ASPECTS OF

THE ORGANIZATION OF REDUNDANCY RULES

I N THE LEXICON

Raghavachar i A m r i t a v a l l i

M . A . , Bangalore U n i v e r s i t y , 1972

M . A . , Simon F r a s e r U n i v e r s i t y , 1977

A THESIS SUBMITTED I N PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

i n t h e Department

of

Languages, L i t e r a t u r e s , and L i n g u i s t i c s

@ Raghavachar i A m r i t a v a l l i 1980

SIMON FRASER UNIVERSITY

29 August 1980

A l l r i g h t s r e s e r v e d . T h i s work may n o t be reproduced i n whole o r i n p a r t , by photocopy

o r o t h e r means, w i t h o u t p e r m i s s i o n of t h e a u t h o r .

APPROVAL

Name : Raghavachari ~ m r i t a v a l l i

Degree: Doctor of Phi losophy

T i t l e o f D i s s e r t a t i o n : Aspects o f t h e O r g a n i z a t i o n o f Redundancy

Rules i n t h e Lexicon.

Examining Committee:

Chairman : Ross Saunders

Richard C. DeArmond S e n i o r S u p e r v i s o r

B r i a n E. Newton

'. Thomas Wasow E x t e r n a l Examiner A s s o c i a t e P r o f e s s o r S t a n f o r d u n i v e r s i t y S t a n f o r d , C a l i f o r n i a , U.S.A.

PARTIAL COPYRIGHT LICENSE

I hereby g r a n t t o Simon F ra se r Un ive r s i t y t h e r i g h t t o lend

my t h e s i s o r d i s s e r t a t i o n ( t he t i t l e of which i s shown below) t o u s e r s

of t he Simon F ra se r Univers i ty Library , and t o make p a r t i a l o r s i n g l e

copies on ly f o r such u s e r s or i n response t o a r eques t from the l i b r a r y

of any o the r u n i v e r s i t y , o r o the r educa t iona l i n s t i t u t i o n , on i ts own

beha l f o r f o r one of i t s u s e r s . I f u r t h e r agree t h a t permission f o r

mu l t i p l e copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be gran ted

by me or the Dean of Graduate S tudies . It is understood t h a t copying

or pub l i ca t i on of t h i s t h e s i s f o r f i n a n c i a l ga in s h a l l no t be allowed

without my w r i t t e n permission.

T i t l e of Thesis I ~ i s s e r t a t i o n :

Author :

( s igna tu re )

(name )

( d a t e )

ABSTRACT

This thesis examines the lexical redundancy rules for

derivationally related words, and draws two theoretically

relevant conclusions: that subcategorizational correspondences

must be captured by lexical rules stated in terms of thematic

functions, and that the rules for semantic and morphological

redundancies are independent.

Subcategorizational correspondences in sentences and noun

phrases are currently captured (by the X-bar theory of phrase

structure) by generalizing the grammatical relations of the S to

the NP. This strategy is shown to fail for deadjectival

nominals, and to be inappropriate for deverbal nominals. In

Chapter Two, an apparently irregular subcategorizational pattern

is shown to arise for a semantically coherent class of verbs:

causative verbs whose objects are "Experiencers. " Crucial evidence comes from verbs which take both Experiencer and

non-Experiencer objects, which are shown to have two patterns of

nominalization.

In light of these data, the role of thematic functions in

lexical rules is examined in Chapter Three. A distinction

between "majorw and ltminorW lexical rules is retained, and two

possible accounts of minor rules are compared: one where both

grammatical and thematic information are accessed, another where

only thematic information is accessed. The latter hypothesis is

C

iii

argued to be superior.

In Chapter Four, data are presented showing the

cross-classification of morphological and semantic relationships

between words, and this is shown to be problematic for the

theory of word based morphology advocated by Aronoff. A model of

the lexicon is developed which expresses the

cross-classification.

for

Andu and Anna

ACKNOWLEDGEMENTS

I wish t o thank t h e Canadian Commonwealth S c h o l a r s h i p and

F e l l o w s h i p Committee f o r g i v i n g m e t h e o p p o r t u n i t y t o t r a v e l t o

Canada and s t u d y a t t h i s u n i v e r s i t y , and t h e C e n t r a l I n s t i t u t e

o f E n g l i s h and F o r e i g n Languages, Hyderabad, I n d i a , f o r g r a n t i n g

m e l e a v e f o r t h e d u r a t i o n of my s t u d i e s .

To t h e members of my s u p e r v i s o r y committee, P r o f e s s o r s

R ichard DeArmond and B r i a n Newton, I am i n d e b t e d f o r t h e i r

gu idance , t h e i r p a t i e n c e and encouragement , and t h e i r k indness .

My s t u d y of t h e l e x i c o n was i n i t i a t e d by J o e l Hust ; my

t h a n k s t o him.

I a m g r a t e f u l t o P r o f . Wasow f o r h i s comments on a n e a r l i e r

v e r s i o n of some of t h e material p r e s e n t e d i n t h i s t h e s i s .

My t h a n k s a l s o t o my f r i e n d s a t SFU and e l s e w h e r e : ~ e r m i n ,

R i t a , Margarita, Terence , M a l i l l o , Mohan and T a r a , f o r s h a r i n g

w i t h me t h e e x c i t e m e n t o f new i d e a s , and f o r t h e i r l o v e and

f r i e n d s h i p .

TABLE OF CO-8 . . Approva 1 .............................................................. i i

.. . Abstract ............................................................. 111 Dedication ............................................................. v Acknowledgements ...................................................... v i

CHAPTER ONE INTRODUCTION .......................1.................*............ 1

1 . Overview ........................................... 1 2 . L e x i c a l Redundancies i n Aspects and Remarks ........ 6

3 . The wLoca lneasw of L e x i c a l Ru les .................. 16

FOOTNOTES TO CHAPTER ONE .............................. 5

CHAPTER TWO THE DESCRIPTIVE INADEQUACY OF GRAMMATICAL RELATIONS ..................................................... 30 1.An Al leged P a r a l l e l i s m ............................. 30 2 . The S u b j e c t l e s s n e s s o f P r e d i c a t e A d J e c t i v e s ....... 32

3 . An " I r r e g u l a r w Class of Deverbal Norninals ......... 49

FOOTNOTES TO CHAPTER TWO ......... .................... 64

CHAPTER THREE THE EXPLANATORY FAILURE OF GRAMMATICAL FUNCTIONS ...................................................... 72

1 . Two Hypotheses .................................... 72

2 . Major and Minor L e x i c a l Ru les ..................... 75

3 . A Cross-Morphological R e g u l a r i t y .................. 83 .................... 4 . The Themat ic -Syn tac t i c Mapping 99

P o s t s c r i p t : Thematic F u n c t i o n s ...................... 110

FOOTNOTES TO CHAPTER THREE ............ .............,12 2

v i i

CHAPTER FOUR MORPHOLOGICAL AND SEMANTIC ASPECTS OF WORD

FORMATION

1 . Preliminaries .................................... 135 2 . The Cross Classification of Morphology and

Semantics ........................................ 137 3 . A Model .......................................... 155 4 . Implications for Productivity and Semantic

Coherence ........................................ 171 5 . Residual Issues .................................. 177 6 . Concluding Remarks ................................ 190

........................... FOOTNOTES TO CHAPTER FOUR 193

BIBLIOGRAPHY ................................................. 206

CHAPTER ONE

INTRODUCTION

1. Overview - In 1970, Chomsky argued in Remarks - on Nominalization for a

substantial reduction in the power of transformations. The

revision of syntactic theory that he proposed was to isolate and

remove from the transformational component all matters

pertaining to derivational morphology, and to concomitantly

enrich the phrase structure and the lexicon, in order to provide

the grammar with alternative mechanisms for capturing

generalizations which were now excluded from the purview of

transformations. The theoretical position outlined in Remarks - on

Nominalization has come to be known as the Lexicalist

Hypothesis, and in the last decade it has had important

consequences for research in two major, hitherto neglected,

areas: the base, and the lexicon. In this thesis we shall be

concerned mainly with the lexicon, in particular, with the

notion Hlexical redundancy rule." I shall propose a hypothesis

regarding the mode of operation of one type of lexical

redundancy rule, and follow up the consequences of my hypothesis

for the theory of word formation. /

Remarks - on Nomina l i za t ion ( h e n c e f o r t h Remarks) c o n s t i t u t e s

t h e f i rs t acknowledgement w i t h i n t h e t h e o r y of t r a n s f o r m a t i o n a l

grammar t h a t t h e i n t e r n a l o r g a n i z a t i o n of t h e l e x i c o n might be

t h e o r e t i c a l l y i n t e r e s t i n g . We may t r a c e t h e e v o l u t i o n o f t h e

l e x i c o n as a l e g i t i m a t e f i e l d o f i n q u i r y from S y n t a c t i c

S t r u c t u r e s (Chomsky 1957), where t h e l e x i c o n i s e n t i r e l y a b s e n t

( l e x i c a l items b e i n g i n t r o d u c e d by c a t e g o r i a l r u l e s o f t h e

base), through Aspects of t h e Theory of Syntax , where i t i s -- - t a k e n t o be "simply a n unordered l i s t of a l l l e x i c a l f o r m a t i v e s m

(Chomsky 1965:84) t o g e t h e r w i t h t h e i r " e s s e n t i a l l y

l d i o a y n c r a t i c w p h o n o l o g i c a l , s y n t a c t i c and s e m a n t i c f e a t u r e s , t o

t h e l e x i c o n of Remarks, w i t h i t s p r o v i s i o n f o r t h e 9 9 n e u t r a l w

l e x i c a l e n t r y , u n s p e c i f i e d as t o s y n t a c t i c c a t e g o r y , w i t h i n

which are o r g a n i z e d words o f t h e same morpho log ica l f l family" and

t h e f e a t u r e s s h a r e d by them. I l l u s t r a t i n g how such a l e x i c a l

e n t r y can be g i v e n f o r verb-noun p a i r s l i k e r e f u s e - r e f u s a l ,

d e s t r o y - d e s t r u c t i o n , Chomsky s u g g e s t s t h a t t h i s is t h e g e n e r a l

s i t u a t i o n w i t h i n t h e l e x i c o n :

We can e n t e r r e f u s e i n t h e l e x i c o n a s an i t e m w i t h c e r t a i n f i x e d s e l e c t i o n a l and s t r i c t s u b c a t e g o r i z a t i o n f e a t u r e s , which i s free w i t h r e s p e c t t o t h e c a t e g o r i a l

- f e a t u r e s [noun] and [verb] . F a i r l y i d i o s y n c r a t i c morpho log ica l r u l e s w i l l de te rmine t h e p h o n o l o g i c a l form o f r e f u s e , d e s t r o y , e t c . , when t h e s e items a p p e a r i n t h e noun p o s i t i o n . The f a c t t h a t r e f u s e t a k e s a noun p h r a s e complement o r a reduced s e n t e n t i a l complement and d e s t r o y on ly a noun p h r a s e complement, e i t h e r a a a noun o r as a v e r b , i s e x p r e s s e d by t h e f e a t u r e s t r u c t u r e of t h e n n e u t r a l w l e x i c a l e n t r y , as a r e s e l e c t i o n a l p r o p e r t i e s . . . .

Let us propose, then, as a tentative hypothesis, that a great many items appear in the lexicon with fixed selectional and strict subcategorization features, but with a choice as to the features associated with the lexical categories noun, verb, adJective. (Chomsky 1970:lgO)

With the evolution of the lexicon from a mere list

(presumably of morphologically primitive words only) into a more

extended component with internal structure, questions arise as

to how much and what kind of information there is in the

lexicon, what information is idiosyncratic and what predictable,

and how the predictability is to be encoded. The "neutralv

lexical entry was proposed by Chomsky as one such device for

encoding predictability. In addition, Chomsky (1970) suggested

that regularities in the lexicon be expressed by lexical

redundancy rules.

