no
no
no
no
yes
yes
yes
no
yes
no
no
yes(4)
no
yes(l)
yes
no
THAT
no
yes
yes
no
yes
no
yes
no
ND
no
yes
yes
yes
yes
yes
yes
ND
ND
no
yes
no
yes
yes
Dutch
English
Old English
Bavarian German
Hausa
Hungarian
Icelandic
Italian
Norwegian
Papiamentu
Portuguese
Quechua
Ancash
Huanca
Imbabura
Russian
Serbo-Croatian
Spanish
Finnish Swedish
pro
EXE
SI
THAT
no
no
no
no
yes
yes
no
yes
no
no
yes
yes
no
ND
ND
yes
yes
yes
yes
no
ND
yes
no
no
ND
ND
yes
yes
ND
yes
no
ND
yes
yes
no
yes
yes
yes
no
yes
yes
yes
yes
yes
yes
yes
yes(1)
yes
yes
yes
no
ND
ND
ND
yes
ND
yes
no
yes(l)
yes(l)
yes(1)
ND
ND
yes
no
no
no
no
no
no
yes
yes
ND: no data
(1): main clause only
chart 6: the van der Auwera sample
(90)
yes-yes
pro-EXE
pro-SI
pro-THAT
EXE-SI
EXE-THAT
SI-THAT
2.3.3.1
yes-no
6
8
4
5
6
4
0
2
5
1
2
1
no-yes
no-no
3
0
7
0
2
3
3
5
2
3
1
4
ND
4
6
0
6
4
6
Testing my results
only five cells are empty, though no one survey lacks all
five.
149
(91)
pro:yes-EXE:no
pro:no-SI:yes
EXE:no-SIryes
EXE:no-THATryes
SIryes-THATrno
0
0
0
2
1
chart 2
Gilligan
0
9
1
0
0
0
1
1
0
1
Almost as
150
In
In the absence
Taraldsen
b.
(som)
Merit snakker
i
we know
med
the man
that M
talk
i
with
'we know the man that Mary is talking with1
Essentially, Taraldsen's analysis treats the null form of
the complementizer as the basic form; thus, the obliga
toriness of the 0 --> som rule for subject relative
clauses constitutes a complementizer-changing rule.
The
152
As mentioned in
(93)
yes-yes
chart 2 lgs
9
van der Auwera
8
pro-SI correlations
yes-no
no-yes
5
1
9
2
0
no-no
5
Whe
veys, these attain moderate success: Rizzi's SI:noTHAT ryes counterexamples include six verb-movement lan
guages (Dutch A, Bavarian German, Irish, Norwegian, Fin
nish Swedish, and West Flemish) and Brazilian Portuguese;
Taraldsen's EXE:yes-THAT:no counterexamples include
154
Though my sample
Conclusions
Although the counterexamples which are uncovered in
Still,
pro
> EXE
b.
SI
c.
SI
--> THAT
d.
155
There, as we
To claim that
The predicted
Safir's
biconditional statement.
When only implicational statements relate phenomena,
however, the constraints on the analysis are much weaker.
A statement like SI --> EXE actually reveals very little
about EXE, only that the antecedent (SI) may occur with it
or not.
For exam
The topic of
None of the
As the null
of constraints.
159
3.0
Once again,
Re
Huang,
There
In
There
In every instance
His
Second,
In this
163
redundantly encoded.
from
(a'mi) i-ka
ba dumi1ni
I
ls-AOR go sleep
'I will go to sleep'
*(bo) ka ba dumi'ni
you AOR go sleep
'you will go to sleep'
164
0 mana xwr-^m
apple eat-lsm
11 eat the apple 1
(Huang 1984:536)
b. *za 0 xwr-dm
I
eat-lsm
I eat i t 1
(4) a.
ma 0 wd-xwar-a
I
PRF-eat-3sf
I ate i t
b. *0 mana wd-xwar-a
apple PRF-eat-3sf
I have eaten the apple1
In Ecua
b.
(Cole 1982:34)
lacks AGR and so null subjects are ruled out by the NIC.
Taraldsen*s hypothesis was originally developed with only
null subjects in mind, though the extension of its ideas
to other null pronouns seems clear enough.
A null
As noted
In the remainder of
b.
0 lai-le
come-ASP
I/you/he/etc. came
Lisi
hen
xihuan 0
i
Lisi very like
Lisi likes me/you/him/etc. very much
167
(6) c.
Zhangsan
shuo [0
bu renshi Lisi]
i
i /j
Zhangsan say
not know
Lisi
Zhangsan said that I/you/he/etc. did not know
Lisi
d.
Zhangsan
shuo [Lisi
bu
renshi 0
]
i
j
*i/*j /k
Zhangsan say
Lisi not know
Zhangsan said Lisi did not know me/you/him/
etc.
The fact that the identity of the gaps in (6abd) can only
be determined through discourse context (as well as other
facts concerning their interpretation, to be discussed in
chapter four) prompted Huang to analyze them as variables
which are bound at s-structure by a null topic.
The same
This conclusion is
168
The
Furthermore, his
nonrich AGR.
169
mich ha-t
*(sie)
tief verletzt
1/ACC have-PRES/3s she/NOM deep wound
she has wounded me deeply
Yet Bavarian
doomed to failure.
Another conceivable option is to ignore the German
type, since it is extremely rare, perhaps making it an
exception to the null pronoun parameter.
This is hardly
null pronouns.
Taraldsen's hypothesis
optionally null pronouns <--> AGR;
obligatorily lexical pronouns < > lack of AGR
2)
Taraldsen-Pesetsky's hypothesis
optionally null thematic pronouns <--> rich AGR;
null nonthematic pronouns <--> weak AGR;
obligatorily lexical pronouns <--> lack of AGR
3)
Rizzi's hypothesis
optionally null thematic pronouns <--> person
and number specification;
null nonthematic pronouns --> person and number
specification or no specification at all
4)
Huang's hypothesis
optionally null thematic subjects <--> rich AGR
or no AGR;
optionally null thematic nonsubjects <--> rich
AGR;
obligatorily lexical pronouns <--> degenerate AGR
171
3.2
There is sig
In this survey,
As a result, the
a nonthematic antecedent.
The essen
Nul1 pronouns
In this section, I consider a syntactic definition of
In the
173
An adjunct cannot
elements.
Given this
I will
b.
zu-k
(ar-i)
liburu-a irakur-ri
you-ERG she-DAT book-ACC read-lprt
d-io-zu
3sACC-3sDAT-2sERG
'you have read the book to her'
c.
iddo (fe)
to/3sm him
'to him'
For
175
b.
yo la
yo le
(min-un) auto-ni
I-GEN
car-lsPOSS
'my car'
176
Never
First of all, it
Furthermore, it ignores
quien
0
i
com-i-o
la
manzana
who
eat-PRET-3s the apple
'who ate the apple?'
177
For
b.
178
3.2.1.3
Besides, depending
(16) a.
[PRO
i
to conclude what
i
follows]
b. *this leads 0
be made for other verbs which commonly 'drop' their objects, e.g., eat, drink.
179
This distinction
(18) a.
[a PRO
i
concludere
i
quanto segue]
b.
questo conduce 0
[a PRO
i
concludere quanto
i
segue]
The amended interpretation test also apparently makes
the correct prediction concerning constructions in which
some missing nominal element may be inferred from the
context.
180
(19)
181
(22) a.
b.
On
3.2.1.4
It is
Note
However, these
Nevertheless,
184
3.2.2
Agreement (reprise)
The bias of the previous survey led to the adoption
Simply
Specifically, the
The
These sub
Examples of direct
185
Summary
In the survey of the following section,the cor
Only those
(24)
3.3
Results of survey II
This survey has as its range the same one hundred
The precise
188
Sthem
EXE
Dobj
0-A
Iobj
Poss PPobj
0+A1
<
i
o
*
chart seven:
Simp
<
i
o
*
0-A
Alyawarra
American Sign
Language
0(A)
0(A)
Angami
Arabic
0+A^
Classical
*0-A
Babungo
0+A
Basque
0+A
Bhoj puri
0+A
Big Nambas
0+A
Blackfoot
0+A
Bobangi
0-A
Burmese
Canela-Kraho 0+A
Cape York
*0-A
Creole
0-A
Caviteno
0+A
Chamorro
0+A
Cherokee
0+A
Chorote
0+A
Chukchi
0+A
Cocama
0+A
Daga
0+A
Dani
0-A
Duka
Egyptian
0+A
Middle
Engenni
*0-A
0+A
Ewondo
Finnish
0+A
0+A
Fore
0+A
Fula
0-A
Garo
Georgian
0+A
0+A
Golin
0+A
Gonj a
0+A
Grebo
Guarani
0+A
*0-A
Guaymr^
0+A
Hausa
Heiltsuk
0+A
0+A
Hixkaryana
0+A
Hua
0+A
Iai
*0+A
Icelandic
Indonesian
0-A
Betawi
Italian
0+A
0+A
Izi
Sinf
0-A
0-A
NC
NC
0-A
0(A)
0-A
0-A
NC
NC
NC
NC
0-A
NC
NC
NC
0-A
0-A
0-A
0-A
0+A
0-A
0-A
0-A
0-A
0-A
0(A)
0+A** 0+A
0-A *0-A
0+A
NC
NC *0-A
0+A^
NC
0+A
0+A
0+A
0+A
0-A
NC
0+A
NC
*0-A
*0-A
0+A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
0-A
0-A
0-A
NC
NC
0-A
0-A
NC
0-A
0-A
0-A
0-A
0-A
0+A
0+A
0+A
0-A
0+A
0+A
0-A
0-A
0-A
NC
NC
0-A
NC
NC
NC
NC
0-A
*0-A
0-A
0-A
0+A*
*0-A
0+A
0+A
0+A
0+A
*0-A
*0-A *0-A
*0-A *0-A
*0-A
0+A
0+A3 0+A6
*0-A
0+A
*0-A
0+A
*0-A
0+A
0+A
0+A6
0+A
*0-A
*0-A *0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A
0+A
*0-A
NC
NC
NC
0(A)
NC
0-A
0-A
0-A
NC
NC
NC
0-A
NC
0-A
NC
0-A
NC
NC
0-A
0-A
0-A
0-A
0-A
0+A
0-A
0-A
0+A
0+A
0-A
0-A
0+A
0-A
0+A
0-A
0+A
0+A
0-A
0-A
0+A
NC
NC
0+A
NC
0-A
0-A
0+A
NC
NC
NC
NC
0-A
0+A
NC
NC
0+A
NC
0+A*
0+A
*0-A
0+A
*0-A
0+A*
0+A
0-A
0+A
*0-A
0+A*
0+A
0+A
*0-A
0-A
0+A
0+A
0+A
*0-A
*0-A
0+A
*0-A
0+A
*0-A
0+A3
*0-A
*0-A
0+A
*0-A
0+A3
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A*
*0-A
*0-A
0+A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A
0+A7
*0-A
0+A7
*0-A
0-A
0-A
0-A
0-A
0-A
0-A
0-A
0-A
NC
*0-A *0-A
0(A) 0+A
0+A *0-A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