The notion wlexical redundancy rulew is not unique to

Remarks; it had been introduced earlier, in Aspects -- of the

Theory - of Syntax (henceforth Aspects). However, with each

modification in the conception of the lexicon and its contents,

different types of redundancies are uncovered that are

apparently lexical, and the roles envisaged for lexical

redundancy rules change and multiply. We shall see that at

present the notion of a lexical redundancy is invoked for a

variety of phenomena, and the types of lexical rules we can now

identify are rather different in form and function from the

lexical redundancy rules originally proposed in Aspects. The

situation of the lexical rule today is analogous to that of the

transformational rule in the early, pre-lexicalist, phases of

generative grammar. Since the theory at that time provided

little alternative to the transformational rule for

characterizing the many kinds of distributional regularities

observed in a language, this one rule-type was employed to

capture diverse kinds of generalizations. What Chomsky attempted

in Remarks was to separate out from a number of allegedly

transformational phenomena those that could legitimately be so

characterized. This precise articulation of the domain of

transformations was only the first step towards arriving at a

narrower and more restrictive definition of transformational

rules; subsequent research has postulated many other conditions

on the form and function of these rules -- compare, for example, the typology of transformations presented in Emonds (1970,

1976), and the proposals of Chomsky in Conditions - on

Transformations (1973) and subsequent work.

Much of the ground that the transformational rule has lost

has been gained by the lexical redundancy rule. The principal

outcome of Remarks was to transfer to the domain of the lexical

rule from that of the transformational rule the regularities

exhibited by derivationally related words. The framework

provided by Remarks and the typology of transformations proposed

by Emonds also paved the way for a further erosion of the domain

of transformations, widely proposed in the current literature:

the elimination of all structure-preserving transformations in

favor of lexical redundancy rules. While this negative

characterization of the domain of the lexical rule is not

perhaps entirely justified, it does serve to indicate the rapid

and successive extensions in the functions accorded to these

rules, and to point out the need for any discussion of lexical

redundancies to clarify its assumptions.

We may distinguish three stages in the evolution of the

lexical rule: the Aspects stage, the Remarks stage, and the

post-Remarks stage. My object will be to show that these three

evolutionary stages correspond to three distinct rule-types (or

sub-types). My main concern will be with the Remarks-type

lexical rule, and the hypothesis I present applies to rules of

this type. I shall argue that the Remarks-type rule differs from

the Aspects-type rule in crucial respects. I hope to demonstrate

that the distinction between the Remarks-type and the

post-Remarks type (1.e. the kind of lexical rule postulated by

Bresnan 1978) can be motivated on grounds other than theoretical

conservatism. For this, I shall draw support from the insights

of Wasow (1977, 1980).

2. L e x i c a l Redundancies i n Aspects and Remarks - - - I n Aspec t s , Chomsky (1965:167) g i v e s t h e l e x i c a l redundancy

r u l e (I), and i t s i n t e r p r e t a t i o n :

(1) ( = h i s ( 2 6 ) ) [ t - NP Manner] ---> [t-NP]

t o be I n t e r p r e t e d i n t h e f o l l o w i n g manner: i f ( D , C ) i s a l e x i c a l e n t r y w i t h d i s t i n c t i v e f e a t u r e m a t r i x D and complex symbol C c o n t a i n i n g [t - NP Manner], t h e n C i s r e p l a c e d by C 1 , which c o n t a i n s each s p e c i f i e d f e a t u r e [@F] o f C , where F # [ t - NP], and a l s o t h e s p e c i f i e d f e a t u r e [t - NP] .

For example, t h e v e r b - h i t might be e n t e r e d i n t h e l e x i c o n

as ( 2 ) :

(2)(=Chomsky1s ( 2 8 1 1 ) ) ( h i t , - [t - NP Manner, . . .I) By r u l e (I), supplemented by a g e n e r a l conven t ion (op. c i t . : 165)

which n e g a t i v e l y s p e c i f i e s t h e e n t r y f o r a l l f e a t u r e s i t is n o t

p o s i t i v e l y s p e c i f i e d f o r , w e d e r i v e a l e x i c a l e n t r y ( 3 ) f o r - h i t :

( 3 ) (=Chomskyls ( 2 9 1 1 ) ) ( h i t , - [t - NP Manner, t - NP,

- Manner, - - -3 . I >

That is , - h i t must a lways a p p e a r w i t h a d i r e c t o b j e c t , b u t i t can

a p p e a r w i t h o r w i t h o u t a manner a d v e r b i a l .

The l e x i c a l redundancy r u l e (1) i s e s s e n t i a l l y t h e

s y n t a c t i c ana logue of redundancy r u l e s i n phonology. The l e x i c a l

e n t r y f o r a l e x i c a l i t e m i s minimal ly s p e c i f i e d f o r i t s r e l e v a n t

f e a t u r e s ; on t h e b a s i s o f t h i s minimal f e a t u r e s p e c i f i c a t i o n ,

t h e r u l e f i l l s i n p r e d i c t a b l e f e a t u r e s i n t h i s l e x i c a l e n t r y ,

f o r t h e same l e x i c a l item.1 - The redundancy r u l e above i s concerned w i t h c o n t e x t u a l

f e a t u r e s , and i t i s hypo thes ized t o be l a n g u a g e - p a r t i c u l a r

(Chomsky 1965:168) . There i s one o t h e r t y p e o f word- In te rna l

redundancy c o n s i d e r e d i n Aspects : t h i s i n v o l v e s i n t r i n s i c

f e a t u r e s t h a t a r e f i l l e d i n by u n i v e r s a l redundancy conven t ions .

(Note t h a t t h e s e are n o t redundancy r u l e s . ) Where a l e x i c a l I tem

c o n t a i n s i n h e r e n t f e a t u r e s t h a t are h i e r a r c h i c r a t h e r t h a n

c r o s s - c l a s s i f y i n g , f e a t u r e s h i g h e r up on t h e h i e r a r c h y can be

p r e d i c t e d , g i v e n f e a t u r e s lower down. Chomsky t h e r e f o r e s u g g e s t s

(1965:165):

L e t u s s a y t h a t t h e sequence of s p e c i f i e d f e a t u r e s ([@I F l ] , . . . ,[o(n Fn]) (d l= t o r - ) i s a h i e r a r c h i c se uence w i t h r e s e c t t o t h e &rammar G i rP- 1 P i 1 i s t h e b c - d e a t u r e - ? h K t l y d o m i n ~ t i f i ~ ' ~ ~ i + l F i t l ] , f o r e a c h i < n , i n G. . . . Where such r e l a t i o n s h i p s o b t a i n , w e can u t r l i z e them t o s i m p l i f y l e x i c a l e n t r i e s

H e s u g g e s t s t h e f o l l o w i n g conven t ion ( i b i d ) :

Suppose t h a t ( [ d 1 F1 1 , . . . , [o(n Fn] ) is a maximal h i e r a r c h i c sequence w i t h r e s p e c t t o t h e grammar G , and t h a t ( D , C ) i s a l e x i c a l e n t r y of G , where C c o n t a i n s [dn Pn]. Then C i s extended a u t o m a t i c a l l y t o C 1 c o n t a i n i n g C a l o n g w i t h a l l o f t h e s p e c i f i e d f e a t u r e s [dl F i ] , f o r each 1, 1 - <i <n.

Thus g i v e n t h e s u b c a t e g o r i z a t i o n r u l e s (4) (Chomsky 1965 :82) ,

w e can s i m p l i f y t h e l e x i c a l e n t r y f o r boy from ( 5 ) t o (61, t h e

f e a t u r e s [+N] and [+Animate] b e i n g p r e d i c t a b l e (Chomsky

1965:166).

( 5 ) (boy, [ t N , +Common, +Animate, +Human, +Count . . -1)

I n g e n e r a l , t h e n , t h e l e x i c a l r edundanc ies d i s c u s s e d i n Aspects

a r e concerned w i t h t h e f e a t u r e m a t r i x o f a s i n g l e l e x i c a l i tem.

There a r e , however, i n d i c a t i o n s ( a l r e a d y ) i n Aspects t h a t

l e x i c a l s t r u c t u r e s and l e x i c a l r edundanc ies may t u r n o u t t o be

more complex. Thus Chomsky writes:

If we r e g a r d a l e x i c a l e n t r y as a s e t of f e a t u r e s , t h e n items t h a t a r e similar i n sound, meaning o r s y n t a c t i c f u n c t i o n w i l l n o t be r e l a t e d t o one a n o t h e r i n t h e l e x i c o n . For example, t h e I n t r a n s i t i v e "groww o f " t h e boy grewn o r "corn grows, and t h e T r a n s i t i v e ngrowfl o f "he grows corn1' would have t o c o n s t i t u t e two s e p a r a t e l e x i c a l e n t r i e s , d e s p i t e t h e meaning r e l a t i o n t h a t h o l d s between them . . . The same would be t r u e o f "dropm i n " t h e p r i c e dropped," "he dropped t h e b a l l , " "he dropped t h a t s i l l y p r e t e n s e " ; o r of wcomrnandN i n t h e example d i s c u s s e d on p.119, and i n innumerable o t h e r c a s e s o f many d i f f e r e n t k inds . A l t e r n a t i v e l y , such r e l a t i o n s h i p s can be e x p r e s s e d by t a k i n g a l e x i c a l e n t r y t o be a Boolean f u n c t i o n of f e a t u r e s . Although i t i s l i k e l y t h a t such a m o d i f i c a t i o n o f t h e t h e o r y of l e x i c a l s t r u c t u r e i s n e c e s s a r y , i t raises many problems o f f a c t and p r i n c i p l e t o which I have no answer, and I t h e r e f o r e c o n t i n u e t h e e x p o s i t i o n wi thou t d e v e l o p i n g it. (Chomsky 1965:214, f n . 1 5 )

The c a s e of command t h a t Chomsky refers t o above p o s e s t h e

problem o f c a p t u r i n g a n a p p a r e n t dependency between t h e

s e l e c t i o n a l f e a t u r e s o f t h e s u b j e c t and o b j e c t o f t h i s word.

T h i s dependency i s i l l u s t r a t e d below:

( 7 ) a . H e commanded t h e p l a t o o n .

b. *His d e c i s i o n t o r e s i g n h i s commission commanded t h e

p l a t o o n .

(8 ) a. H e commanded o u r r e s p e c t .

b. H i s d e c i s i o n t o r e s i g n h i s commission commanded o u r

r e s p e c t .

Chomsky n o t e s t h a t command when i t t a k e s a n a b s t r a c t o b j e c t (as

i n ( 8 ) ) h a s a " d i f f e r e n t though n o t t o t a l l y u n r e l a t e d s e n s e w

from command i n ( 7 ) ( 1 9 6 5 : l l g ) . He t e n t a t i v e l y s u g g e s t s t h a t t h e

s e l e c t i o n a l f e a t u r e s f o r command s p e c i f y t h e dependency between

n o n - a b s t r a c t o b j e c t and non-abs t rac t s u b j e c t . Thus command might

be g i v e n t h e two f e a t u r e s [[+Animate] Aux - D e t [+Animate]]

and [ [ + A b s t r a c t ] Aux - D e t ( + A b s t r a c t ] ] , b u t n o t t h e f e a t u r e

[ [ + A b s t r a c t ] Aux - D e t [+Animate]]. However, Chomsky

acknowledges t h a t

t h e grounds f o r t h i s d e c i s i o n a r e ve ry weak, s i n c e a c r u c i a l q u e s t i o n -- namely, how t o e n t e r l e x i c a l items w i t h a range o f d i s t i n c t b u t related s y n t a c t i c and semant ic f e a t u r e s -- i s f a r from s e t t l e d . (1965: l lg-120) .

Within t h e framework of Remarks, where in t r a n s f o r m a t i o n s

were p r o h i b i t e d from per fo rming d e r i v a t i o n a l morphology, t h i s

q u e s t i o n cou ld no l o n g e r be l e f t u n s e t t l e d . I t became n e c e s s a r y

t o c a p t u r e r e g u l a r i t i e s i n t h e f e a t u r e m a t r i c e s of

morpho log ica l ly r e l a t e d words, each of which was p r e s e n t i n t h e

l e x i c o n i n i ts own r i g h t . Thus w e p o i n t e d o u t i n t h e p r e v i o u s

s e c t i o n t h a t t h e " n e u t r a l n l e x i c a l e n t r y f o r d e r i v a t i o n a l l y

r e l a t e d words i s a t t h e c o r e of t h e Remarks l e x i c o n . A l e x i c a l

e n t r y is no l o n g e r s imply a c o n j u n c t i o n o f t h e f e a t u r e s f o r a

s i n g l e l e x i c a l i tem; i t i n c o r p o r a t e s d i s j u n c t i v e c h o i c e s o f

f e a t u r e s , w i t h each set o f c h o i c e s l e a d i n g t o a " s e p a r a t e b u t

r e l a t e d n l e x i c a l item. Hust (1978) i l l u s t r a t e s s c h e m a t i c a l l y t h e

s t r u c t u r e o f such a l e x i c a l e n t r y . (L - refers t o t l e x i c a l

e n t r y 1 . )

Phono log ica l , s y n t a c t i c and semant ic f e a t u r e s

common t o a l l branches of L

Other f e a t u r e s s p e c i f i e d Other f e a t u r e s s p e c i f i e d f o r L , i f L i s s p e c i f i e d f o r L, i f L is s p e c i f i e d w i t h t h e f e a t u r e - f w i t h t h e f e a t u r e g

He s ta tes : " t h e f e a t u r e complexes which r e s u l t a t t h e t e r m i n a l

nodes i n such a b ranch ing diagram a r e t h e words ( l e x i c a l i t e m s )

of t h e l anguage , s u b j e c t t o l e x i c a l i n s e r t i o n . "

Such a m o d i f i c a t i o n o f l e x i c a l s t r u c t u r e a l l o w s us n o t

mere ly t o e x p r e s s r e g u l a r i t i e s between m o r p h o l o g i c a l l y r e l a t e d

words, such as r e a d and r e a d a b l e ; i t a l s o o f f e r s a method o f

d e a l i n g w i t h words t h a t are a p p a r e n t l y t h e nsamell word, but

which have a range of r e l a t e d meanings and s y n t a c t i c p r o p e r t i e s ,

such as d r o p and command i n t h e examples g i v e n by Chomsky. We

can now s a y t h a t command i n ( 7 ) i s a s e p a r a t e l e x i c a l item from

command i n ( 8 ) , but r e l a t e d t o i t . Such a n a n a l y s i s o f command

i s mot iva ted i n d e p e n d e n t l y of semant ic c o n s i d e r a t i o n s i n Hust

(1978) . R e c a l l t h a t i f command i s t o be t r e a t e d as a s i n g l e

l e x i c a l item, i t i s n e c e s s a r y f o r s e l e c t i o n a l f e a t u r e s t o encode

s u b j e c t and o b j e c t dependencies . Hust shows t h a t t h e i n c l u s i o n

of such dependenc ies l e a d s t o a n e x p o n e n t i a l i n c r e a s e i n t h e

number o f s e l e c t i o n a l f e a t u r e s i n t h e grammar.2 H e proposes

i n s t e a d t h e f o l l o w i n g ( p a r t i a l l y s p e c i f i e d ) l e x i c a l e n t r y f o r

command, where each t e r m i n a l node r e p r e s e n t s a l e x i c a l i t e m , and

t h e f e a t u r e s s p e c i f i e d f o r a node " p r e c i p i t a t e w o n t o t h e nodes

dominated by it.

command

t NP

-[-human] - (by precipitation)

+ NP - + NP -

This entry shows that when command takes an object like platoon,

it must have a human subject; but when It takes an abstract

object, it is free with respect to subject selection. Hust

points out that "the fact that differences in the meaning of

command correlate with differences in the syntactic feature . constellation lends support to the analysis proposed here,

since, in general, each terminal node in a branching lexical

entry will have certain idiosyncratic features peculiar to just

that lexical item.11 In this and other cases, Hust stresses that

the structure of his lexical entries is not motivated by

semantic conalderations, but that semantic facts follow

neveptheless from his analysis.

We shall in the course of our investigation encounter many

such cases, where non-semantic considerations motivate a

differentiation of homonymous lexical items, and semantic facts

follow from thia differentiation. The differentiation of

homonymous l e x i c a l items i n f a c t t u r n s o u t to b e c r u c i a l f o r t h e

a n a l y s i s of l e x i c a l r edundanc ies , 1.e. f o r t h e e x t r a c t i o n of

r e g u l a r i t i e s i n t h e s u b c a t e g o r i z a t i o n a l frames of

morpho log ica l ly r e l a t e d words. I w i l l show, f o r example, t h a t

t h e v e r b d e p r e s s i n i t s p h y s i c a l s e n s e canno t be t r e a t e d a s t h e

same word a s d e p r e s s i n t h e emot iona l s e n s e . S i m i l a r l y , I w i l l

show t h a t a noun l i k e amusement, i n t h e s e n s e o f a n emot iona l

s t a t e , must be a d i f f e r e n t word f rom t h e noun ( a n ) amusement, i n - t h e s e n s e o f something which can t r igger t h a t emot iona l s t a t e .

Such d i s t i n c t i o n s must be main ta ined r e g a r d l e s s of whether w e

adop t t h e f t n e u t r a l w l e x i c a l e n t r y , o r t h e f u l l y s p e c i f i e d

l e x i c a l e n t r y w i t h t t c o s t l e s s w i n f o r m a t i o n ( s u g g e s t e d by

Jackendoff 1975) . -

The complex i ty of l e x i c a l s t r u c t u r e r e v e a l e d by such c a s e s

h a s a b e a r i n g on t h e t h e o r y of word f o r m a t i o n , as we s h a l l see

i n Chap te r Four. The e x i s t e n c e o f more t h a n one word d r o p o r

d e p r e s s r e v e a l s , moreover, a d i f f e r e n t f a c e t o f t h e n o t i o n

" s e p a r a t e b u t r e l a t e d l e x i c a l item." The r e l a t i o n between d r o p

on a p h y s i c a l s c a l e and d r o p on a n a b s t r a c t s c a l e o f v a l u e s

( such as p r i c e ) is , f o r example, an i n t u i t i v e l y r e g u l a r one

which i s i n s t a n t i a t e d i n many o t h e r c a s e s . However, o u r

i n v e s t i g a t i o n o f t h e l e x i c a l r e l a t i o n s h i p must h e r e base i t s e l f

n o t on r e g u l a r i t i e s i n t h e s u b c a t e g o r i z a t i o n a l c o n t e x t s of t h e

r e l a t e d words, bu t on t h e c o n c e p t u a l s t r u c t u r e t h a t p e r m i t s t h e

observed semant ic r e l a t i o n s h i p s . Jackendof f (1972,1976,1978)

provides very interesting analyses of such cross-field

generalizations in terms of thematic functions. The development

of such a theory of lexical relationships might vindicate our

decision to postulate many homonymous lexical items in such

cases, rather than a single lexical item with a range of

interdependent syntactic and semantic properties.

The expansion of the scope of the lexical entry in Remarks

was accompanied by a corresponding enlargement in the scope of

the lexical redundancy rule in this model. Thus Hust (1978 :76)

writes :

Within the framework of the Lexicalist Hypothesis . . ., lexical redundancy rules seem to take on quite a different function . . . In addition to filling in feature specifications for a given lexical item on the - basis of the minimal feature specification of the lexic,al entry, lexical redundancy rules are proposed as a means to account for regularities regarding selectional restrictions, strict subcategorization features, etc., which hold between pairs of derivationally related lexical items.

The lexical redundancy rule thus becomes a tool for analyzing

the relationship between a base word and its morphological

derivative, and the theory of lexical redundancies may now be

viewed as the inverse of a theory of word formation.

I shall show that this type of lexical rule is very

different from the lexical rule of Aspects. However, the

formalism originally proposed by Chomsky (1970) for such rules

makes them appear essentially similar to rule (1). Consider for

example Chomskyls p r o p o s a l f o r t h e l e x i c a l r u l e which c a p t u r e s

t h e p a r a l l e l i s m i n t h e o b J e c t s of v e r b s l i k e read and t h e

s u b j e c t s o f t h e c o r r e s p o n d i n g a d j e c t i v a l p r e d i c a t e s l i k e

r e a d a b l e :

. . . it can be fo rmula ted as a l e x i c a l r u l e t h a t a s s i g n s t h e f e a t u r e [ X ] t o a l e x i c a l item [V-able] where V h a s t h e i n t r i n ~ s e l e c t i o n a l f e a t u r e [ - T. (Chomsky 1970:213)

S t a t e d i n t h i s f a s h i o n , t h e r u l e a p p e a r s t o b e "only a ve ry

minor e x t e n s i o n o f t h e form and f u n c t i o n o f l e x i c a l redundancy

r u l e s proposed i n Aspects" (Hust 1978 :78) ; f o r i t t a k e s as i n p u t

one s u b c a t e g o r i z a t i o n f e a t u r e , and o u t p u t s a n o t h e r

s u b c a t e g o r i z a t i o n f e a t u r e . Moreover, i f r e l a t e d l e x i c a l items

a r e p a r t o f a s i n g l e , " n e u t r a l n l e x i c a l e n t r y , t h e Remarks t y p e

r u l e s cou ld s t i l l be c o n s i d e r e d a s r u l e s which f i l l i n

p r e d i c t a b l e f e a t u r e s w i t h i n a s i n g l e l e x i c a l e n t r y .

One of my main arguments i n t h i s t h e s i s (however) w i l l b e

t h a t t h e Remarks t y p e l e x i c a l r u l e shou ld n o t be f o r m u l a t e d a s a

mapping between s u b c a t e g o r i z a t i o n a l f e a t u r e s as such , bu t as a

mapping between argument s t r u c t u r e s .

3. The wlocalness~ of Lexical Rules. -- - The redundant relation between lexical items that the

lexical rules in Remarks try to capture concerns the

regularities in their subcategorizational frames. Thus the fact

that the object of transitive grow is selectionally parallel to

the subject of intransitive grow, or that the object of - read is

selectionally parallel to the subject of - be readable, is of

interest to this type of rule. The obvious way to capture these

relations would seem to be to postulate a rule which operates on

the subject, obJect etc. of an input word, to assign its

features to the subject, object etc. of a derived word. This is

the natural interpretation of the rule L- r X] ---> [ X ] for

verbs and their -able adjectival derivatives in Chomsky

(1970:213); for Chomsky here assumes the framework of Aspects,

wherein selectional restrictions are defined on heads of phrases

linked by grammatical relations. His rule assigns the

selectional features of the verb's object to the subject of the

corresponding adjectival predicate. It has thus been widely

assumed in current research that the lexical rule of Remarks

operates on grammatical relations; the clearest articulation of

this position is in Wasow (1977).

I shall argue that such a formulation of these lexical

rules is fundamentally in error. I show that a rule which refers

to grammatical relations at all is incapable of explaining, or

predicting, the full range of subcategorizational

correspondences in the frames of derivationally related words. I

propose instead that lexical rules relate the argument

structures of words, specified in terms of thematic functions.

The observed correspondences in subcategorizational frames are

viewed as the outcome of general principles governing the

assignment of thematic structures to syntactic frames, as

suggested by Anderson (1977).

A maJor argument for formulating lexical rules of the

Remarks-type in terms of grammatical relations is the necessity

of characterizing the "localw property of these rules. This

argument is given by Wasow (1977), in the context of an

insightful examination of the differences between lexical rules

and transformations (structure-preserving or otherwise). I quote

below the third of Wasow1s five criteria for distinguishing

lexical redundancy rules, and his Justification of this

criterion.

Lexical Rules Transformations

nlocalH; involve only need not be "localn; NPs bearing formulated in terms of grammatical relations structural properties to items in question of phrase markers

. . . lexical rules must be more "localw than transformations. Whereas transformations are mappings between entire phrase markers, lexical redundancy rules are mappings only between lexical items. Hence, lexical rules ought not to be able to refer to aspects of the environments In which the lexical items appear, other than those aspects that must for independent reasons be included In the lexical entries anyway. . . .

. . . one rather strong hypothesis to put forward is that the only elements of a verb's (footnote omitted) environment that may enter into the statement of lexical redundancy rules are the NPs bearing deep structure grammatical relations to it (viz., its subject, direct object and indirect object). In an intuitive sense that is hard to pin down, these NP1s are the elements of a verbls environment most closely associated with it. There can be little doubt that they must enter into the statement of contextual features; indeed, informal statements of both selectional restrictions and strict subcategorization are typically formulated in terms of these relations . . . Thus, I claim that a natural way of stating the lllocalnessn property of lexical rules is to insist that they be llrelationalll in this sense. I assume that transformations, in contrast, are defined in the usual way, in terms of structural relations of phrase markers (footnote omitted). (Wasow 1977:330)

Although Wasow adopts this as a working hypothesis, he in fact

leaves the question open. In the concluding section of his

paper, he acknowledges the possibility of llsome other way of

characterizing the 'localness* property of lexical rulesf1

(1977:354) (possibly in terms of thematic functions), stressing

only that tlCriterion 3 or some other llocalnessl condition is

needed. l1

Let us briefly consider the evidence Wasow provides that

some l*localnessw condition is necessary. Consider the rule which

relates transitive and intransitive verbs:

(10) a. (=his (5a)) John showed hostility.

b. (=his (5b)) Hostility showed.