0+A
0+A
*0-A
0+A
0+A
0+A
*0-A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
*0-A
0+A6
*0-A
*0-A
0+A
0+A
0+A1
0-A6
*0-A
0+A
*0-A
0+A
0+A
*0-A
0+A
*0-A
0+A
0+A*
0+A6
*o -a '
0+A^ *0-A
*0-A *0-A
*0-A *0-A
Sthem
0+A
Kamba
0+A
Kewa
0+A*6
Kiowa
0+A
Kobon
0+A
Kpelle
0+A
Kwaio
0+A
Lenakel
0+A
Luganda
0+A
Madi
0-A
Malagasy
0-A
Malayalam
0+A12Mam
0+A
Manam
0-A
Mandarin
0+A
Mixteco
0+A
Mundari
0-A
Murut
0+A
Mwera
0+A
Nama
0+A
Navaho
0+Alz"
Ngandi
Nimboran
0+A
0+A***
Noni
0+A
Nupe
0+A
Olo
0+A
Paamese
0+A
Papago
Papiamentu *0-A
Quechua
Cochabamba 0+A
0-A
Rao
0+A
Sakao
Sao Tome
0+A***
Creole
Sara-Ngambay 0+A
0+A*2
Sawu
0+A
Shona
0+A
Siroi
0+A
Somali
0-A
Sre
0+A
Swahili
*0-A
Tagalog
0+A
Tamazight
0-A
Thai
0(A)
Tibetan
0+A
Tigak
0+A
Tiwi
0+A
Tolkapaya
Sinf
Simp
EXE
Dobj
NC
0-A
NC
0-A
NC
NC
NC
0-A
0-A
NC
0-A
0-A
NC
NC
NC
NC
NC
0-A
NC
NC
NC
0-A
0-A
NC
NC
NC
NC
0-A
0-A
0-A
0+A
0-A
0-A
0-A
0-A
0-A
0+A
0-A
0-A
0-A
0+A
0-A
0-A
0+A
0-A
0-A
0-A
0+A
0+A
0-A
0-A
0-A
0+A
0-A
0(A)
0-A
NC
NC
NC
NC
NC
NC
0+A
0-A
NC
NC
NC
NC
NC
0-A
NC
0-A
0-A
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
0-A
0+A *0-A
0-A *0-A
0+A^
0+A
0-A *0-A
0+A *0-A
0+A *0-A
*0-A *0-A
0+A
0+A
*0-A *0-A
0-A *0-A
*0-A *0-A
0+A1*- *0-A
0+A*
o +a 3
0-A *0-A
*0-A *0-A
0+A1*
0+A
0-A *0-A
0+A *0-A
0+A *0-A
0+A *0-A
0+A1*^ 0+A*
*0-A *0-A
*0-A *0-A
0+A *0-A
0+A *0-A
0+A *0-A
0+A5 0+A*
*0-A *0-A
0-A
NC
NC
0-A
0-A
0+A
0-A
0-A
NC
0-A
0-A
0-A
0-A
0-A
NC
NC
NC
0-A
0-A
NC
NC
0-A
0-A
0-A
0-A
0+A
0+A
0-A
0-A
*0-A^
0-A
0-A
0-A
0(A)
0-A
0+A
NC
NC
NC
NC
NC
NC
NC
0+A
NC
NC
0-A
0-A
0+A
0-A
NC
NC
NC
NC
lobj
Poss PPobj
*0-A
*0-A
0+A
0+A
0+A4
0+A4
0+A
*0-A
0+A4
0+A
0-A
0+A
0+A4
0-A4
0+A
0+A1
*0-A
*0-A
*0-A
0+A
0+A
0+A
0+A1
0+A
*0-A
*0-A4
0+A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A11
0+A*1
*0-A
0+A7
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A
*0-A
0+A
*0-A
*0-A
*0-A
0+An
0+A
*0-A
0+A
0-A
0+A
*0-A
*0-A
*0-A
0+A *0-A
*0-A *0-A
0+A4 o +a m
*0-A
0+A*
0+A1*
0+A5
0+A*
*0-A
*0-A
0+A*
*0-A
0+A
*0-A
0-A
0+A
0+A
0+A
*0-A
0+A*
*0-A
0+A*
0+A*
*0-A
*0-A
0+A*
*0-A
0+A
*0-A
*0-A
*0-A
0+A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
0+A1
*0-A
*0-A
0+A4
*0-A
0+A
*0-A
0+A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+An
*0-A
*0-A
0+A**
*0-A
*0-A
Sthem
Tondano
0+A
Turkana
0+A16
Turkish
0+A
Vietnamese
0-A
Warao
*0-A
0+A
Waskia
West
Greenlandic 0+A
*0-A
Yebamasa
Yessan-Mayo 0-A
0+A
Yoruba
Sinf
Simp
EXE
Dobj
NC
0-A
0-A
0-A
0-A
NC
0+A
0+A
0-A
0-A
0-A
0+A
NC
NC
0-A
0-A
NC
NC
*0-A
0+A
16 0+A16 0+A6
0-A *0-A
0+A
0-A *0-A *0-A
*0-A *0-A *0-A
0+A17 0+A6
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0-A
0-A
NC
0-A
0+A
0-A
0-A
0-A
NC
NC
NC
NC
0+A
*0-A
0-A
0+A
*0-A
*0-A
*0-A
*0-A
Iobj
Poss PPobj
0+A
o +a
*0-A
*0-A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
tion.
According to
ment encodes the goal and source NPs, and so an 0+A entry
191
All
b.
b.
a-tyj!
2s-strong
'be strong!'
ere!
sing
'sing!'
b.
il
bel
tempo
la
invoglia 0
i
i
the nice weather 3sfCL induce
[a PRO restare]
i
to
stay
'the nice weather induces her to stay'
il
bel
tempo
invoglia 0
i
the nice weather induce
[a PRO restare]
i
to
stay
'the nice weather induces [one] to stay'
These too are discussed in greater detail below.
193
As
to main clauses.
3.4
Discussion
The information in chart seven can be briefly
76
1
32
12
56
26
55
21
17
50
72
21
18
0
3
0
*0+A
2
0
0
2
0
0
0
0
NC
9
0
1
0
28
74
42
79
0
50
0
67
0
0
0
0
>
0-A
0
1
Sthem
Sinf
Simp
EXE
Dobj
Iobj
Poss
PPobj
0+A
agreement
yes
no
null
0+A
0-A
*null *0+A
*0-A
site
The point of this discussion is to verify that each of the
cells filled in (30) is real.
These in
195
3.4.1.1
agreement
yes
no
null
76 17
Sthem
*null
total: 104
four doublets: three 0(A)-ASL
Angami
Tibetan
one 0+A -Classical Arabic
The type of language with null thematic subjects and
agreement, along with its counterpart, where both are
lacking, forms the basis for all work on null subjects.
Perhaps surprising is the frequency of languages like
Italian (Rizzi 1982:117).
(33)
0 verr^
come/FUT/3s
he/she/it will come'
(34)
Is
2s
3s
lp
2p
3p
* (m i )
I
*( b o )
*(&)
*(nos)
*(boso)
*(nan)
ta
PROG
ta
ta
ta
ta
ta
bini
come
bini
bini
bini
bini
bini
'I am
etc.
Zhangsan
shuo [0
i
bu
renshi Lisi]
Z
say
NEG know
L
'Zhangsan said that he did not know Lisi'
I count seventeen languages of this sort in my sample, all
of which occur in south and east Asia or Oceania.
The fourth type has the least number of examples.
The data is unmistakable, however.
Icelandic, for in
3.4.1.2
Is
2s
3s
lp
2p
3p
*(eg)
*(/&u)
*(hann)
*(vi%)
*(i^&)
*(eir)
hef, hefi
hefur, hefir
hefur, hefir
h&fum
hafi^
hafa
'I have'
etc.
aspect.
In many in
It is generally more
198
verb forms are not relevant and have been excluded from
the four-cell matrix.
Given these definitions and restrictions, only two
cells are exemplified in the survey.
(38)
agreement
yes
no
null
50
*null
Sinf
total: 101
NC: 50
one doublet: 0(A)-Finnish
Among the relevant forms, that most commonly encountered,
e.g., in the Swahili example below, is the bare infini
tival without agreement.
(39)
Juma a-na-taka
ku-ondoka
J
3s-PRES-want INF-leave
'Juma wants to leave'
poika
yritta-a
i
PRO
kirjoitt-aa kirje-en
i
boy
try-PRES/3s
write-INF
letter-ACC
'the boy is trying to write the letter'
b.
case.
Imperative subjects
agreement
yes
no
null
32
72
*null
Simp
total: 105
five doublets: four 0(A)-ASL
Canela-KrahS
Papago
Tigak
one *0-A*** -Tagalog
It is also like the previous null pronoun because I have
not adequately defined the site beforehand.
part, I have relied upon a functional definition of imperatives--they are the forms used when giving an order-and in most languages, this suffices.
200
0 e-ju-na
ape!
2s-come-P0L here
come here please!
(43)
0 odota t&nne
wait here
wait here!
The significance of
alis 0 (na)
leave
now
leave now
Nonthematic arguments
201
However, in this
case, the NC entries do not reflect the lack of nonthematic subjects in these languages; rather, the high
number of languages classified as irrelevant is more the
result of incomplete source materials.
With this caveat, there are only three different
types exemplified in the languages of the survey.
(46)
agreement
yes
no
null
12
21
*null
EXE
total: 102
NC: 67
two doublets: 0+A -Classical Arabic
0+A -Icelandic
Again, the null variety accompanied by agreement is very
familiar from studies of the Pro-drop parameter.
An exam
b.
(50) a.
b.
Direct objects
203
(51)
agreement
yes
no
null
56
18
*null
28
Dobj
total 102
two doublets: 0(A)-ASL
Italian
Only the no-yes cell is empty, but this type is found only
with subjects, as a careful scan of the table in (30)
reveals.
Null direct objects with agreement are found in
Cochabamba Quechua, for instance.
(52)
baba
fun *(awa) ni owo
father gave
us
in money
'father gave us money'
il bel tempo
invoglia 0 a restare
the nice weather induce
to stay
'the nice weather induces [arb] to stay'
204
3.4.1.6
agreement
yes
no
null
26
*null
74
Iobj
total: 100
Only two types exist: languages with agreement and null
pronouns, e.g., Basque (Saltarelli et al 1983:486); and
those lacking both, e.g., Guarani (Canese 1983:57).
(58)
zu-k
(a-ri) liburu-a irakur-ri d-io-zu
you-ERG her-DAT book-ACC read-lPRET 3sA-3sD-2sE
'you have read the book to her'
(59)
In Mandarin, an overt
b.
wo song ta
0du 0
i
3 i
3
I send him book read
fI sent hima book
to read
shu
0 du 0
i
3
Possessive pronouns
agreement
yes
no
null
55
Poss
*null
0 42
total: 100
Quite a few languages have agreement with this argument,
though there is a significant amount of variation in the
productivity of this agreement within these languages.
206
b.
(sin-un) lapse-si
2s-GEN child-2sP0SS
'your child'
Pekka
katkais-i-0
i
Pekka break-PST-3s
'Pekka broke his arm'
c.
(63) a.
arm-3sP0SS
(min-un) auto-ni
ls-GEN car-lsPOSS
'my car'
mama-na
0
father-lsPOSS
'my father'
b.
yam-e
0
eye-3sP0SS
'his eye'
c.
gutut ne-ga
story I-GEN
'my story'
(64) a.
kSte-ns&
i
0 boko-ii)
brother-2sP0SS
'my brother'
b.
Samu-a? kasur
Samu-GEN guilt
'Samu's guilt'
c.
Soma Munda-?
uri?
some Munda-GEN cattle
'some Munda's cattle'
In Imbabura
207
(pay-pa) wayi-n
he-GEN house-lsPOSS
'his house'
Imbabura Quechua
Ancash Quechua
As
(67)
Zhangsan, baba
hen youqian
Zhangsan, father very rich
'Zhangsan, [his] father is very rich'
tule, sanoi
come say
'come, said
into the
3.4.1.8
Adpositional objects
208
(68)
agreement
yes
no
null
21
*null
79
PPobj
total: 100
At least a few of the languages in the yes-yes cell have
null objects of true adpositions, e.g., Irish (McCloskey &
Hale 1985:507),
though others
in theyes-yes cell,
Hixkaryana(Derbyshire 1979:116)
e.g.,
appear tohavenominal
bhi me ag
caint leofa
(3 inne
was I PROG talk with/3p
yesterday
'I was talking to [them] yesterday'
(70)
eryewtoako (3 ro-hana-ka
sit down
IsPOSS-side-to
'sit down at my side!'
Summary
+
+
+
+
+
+
+
+
*0-A
*0+A
0-A
+
+
+
+
+
+
-
+
-
+
-
+
-
+
+
+
+
+
209
3.4.2
In this as in the
I have
e.g., one 0(A) is counted as one 0+A and one 0-A, and
doublets indicated by the diacritics in chart seven are
resolved, the results reduce to the table in (72), below.
0+A
Sthem
Sinf
Simp
EXE
Dobj
Iobj
Poss
PPobj
14
2
0
11
8
6
6
4
0-A
1
17
21
3
8
1
1
1
*0+A
9
0
0
5
0
0
0
0
*0-A
2
0
0
6
10
14
14
16
NC
0
4
0
2
0
0
0
0
First, the
This
210
Sthem
Arabic
Bani-Hassan 0+A
0+A
Levantine
0+A
Chamorro
Dutch: A
*0+A
B
*0+A
English
*0(A)
West Flemish 0+A*'
French
0+A
modern
old
0+A*
German
*0+A
standard
Bavarian
0+A3*
0+A
Hausa
Hebrew
0+A
*0+A
Icelandic
Irish
0+A
Italian
0+A
Mandarin
0-A
Quechua
0+A
Imbabura
Portuguese
European
0+A
Brazilian
0+A
Spanish
0+A
Swedish
*0-A
EXE
Dobj
Iobj
Poss PPobj
0-A
0-A
0-A
0-A
0-A
0-A
0-A
0+A1
0+A
NC
*0-A
0-A*
*0+A
0+Ax
0+A
0+A
0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A
0+A
0+A
*0-A
*0-A
*0-A
*0-A
0+A
0+A
*0-A
*0-A
*0-A
*0-A
*0-A
0-A
0-A
0-A
0-A
*0-A
0(A)
0+A* *0-A
0+A
0+A
*0-A
*0-A
0-A
*0-A
0-A
0-A
0-A
NC
0-A
0-A
0-A
NC
0-A
0-A
0-A
0-A
0-A
0-A
0-A
0-A
0-A* 0-A
0-A* *0-A
0+A
0-A
0+A
0+A
0+A *0-A
NC *0-A
0+A
0(A)
0-A
0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
0+A
*0-A
*0-A
*0-A
0+A
0+A
*0-A
0+A
*0-A
0-A
*0-A
*0-A
*0-A
0+A
*0-A
0+A
*0-A
*0-A
0-A
0+A
NC
*0-A
*0-A
*0-A
0(A)
0(A)
0-A
0-A
0-A
0-A
0-A
0-A
0+A
0+A
0+A
*0-A
0(A) 0+A
0(A) 0+A
0(A)3 0+A
*0-A *0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
*0-A
Sinf
Simp
NC
NC
0-A
0-A
0-A
0-A
0-A
0+A
0+A
76
1
32
12
56
26
55
21
0-A
17
50
71
21
18
0
3
0
*0+A
2
0
0
2
0
0
0
0
*0-A
9
0
1
0
28
74
42
79
NC
0
50
0
67
0
0
0
0
211
The first of
During
212
la
fiesta?
i
Juan bring beer
to the party
will Juan bring beer to the party?
b.
su
novia
me dijo que
traerlTa 0
i
his girlfriend me told that bring
his girlfriend told me that he would bring
some
One additional point which arises only in connection
with the smaller survey merits attention.
This is the
j 'ai
deja
repu
1 invitation
ls+have already received the+invitation
I have received the invitation already
It furthermore satisfies
213
je n'ai
pas encore paye la note
Is NEG+have NEG yet
paid the bill
'I haven't yet paid the bill1
[ [ 0] [
je+ai]
IP NP
INFL
214
French has
lo
diro10 alia
polizia
i
i
i
everything 3smD0.CL I/say
to/the police
'everything, I told it to the police1
b.
tutto
tut
0 1' e
capita' de nut
i
i
everything
3s+has happen in night
'everything happened in the night'
215
*tout
In
b.
la canta e
*(la) balla
3sf sing and 3sf dance
'she sings and (she) dances'
elle chante et danse
3sf sing
and dance
'she sings and dances'
216
Where
In the
Sthem
Sinf
Simp
EXE
Dobj
Iobj
Poss
PPobj
0+A
0
1
>
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
*0+A
-
+
-
*0-A
+
-
+
+
+
+
+
+
217
Taraldsens generalization
Note, however,
Huang's generalization
These classifica
Unlike
Where there is no
219
Even
Maria
sa
i
ku
*(e)
ta sabi
i
Unlike
220
3.5.1.3
Is
2s
3s
lp
2p
3p
ik
werk
jij
werkt
hij,zij, het werkt
wij
werken
jullie
werken
zij
werken
'I work'
etc.
It
221
(87)
Is
2s
3s
lp
2p
3p
*(eg)
*(/u)
*(hann)
*(vi^&)
*(y&i*&)
*(^eir)
hef, hefi
hefur,hefir
hefur,hefir
htifum
hafi%
hafa
(88)
Is
2s
3s
lp
2p
3p
(yo)
cantaba
(tu)
cantabas
(l,ella)
cantaba
(nosotros)
cantabamos
(vosotros)
cant^bais
(ellos,ellas) cantaban
'I sang'
etc.
As noted in chapter
I discuss
223
224
Furthermore, as I will
In other
225
226
4.0
In
Section 4.4
4.1
Identification by agreement
The main idea behind all definitions of agreement is
Identification is
The
Note that
Is
2s
3sm
3sf
lp
2p
3p
*(yu)
*(mu)
*(!)
*(iso)
*(mani,yua)
*(mua)
*(ina)
basaja
basaja
basami
basamo
basaja
basaja
basama
'I sing'
etc.
229
Is
2s
3sm
3sf
liom
leat
leis
leithi
!with
'with
'with
'with
me'
you'
him'
her'
Is
2s
3s
(min&)
(sin&)
(h&n)
laula-n
laula-t
laula-a
'I sing'
'you sing'
'he/she/it sings'
230
c.
(6) a.
S'
b.
c.
His use of
231
One might as
In Mojave
In West Flemish,
b.
.da-nk
.da-j
.da-tje
.da-se
.da-t
.da-me
.da-j
.da-nze
(ik)
(gie)
(jij)
(zie)
(et)
(wunder)
(gunder)
(zunder)
kommen
komt
komt
komt
komt
kommen
komt
kommen
'that
'that
'that
'that
'that
'that
'that
'that
I come'
you come'
he comes'
she comes'
it comes'
we come'
you come'
they come'
man-ma'udai (helm)
pl-ride
we
'we rode'
233
(11)
Structural conditions
When instances of identification by morphological
b.
(12) c.
d.
e.
governs
if and only if
1) o/ m-commands /& , and
2) there is no
, a barrier for
that ^ excludes oC
(14) a.
b.
c.
, such
m-commands
if and only if
1) oC. does not dominate
, and
2) every maximal projection that dominates
dominates oi.
excludes fe if no segment of o(, dominates^
is dominated by /& only if it is dominated
by every segment of fe
For example,
not with the verb per se but with the inflection node
(INFL) which is head of the sentence (IP).
235
In Spanish object-clitic
Juan [ [ l a
am-a]
0 ]
VP V i
i
J
3sfCL.O love-3sS
'Juan loves her'
...[
da-nk
[ (ik) kommen]]
CP
IP
that-lsS
I
come
'...that I come'
of any maximalprojection
it
236
0 fui-mos
invitados a la
fiesta
be/PRET-lp invite
to the party
'we were invited to the party'
zu-k
a-ri
liburu-a irakurri d-io-zu
2s-ERG 3sf-DAT book-ACC read
3sDO-3sIO-2sS
'you have read the book to her'
(20) has the representation similar to (21) at s-structure, i.e., before cliticization at PF.
(21)
238
4.1.3
feature-identified.
239
4.1.4
Summary
The
two are
b.
240
According
If
b.
The point is
4.2
In
For
b.
0
kanjian 0
le
I/you/he/etc. see
I/you/he ASP
!I saw him
Zhangsan
, 0 baba
hen
youqian
i
i
Z
father very rich
'Zhangsan, his father is very rich'
Zhangsan i zai xi
0 shou
i
i
Zhangsan
at wash
hand
'Zhangsan is washing his hands'
b. *Zhangsan
kanjian-le 0
i
shu
i
Zhangsan see-ASP
tree
'Zhangsan saw his trees'
After all, as Huang notes, syntax is otherwise not sensi!tive to the difference between kinship terms and other
!NPs; furthermore, outside the interpretive facts there is
no evidence that null pronouns exist in these construc
tions.
j
i
]
j
I
Huang
Zhangsan
zai xi
baba
i
i,*j
Zhangsan at wash
father
'Zhangsan is washing his father'
244'
(25) b.
Zhangsan
zai xi
shou
i
i, j
Zhangsan at wash
hand
'Zhangsan is washing his/artificial hands'
c.
Zhangsan
zai xi
i
che
i, J
Zhangsan at wash
car
'Zhangsan is washing his car'
(26) a.
b.
c.
Zhangsan , 0
baba
da-le
0
i
i,*j
i
Zhangsan
father hit-ASP
'Zhangsan, his father hit him'
Zhangsan , 0
shou xia-dao le 0
i
i
Zhangsan
hand scare
ASP
'Zhangsan, his/artificial hands scared him'
Zhangsan , 0
che ya-dao
le 0
i
i ,3
i
Zhangsan
car press-fall ASP
'Zhangsan, his/a car ran over him'
0
kanjian 0
le
I/you/he/etc. see
I/you/he ASP
'I saw him'
245
Where there is a
Zhangsan
shuo 0
i
kanjian Lisi le
i
Zhangsan say
see
Lisi ASP
'Zhangsan said that he saw Lisi'
Huang's inspiration for the pronoun-variable distinc
tion is the fact that only the coreferential subordinate
subject has a fixed reference, whereas the other gaps are
dependent upon discourse: the gaps in (21a), for example,
might be interpreted as any pronoun.
These discourse-
TOPIC
TOPIC
i
'TOPIC
0
3
kanjian 0
i
see
saw TOPIC '
i
3
le
3
ASP
246
(29) a.