This rule must relate the object of the transitive verb to the

subject of the intransitive verb. The number of idiosyncratic

exceptions to the rule argue that it is lexical, not

transformational. Thus drop but not lower, shatter but not

demolish, - move but not transport, exhibit this alternation.

The fact to be explained is that even a verb which normally

exhibits this alternation fails to do so in certain

constructions. Thus show when it occurs In the -9

accusative-infinitive ("raised to objectw) construction, has no

Intransitive counterpart.

(11) a. (=his (5c)) John showed hostility to be a result of

cold weather.

b. (=his (7b))*Hostility showed to be a result of cold

weather.

Similarly, drop does not exhibit the causative alternation in

the ndouble objectw construction (13), although it does so

otherwise ((12)).

(12) a. (=his (8a)) They dropped the rope 100 feet.

b. (=his (8b)) The rope dropped 100 feet.

(13) a. (=his ( 8 ~ ) ) They dropped John the rope.

b. (=his (lob)) John dropped the rope. (#a)

Thus the transitive-intransitive rule does not simply relate any

immediately post-verbal NP to the subject. It is in addition

sensitive to some information in the verb's lexical entry: It Is

l1 local. l1

Wasow gives examples of two other lexical rules which

relate objects to subjects: the -able rule, and the adjectival

passive rule. Once again, these rules fail to apply in

constructions like the accusative-infinitive and the double

object. Thus there is a consistent contrast in the acceptability

of the verbal passive, which (Wasow hypothesizes) is a

transformational rule, and the adjectival passive, which is a

lexical rule, when the passive applies to these two

constructions. This contrast is illustrated below. The tests for

an adjectival passive are prefixability with - un-, and

cooccurrence with verbs like seems, sounds, etc.

(14) a. (=his (62b)) John is known to be a communist.

b. (=his (62c)) *John is unknown to be a communist.

(15) a. (=his (56b)) Bill was told (the story).

b. (=his (56d)) *Bill was untold (the story).

Consider also (16), which shows that the adjectival passive

cannot prepose a chunk of an idiom in immediately post-verbal

position, although the verbal passive can.

(16) a. (=his (64a)) Advantage is easily taken of John.

b. (=his (64a)) *Advantage sounds easily taken of John.

What is it about these constructions that prohibits the

lexical rules from applying to them? Wasow points out that in

all these constructions, the immediately post-verbal NP is not

the verb's direct object. In the accusative-infinitive

construction, the post-verbal NP is arguably the subject of the

complement clause. In the double object construction, the

indirect object of the verb precedes its direct object. For

idiomatic verbs like - take advantage -9 of a structure like

[[take] [advantage] of] is plausible, in which case the direct V N V

object would be not the idiom chunk advantage, but the NP

following the entire idiom - take advantage - of.

We can now account for the above facts by hypothesizing the

lexical rules to be llrelationalll; they relate only direct

objects to subjects. Even if some other NP occupies the normal

direct object (immediately post-verbal) position, they cannot

relate this NP to the subject.3

But an alternative explanation of these facts in terms of

thematic functions is suggested by Anderson (1977) . We know that

the lexical entry of a verb contains a semantic representation

in terms of thematic functions (Jackendoff 1972); it also

contains information correlating thematic functions with

syntactic positions. Suppose now that the lexical rules we

considered were sensitive to the notion Theme. Then they would

apply to direct objects, since direct objects are generally

Themes. But they would not apply to the post-verbal NPs in the

other constructions, for

Indirect Objects are Goals (usually); idiom chunks have no thematic status, and NPs from a lower clause bear no thematic relation in a higher one. (Anderson 1977:371)

(Anderson's hypothesis that the rule relates Themes is also

consistent with Wasow1s observation that (13a) and (13b),

although they are both grammatical, must not be related because

"John - in [ (13b)J is the agent, not the goal of drop1' (op. cit.:332). In (12), on the other hand, - the rope would be

identified as the Theme in both sentences, for it is the entity

that moves.)

The lllocalnessv of lexical rules is therefore not a

compelling argument for requiring these rules to be formulated

in terms of grammatical relations. 4 However, given that

grammatical relations and thematic functions can equally well

account for the same range of facts, there is no compelling

evidence (at this point) for deciding between them either. I

shall in the following chapters present some new evidence that

bears on this question.

The obvious problem for any theory of lexical redundancy

rules which appeals to thematic functions is the problem of

adequately defining these functions. However, I believe that

this problem is not insurmountable. Moreover, the definition

problem is not unique to thematic functions; for defining the

appropriate range of grammatical relations for lexical rules

t u r n s o u t t o be s i m i l a r l y p r o b l e m a t i c . Chomsky (1965) p r e s e n t e d

a s t r i c t d e f i n i t i o n of " ~ u b j e c t ' ~ and l l o b j e c t " i n te rms of

s t r u c t u r a l c o n f i g u r a t i o n s , b u t ( a s Wasow n o t e s ) i t i s n o t

obv ious t h a t t h e n o t i o n " i n d i r e c t o b j e c t " can be s i m i l a r l y

d e f i n e d . (Wasow1s 1977 model i m p l i c i t l y r e j e c t s Chornskyls

d e f i n i t i o n s : compare h i s h y p o t h e s i s t h a t h e l p and thank t a k e

on ly i n d i r e c t o b j e c t s . ) F u r t h e r , if l e x i c a l r u l e s a r e ' t o be

f o r m u l a t e d i n te rms of grammat ica l r e l a t i o n s , i t becomes

n e c e s s a r y t o d e f i n e t h e s e r e l a t i o n s f o r c a t e g o r i e s o t h e r than

ve rbs . Thus Wasow w r i t e s :

My C r i t e r i o n 3 r e q u i r e s t h a t NPs mentioned by l e x i c a l r u l e s b e a r grammat ica l r e l a t i o n s t o t h e i t e m s undergoing t h e r u l e s . T h i s r e q u i r e s e i t h e r t h a t a l l l e x i c a l r u l e s i n v o l v e v e r b s , o r t h a t grammat ica l r e l a t i o n s be d e f i n e d f o r i t ems of o t h e r c a t e g o r i e s . The fo rmer a l t e r n a t i v e i s probably u n t e n a b l e ( s i n c e , e .g . , t h e r e a r e d e - a d j e c t i v a l nouns ) , and t h e l a t t e r a l t e r n a t i v e f u r t h e r c o m p l i c a t e s t h e a l r e a d y d i f f i c u l t problem of d e f i n i n g grammat ica l r e l a t i o n s . (Wasow 1977:354)

The o n l y e x p l i c i t p r o p o s a l s t h a t have s o fa r been made f o r

e x t e n d i n g grammat ica l r e l a t i o n s t o c a t e g o r i e s o t h e r t h a n v e r b s

a r e t h o s e o f Jackendof f (1974a, 1 9 7 7 ) , working w i t h i n t h e X-bar

t h e o r y of p h r a s e s t r u c t u r e ; and i t i s w i t h a c o n s i d e r a t i o n of

t h e problems f a c e d by t h i s a t t e m p t t h a t we w i l l b e g i n o u r

i n v e s t i g a t i o n . The l e x i c a l r u l e s we w i l l examine a r e t h e r u l e s

f o r t h e n o r n i n a l i z a t i o n s of v e r b s and a d j e c t i v e s . The o p e r a t i o n s

performed by t h e s e r u l e s have s o fa r been assumed t o be ve ry

s i m p l e : they t a k e a s i n p u t t h e v e r b ' s o r t h e a d j e c t i v e ' s s u b j e c t

o r o b j e c t , and a s s i g n t h e s e same NPs t o t h e noun, a s i t s

l l sub jec t l l o r l lobjec t l l . The d i s c u s s i o n h a s t h u s c e n t e r e d n o t s o

much on t h e r u l e s themse lves , a s on how t o c a p t u r e t h e

s u b c a t e g o r i z a t i o n a l l l p a r a l l e l i s m u which a p p a r e n t l y r e s u l t s

between t h e S and t h e NP. I show t h a t t h e r e i s no such g e n e r a l

p a r a l l e l i s m i n t h e S and t h e NP, and t h a t t h e n o m i n a l i z a t i o n

d a t a are much more complex.

FOOTNOTES - TO CHAPTER ONE

It may be questioned whether a rule like (1) is

necessary, since the information it provides could equivalently

be given by a contextual feature with an optional element, e.g.

c +- N~(~anner)]. However, there are problems with using

optional elements in contextual features, in the Aspects model.

This is because

a set of frames in which the symbol A occurs imposes a corresponding subclassification on A, with one subclassification corresponding to e a c h s t e d context. (Chomsky 1965:94, emphasis added)-

In Aspects, transitivity is treated as a contextual feature.

Thus given Chomskyls lexical entry (loc. cit. ) for grow,

(i) (=his (41)) &row, C+V, + - NP, + - #, + A d j ]

nothing would prevent the abbreviation of the contextual

features for transitive and intransitive grow as [+ - (NP)#], if

parenthesized elements were allowed. But with this abbreviation,

the transitive-intransitive subclassification would be lost.

I have argued elsewhere (Amritavalli 1979) that

transitivity is not a contextual feature, but an intrinsic

feature of the verb. If so, transitive and intransitive grow

would be treated as two lexical items, the contextual features

for which could not be collapsed. Further evidence is available

to support treating transitive and intransitive verbs as

separate lexical items. Thus Hust (1978) notes that intransitive

grow, but not transitive grow, can occur with a quantifier

phrase :

(ii) The corn grew a foot.

(iii) *John grew the corn a foot.

Similarly, DeArmond (1980) observes that we can choose a

locative prepositional phrase for open when it occurs in the

frame - #, but not when it occurs in the frame - NP : (iv) The door opens onto the patio.

(v) *John opens the door onto the patio.

If we can thus distinguish the true optionality of contextual

elements for a word from the case where optionality is a symptom

of a differentiation in lexical items, it becomes possible to

abbreviate contextual features for a given lexical item, and

rules like (1) become unnecessary. We may note that informal

statements of contextual features in the literature have

utilized abbreviatory conventions freely, and few other

redundancy rules like ( 1) have been posited.

It would also result in the definition of a false grammatical

relation "Subject-Object". In the Aspects theory, any two

grammatical functions can potentially define a grammatical

relation. According to Chomsky, what distinguishes the

"irrelevant psuedorelation Subject-Object" from the "legitimate

and traditionally recognized grammatical relation Subject-Verbw

is the irrelevance of Subject-selection to Object-selection:

"the choice of Main-Verb is determined by the choice of Subject

and Object, though Subject and Object are in general chosen

independently of one another and, correspondingly, have no

grammatical relation of the sort in question holding between

themw (Chomsky 1965:73-74).

3 Wasow here makes the assumption that Dative Movement and

Raising to Object are not transformations. If they were, the

inapplicability of lexical rules to indirect objects and

I1raised" objects could be explained without reference to

nlocalness.~ Since lexical rules precede transformations, the

relevant NPs would not be in immediately post-verbal (direct

object) positlon when the lexical rules applied.

Notice however that the wlocalnessll condition is still

needed to account for (16b). A reference to relational

information also appears necessary to rule out (ii) and (iv)

below, where - John is generated in immediately post-verbal

position:

(i) Someone helped John.

(ii)*John seems helpable/(un)helped.

(iii) Someone thanked John.

(iv)*John appears thankable/(un)thanked.

Wasow hypothesizes that help and thank take only indirect

objects. He observes that the cognate German verbs helfen and

danken take dative objects, and that the preposition which

appears in the corresponding nominalization is - to, usually a

marker of indirect objects:

(v) Our help ?to the hostess went unacknowledged.

thanks *of

Anticipating a little, note that Anderson's alternative

explanation in terms of thematic functions is possible here as

well. Thus if the objects of help and thank are analyzed as

Goal, a lexical rule sensitive to the notion Theme would not

apply to them.