Zhangsan
,Lisi shuo 0
i
hen
xihuan wo
Zhangsan
Lisi say
very like
I
'Zhangsan, Lisi said he like me very much'
b.
Zhangsan
Zhangsan
'Zhangsan,
xihuan
i
Lisi say I very like
Lisi saidI like him very much'
Lisi shuo 0
hen xihuan wo
Lisi say
very like
I
'Lisi said you/he/etc. like me very much'
b.
247
If that object
(32)
Lisi
shuo ni
kanjian-le ta
L
say you see-ASP
he
'Lisi says that you say him'
1
i
The behavior of the gaps in (33), Huang notes, is precise- j
ly parallel to the phenomenon of strong crossover.
!
!
(34)
*who
did John
i
'
i
f
variableand variables may not be bound (by another argu!ment)according to binding condition C.
j
|
i
I
|
i
248
;
1
j
i
j
j
j
i
,
*
John
ta
i
shuo ni
bu
ken
bangzhu 0
i
J
he say you not will help
John, he said that you will not help him
b.
haizi
yiwei mama
i
hao
zeguai
le
i
|
child think mother will reprimand
ASP
I
the child thinks his mother will reprimand him|
i
Other counterexamples challenge Huangs subjacency claims. ;
Again, Xu allows, it is true that there are instances in
which an object gap inside an island may not be proper,
e.g., Huang's example (31b) above, Xu mentions others in
249
Li xiaojie
zhao bu
zhao [ [ 0 ken qu 0
NP CP j
i
L Miss
find NEG ASP
will wed
de] nanren ]
'
,
man
'Miss Li can't find a man who will marry her'
i
i
Xu's alternative is FEC, the free empty category, the ;
FEC
250
they prove not that the ECs are not variables, but that
the variables are not created by movement.
As for the
These are
251
(37) a.
b.
compra-ste
cafe ?
i
buy-PRET/2s coffee
'did you buy coffee?'
si, compre 0
i
'yes, I bought (some)'
252
Since
Zhangsan
shuo 0
i
xiawu
hui
lai
Z
say
afternoon will come
'Zhangsan says that he will come this afternoon1
Noting that the requirement for a subject antecedent in
the next higher clause which binds the null pronoun is
extremely similar to the phenomenon of obligatory control
PRO, e.g., in (39), Huang suggested a single analysis for
the two constructions.
(39)
wo
zhunbei 0
i
mingtian lai
i
I
prepare
tomorrow come
'I expect to come tomorrow'
253
That is, Huang proposed that the null pronouns in (38) and
(39) are identified by a Generalized Control Rule (GCR).
(40)
*Zhangsan
kanjian 0
i
le
i
Zhangsan see
ASP
'Zhangsan saw himself'
This coindexing violates binding condition B, which states
that a pronoun must be free within a specified domain,
e.g., the minimal sentence containing it.
However, Huang's GCR also has a number of short
comings.
If (40)
(42)
Zhangsan , [ [ 0 0
changge de] shengyin ]
i
NP CP i j
j
Zhangsan
sing
voice
hen haoting
very good
'Zhangsan, the voice with which he sings is
good'
Huang argues that the GCR coindexes the subject gap with
the topic, and thereby explains why only subject pronouns
appear inside complex NPs, but the analysis requires that
he alter the definition of 'closest nominal element', such
that it ignores the head of the relative clause.
There is a third, more compelling reason to seek an
alternative analysis, the fact that Huang's identifier,
the closest nominal element, does not govern the null
pronoun it identifies.
in
Zhangsan
shuo 0
i
xiawu
hui
lai
Z
say
afternoon will come
'Zhangsan says that he will come this afternoon'
wu xiansheng
Wu [title]
I know
this
'Mr. Wu, I know this man'
b.
man
shige li
wuge lan
le
i
i
ten
pear five spoil ASP
'of the ten pears, five spoiled'
The
256
yinwei shige li
wuge lan
le...
because ten
pear five spoil ASP
'because, of the ten pears, five spoiled...'
^ - C P ___
SPEC_____________ __
C_______________ ___
TOP
J^IP
NP
I'
I
yinwei shige li wuge
lan le
Zhangsan
shuo 0 xiawu
hui lai
i
Z
say
afternoon will come
'Zhangsan says that he will come this afternoon'
i
(46)
Zhangsan
i
shuo [ [ 0
CP IP i
This
This
258
No such
Huangs GCR
259
(48)
(49)
*(mi)
*(bo)
*(&)
*(nos)
*(boso)
*(nan)
(50)
mi
ta bini
ta bini
ta bini
ta bini
ta bini
ta bini
kier PRO
i
'I am coming'
etc.
bai
i
I
want
go
'I want to leave'
Still, null thematic subjects are ungrammatical even when
coreferential with a superordinate subject.
(51)
Maria
sa
ku
i
M
know
'Mary knows
*(e)
ta sabi
i
that she be smart
that she is smart'
unexplained.
260
Nevertheless, the
261
(52) a.
guiya[
he
[na
esti na patgun
NP
i
this L child
bai in-na'puti e
[ i ma'a'nao
0 e
CP
j i
C sSU-afraid
]]]
i
that IpSU-hurt
here is the child that is afraid that we will
hurt'
b.
ha-tattiyi
si Rosa
3sSU-follow
Rosa the boy
there LOC house
[ni ha-fa'nu'i 0 0 gi ma'pus na
Huebis]]
C 3sSU-show
LOC PST
last Thursday
'Rosa followed the boy to the house that she
had shown him last Thursday'
She adds that, unlike objects moved by overt WH-movement,
they do not trigger WH-agreement, either (Chung
1984:124,125).
(53)
hafa
f-in-ahan-na
si Antonio e
what WH.0B-buy-3sSU
Antonio
'what did Antonio buy?'
(54) a. *kao ni-lalatde-nna
0
Q
WH.OB-scold-3sSU
'did she scold them?'
b.
kao ha-lalatdi 0
Q
3sSU-scold
'did she scold them?'
262
Juzi
nin [Marya 0
i
juyanata]
Juan
yuyan [[chay 0
!
1
rijsishka] runa
i
j i
J
Juan think
that
knew
man/ACC
mirkadu-pi kashka-ta
market-in was-ACC
Juan thinks that the man who knows him was
in the market
c.
Juan
j
I
Maria-wan rikushka-ta]
!
1
j
i
i
Juan think that man
Maria-and saw
Juan thinks that man saw him and Mary
This behavior is inconsistent with variables of course,
but expected of pronouns.
The evidence suggests that Japanese is more like
Chamorro and the languages that Cole discussed than like
Mandarin.
Hanako-wa
[Taroo-ga
butta] to
ita
1
1
Hanako-TOP Taroo-NOM
hit
COMP say
Hanako said that Taroo hit her
j
!
I
i
For in
John-wa [ [ 0 0
issyoo hazi-o
kakaseru]
i NP CP j i
John-TOP
life
shame-ACC heap
dameonna-to ]
kekkonsita
3
no good woman-with marry
John married a no-good woman who will
embaress him his whole life'
daremo
[ [ 0 0
katta] susi-o ] tabenakatta!
NP CP i j
j
noone
bring sushi-ACC eat
'nobody ate the sushi he brought'
i
b. *daremo
264
His solu
In Mandarin, he
at all.
to this subsection, I
In Mandarin,
Though this
266
4.3
Sax
tried [PRO
i
Jeff
i,*j
asked Bill how painful it would be
i
3
PRO
to shave his sunburned face]
i, j ,k
267
(59) c.
Coreference
When there is no
268
Consideration of the
As for
The
This classifi
269
The interpreta
OC PRO
I return to this
First and
The null
In a constrained theory
272
et Kiryat-sefer u-lexad-a]
ACC Kiryat-sefer and-take-it
et Axsa bit-i
le-isha
ACC Axsa daughter-my to-wife
Kiryat-sefer and capture it,
my daughter as wife
In section
The
273
Clark (1985),
In sentences with
[
very rarely [ it is [ easy [ OP
CPP
i IP
AP
CP i
[t to please John]] x ]]]
i
i
a true
baseball.
(64)
274
The insertion
of always into (64) does not alter the fact that the
sentence could still be true even if it referred only to
John.
(66)
275
c.
(70) a.
to finish
i
i, J
it is rarely fun PRO to shave myself
b.
c.
Paradoxi
276
In that chapter, I
featureless identifier.
Clearly,
is fun [
[ 0 ][
to][
play baseball]
IPi NP i INFL i VP
[PRO
to love me]
i
Obligatory control
The criterion I have used to diagnose OC PRO is the
279
[PRO
[while PRO
a
3
running in place], J
gave B
reward
b.
gave B
a-
list
c.
saw B
J
i
a.
likes [PRO
J
i
b.
walking downtown]
i/ *3
being a movie star]
[while PRO
j
i,*j
to be a movie star]
wants [PRO
i
c.
persuaded B
J
i
d.
promised B
[PRO
j
to leave]
*k
3
i,*j,
(76)
it is essential to me [PRO
to win the race]
i
i,*j
(77) a. here is your new razor [PRO
to shave
i
i,*j
with e ]
J
[PRO
to leave]
i-j ,*k
b. I
witnessed J's
attempt [PRO
to escape]
i
j
In some of the structures in (74) through (77), it is
indeterminate which of the antecedents in the next higher
clause is the controller.
280
Jasked B
[PRO
i
3
i,*j
himself]
to be allowed to shave
b.
Jasked B
[PRO
i
3
*i,j
to shave himself]
281
j
This proposal, which thus far relies heavily upon the '
i
j
1 analysis presented in Williams (1981), answers two of the
i
questions posed above. The shared indexing is obligatory
because the predicate-linking rule is obligatory.
Fur
we
tried [
[
PRO [
to ] [ win
CPi IPi
i INFL i
VP
the race ]]]
1
I
I return to this
!
I
OC PRO be thematic.
When
a predi-
| cate seeks a subject due to the exigency of the predicatej linking rule, it may be coindexed with any of the arguiment-positions which it c-commands.
Where that
A-position
I
So|
i
also is it OC in constructions where there is more than a I
I
single thematic antecedent, e.g., (74).
In these latter
|
i
cases, the choice of subject is apparently determined by
t
i
i
|
beis OC
;
i
283
(82)
*PRO
rained
i
?PRO
read a book
i
It is clear
those
(84) c.
The only
ject, however.
In particular, it is
And as Mohanan
In Hindi, lexical
]
l
ii
!
(Mohanan 1985:643).
(86)
a. baccoo
ne[ ek duusre
ki
taswiire] dekhll
i
i
children ERG each other GEN picture
saw
'the children saw each other's pictures'
b. *bacco6
ne soca
1
ki
[[ek
duusre
.
1
ki
i
i
|
Jon
ne bil se [PRO
haathi ko maarne ka]
i
j
i,*j
John ERG Bill ACC
elephant ACC beat/INF
vardaa kiyaa
promise make
'John promised Bill to beat the elephant'
j
|
I
i
i
286
(88) a.
Jon
skipa i Haraldi
bi ja Marie
klipa
i
3
k
John order
Harold
ask
Mary
pinch
sig
i, j ,k
self
'John ordered Harold to ask Mary to pinch J/H/M'
b.
Jon
skipa i Haraldi
[PRO
lofa
Marie
i
j
*i,j
k
John order
Harold
promise Marie
a [PRO
klipa Onnu]]
*i,j,*k
pinch Anna
'John ordered Harold to promise Mary to pinch
Anna'
Yet as (88b) demonstrates, the controller of OC PRO is not
allowed the same range of possible antecedents.
In addition, this analysis explains Lebeaux's linked
readings for PROs in the same sentence, e.g., (89), with
out recourse to a null operator, which Lebeaux suggest is
adjoined to the main clause, whence it binds both PROs.
(89)
[PRO
to love me]
i
level at which predication relations are defined, i.e., dstructure, the structure of (89) is as in (90).
(90)
is [PRO
i
to know me]
i
[PRO
to love me]
i
Aoun
the subject,
b.
(93) a.
hayi t-um-aitai i
lepblu
who
read the book
'who read the book?'
*hayi ha-taitai i
lepblu
i
taotao ni tumaitai i
lepblu
the person who read
the book
'the person who read the book'
b.
guahu tumaitai i
lepblu
I/FOC read
the book
'it was I who read the book'
c.
ha-taitai 0 i
lepblu
read
the book
'he/she read the book'
Aoun's
Nonobligatory control
Despite the recent attempts in the GB literature to
John
b.
John
[it is possible e
i
i
to feed himself]]]]
i
[PRO
i
289
*John
i
PRO
j
to feed himself
i,*j
The indexing in
Without this
This
In
290
(97) a.
told B
i
b.
c.
d.
they
e.
i
f.
[
a man [PRO to talk to e ]] finally arrived
NP
i
i
d.
J
J
asked B
i, 3
[what PRO
to do]
to do]
i
j
i,j,k
I bought a table [on which PRO
to place
i
J
J
the flowers e ]
3
J bought a book on [how PRO
to succeed]
i
i, 3
291
That is, CP
is the predicate.
However, by (85), CP is only a predicate when its
head is unspecified for features.
shouted [PRO
i
to leave]
i, j
292
(101) a.
b.
!
i
!
interpretation
1
i
]
In the second example, either of two sources might be j
Jthe cause of the definite interpretation of PRO, i.e., if
NOC PRO is parallel to a lexical pronoun.
The antecedent
!
i
i
I
j
,
|
b.
they
thought that J & B said that [their
!
i
j
i,j,k 1
feeding each other] would be difficult
!
i
The subject of the gerund in (102a) might be J & B, they,
or even the union of both groups.
!
i
In
fact, this is
the the only point which distinguishes them: a NOC PRO may :
;be bound by a linguistic antecedent.
however.
That is,
a.
b.
should be interpretations in which NOC PRO is coreferential with thematic elements which do not c-command it,
parallel to the following examples involving lexical pro
nouns .
(104)
a. for them
children
i
b. for them
i ,3
would be the highlight of J and B s
i
celebration
c. their
i, j
that B and K
forever
294
(105) a.
PRO
b.
PRO
i
to go to dinner in Beverly Hills would
i, 3
be the highlight of J and B's
c.
PRO
celebration
i
feeding each other became a problem
that B and K
forever
b.
(109) a.
PRO
b.
his
A null unspeci
For it to be interpreted,
Thematic control
These
b.
Instead, Jaegglis
In (112b),
In my (112) example,
298
Summary
My analysis of PRO can be summarized as follows.
PRO
It is also
If the spe
If
299
The domain of
Unlike
300
4.4
301
buy
for which of his children did Maria buy i t
302
(116) b.
OP
[a Joana viu 0
i
Joana saw
on TV
last
night
Joana saw him on television last night
No such restriction constrains the distribution of null
t
arbitrary objects in Italian, however (Rizzi 1986:514).
(117)
|vious one because LF movement does not obey the doublyfilled COMP filter; since movement at LF generally affects
non-WH quantifiers, this also lends itself to the view
that arbitrary interpretation is an instance of universal
quantification (cf. Epstein 1984).
b.
c.
303
But first an
i
!
i
]
above.
304
rend 0
intelligent
i
therapy
which makes
intelligent
est une therapeutique qui
rend 0 sur de soi
is a
therapy
which make
sure of self
'a therapy which makes 0 intelligent is a therapy
which makes 0 self-confident'
Furthermore, he relates the failure of the indexing in
(120a) to Weak Crossover effects in (119b) (Authier
!1987a:10).
i
(120) a. *son
b.
ambition rend 0
vulnerable
i
i
one's ambition make
vulnerable
'one's ambition makes 0 vulnerable'
;
I
*who
i
j
i
upset e
i
j
i
[PRO
b.
who
i
i
!
|
'
,
j
J
j
j claims about the quantificational force of PRO.
i
j
For in-
(123)
krapula usea-sti
saa 0 sairaa-ksi
hangover frequent-ADV make
sick-TRNS
'hangovers frequently make 0 sick'
sick.
I
As for the facts in (120) and (121), these seem to be '
part of a more general property of arbitrary null objects,
i.e., that the only two nominals which a null arbitrary
object may be coindexed with are another arbitrary null
object, e.g., in (119) above, and an arbitrary PRO, i.e.,
free PRO without a pragmatically induced interpretation.
(This Italian example is from Rizzi 1986:511).
(124)
un illusione [PRO
sperare [che
un buon
i
be a illusion
hope
that a good
pranzo possa riconciliare 0
con se stessi
i, j
meal
can
reconcile
with oneself
'it is an illusion to hope that a good meal can
reconcile 0 with oneself'
A null arbitrary object may not corefer with any nonarbitrary NP (example from Authier, p.c.).
(125)
*Jean
trouve qu'
un bon
psychanaliste peut
i
Jean think that a good psychoanalyst can
reconcilier 0 avec lui-meme
i
reconcile
with himself
'Jean thinks that a good psychoanalyst can
reconcile him with himself'
And it is not apparent how a variable analysis would
account for this restriction.
*Jean
trouve [OP
i
[qu'...e
i
...]]
i
John
i
is easy [OP
i
[PRO to please e ]]
i
However, the
The thrust of my
supports my claim.
308
identi
the appro
309
Let us
This linking
An object PRO,
Thus, it is unable to
A PRO subject
That is,
310
In
b.
311
In chap
312
The lexicalist
Three major
The result is
INFL
V
NP
John
-ing refus- the offer John refusal of the offer
'John's refusing the offer'
'John's refusal of the offer
This difference has been assumed in nearly all transforma
tional generative analysis of the past decade.
It is commonly observed that arguments of a verb may
be missing in derived nominals, e.g., in the following
examples from Williams (1985).
(133) a.
b.
c.
admiration of him
d.
of attempt
PRO
attempt
to leave
b.
c.
315
INFL
the
-tion
invaders
VP.
V
NP
destroy
Rome
marks the subject of IP in situ (cf. the rule of ofinsertion which operates on objects), that subject may
participate in obligatory control in a manner parallel to
subjects of other OC constructions.
Furthermore, it is
317
(142) c.
d.
the destruction
b.
a sneeze
by a stuffy bureaucrat
c.
a prayer
d.
a single
by Chili Davis
b.
c.
Note how the fact that NOC and OC readings exist for these
null subjects is consistent with the fact that the PRO
318
This
Such a rule is
[ e t
i
[ t of Mary]]
i
(147).
319
who
This prediction
Rome 1s destruction
320
the destruction
b.
c.
(153) a.
b.
role
the
321
This
322
To sum
;
!
;
i
b. lorsque je vois un banc , je saute par-dessus 0 I
i
i;
when
I see a bench
I jump over
;
whenever I see a bench, I jump over it
c.
ce
!
I
*le
bebe
a bave
dessus 0
i
the baby slobbered upon
the baby slobbered upon itself
By default, it is a null pronoun.
i
324
dans la
foule et
quand les
i
there is
a
woman in
the crowd and when the
lumieres se sont eteintes, le meurtier a tire
lights
went out
the murderer
shot
dessus 0
i
upon
1there was a woman in the crowd and when the
lights wnet out the murderer shot at her1
This hypothesis is compatible with Z-Hs syntactic obser
vations as well.
binding condition A.
(161)
at the
325
b.
Note the
a. dans
sur
sous
dedans
dessus
dessous
b. autour de
pres de
% 1 1exterieur
autour
pres
^ 1"exterieur
c . avec
derri&re
devant
de
avec
derri^re
devant
Com
a. *je m'assieds
toujours sur la
b. *je m'assieds
toujours la sur
sur moi
b. *moi-dessus
c. *je-dessus
At the very least, the locative interpretation is closely
tied to these morphological changes.
327
However, be
in
c.
im bin kik-im
yu lasnait
he PST kick-TRAN you last night
?he kicked you last night1
dog i bin bait-im
0
dog
PST bite-TRAN
the dog bit him/they'
4.4.4
you go on ahead!
go on ahead!
you go on ahead!
somebody go on ahead!
This removes
In
330
(171) a.
b.
It might also be
Where there is
332
(172)
skamje dy/*my/*him
shame you/me/him
fbe ashamed of yourself/myself/himself'
(173) a.
b.
c.
In the absence of
in the world.
333
Yet to be analyzed is
I return to
j
!4.4.5
(175) a.
b.
[ Mary
NP CP
owiza
wa kage] 0
i
!
1
ki] he ophewathu
i
!
i
IHRC (176a) as it is out of a headed relative clause
|
i
(176b), a point which is easily captured by the assumption!
I
of a null head but anomolous otherwise (Cole 1987:286,287).
i
(176) a. *[[Juan [[0 sisa-kuna-ta japi-shka] 0 ]
|
i
i
Juan
flower-PL-ACC pick-NML
|
gushta-j] warmi ] juyaylla-mari
i
j
I
like-NML woman
beautiful-VAL
'the woman that Juan like the flowers that
|
picked is beautiful'
j
334
b. *[[Juan [[ 0
0
i
japi-shka] sisa-kuna
j
]
J
Juan
pick-NML
flower-PL
gushta-j] warmi ] juyaylla-mari
like-NML woman
beautiful-VAL
the woman who Juan likes the flowers that
picked is beautiful
!
I
i
! inside IHRCs.
pi sni] 0
ki/cha] hena
iyokipi
[[suka wazini
ophewathu] 0
i
cha] sape
dog not-a
IsSU/buy
IND
'no dog that I bought is black'
In (177b), on the other hand, a matrix
sni
j
'
i
black NEG
negation marker
j
I
1
clause.
j
But this
head at s-structure?
Cole notes that IHRCs are found only in languages in
which null anaphora (i.e., null pronouns) are generally
licit.
The
Yet
(178) a.
b.
for the
the speci
direct.
Still, it can be maintained that the null head is
identified by that operator.
The index on CP
337
4.5
338
llam-an
a la puerta
call-PRES/3p at the door
'Arb is knocking at the door'
subject have
339
(181) b. *yo
dije que 0
llaman a la puerta
i
*tu dijiste que 0 llaman a la puerta
i
i
*el dijo que 0 i llaman a la puerta
i
i
*nosostros dijimos que 0 llaman a la puerta
i
i
*vosotros dijisteis que 0 llaman a la puerta
i
i
ellos dijeron que 0 llaman a la puerta
i
i
'they said that they are knocking at the door'
i
b.