4 A weaker argument (in Wasow1s 1977 framework) for utilizing

grammatical relations in lexical rules is that this serves to

differentiate lexical from transformational rules. However,

Anderson points out that it is - a priori doubtful if the locus of differentiation for transformational and lexical rules lies in

the sensitivity of the latter rule-type to grammatical

relations. For although grammatical relations do not directly

figure in the statement of transformations, they do figure in

conditions on the applicability of transformations (so that a

transformation might, for example, apply differentially to

subjects and non-subjects). Anderson concludes that ll(t)he

criterion of sensitivity to grammatical relations (at least to

llSubjectll) is therefore not a sufficient condition for the

assignment of a r u l e t o the category of lexical redundancy

ru les" ( 1977 ~ 3 6 6 ) . (See a l s o DeArmond (1980) ) .

CHAPTER TWO

THE DESCRIPTIVE INADEQUACY OF GRAMMATICAL RELATIONS

1. An Alleged Parallelism - The system of phrase structure known as X t Syntax,l

developed by Jackendoff (1974a, 1977), receives its impetus from

the Lexicalist Hypothesis, which calls for a lexical account of

cross-categorial parallelisms. Since the Lexicalist Hypothesis

was first articulated (in Chomsky 1970) with respect to the

relation between sentences and derived nominals, X f Syntax takes

the expression of subcategorizational parallelisms in Ss and NPs

to be one of its major areas of concern. In particular, there

are in related Ss and NPs regular correspondences in the

syntactic positions on which an identical set of selectional

restrictions must be imposed. X 1 Syntax attempts to account for

such correpondences by generalizing the grammatical relations of

the S to the NP, with a view to generalizing the projection

rules which enforce selectional restrictions on corresponding

positions in the S and the NP. Since grammatical relations are

defined (as in the standard theory) by phrase structure

configurations, subcategorizational parallelism Is ultimately

sought to be expressed in terms of structural parallelism.

In this chapter and the next, I present evidence that the

subcategorizational correspondences in Ss and NPs cannot be

adequately expressed in terms of generalized grammatical

relations, or even as a mapping between (two sets of)

grammatical functions. I argue for an alternative theory of

subcategorizational correspondences, based on (a refined notion

of) thematic functions, and general principles assigning

thematic functions to syntactic positions. In section 2, I

consider one of the two major cases of parallelism that the

generalization of grammatical relations purports to account for

(namely, the case of Ss with adjectival predicates, and

deadjectival derived nominals); and show that

subcategorizational parallelism is not matched by structural

parallelism in this case. Thus an explicit formulation of

projection rules, in the manner envisaged by X t Syntax, reveals

that the rules for adjectives and deadjectival nouns cannot be

generalized by referring to a common set of grammatical

relations.

In later sections, I demonstrate that there is no general

parallelism in grammatical relations in the case of Ss with

verbal predicates and deverbal derived nominals (the second

major case). A group of apparently idiosyncratic exceptions to

the postulated parallelism is examined, and the thematic

function Experiencer is seen to be crucially involved in all

these cases. I argue that the pattern of correspondence in the

subcategorizations of the verbs and deverbal nouns under

consideration is not an irregularity in the language, but an

instance of a regular subcategorizational pattern which obtains

wherever this particular combination of thematic functions

occurs. I show that a constant pattern of correspondences in the

subcategorizational frames of verbs, nouns and adjectives can be

predicted for a constant set of thematic functions, irrespective

of whether it is the verb, the adjective or the noun which is

morphologically 'basic.'

It is seen that the generalizations that can be made about

subcategorizational correspondences in Ss and NPs are not

amenable to an approach which compares only pairs of

subcategorizational frames (at a time). But the lexical rules in

a theory based on grammatical relations (advocated in X' Syntax)

can relate only the subcategorizational frames of a pair of

words such that one is morphologically derived from the other,

and are therefore incapable of capturing the required type of

generalizations.

2. The Subjectlessness of Predicate Adjectives -- - Jackendoff (1974a, 1977), discussing the arguments against

a transformational derivation of a deverbal nominal like John's

criticism -- of the play from a sentence like John criticized the

play, observes that

(o)ne of the stronger arguments for deriving this nominal from a sentence is that the selectional restrictions parallel those of the sentence John - criticized the play; the range of noun phrases that can - occupy possessive position in the derived nominal is identical with the range of possible subjects in the sentence, and the range of noun phrases following of is - identical with the range of direct objects in the sentence. In order to capture this generalization for a wide range of nominals, we do not want to state twice in the grammar the selectional restrictions shared by criticize and criticism; nor do we want to state twice the set of rules which enforce these selectional restrictions, once for NP and once for S. (Jackendoff 1974a:9, 1977:16)

A non-redundant specification of the selectional features of

criticize, criticism (as of other lexically related words) can

be given in a llneutralll lexical entry (cf. Chomsky 1970, Hust

1978; and Jackendoff 1975 for an alternative formalization).

Regarding the non-redundant enforcement of identical selectional

restrictions in the S and the NP, Jackendoff suggests that

along with a generalization of the base rules. . ., there is a corresponding generalization in the projection rules. For example, the notion "subject ofw can, where semantically appropriate, be generalized to possessive NPs, and "object of" can be generalized to postnominal NPs. This generalization ensures that the rules enforcing selectional restrictions on criticize and criticism are in fact the same rules in 3s and NPs, and they need not be stated twice. (Jackendoff 1977:16)

A similar selectional parallelism is seen in the S John is -- perverse, and the (deadjectival) nominal John's perversity. Once

again,

we must account for the generality of selectional

restrictions, e.g. for the fact that the subject in (2.37) (i.e. John's be in precisely the same way that perversity restricts the genitive phrase in (2.38) (1.e. John's perversity, R.A.). (Jackendoff 1977:21)

Once again, the subject-possessive NP parallelism is to be

expressed in terms of the generalized notion "subject of," and a

generalized projection rule referring to this notion:

Since the genitive NP in (2.38) is to be considered the subject of (2.38), the rules which impose selectional restrictions on grammatical subjects will apply equally in (2.37) and (2.38). Thus, as in section (2.1) (1.e. the case of verbs and deverbal nouns, R.A.), the Lexicalist Hypothesis requires a cross-category generalization to be expressed, but this time in terms of adjectives and nouns. (op.cit.:21)

Let us examine how the proposed generalization of grammatical

relations and projection rules works for verbs and deverbal

nouns. Jackendoff first modifies the structural configurations

of S and NP, to ensure a parallelism in the position of the

"subjectw in both.2 An important innovation in the interests of

structural parallelism is the identification of S with V"'.

Compare the configurations (la, lb) with (2a, 2b) below. An

obvious difference between (la) and (lb) is that the subject NP

in (la) is not dominated by the V w whose head V imposes a

selectional restriction on it; whereas in (lb), the possessive

NP is dominated by the N" whose head N selectionally restricts

it:

John p a s t c r i t i c i z e

s u b j e c t s p e c i f i e r head NP of V t t V

1 : I J o h n ' s c r i t i c i s m

Poss-NP head N s p e c i f i e r

of N t t

I n ( 2 a ) , S i s V V t t . The s u b j e c t NP i n ( 2 a ) and t h e p o s s e s s i v e NP

i n ( 2 b ) are b o t h dominated by t h e X t t l c a t e g o r y whose l e x i c a l

head - X ( V I n ( 2 a ) , N i n ( 2 b ) ) s e l e c t i o n a l l y r e s t r i c t s them.

John past briticize

subject head V NP

specif ier of V"'

John s criticism

Poss-NP head N specif ier of N1ll

A generalized definition which picks out the subject of the

sentence and the possessive NP in the nominal can now be given

as [N1ll, X1ll] (the N1ll immediately dominated by an X1ll). In

Jackendoff1s system of phrase structure, nouns and verbs share

the feature [tSubj], and he therefore gives the generalized

definition for "subject off1 as [N"', [tS~bj]"~] (op.cit.:41).

The generalized definition of llobject off1 is similarly given as

[N1ll, [sSU~J]~] (the N1ll immediately dominated by N1 or V1)

(op. cit. ~42).

Jackendof f does n o t f o r m a l i z e t h e p r o j e c t i o n r u l e s f o r

c r i t i c i z e and c r i t i c i s m , t o d e m o n s t r a t e i n what r e s p e c t t h e s e

r u l e s " a r e i n f a c t t h e same r u l e s i n S s and NPs," once t h e

n o t i o n s " s u b j e c t o f " and " o b j e c t o f " a r e g e n e r a l i z e d . But w e may

f o r m a l i z e t h e r u l e f o r t h e i r w s u b j e c t s n a s i n ( 3 ) :

( 3 ) Given a l e x i c a l e n t r y wi th t h e f e a t u r e [ t S u b j ] ( N o r

V) :

s u b s t i t u t e t h e r e a d i n g o f [ N 1 * ' , [ + S u b j l l l l ] ( " i t s

s u b j e c t " ) , where [ t S ~ b j ] ' ~ ' i s t h e p h r a s a l c a t e g o r y

w i t h t h e l e x i c a l e n t r y a s head, f o r argument - x i n t h e

e n t r y ' s f u n c t i o n a l s t r u c t u r e . 3

The p r o j e c t i o n r u l e f o r t h e " o b j e c t s " of c r i t i c i z e and c r i t i c i s m

can be f o r m u l a t e d a l o n g t h e same l i n e s . Thus t h e

s u b c a t e g o r i z a t i o n a l p a r a l l e l i s m s of v e r b s and d e v e r b a l nouns a r e

c a p t u r e d by g e n e r a l i z e d p r o j e c t i o n r u l e s r e f e r r i n g t o t h e i r

n ~ ~ b j e ~ t ~ " and " o b j e c t s . "

L e t us now t r y t o g i v e a similar g e n e r a l i z e d p r o j e c t i o n

r u l e f o r a d j e c t i v e s and d e a d j e c t i v a l nouns, as s u g g e s t e d by

Jackendof f (1977:21) . Consider f i r s t t h e s t r u c t u r e s ( 4 a ) and

( 4 b ) -

P r e s I

p e r v e r s e

s u b j e c t head p r e d i c a t e NP V a d j e c t i v e

John s p e r v e r s i t y

Poss-NP Head N s p e c i f i e r

o f N t f 1

I t i s immedia te ly a p p a r e n t t h a t t h e s t r u c t u r a l r e l a t i o n between

John and p e r v e r s e i n ( 4 a ) i s n o t t h e same a s t h e s t r u c t u r a l

r e l a t i o n between -- J o h n l s and p e r v e r s i t y i n ( 4 b ) . The

n o n - p a r a l l e l i s m between ( 4 a ) and ( 4 b ) i s similar t o t h e

n o n - p a r a l l e l i s m between ( l a ) (where in S was n o t t r e a t e d a s

v f t ' ) , and ( l b ) . I n ( 4 a ) , a s i n ( l a ) , t h e s u b j e c t N ' " i s

o u t s i d e t h e X " ' whose head X imposes a s e l e c t i o n a l r e s t r i c t i o n

on i t . I n ( 4 b ) , a s i n (lb), t h e p o s s e s s i v e N u 1s dominated by

t h e X I 1 ' whose head X imposes a s e l e c t i o n a l r e s t r i c t i o n on i t .

Rule ( 5 ) r e p r e s e n t s my a t t e m p t t o s t a t e t h e p r o j e c t i o n

r u l e s f o r p e r v e r s e and p e r v e r s i t y . L e t u s t a k e nouns t o be

[+Subj , -Obj], a d j e c t i v e s t o be [-Subj, -Obj], and v e r b s t o be

[+Subj , +Obj] ( f o l l o w i n g Jackendof f 1977:32-33). The r u l e has

two s u b p a r t s . Subpar t ( i ) a p p l i e s i n t h e NP, i .e. ( 4 b ) . I t i s

s i m i l a r t o r u l e ( 3 ) . Subpar t (ii) a p p l i e s i n t h e S , i . e . (4a) .