340
(185)
Condorav
342
llam-an a la puerta
call-PRES/3p at the door
Arb is knocking at the door
It cannot
may choose a
identifies
b.
h&n-en vaimo-nsa
3s-GEN wife-3sP0SS
his wife
0 vaimonsa
'ones wife'
343
344
min& kirjoita-n
sinSi kirjoita-t
hSn kirjoitta-a
me
kirjoitai-mme
te
kirjoita-tte
he
kirjoitta-vat
'I write'
etc.
Yet these agreement forms are not equal for the purpose of
identifying a null subjects, however.
A null pronoun is
(minS) kirjoit-a-n
kirje-et
I
write-PRES-ls letter-ACC/PL
'I write letters'
b.
345
Pekka
tiet&& ett& 0
on oikeassa
i
Pekka know
that be right
'Pekka knows that he is right'
i
b.
Peka-n
&iti
i
tiet&& ett& 0
j
on oikeassa
*i,J
P-GEN
aunt
know that
be right
'Pekka's aunt knowsthat she is right'
It is also possible for the null third person pronoun to
remain indefinite in the presence of a possible binder,
however.
(191)
Joulupukki
n&k-e-e
i
kun 0
tek-e-e
,
arb
see-PRES-3s when
do-PRES-3s
pahaa
bad
'Joulupukki(=Santa Claus) sees when one does
bad'
Basically the same situation is evidenced in the
Finnish nominal agreement paradigm as well: an overt third
person pronoun yields a definite interpretation and a null
pronoun with third person agreement is indefinite, except
when it is interpreted as coreferential with some ccommanding NP.
(192) a.
b.
(193)
h&n-en vaimo-nsa
3s-GEN wife-3sP0SS
'his wife
0 vaimonsa
'one's wife'
Pekka
voi vietellS [0
i
vaimo-nsa]
i
In this lan
b.
talila ma'amina se
0 hiclixa
ba-bxina
i
i
Talila believe that
succeed/PST/3sf in-test
'Talila believes that she passed the test'
dani
bikes me-talila
i
se
j
tavo
*i,j
Dani ask
to-Talila that
come/FUT/3sf
'Dani asked Talila if she will come'
There are two differences between Hebrew and Finnish,
however: a third person singular pronoun in Hebrew is
never licit by itself, i.e., the null pronoun in (194b)
does not receive an indefinite interpretation in Hebrew;
and, as a result, the coreference in (195) is obligatory
in Hebrew.
Borer (1986) suggests that the above range of facts
can best be accounted for by assuming that the third
person singular INFL of past and future tenses, like the
INFL of OC infinitival complements, is anaphoric and so
requires an antecedent.
(196)
*dan
1amar le-Talila
i
se
0 tamid
3
that
be late/PRES/3sf
i
J
third per
i
j
I
distinct in Hebrew, e.g., the antecedent of PRO is deter- !
mined absolutely whereas the controller of the
son
third per
348
(197) a.
b.
Dan
'amar le-Ran 0
la-vo
i
j *i/j
Dan say
to-Ran
INF-come
'Dan told Ran to come'
Dan
'amar le-Ran
se
0
yavo
j
i,j
Dan say
to-Ran
that
come/FUT/3sm
'Dan told Ran that he(Dan or Ran) will come'
i
a.
b.
jokainen
tietSS. ettS on suotavaa ettS
everyone know
that be advisable that
0 tule-e
aikaisin
come-PRES/3s early
'everyone knows that it is advisable that one
*come early'
'every X, X a person, knows that it is advisable
that X come early'
tied&n ett& 0
i
on
oikeassa
Is
know
that
be/3s right
'I know that I am right1
349
b. *sin&
tied&t ettS 0
i
on oikeassa
i
2s
you know that you are right
(190) a.
:
Pekka
tiet&S. ett& 0
i
on oikeassa
i
j
j
Pekka know
that
be right
Pekka knows that he is right
!
Besides,according
tothe theory
I
|nounsI havedeveloped,
!
!
they pronouns
cannot bearbitrary
When
I
\
b.
j
j
Peka-n
j
| Note that this analysis achieves the correct interpretsi
The interesting
j
I
j
i
1
I
t
j
(1980:763).
omrim
se
hem
mesugaim
i /J
say/3p that they
crazy
'they say that they are crazy'
i
b.
hem
yadu se
yefatru
et
Dan
i
i, j
they know that
would fire/3p ACC Dan
'they knew that they would fire Dan'
It is no doubt significant that Finnish lacks indefinite
null pronouns with third person plural agreement.
Assume, therefore, a parameter such that indefinite
null pronouns may be accompanied by third person singular
agreement or third person plural agreement.
Finnish is
set for singular agreement; Hebrew and Spanish are set for
plural agreement.
First, whereas
[+anaphoric],
Third, in addition
4.6
No changes were
354
Take, for
A transi
355
By my analysis, this
The exten
356
5.0
(1), the prediction was that only the 0+A (null pronoun
with agreement) and *0-A (obligatorily lexical pronoun
without agreement) columns should be filled by positive
(+) entries and that these columns should uniformly be
positive.
<
i
o
Sthem
Sinf
Simp
EXE
Dobj
Iobj
Poss
PPobj
0+A
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
*0+A
+
-
+
-
*0-A
+
-
+
+
+
+
+
+
357
In
Sthem
Sinf
Simp
EXE
Dobj
Iobj
Poss
PPobj
+
+
+
+
+
+
+
+
O-ID
*0+ID
*0-ID
+
-
(+ )
(+ )
-
+
+
+
+
358
It is interesting
Is
2s
3s
lp
2p
3p
*(eg)
*(/*u)
*(hann)
*(vi^)
*(^i%)
*(^eir)
hef, hefi
hefur, hefir
hefur, hefir
httfum
hafi^
hafa
'I have'
etc.
Yet
359
PRO,
How
Since
That
It is concluded that
361
There,
I conclude this
of
5.1
Principles of licensing
A priori, it is not clear what a licensing principle
might be.
363
5.1.1
Licensing of ECs
This conclusion is
involved in the licensing of null pronouns; after all, NPtraces and WH-traces must be properly governed according
to the ECP, and PRO (again, for Rizzi and others) is
licensed by way of being ungoverned.
I too have recognized that government is involved in
the analysis of null pronouns, though in the previous
364
The
365
However,
This analy
In
366
less motivated: he argued that Case assignment at sstructure causes lexicalization at PF, and that Caseassignment in NSLs may be delayed until LF, which does not
feed PF.
These are not the only possible hypotheses concerning
Case and licensing, of course.
b.
0 ne
e
stato fotografato uno nudo
of-them has be
photograph
one nude
'one of them has been photographed nude'
EXE vengono fotografati 0 nudi
are
photograph
nude
'they
are photographed nude'
DEF/*ARB
Though the
Second, it also
In Chomsky (1981),
369
5.1.2
a. an NP is nominative if governed by
INFL [+AGR]
b.
c.
an NP is oblique if governed by P
(7)
(8) a.
b.
The rule of
However, we
370
Imper
Note
In particular, it is only
WH-traces, in particular, are commonly CaseThis status follows from standard assumptions
Since it seems undeniable that WH-trace is Casemarked (or at least the A'-chain itself, cf. the French
372
This
|
i
I
tion to lexicalization.
i
j
(15)
Case
a.
b.
c.
d.
lexical NP
WH-trace
null pronoun
PRO, NP-trace
Features
+
+
Lexical
+
jWH-trace is null for him because the Case assigned a WHtrace is absorbed by its antecedent; instead, I maintain
that Case remains with the trace, which explains why
contraction is blocked in (14).
i
\
'
i
Before demonstrating the particulars of how the prin- |
i
i
null pronouns, it is instructive to put my hypothesis in a j
wider perspective.
The starting point for this perspective is the dis
cussion of identification and visibility in the conclusion i
i
of theprecedingchapter.
sis that
an NPmust
There,
bevisible
Iaddressed
at LF
the hypothe- 1
in order to receive
!
j
1
374
The
<
positions within
j
i
a visible A-chain has Case or is headed by PRO. |
(17)
!
I
j
(18) a. *under
the stars to seem the best place to sleep,i
you have to be crazy
b.
:
I
t
i
Case.
it cannot be lexical.
i
As such, the infinitival must be linked to some antece
dent, with the result that the nonthematic subject is
reference.
a predication.
I have mentioned two ways in which an infinitive
might escape being a predicate; if the head of CP is
filled with features via subcategorization, e.g., the way
a sentential complement to a verb like wonder subcate-
i
I
constructions.
This is
^
i
|
b.
(Chomsky 1985:95)
c.
d.
i
!
i
(Fabb 1984:54)
i
In English, it is difficult if not impossible to decide
whether this prediction is correct, for abstract Case
rarely has a surface reflex in this language.
However, in
these
poija-t
ovat opiskelija-t
boy-NOM/PL be
student-NOM/PL
'the boys are students'
b.
saan-ko
puhutella johtaja-a
be able-Qprt speak
manager-PART
itse-&-&n?
self-PART-3sP0SS
'may I speak to the manager himself?
c.
I
|
|
|
i
377
j Case.
|
\
. mystery.
In
Thus, it is
If the arguments of this section are correct, however, there is a one-to-one correlation between Case and
!
i
hypothesis.
5.1.4
I
I
378
assignment.
:
J
1
!
i
j
nouns.
In so
i
I
j
behavior in particular languages, but also provide a solu-|
arguing, I not only explain otherwise puzzling syntactic
tions, one for subjects and the other for direct objects,
though I assume that the arguments and conclusion extend
I
i
t
j
INFL.
through predication.
in section 5.4.
It has long been held that subjects occur in the
specifier of IP position, e.g., in (23a).
Given the
I
j
i
(23) a.
^IP
SUB J
^ 1 '^
INFL
b.
^-IP
SPEC
|
SUBJ V P !
I
The evidence I present in favor of the subject com
VP
^-1
INFL
!
i
Dutch (including the nonNSL dialect A), like many of |
the Germanic languages, has constructions in which a non! thematic subject cooccurs with a thematic subject.
I
.
'
J
380
(24)
|
I
|
|
!
The first
Thiersch
adjoined position.
All
381
position as well.
I therefore
VPconsituent, asargued
inMcCloskey
there
I
is aj
(1983). Thus,
|
i
i
i
i
382
Sproat's
The
i
I
derivation thus also requires the adjunction of INFL to S, i
i
the result of all the movements being something like (25). !
(27)
V + INFL
NP
Tf
NP
i
V-INFL movement apparently operates in many languages
(cf. Koopman 1984b, Koster 1976, Torrego 1985) and it
follows general principles of movement (cf. Chomsky 1986).
But there is
INFL-adjunction rule.
According to
status ofSproat's
Chomsky(1986), the
[
[ V [NP [ INFL [
t
CP C ? i IP
I
VP' i
[ t ...]]]]]]
VP i
i
i
383
INFL
This
To review,
it has been posited that the verb (like the noun) assigns
384
right.
The assump-
I
the government domain of V according to the definition of 1
government argued for in Chomsky (1986) and adopted in thej
current work (cf. section 1.2).
I noted ;
In general,
j
!
I
ungrammatical.
Mandarin.
i
j
j
|
Romes destruction
5.2
Subject pronouns
i
i
j
A-position hypothesis.
Parameters concerning
!
I
these parameters with the directionality of Case parameter:
i
;
is shown to predict those types. The section concludes
with a summary of subject licensing.
I
5.2.1
!
i
,
itypes,
j
j
Where appro-
i
diacritics under EXE in chart 9) and nonNSLs (the four *0A and *0+A cells under EXE in chart 9), respectively.