Note t h a t h e r e t h e r u l e must r e f e r t o t h e " s u b j e c t o f u t h e V"',

r a t h e r t h a n o f t h e A " ' domina t ing t h e A ( t h e l e x i c a l e n t r y . )

( 5 ) Given a l e x i c a l e n t r y w i t h t h e f e a t u r e [-Obj] ( N o r A ) :

i. if t h e e n t r y i s a l s o [+Subj] ( i . e . N ) ,

s u b s t i t u t e t h e r e a d i n g of [Nw', [ + S u b J I H t ] ( " i t s

s u b j e c t f 1 ) , where [+Subj] i s t h e p h r a s a l c a t e g o r y

w i t h t h e l e x i c a l e n t r y a s head , f o r argument - x i n t h e

e n t r y ' s f u n c t i o n a l s t r u c t u r e .

il. i f t h e e n t r y i s a l s o [-Subj] ( i .e . A ) ,

s u b s t i t u t e t h e r e a d i n g of [NIf', [ + S u b j I u ' ] , where

[ + S u b j I t 1 ' i s t h e p h r a s a l c a t e g o r y whose head s t r i c t l y

s u b c a t e g o r i z e s t h e X t f l c a t e g o r y of which t h e l e x i c a l

e n t r y i s t h e head ( " t h e s u b j e c t of t h e v e r b which

s t r i c t l y s u b c a t e g o r i z e s t h e AP dominat ing t h e

a d j e c t i v e " ) , f o r argument - x i n t h e e n t r y ' s f u n c t i o n a l

s t r u c t u r e .

There i s no s e n s e i n which ( 5 ) can be regarded a s a s i n g l e

p r o j e c t i o n r u l e . The p o i n t i s t h a t s i n c e t h e r e i s no p h r a s e

s t r u c t u r e c o n f i g u r a t i o n ( 6 a ) c o r r e s p o n d i n g t o ( 6 b ) and ( 6 c ) ,

and t h e r e f o r e t h e d e f i n i t i o n [ N w l , [ - O b j ] l l l ] does n o t p i c k o u t

a n y t h i n g i n t h e c a s e o f a d j e c t i v e s , t h e r u l e s f o r a d j e c t i v e s and

d e a d j e c t i v a l nouns canno t be g e n e r a l i z e d by r e f e r r i n g t o a

grammat ica l r e l a t i o n between a n X and t h e I N w t , X v l ] , as t h e

r u l e s f o r v e r b s and d e v e r b a l nouns can be.

I t i s i n t e r e s t i n g t h a t e l sewhere i n X 1 Syntax , Jackendof f - (1977:32, f n . 2 ) c i t e s p r e c i s e l y t h e absence o f a p h r a s e

s t r u c t u r e c o n f i g u r a t i o n ( 6 a ) i n s u p p o r t of a n a n a l y s i s

( i n d e p e n d e n t l y s u g g e s t e d i n Jackendof f ( 1 9 7 4 b ) ) where t h e

s e l e c t i o n a l r e s t r i c t i o n of a p r e d i c a t e a d j e c t i v e on t h e s u b j e c t

of t h e s e n t e n c e i s e n f o r c e d by a r u l e r e f e r r i n g t o t h e t h e m a t i c

r e l a t i o n Theme, and n o t t o t h e grammat ica l r e l a t i o n " s u b j e c t

o f " :

. . . i t i s o f t e n assumed t h a t John i n John i s t a l l i s - --- t h e s u b j e c t o f t a l l . T h i s a s sumpt ion i s i n c o r r e c t . Although t a l l imposes a s e l e c t i o n a l r e s t r i c t i o n on t h e NP i n - NP - m a l l -9 t h e NP b e a r s t h e g rammat ica l r e l a t i o n " s u b j e c t o f " t o t h e v e r b -3 be n o t t o t h e a d j e c t i v e . . . . An NP b e a r i n g t h e " s u b j e c t o f n r e l a t i o n t o a n a d j e c t i v e would have t o be c o n t a i n e d i n t h e AP, a s t h e s u b j e c t of a noun i s c o n t a i n e d i n t h e NP. S i n c e t h e r e i s no AP * J o h n ( ' s ) f e a r f u l co r respond ing t o t h e NP J o h n ' s f e a r , f o r example, we conc lude t h a t a d j e c t i v e s do n o t have s y n t a c t i c s u b j e c t s . For a d i s c u s s i o n o f how t h e s e l e c t i o n a l r e s t r i c t i o n i s Imposed w i t h o u t a grammat ica l r e l a t i o n , s e e Jackendof f (1974b) . . .

I n t h e a r t i c l e t o which he r e f e r s u s , Jackendof f p roposes a

Complex P r e d i c a t e Rule (CPR). Given a s e n t e n c e l i k e ( 4 a ) , John

i s p e r v e r s e , t h e CPR combines t h e i n t e r p r e t a t i o n o f t h e - a d j e c t i v e p e r v e r s e w i t h t h a t o f t h e v e r b be, t h e l a t t e r a c t i n g

"simply as a s e m a n t i c a l l y unmarked p l a c e h o l d e r , s o t h a t t h e CPR

can a p p l y and c r e a t e a v e r b a l one-place f u n c t i o n o u t o f t h e AP"

( Jackendof f 1974b:502). Once t h e CPR a p p l i e s ,

( t ) h e s u b j e c t , s t r i c t l y s u b c a t e g o r i z e d by be, can now be t a k e n as NP1 i n t h e new f u n c t i o n and i s i n s e r t e d by Argument S u b s t i t u t i o n , s u b j e c t t o t h e s e l e c t i o n a l r e s t r i c t i o n s o f t h e a d j e c t i v e . ( l o c . c i t . )

S i n c e i n t h i s i n s t a n c e i t i s t h e l l sub jec t l ' t h a t t h e p r o j e c t i o n

rule of Argument Substitution inserts into the functional

structure of the complex predicate - be perverse, it might appear

that the notion '!subject off1 is again useful for generalizing a

projection rule cross-categorially. The case of adjectives and

deadjectival nouns appears to differ from that of verbs and

deverbal nouns only in that in the first case, a complex

predicate V-AdJ must be created before the projection rule picks

out the llsubject of l1 the V or the N for argument substitution.

If so, the claim that "the rules which impose selectional

restrictions on grammatical subjects will apply equally" in John

is perverse and John's perversity could be substantiated. -

Unfortunately for this claim, Jackendoff' (1974b:488ff.) provides

independent evidence that argument substitution into complex

predicates must refer to thematic functions.

The CPR was originally motivated for the interpretation of

complex predicates like put/pin/place/fix - the blame - on and

take/get/receive/accept - the blame - for, which proceeds in the

following manner. The functional structure of the noun blame is

specified for (at least) two arguments: a 'blamer,' and a

'blamee. But in John put - the blame -- on Bill and in Bill took the --7

blame, the NPs which must be substituted into the noun's

functional structure do not appear as the syntactic tlsubjecttl or

llobjectw of the noun; they stand in grammatical relations to the

verbs put and take. The situation is thus similar to that of the

predicate adjective perverse. The CPR now applies to create the

complex p r e d i c a t e s put t h e blame ( o n ) and t a k e t h e blame ( f o r ) , - -- f o l l o w i n g which t h e arguments o f t h e v e r b s can be "borrowed1' f o r

i n s e r t i o n i n t o t h e noun ' s f u n c t i o n a l s t r u c t u r e .

But obse rve t h a t i f t h e v e r b i s put, i t s p r e p o s i t i o n a l

o b j e c t must be s u b s t i t u t e d f o r t h e Iblarneet a rgument , w h i l e i f

t h e v e r b i s t a k e , i t s s u b j e c t must be s u b s t i t u t e d f o r t h e same

argument. Jackendof f t h e r e f o r e a r g u e s t h a t t h e p r o j e c t i o n r u l e

does n o t s imply r e f e r t o grammat ica l r e l a t i o n s i n such c a s e s .

R a t h e r , i t r e f e r s t o t h e t h e m a t i c r e l a t i o n s of t h e ve rb . The

Goal i s a lways chosen a s t h e 'blamee. For put, t h e Goal a p p e a r s

a s a p r e p o s i t i o n a l o b j e c t , whereas f o r -, t a k e t h e Goal a p p e a r s a s

t h e s u b j e c t . Thus t h e a p p r o p r i a t e NPs ar9e chosen a s 'blameel i n

bo th c a s e s .

I t i s i n f a c t t h e t h e m a t i c b a s i s of argument s u b s t i t u t i o n

i n t o complex p r e d i c a t e s t h a t s e r v e s t o recommend t h e e x t e n s i o n

o f t h e CPR t o d e a l w i t h t h e s e l e c t i o n a l r e s t r i c t i o n s of

p r e d i c a t e a d j e c t i v e s , i n Jackendoff (1974b) . Jackendof f g i v e s

two arguments a g a i n s t b a s i n g argument s u b s t i t u t i o n on

grammat ica l r e l a t i o n s , i n t h e c a s e of p r e d i c a t e a d j e c t i v e s .

F i r s t l y , t h e NPs t h a t p r e d i c a t e a d j e c t i v e s r e s t r i c t a p p e a r i n

d i v e r s e s y n t a c t i c p o s i t i o n s , depending on t h e s e m a n t i c s t r u c t u r e

of t h e ve rb . Thus t h e a d j e c t i v e p e r v e r s e r e s t r i c t s t h e s u b j e c t

i n ( 7 a ) and ( 7 b ) , bu t i n ( 7 c ) , i t r e s t r i c t s t h e o b j e c t :

( 7 ) a . John i s p e r v e r s e .

b. John impresses (strikes) me as perverse.

c. I consider John perverse.

Jackendoff therefore suggests that

the modified NP is always the Theme of the adjective, and so, for the sake of the CPR, it must appear as the Theme of the main verb. (op. cit. :502)

The second argument concerns the range of syntactic positions

that can be selectionally restricted by rules referring to

grammatical relations. Thus Jackendoff (1974b:500) writes:

In the standard formulation of strict subcategorization and selectional restrictions (Chomsky 1965), a lexical item can place restrictions only on the occurrences of NPs and PPs within its major phrase. A verb can determine the nature of its subject and of constituents in the VP; that is, it restricts the arguments in the immediately dominating S. . . . Predicate adjectives present an apparent anomaly in this respect. . . . there is never an NP to the left of the adjective in an AP which functions as "subject. l1 . . . Thinking only of sentences like John is pilty, one is tempted to call the modified the subject of the adjective and to try to state a strict subcategorization (sic) - restriction on this basis. But since John is outside the AP and is in fact the subject of - J b e t h i s proposal goes against the generalization about strict subcategorization that we noted above.

The problem that Jackendoff discusses in the preceding

extract in fact arises only within the framework of X 1 Syntax.

The wgeneralization about strict subcategorization" that he

wishes to capture is the requirement in Chomsky (1965:gg-100)

that strict subcategorization be "strictly local.ll This

requirement does not apply to selectional restrictions in the

standard theory, and with good reason. Recall that in the

standard theory, S is rewritten as NP VP. Now if selectional

restriction were "strictly local,1' it would not be possible to

impose the selectional restrictions of a verb on the subject of

the sentence, since the subject is not dominated by the major

category (VP) which introduces the V.4 But selectional

restriction is not "strictly local,t1 and therefore it is no more

of a problem (for the standard theory) that the subject NP in - NP

be AP structures is outside the AP, than it is that the subject --

NP in -- NP VP structures is outside the VP.

Consider, however, a theory where S is V f t t . In such a

theory, the subject of the sentence is dominated by the major

phrase ( V H t ) that introduces the V. We can now legitimately

claim that the subject-verb selectional restriction is tlocalt

with respect to the X t t l category whose head X specifies the

restriction. The selectional restriction between the possessive

NP and the head N is also rlocalt in this sense. In this

framework it would indeed be an anomaly that a predicate

adjective restricts an NP (namely, the subject) which is not

dominated by the A " ' whose head A specifies the restriction.

Thus a nice constraint on selectional restriction which is made

possible by the identification of S with V t t l also serves as an

argument against a projection rule referring to "subject of,!'

for predicate adjectives.

The discussion of the problem of predicate adjectives is

complicated by the obliteration, in X1 Syntax, of the

distinction between the notions Ngrammatical functionll and

ngrammatical relation," made by Chomsky (1965:71,73). Thus

Jackendoff presents [N1ll, [+~ubj]'~'] as a definition of Ifthe

generalized grammatical relation fsubject-oflll (op. cit.:41,

emphasis added); but in the standard theory, this would be the

definition of a grammatical function. Chomsky defines

grammatical relations derivatively in terms of functional

notions. The grammatical function "subject-of" is defined for

the S, and the grammatical relation "Subject-Verb" is the

relation between the Subject-of the Sentence and the

Main-Verb-of the Predicate-of the Sentence (Chomsky 1965:73).