In nonNSLs, of which there are only four--Dutch A,
English, French and Swedish--in the survey, nonthematic
i
|
subordinate clauses.
!
I
|
(33) a.
b.
388
EXE
Sthem
0(A)
0-A
American Sign Language
0+A(1)
0+A(1)
Arabic, Classical
0+A
0+A
Arabic, Bani-Hassan
*0-A
0-A
Babungo
0+A
0+A
Blackfoot
Bobangi
0+A
0+A
*0-A
0-A
Cape York Creole
0-A
0-A
Caviteno
0(A)
0+A
Chamorro
0+A
0-A
Chorote
*0-A
0-A
Duka
Dutch A
*0+A
*0-A
0-A(2)
Dutch B
*0+A
0+A
0+A
Egyptian, middle
English
*0( A )
*0+A
Finnish
0+A
0+A
0+A(2)
Flemish, west
0+A(2)
French, modern
*0+A
*0-A
French, Old
0+A(2)
0+A(2)
0+A
0-A
Fula
0-A
0-A
Garo
Georgian
0+A
0+A
German, Bavarian
0+A(2)
0-A(2)
German, standard
*0+A
0-A(2)
Guaymi
*0-A
0-A
Hausa
0+A
0+A
0+A
0+A
Hebrew
0+A
0+A
Hua
*0+A
0+A(2)
Icelandic
Indonesian, Betawi
0-A
0-A
Lenakel
0+A
0+A
Italian
0+A
0-A
Luganda
0+A
0-A
Mandarin
0-A
0-A
Mundari
0+A
0-A
0-A
0-A
Murut
0-A
Papiamentu
*0-A
0+A
0+A
Portuguese, Brazilian
0+A
0+A
Portuguese, European
0+A
Siroi
0+A
Spanish
0+A
0+A
Swahili
0+A
0-A
*0-A
Swedish
*0-A
*0-A
0-A
Tagalog
Tamazight
0+A
0+A
0-A
0-A
Thai
0-A
Turkish
0+A
0-A
0-A
Vietnamese
I (1):
main clauses only; lexical in subordinate clauses
I (2):
null in various sites; always
lexical when initialin
|
main clauses only
|
Chart 9: thematic and nonthematic subjects
|
I
I
0+A and
In these
This
In Classical Arabic, to
complementizer ?anna,
(35) a.
b.
I
I
In
f>u
heldur (a'fc) (*^a%) var dansa^ a*S skipinu
you
believe that
it was danced on ship
'you believe that [it] was danced on the ship'
390
>i 1-L]J
Wdt>
-L U
U a i l U B U
on the ship'
b.
c.
var
dansa$
a skipinu
was
it danced on
ship
'was [it] danced on the ship?'
Among
(38) and
3911
(39) a.
b.
(40) a. es ekelt
mir
vor
dir
it disgust me/DAT before you
'you make me sick'
b.
b.
I
|
ze se
hu kolkax satum margiz oti
that he so
dumb annoy me
'it annoys me that he is so dumb'
'
I
!
i
(46) a.
393
(46) b.
(47) a.
i
|
I
1
b.
j
I
j
|
I
;
I
394
This
type is also exemplified in the survey by three verbsecond languages: Old French, Bavarian German and West
] Flemish (as well as Gothic (Wright 1899:134), Old Frisian
! (Markey 1981:170), and modern Frisian (Reuland 1983:32)
I
from outside the survey).
In these verb-second languages,
subjects are optionally null in inversion structures but
invariably lexical in sentence-initial position.
(These
a.
kumm-st (du)?
come-2s you
are you coming?'
i
b.
g gem
given
!
]
lan-
395
b.
i
;
|
i
!
j
I
l
I
I
The result is
ter languages lack agreement morphology, it is quite likely that they lack null thematic subjects because of the
failure of identification.
I
i
j
t
i
397
them henceforth.
follow in (52).
(51)
core NSL (31)
restricted NSL (4)
EXE-NSL (9)
core nonNSL (4)
(52) a.
b.
c.
d.
EXE
null
null/lexical
null/lexical
lexical
pro
opt. null
opt. null
lexical
lexical
core NSL:
core nonNSL:
A fifth
5.2.2.1
Core NSLs
These
i
|
!
j
I
i
assumptions.
First of all, since nonthematic subjects in core NSLs
'
i
i
complement of IP.
I
!
The >
I
discussion of Dutch double-subject constructions in sec\
tion 5.1.4 made it clear that the nonthematic subject
precedes the thematic subject.
Note
Let us there
specifiers of IP.
!
i
]
399
j
j
I
j
If the thematic
i
i
subject were generated in the specifier position, it would I
lack Case, and because movement to the complement position
language.
When a
|
i
thematic subject remains in situ, it is Case-marked and
!
i
lexical; a null subject escapes Case and lexicalization byj
i
j
position.
i
j
j movement.
It is not lexical,
i
|
t
this assertion.
i
structure.
In this
position of IP.
j
!
5.2.2.2
I_ _ _
I
401
That is,
I
<
If it remains in
I
between the thematic and nonthematic subjects, ostensibly |
to transfer the features which would identify the null
j
i
I
1
'
I
402
5.2.2.3
I
1
i
The chal-
I
1
j
j these
I
i
I jects whereas languages of the other type do not.
i
i
!
Nothing in what I have thus far introduced allows the |
i
generation of this kind of variation.
If the directional- [
!
j
i
|
i
j
sections.
i
i
!
'
'
j
I
403
English
b.
Swedish
c.
German
d.
Icelandic
b.
c.
d.
i
|
404
(55) a.
d. *Jon telur
y6a^ hafn einhvern eti^ hakarlinn
John believe
have somebody eat shark
'John believes there to have somebody eaten the
shark'
Third, subjects control agreement to some degree in all
four languages, but only the nonthematic elements in (col
loquial) English and Swedish do so (p.87-88).
(56) a.
b.
c.
es kamen
viele Leute
came/3p many people
'there came many people'
d.
a.
b.
the drainpipe
j
I
405
I
|
i
i
I
(58)
;
i
es [ sind
CP
C'
[
[ PP NP t ] t ]]]
i IP VP
i
i
i
I
I
i
I
b.
[ dass [ [
CP C'
IP VP
wird getanzt]]]]
i
i
The structures in
[ es
CP
...[
CP
[ sind [0 [ P P N P t
]t
C'
i IP
VP
i
[ dass [ 0 [ ...]]]]
C'
IP
VP
]]]
i
5.2.2.4
I
I
!
as to how restrictedNSLs
nominative Case.
(
i
;
!
407
morphology.
The subject of a
!(Rizzi
!
j
!
1982:83).
(62) a.
i
!
I
i
'
;
ii
(64)
|
Il
In the literature, the discussion concerning (61) andj
However,
one might argue that there is a universal rule of V-toINFL movement, such that INFL is always filled and (64)
alone suffices.
|
i
!
Case assigned
j
i
|
!
index.
409
i
only.
' 5.2.2.5
i
i
j
j
die.
By the Case-as-lexicali-
I
they must be Case-marked.
But this must be g-Case, for two reasons.
First,
|
Therefore, it must be thatg-Case
intheEXE-NSLs
!
i
is
are sentence-
410
____
i
1
However, Platzacks
evidence, repeated below, indicates that an inverted nonI thematic subject in German and Icelandic is not lexical.
(54) c.
d.
Case
it is lexical.
(65)
[EXE [ V+INFL
[
[ ...t
] t ]]]
CP
Cf
i IP VP
i
i
I
j
[[ V+INFL [ EXE [
...t [t ]]]]]
CP C'
i IP
VP
i
i
'
operative.
j EXE-NSLs.
in
(Platzack 1983:95).
I
j
(67) a.
b.
(68) a.
b.
j
!
!
!
412
(69) a.
b.
eine bfise
Hexe steuerte
die Frfische bei
a
wicked witch contribute the frogs
with
'a wicked witch contributed the frogs'
es steuerte eine bdse Hexe die FrOsche bei
I
But both !
|
t
However,
j
i
j
t
i
i
!
i
|
i
makes it clear that one more distinction needs to be made. ,
i
To this point, I have posited four types: core NSL,
'
restricted NSL, EXE-NSL, and core nonNSL.
In this last
!
'
I
1
I
1
i
I
i
(71) a.
b.
j
i
(72) a.
(73) a.
j
i
j
are always lexical in Dutch, but only lexical in sentenceinitial contexts in Icelandic.
tions.
in section 5.3.1.
I
i
I
i
Still,
I
j
|
+
j
I
415
i
kinds of double-subject constructions found in Dutch, as
(71) through (73) indicate.
b.
c.
|
j
;
'
|
I
I
|
j
The difference between what I have been calling double! subject constructions and these existentials is signifij cant: only in the former is there both a nonthematic
!subject
of the predicate.
in (75) is the
in
subject of
416
i
!
;
!
!
I
'
i
(75)
sentences.
null
thematic
subjects
null
nonthem
subjects
lexical
nonthem
subjects
yes
yes
yes
no
no
no
yes
yes
yes
yes
double
subject
sentences
7
yes
yes
yes
no
!
j
1
core NSL
restr NSL
EXE-NSL
Dutch A
core nonNSL
yes
yes
no
no
no
a.
b.
c.
d.
e.
f.
gh.
direction
of
G-Case
direction
of
F-Case
thematic
subject
position
right
right
right
right
left
left
left
left
right
left
left
left
right
right
left
left
CPLT
SPEC
CPLT
SPEC
CPLT
SPEC
CPLT
SPEC
I
1
1
1
core NSL
*
!
1
*
restricted NSL
*
I
EXE-NSL; Dutch A
nonNSL
Recall that
,
j
A (b)-type of lan
i
j
j
from j
i
I
|
418
the reasoning went, it was likely that they had the same
i
jstructure as NSLs, the sole difference being in terms of
identification.
5.3
419
it is
j
i
'
!
i
|
1
|
theta roles to the specifier position predicts that there
And accor
I
i
Some
!
i
If licensing of null
:
420
The above array of data makes it appear that the lexicalization of argument small clause subjects follows the line
established by thematic and nonthematic subjects of tensed
clauses.
421
l
_
Ii language.
i
agreement-identified null direct objects; still, it has
!
arbitrary
i
(81) a.
i
b.
c.
null objects.
(minfi) kirjoiti-n kirja-n
I
write-Is
book-ACC
'I wrote the book'
Paul Newman voi vietellfi *(sinu-t)
Paul Newman can seduce
you-ACC
'Paul Newman can seduce you/*one'
j
'
>
I
j
null.
!
I
Ahmed a-li-m-piga
(mtoto)
Ahmed 3sS-PST-3sO-hit child
'Ahmed hit the child/him'
(83)
ne me
derangez pas 0
NEG lsDO.CL bother
NEG
'don't bother m e !'
I
j
j
I
I
I
;
!
i
422
I
Following the
correct one.
j
j
:
i
i
i
;
;
\
423
(86) a.
b.
|
1
\
i
i
|
i
verb governing the NP has agreement which identifies it.
!
i
I
i
;
Thus, the
b.
I
!