This being the case, there is nothing (in the standard theory)

to prevent the postulation of a grammatical relation

vSubject--Predicate Adjective," given appropriate functional

definitions. The subject of the S is thus accessible to the

selectional rules of both V and A . But X1 Syntax apparently

attempts to state selectional restrictions in terms of the

grammatical functions of a single phrasal category. Ss with an

adjectival predicate present a problem for this attempt.

Observe that the point at issue is not merely a matter of

evolving an alternative method of capturing the

subcategorizational parallelisms of adjectives and deadjectival

nouns; though this is not a trivial problem, given the

assumpt ions of X ' S y n t a x - 5 The m o t i v a t i o n f o r i d e n t i f y i n g S wi th

V I t t , and f o r g e n e r a l i z i n g t h e grammat ica l r e l a t i o n s of t h e S t o

t h e NP, i s i t s e l f c a l l e d i n t o q u e s t i o n . There i s no a p r i o r i - reason t o assume t h a t t h e S and t h e NP a r e s t r u c t u r a l l y

p a r a l l e l ; a s H o r n s t e i n (1977) p o i n t s o u t , t h e r e i s i n f a c t

ev idence t h a t they a r e n o t p a r a l l e l . (Thus S s have a u x i l i a r i e s

and complement izers , w h i l e NPs have d e t e r m i n e r s . ) Jackendof f

(1977:38-39, f n . 5 ) d i s m i s s e s t h e s e d i f f e r e n c e s as " i r r e l e v a n t , "

f o r " t h e X I Convention s a y s n o t h i n g a b o u t what t o do w i t h

n o n p a r a l l e l s t r u c t u r e s . . . . t h e X t Convention s a y s s imply t h a t

when p a r a l l e l i s m s e x i s t , t hey must be e x p r e s s e d . Now t h e

observed p a r a l l e l i s m i n t h e S and t h e NP t h a t t h e t h e o r y must

accoun t f o r is i n t h e s u b c a t e g o r i z a t i o n , and i t o b t a i n s f o r S s

w i t h two t y p e s o f p r e d i c a t e s ( v e r b a l , and a d j e c t i v a l ) and t h e

c o r r e s p o n d i n g NPs. By J a c k e n d o f f 1 s argument , t h e p o s t u l a t i o n o f

a s t r u c t u r a l p a r a l l e l i s m would be j u s t i f i e d ( r e g a r d l e s s of o t h e r

s t r u c t u r a l d i f f e r e n c e s ) i f i t succeeded i n e x p r e s s i n g b o t h t h e s e

c a s e s of d i s t r i b u t i o n a l p a r a l l e l i s m . But w e have s e e n t h a t one

c a s e of d i s t r i b u t i o n a l p a r a l l e l i s m remains unexpressed i n

s t r u c t u r a l terms.

Moreover, t h e s t r u c t u r a l p a r a l l e l i s m o f S and NP is

a c h i e v e d a t t h e c o s t o f o b s c u r i n g a g e n e r a l s t r u c t u r a l

d i f f e r e n c e between S and NP, and of c r e a t i n g a n o n p a r a l l e l i s m

between v e r b a l and a d j e c t i v a l p r e d i c a t e s . If S is n o t V t t t , t h e

s u b j e c t of t h e S i s a s p e c i f i e r o f n e i t h e r t h e V n o r t h e A ,

while the llsubjectv of the NP is always a specifier of the N.

The restriction against sentences in the possessive NP position

(but not in the subject position) may be attributed to this

difference (cf. Emonds 1976, Jackendoff 1977:44). The projection

rules for verbs as well as adjectives would have to refer to a

position (namely, the subject of the sentence) outside the major

phrase introducing the V or the A , and there would be no need to

assume a radical difference in the types of projection rules

involved.

In the light of these problems we might consider reverting

to a theory of phrase structure wherein S is not V f f l , and

relinquish the claim that the grammatical relations of the NP

are identical to those of the S. The NP could be allowed its own

set of grammatical functions, wPoss-NP" and "of-NP.'I - However, we

might still try to state subcategorizational correspondences in

terms of a correspondence in the grammatical functions of the S

and the NP. Thus we may claim that the Subject corresponds to

Poss-NP, so that a selectional rule which relates Subject-Verb

or Subject-Adjective in the S will relate Poss NP-head N in the

noun phrase; and that the Object corresponds to the - of-NP, so

that a selectional rule which relates Verb-Object in the S will

relate head N-of - NP in the noun phrase. This is essentially the suggestion in Chomsky (1970:201):6

. . . grammatical relations are defined by configurations in the deep structure, and selectional

f e a t u r e s re la te t h e heads of p h r a s e s t h a t a r e a s s o c i a t e d i n s p e c i f i c grammat ica l r e l a t i o n s . Then t h e words John and p roof a r e t h e heads o f t h e r e l a t e d p h r a s e s s e v e r a l of J o h n ' s and p r o o f s of t h e theorem i n s e v e r a l o f J o h n ' s p r o o f s -- of t h e t h e o r e m y a x t h e same s e l e c t i o n a l ~ e a t u r e t h a t a s s o c i a t e s s u b j e c t and v e r b i n John proved - t h e theorem w i l l r e l a t e t h e s e two i t e m s , d e s p i t e t h e ve ry d i f f e r e n t s y n t a c t i c o r i g i n o f t h e r e l a t i o n s h i p . (Chomsky 1970 :201)

The n e x t s e c t i o n i s a n a t t e m p t t o show t h a t even t h e weaker

c l a i m o u t l i n e d i n t h e p r e c e d i n g p a r a g r a p h must be r e j e c t e d . I

s h a l l a r g u e t h a t a mapping r u l e which r e f e r s t o grammat ica l

f u n c t i o n s ( a t a l l ) i s t h e wrong t h e o r e t i c a l mechanism f o r

s t a t i n g cor respondences of s e l e c t i o n a l r e s t r i c t i o n s between

d e r i v a t i o n a l l y r e l a t e d words; and t h a t ( i n i t s p l a c e ) w e need t o

u s e a r u l e which o p e r a t e s on t h e m a t i c f u n c t i o n s .

3 . An v I r r e g u l a r w C l a s s of Deverbal Nominals -- - Examples of v e r b s and d e v e r b a l nouns l i k e

c r i t i c i z e - c r i t i c i s m , d e s t r o y - d e s t r u c t i o n , which are commonly

d i s c u s s e d i n t h e l i t e r a t u r e , have l e d t o t h e s e t t i n g up of a

paradigm o f co r respondence between ( I ) t h e s u b J e c t of t h e

s e n t e n c e and t h e p o s s e s s i v e NP i n t h e nominal , and (ii) t h e

o b j e c t o f t h e v e r b and t h e - of-complement t o t h e noun. Hence t h e

a t t e m p t t o f o r m u l a t e p r o j e c t i o n r u l e s f o r t h e S and t h e NP i n

terms o f g e n e r a l i z e d grammat ica l r e l a t i o n s . If t h i s paradigm of

correspondence can be shown t o be u n j u s t i f i e d , t h e p r i n c i p a l

r e a s o n f o r g e n e r a l i z i n g t h e grammat ica l r e l a t i o n s of t h e S t o

t h e NP d i s a p p e a r s , I n what f o l l o w s I show t h a t t h e r e i s a

s e m a n t i c a l l y c o h e r e n t c l a s s of v e r b s and m o r p h o l o g i c a l l y

d e v e r b a l nouns f o r which t h e p o s t u l a t e d paradigm i s n o t v a l i d .

F u r t h e r , I show t h a t t h e s e v e r b s and nouns a r e n o t l l i r r e g u l a r l l

i n any way; and t h a t t h e y would i n f a c t be l l i r r e g u l a r l l i f t h e y

conformed t o t h e commonly assumed p a t t e r n of correspondence .

Chomsky (1970:188-189) n o t i c e d t h a t t h e d e v e r b a l nominals

amusement and i n t e r e s t do n o t show t h e l1expectedt1 p a t t e r n of

co r respondence w i t h t h e i r v e r b s :

(8) a. John amused ( i n t e r e s t e d ) t h e c h i l d r e n w i t h h i s

s t o r i e s

b. * J o h n l s amusement ( i n t e r e s t ) o f t h e c h i l d r e n w i t h

h i s s t o r i e s

I n t h e s e c a s e s , t h e s e t of p o s s i b l e o b j e c t s o f t h e v e r b

cor responds t o t h e s e t o f p o s s i b l e l t s u b j e c t s v i n t h e NP:

( 9 ) t h e c h i l d r e n 1 s amusement a t ( i n t e r e s t i n ) J o h n ' s

s t o r i e s 7

It has been s u g g e s t e d by Lakoff ( 1 9 7 0 ) , Jackendof f

(1975 :660-661), and (more r e c e n t l y ) by Wasow (1977 :359, fn . l 5 ) ,

t h a t t h e s u b c a t e g o r i z a t i o n a l p r o p e r t i e s of such nominals be

accoun ted f o r by r e l a t i n g them t o a n a d j e c t i v a l p a s t p a r t i c i p l e

r a t h e r t h a n t o t h e ve rb . Thus ( 9 ) would be r e l a t e d t o ( 1 0 ) :

(10) The children were amused at (interested in) John's

stories.

Wasow first motivates a distinction between "transformational1'

passives and 'llexicalll passives, and shows that the latter are

derived by a category-changing lexical rule. One criterion for

distinguishing a transformational from a lexical rule is that

the latter has idiosyncratic exceptions (Wasow1s Criterion 5).

Since the participles in ( 10) select idiosyncratic prepositions

like -* at - in, instead of the expected a, they are candidates for the lexical rule by this criterion. The nominals in ( 9 ) seem to

follow the adjectives1 idlosyncracy in preposition selection.

Neither the adjectives nor the nominals (unlike the verbs)

permit inanimate subjects.8

I give below a partial list of verb-participle-nominal

triplets for which the paradigm we noted in (8)- (10) seems to

hold.

(11) a. amaze, amuse, annoy, astonish, bore, confuse,

delight, disappoint, disgust, dismay, distress,

elate, embarrass, fascinate, frustrate, humiliate,

hurt, inspire, interest, irritate, (dis)please,

puzzle, relieve, surprise, vex

b. amazed (at), amused (at), annoyed (at), astonished

(at), bored (with), confused (by), delighted (at),

distressed (at), elated (at), embarrassed (at),

fascinated (with), frustrated (at), humiliated (by),

hurt (by), inspired (by), interested (in), irritated

(at), (d1s)pleased (at), puzzled (at), relieved

(at), surprised (at), vexed (by)

c. amazement (at), amusement (at), annoyance (at),

astonishment (at), boredom (with), confusion (*by),

delight (at), disappointment (with), disgust (at),

dismay (at), distress (at), elation (at),

embarrassment (at), fascination (with), frustration

(at), humiliation (*by), hurt (*by), inspiration

(*by), interest (in), irritation (at), (d1s)pleasure

(at), puzzlement (?at), relief (at), surprise (at 1,

vexation (*by)

Relating the nominals in (llc) to the participles in (llb)

neatly accounts for distributional facts. Further, it reinstates

the parallelism between the subject of the sentence and the

possessive NP in the nominal. Note, however, that it constitutes

no kind of explanation of why only a vpassivew nominalization

exists in just these cases09 It is not generally the case that

the existence of an adjectival passive implies that the

corresponding nominal is related to this adjectival passive and

not to the verb. We can argue for adjectival passives in the

case of collected and explored (since we have uncollected and

unexplored, and - un- attaches only to adjectives; cf. Siege1

1973, 1974, and Hust 1977), but collection and exploration must

be related to the verb: - the Church's collection - of tithes, the

Norwegians1 exploration -- of the Antartic. The analysis suggested

by Lakoff, Jackendoff and Wasow invites us to accept the fact

that a verb in (lla) does not have a related nominalization, or

the fact that a participle in (llb) has a related

nominalization, as (in each case) an arbitrary fact about the

individual lexical item. A lexical 'Igap1l needs no further

explanation. (Thus Jackendoff (1975:661), who considers only one

nominal, amusement, of the group of nominals (llc), expressly

states that "the existence of only one of the possible forms of

amusement is an -- ad hoc fact, expressed in the lexic~n.~)

But it is striking that the verbs for which the same -- ad hoc

fact must be expressed form a coherent semantic class. The verbs

in (lla) signify the causation of internal states in animate

(typically human) objects that are capable of experiencing these

states. It would appear that any attempt to explain the paradigm

(1la)- (llc) must mention some such semantic fact. Chomsky

(1970) suggested that this semantic fact is a feature [+cause]

of the verb.10 The explanation however cannot be (just) the

causative nature of the verb, for as Jackendoff (1975:660)

notes, "other causatives do have nominalizations~ (example (12b)

from Jackendof f ) :

(12) a. Gamma rays excited the protons.

b. the excitation of the protons by gamma rays

Let us tentatively hypothesize that the crucial semantic

fact is not the causativity of the verb as such, but the

semantic role of the verb's object. Let us call the animate

objects of the verbs in (lla), which are experiencing internal

states, Experiencers. As a first approximation, we may describe

the facts in (1la)- (llc) as follows: if the object of a

causative verb is an Experiencer, the verb does not have a

"regularw nominalization with a causative interpretation.