I
!
j Two operations derive the surface form of this sentence:
j
I
I
!
tic objects.
j
t
|
i
|
(88)
i
j
i
i
j
in (89).
|
l
(89) a. ne me
d^rangez pas 0
NEG lsDO.CL bother NEG
1don11 bother m e !1
b. une bonnebiere reconcilie 0 avec
soi-meme
a good
beer reconcile
with oneself
'a good beer reconciles ARB with oneself1
Jaeggli (1982,1986b) has independently proposed a
mechanism which achieves the desired results.
When direct
(Jaeggli 1982:14).
(90) a.
!
'
!
425
(90) b.
la
vimos 0
3sfD0.CL we saw
'we saw it'
j
t
|
1
i
426
(91) b.
I
1
|
de Aznei
macht ain mede
the medicine makes one drowsy
'the medicine makes one drowsy'
j
'
j
(92) a.
b.
ritengo
0 probabile che ...
I believe
probable that
'I believe it probable that...'
i
It would therefore appear that VP Case is assigned to |
its left (i.e., to the specifier of VP) whenever INFL in
that language assigns g-Case to the left.
be?
Thus, I
|
i
|the
!be null nonthematic elements inside VP, i.e., in constructions like (93).
To summarize: in general, it is argued that the
parameter settings for g-Case motivated in section 5.2
j
*
!
j
!
i
427
This
428
5.3.2
It is known
When in the
430
(95) a.
gard-es qu' il n en
escap un seus
guard-2s that
NEG+of them escape one only
'see to it that none of them escape
b.
ce
sera domige grant se il en
muert
that be
pity
great if
ofthem die
nul
anyone
it will be a great pity if any of them die
i
These two facts follow directly from the analysis of
I
restricted NSLs given in section 5.2.2.5. Unlike the
other restricted NSLs that I have examined, i.e., Bavarian
German and West Flemish, OFr had obligatorily lexical
subjects in subordinate clauses (Adams 1987:2).
i
(96)
einsi corur-ent 0 par mer tant que il
vindr-ent
thus ran-3p
by sea until
they come-3p
a Cademelee
to Cadmee
thus they ran by the sea until they came to
Cadmee
In section 5.2.2.5, I noted that there were two
options for thematic subjects in subordinate clauses in
the restricted NSLs: if V does not move to INFL, g-Case is
irrelevant, F-Case is assigned to the complement position,
so a pronoun which moves to the specifier position is
null; if there is V-INFL movement, however, g-Case is
uniformly assigned to the specifier and all pronouns are
lexical.
(Note
in subordinate clauses.
431
Only
The
subject to leveling.
(97)
singular
first
second
third
-0,-e
-(e )s
-(e )t
plural
-(i )ons
-(i)ez
-ent
432
the
By the end of
singular
first
second
third
-0
-0
-0
plural
-ons
-ez
-0
433
a.
b.
c.
d.
e.
f.
gh.
direction
of
G-Case
direction
of
F-Case
thematic
subject
position
right
right
right
right
left
left
left
left
right
right
left
left
right
right
left
left
CPLT
SPEC
CPLT
SPEC
CPLT
SPEC
CPLT
SPEC
core NSL
*
*
restricted NSL
*
EXE-NSL; Dutch
nonNSL
Due to the
434
As a result, lexical
The directionality of
435
I 5.4
my main task is rather to make comparisons with my analysis on a more item-specific level, i.e., in terms of the
more limited scenarios covered in the competing analyses.
I
It might be supposed that my concentration upon a crosslinguistic expanse of data has made my analysis less
applicable to these specific tasks, but I hope to convince
the reader that my analysis also enjoys a greater adequacy
on the more item-specific level.
I have organized this discussion of competing
analyses by the three phenomena which I have argued must
be at the core of any theory of licensing: the lack of a
lexical alternative to PRO, the failure of identification
in Icelandic, and the distribution of lexical nonthematic
subjects.
frequently.
Each competing analysis will be judged in two ways,
The first
j
j
^
1
1
i
In the recent
!
i
anaphor, which entails that it is subject to the contradictory binding conditions of both ECs.
The ploy is to
I
!
i
the other empirical.
governed.
That is,
it is an ad hoc stipulation.
Furthermore, these
j
i
!
The major
I
i
j
case.
The
j
t
value.
some awareness of the conditions under which a null pronoun may be identified as arbitrary in reference.
Though
l
j
i
|
I
j
i
The other two phenomena addressed in previous
|
I
I
i
I
'
i
|
1
!
;
440
i
>
I
|
First, by
The
By assigning
I
441
; licensing include PRO in a general theory of null pronouns, and both posit the lack of Case to be important to
!
|
null pronouns.
There are, of course, major differences between the
two theories in the way that Caselessness is attained.
For Jaeggli, a null direct object is Caseless because the
clitic identifier (at least in Spanish) demonstrably ab
sorbs the objective Case.
type of language.
!
i
Thus, there
^
i
i
Although it is
j
i
j
I
of licensing heads.
I have also briefly mentioned the analysis in Adams
(1986), where it is proposed that the position of a null
i
j
|
!
j
I
,
|
!
I
i
j
i
j
i
|
!
t
Since she
\
i
;
I
I
'
!
444
As such, it is
uninteresting.
Three analyses remain to be discussed, each of which
directly addresses nonthematic pronouns: Travis, Safir,
and Reuland.
To
if VP and NP are
i) generated adjacent to one another, or
ii) adjacent because INFL is empty,
then NP is the complement of VP
445
a. es/*0 INFL t t VP
b.
XP INFL *es/0 t VP
a. PN/*0 INFL VP
b.
INFL PN/0 VP
i
j
I
Note that this mechanism might be extended to PRO, since a |
c.
INFL VP PN/0
|
i
1
Note that
i
i
subject inversion.
I
Travis1 analysis has other shortcomings as well.
!
,
Ibegin with, her contention that subjects in NSLs are
To
i
<
j
Like all
How
analysis.
Safirs line of reasoning begins with the assumption
I
|
|
|
I
I
j
i
I
j
I
447
(103)
insert es as required
i
iwithout.
S
j
Nevertheless, my analysis is
j
j
j
the j
448
Case.
And if
My Case-
circular.
i
i
r
\
i
^
!
!
j
i
|
Fur
I
j
>
t
i
!
j
450
e.g., Icelandic.
451
!
j
Furthermore, nominative
I
'
\
!
This i
i
provides another context (besides the Case-marked comple- i
ment) in which a thematic subject is lexical, since all
I
j
topics (except those in languages with a discourse-linking!
rule, like Mandarin) are lexical.
il
I
!
This, I
j
\
i
|
I
j
i
j
But it is also
'
j
All
453
That is, ;
In particular, I
!
!
predicted the
454
6.0
Concluding remarks
j
i
I
parameters.
In fact, I believe whole-heartedly that this j
i
dissertation falls clearly within the bounds of the GB
I
research paradigm.
Yet I insist there is a difference.
j
i
j
First is the
matter of perspective.
Even in those
|
j
|
1
!
\
i
455
In
!
This relates to another major difference between the
i
I
I
i
| approach taken in this dissertation and that approach
| characteristic of most other GB theory: the utility of
1
j surface data. As noted numerous times, the GB attitude
I
towards the surface is one of distrust, since all impor
tant generalizations are deemed to be abstract.
The
'
i
essential to this entire enterprise.
!
!
How
|
i
If there is one aspect I
I
flattery, I believe, it is
i
I
I
I
BIBLIOGRAPHY
458
459
I-Subjects. LI 17.375-416.
j --- .
460
et
i
-- .1982.
1
1977.
LI
461
]
-- . 1984. Identifiability and Null Objects in Chamorro.
In Brugman and Macauley, 116-130.
. 1987. The Syntax of Chamorro Existential Sentences.
In Reuland and ter Meulen, 191-225.
Clark, Robin. 1985. Boundaries and the Treatment of
Control. Los Angeles: University of California
dissertation.
Cole, Peter. 1982. Imbabura Quechua. Amsterdam: North1
Holland.
. 1987a. Null Objects in Universal Grammar. LI 18.597612.
. 1987b. The Structure of Internally Headed Relative
Clauses. NLLT 5.277-302.
Comrie, Bernard. 1981. Language Universals and Linguistic
Typology. Chicago: University of Chicago Press.
. 1984a. Language Universals and linguistic argumenta
tion: a reply to Coopmans. JL 20.155-163.
. 1984b. Subject and Object Control: Syntax, Semantics,
Pragmatics. In Brugman and Macauley, 450-464.
and Norval Smith. 1977. Lingua Descriptive Series:
Introduction and questionaire. Lingua 42.1-72.
.Condoravdi, Cleo. 1987. Arbitrary Reference, pro and Bare
Plurals. Papers from the 23rd Annual Meeting of the
!
CLS, ed. by Barbara Need, Eric Schiller, and Anna
Bosch, 18-30. Chicago: CLS.
jcontreras, Heles. 1976. A theory of word order with spej
cial reference to Spanish. Amsterdam: North-Holland.
Cook, William Hinton. 1979. A Grammar of North Carolina
Cherokee. Yale University dissertation.
Coopmans, Peter. 1983. Review of Comrie 1981. JL 19.455473.
. 1984. Surface Word Order Typology and Universal Gram
mar. Lg. 60.55-69.
i
Corum, Claudia; T. Cedric Smith-Stark; and Ann Weiser,
I
eds. 1973. Papers from the 9th Annual Meeting of the
CLS. Chicago: CLS.
462
H.
464
466
468
14.421-446.
. 1986. Phrase Structure, Extractions and Binding.
University of London ms.
Mateus, Maria Helena Mira; Ana Maria Brito; Ines Silva
Duarte; and Isabel Hub Faria, eds. 1983. Gramatica da
Lingua Portuguesa. Coimbra: Livraria Almedina.
Matthews, G. Hubert. 1965. Hidatsa Syntax. The Hague:
|
Mouton.
I
McCawley, James D. 1970. English as a VSO Language. Lg.
46.286-299.
469
I
I
\
and Kenneth Hale. 1984. On the syntax of personnumber inflection in Modern Irish. NLLT 1.487-534.
McGregor, Donald E. and Aileen R.F. McGregor. 1982. Olo
Language Materials. Canberra: LCC.
Meier, Paul; Inge Meier; and John Bendor-Samuel. 1975. A
Grammar of Izi, an Igbo Language. Norman: SIL at the
University of Oklahoma.
Merrifield, William R. 1968. Palantla Chinantec Grammar.
I
Mexico, D.F.: National Institute of Anthropology.
Milsark, Gary. 1985. Case Theory and the Grammar of
|
Finnish. In Berman et al., 319-331.
Mohammad, M.A. 1984. Subject Extraction in Arabic. USC
ms.
Mohanan, K.P. 1983. Functional and Anaphoric Control.
LI 14.641-674.
. 1985. Remarks on Control and Control Theory. LI
16.637-648.
Montalbetti, Mario. 1984. After binding. MIT dissertation.
Moravcsik, Edith A. 1978. Agreement. In Greenberg et al.,
volume 4, 331-374.
Murane, Elizabeth. 1974. Daga Grammar. Norman: SIL at the
University of Oklahoma.
Nguyen, Dang Liem. 1975. Vietnamese Grammar: A Combined
Tagmemic and Transformational Approach. Canberra:
LCC.
470
I
Penchoen, Thomas G. 1973. Tamazight of the Ayt Ndhir. Los
Ange1e s : Undena.
i
471
'
474
r- - - - - - - - - - - - - - - - - - - :
475
of
476