Instead, the subject of the verb appears, if at all, as a

prepositional object in the NP; and the verb's object appears as

"subject. Further, there is an adjectival passive which takes

the verb's object as subject. The nominal therefore appears to

be subcategorizationally parallel to the adjectival passive.

When considered more closely, the causative verb excite (of

(12)) and its nominalizations can be seen to provide a

particularly neat confirmation of our hypothesis. The verb

excite can take either inanimate objects, as in (12a), or

animate objects, as in (13a):

(13) a. The news excited Mary.

In (13a), Mary is an Experiencer. The interesting fact is that

(13a) does not have the nominalization (13b) (compare (12b)):

( 1 3 ) b. * t h e e x c i t a t i o n ( e x c i t e m e n t ) o f Mary by t h e news

R a t h e r , we have t h e d e r i v e d nominal (14a), which i s

s u b c a t e g o r i z a t i o n a l ~ l y p a r a l l e l t o ( 1 4 b ) :

( 1 4 ) a . Mary's e x c i t e m e n t ( * e x c i t a t i o n ) a t t h e news

b. Mary was e x c i t e d a t t h e news.

Observe t h a t t h e p a r t i c i p l e e x c i t e d - a t does n o t a l l o w inan imate

s u b j e c t s .

( 15 ) *The p r o t o n s were e x c i t e d a t gamma r a y s .

For e x c i t e , t h e r e a r e two d i f f e r e n t morpho log ica l forms of

t h e nominal ( e x c i t e m e n t , e x c i t a t i o n ) , each w i t h i t s s p e c i f i c

d i s t r i b u t i o n a l p r o p e r t i e s . There a r e o t h e r v e r b s whose nominals

have o n l y one morpho log ica l form, bu t t h e nominal once a g a i n

shows d i f f e r e n t s u b c a t e g o r i z a t i o n s depending on whether t h e

o b j e c t o f t h e v e r b i s inan imate , o r a n an imate Exper iencer . If

t h e o b j e c t i s i n a n i m a t e , t h e s u b c a t e g o r i z a t i o n o f t h e nominal

f o l l o w s t h a t o f t h e verb . If t h e o b j e c t i s an imate , t h e r e i s a

l e x i c a l l y r e l a t e d p a s t p a r t i c i p l e , and t h e s u b c a t e g o r i z a t i o n o f

t h e nominal f o l l o w s t h a t o f t h e p a r t i c i p l e .

(16) a . The a i r p r e s s u r e d e p r e s s e s t h e l e v e r .

b. t h e a i r p r e s s u r e ' s d e p r e s s i o n of t h e l e v e r

(17) a . The weather d e p r e s s e d Mary.

b. *the weather's depression of Mary

(18) a. Mary was depressed at the thought of going home.

b. Mary's depression at the thought of going home

(19) a. *The lever was depressed at the air pressure.

b. *the lever's depression at the air pressure

(20) a. The machine's rotary action agitates the soap

solution.

b. the agitation of the soap solution by the machine's

rotary action

(21) a. The (aide's) resignation agitated the official.

b. *the (aide's) resignation's agitation of the

official

(22) a. The official was agitated at (over) the aide's

resignation.

b. the official's agitation at (over) the aide's

resignation

(23) a. *The soap solution was agitated at (over) the

machine's rotary action.

b. *the soap solution's agitation at (over) the

machine's rotary action

( 2 4 ) a. The lexical entry satisfies the non-distinctness

requirement.

b. the satisfaction of the non-distinctness requirement

by the lexical entry

(25) a. The dress satisfies Mary.

b. *the satisfaction of Mary by the dress

(26) a. Mary is satisfied with the results.

b. Mary's satisfaction with the results

( 2 7 ) a. *The non-distinctness requirement is satisfied with

the lexical entry.

b. sthe non-distinctness requirement's satisfaction

with the lexical entry

(28) a. Indiscriminate mining exhausted the country's

resources.

b. the exhaustion of the country's resources by

indiscriminate mining

(29) a. The walk exhausted Mary.

b. *the walk's exhaustion of Mary

(30) a. Mary was quite exhausted (by the wa1k);ll the walk

left Mary exhausted; Mary seems very exhausted.

b. Mary's exhaustion (*by the walk)

(31) 'Indiscriminate mining left the country's resources

very exhausted; +the country's resources seem very

exhausted.

We have an interesting problem here. From the point of view

of lexical insertion, verbs like excite, exhaust, depress,

agitate and satisfy are free to take either animate or inanimate

objects. That is, these verbs need not be specified for the

features [t [+Animate]] or [t [-Animatell.12 But their

nominalizations differentiate between the animate and inanimate

NPs following these verbs. If this NP is inanimate, it is

assigned the - of-NP position in the noun phrase (and the subject

is assigned the Poss-NP position.) This is the "expectedv

correspondence. But if this NP is animate, it is assigned the

Poss-NP position, and the subject of the S fails to appear in

the NP, or appears as a prepositional object. This is the

"skewedu correspondence that we noticed between the verbs in

(lla) and the nouns in (llc). The basis for this differentiation

is obviously not a difference in the grammatical functions of

these NPs, which in both cases remain the verb's direct object.

Thus our model of a vsimplev mapping from the grammatical

functions "Subject of the Sentence" and "Object of the Verbn to

the grammatical functions llPoss-NPu and "of - -NPn apparently

breaks down.

Observe t h a t t h e problem i s n o t avo ided by p o s t u l a t i n g two

nominals , ( e . g . ) d e p r e s s i o n and d e p r e s s i o n , and r e l a t i n g one

of them t o t h e p a r t i c i p i a l a d j e c t i v e and t h e o t h e r t o t h e ve rb .

T h i s l l s o l u t i o n v o n l y s h i f t s t h e problem t o a n o t h e r r u l e of t h e

grammar. For w e n o t i c e t h a t t h e r u l e f o r t h e p a r t i c i p i a l

a d j e c t i v e must a l s o be s e n s i t i v e t o t h e f e a t u r e o f animacy of

t h e v e r b ' s o b j e c t . The s u b j e c t s of t h e p r e d i c a t e s - be e x c i t e d at,

be d e p r e s s e d at, be s a t i s f i e d w i t h , be a g i t a t e d a t and seem - - - exhaus ted must be an imate ; they cor respond t o o n l y t h e an imate

o b j e c t s o f t h e r e l a t e d v e r b s . T h i s means t h a t t h e r u l e r e l a t i n g

t h e v e r b and t h e p a r t i c i p i a l a d j e c t i v e must d i f f e r e n t i a t e

between t h e an imate and i n a n i m a t e o b j e c t s of t h e ve rb .

The problem becomes n i c e l y f o c u s s e d when w e l o o k a t t h e

fo rmal i sm o f l e x i c a l redundancy r u l e s . The g e n e r a l schema f o r

such r u l e s h a s been s u g g e s t e d by Chomsky ( 1 9 7 0 ) , f o r v e r b s and

t h e i r - ab le - d e r i v a t i v e s ( r e a d - r e a d a b l e ) , - and i t h a s been f u r t h e r

developed by Hust (1978) :

. . .a s u b r e g u l a r i t y . . . r e g a r d i n g s e l e c t i o n a l r u l e s i n t h e c a s e o f - a b l e . . . can be f o r m u l a t e d as a l e x i c a l r u l e t h a t ass= t h e f e a t u r e [ X - 1 t o a l e x i c a l i t e m [V-able] where V has t h e i n t r i n s i c s e l e c t i o n a l f e a t u r e [- X I . (Chomsky 1 9 7 m 1 3 , emphasis added)

The p o i n t t o n o t e is t h a t w i t h i n t h i s formal ism, t h e r u l e s can

mention on ly t h e i n t r i n s i c s e l e c t i o n a l f e a t u r e s o f t h e ve rb .

There fo re , if t h e v e r b ' s s e l e c t i o n a l f e a t u r e s s a y n o t h i n g a b o u t

t h e animacy of t h e d i r e c t o b j e c t , t h i s i n f o r m a t i o n w i l l n o t be

available to the lexical redundancy rules. In the framework of

Jackendoff (1975 ) , we must similarly differentiate between the

[ +- [+Animate]] and the [ + [-Animate]] subcategories of

these verbs. Jackendoff gives fully specified lexical entries

which are related by lexical redundancy rules. We must relate

excitation to excite [ + [-Animate]], and excitement to excite

[+- [+Animate]]; so also for the two senses of depression,

agitation, satisfaction, and exhaustion. We thus have a

situation where a selectional feature irrelevant to lexical

insertion must nevertheless be specified in order to capture

lexical redundancies. That is, verb subcategorization must

proceed beyond the requirements of lexical insertion, for these

verbs.

Given that it is necessary to set up [t [+Animate]] and

[ +- [-Animate]] subcategories of these verbs, it is now

possible to treat each subcategory as a separate lexical item.

But now (it could be pointed out) it is again possible to state

the required lexical redundancy rules simply in terms of

grammatical functions. The rules would not have the problem of

differentiating between the animate and inanimate objects of the

verbs, since this differentiation would be made available to

them as part of the subcategorization of different lexical

items. However, statements of correspondences in grammatical

functions once again conceal regularities in the data. It is not

accidental that of ten verbs (five homonymous verb pairs), it is

the five [ + [+Animate]] verbs which have a nominalization

with the object appearing in the Poss-NP position. Observe that

these animate objects are Experiencers; our hypothesis predicts

that they should appear in the Poss-NP position. Again, it is

only the five [ + [+Animate]] verbs that have adjectival

passives. If lexical redundancy rules referred only to

grammatical functions, we would not expect such a systematic

separation of five homonymous verb pairs into the classes

[ +- [+Animate]] and [ + [-Animate]] for purposes of rule

application, coincidental with a difference in the semantic role

of their objects. Thus regardless of whether excite, exhaust,

depress, agitate, and satisfy are each treated as one lexical

item or two, an explanation of the subcategorizational

correspondences in these cases cannot be given unless the

semantic role of their object NPs is taken into account.

Observe that it is the semantic role of the object, and not

its animacy or inanimacy per - se, that is relevant for lexical

redundancies. We may contrast the examples above with criticize

-criticism, where the inanimacy or animacy of the object does

not correlate with changes in its semantic role, and is (thus)

irrelevant for lexical redundancies. Compare John's critici~m - of

the play, John's criticism of Mary. Note also that the semantic - -

roles (thematic functions) to which lexical redundancy rules are

here claimed to be sensitive are much finer than notions like

Theme. Thus, for Jackendoff (1972), both - the lever and Mary as

objects of depress in (16a) and (17a) would be Themes; since

Themes include NPs whose location or change of location is

asserted by the predicate, and the notions vvlocationw and

"change of locationv1 are generalized to cover both physical

space and "abstract" space (such as points on an emotional

scale). Thus adjectives can function as abstract locations, and

the subjects of such adjectival predicates are considered Themes

(op. cit. :29-31). But, as we see, this gross classification will

not do for us.

The finer subcategorization of excite, exhaust, depress,

agitate and satisfy allows us to include the [+ [+Animate]]

subcategory of these verbs with the verbs in (lla) (all of which

take only animate objects), and to extend our generalization

about Experiencers to this new class of verbs, (llaf). That

finer subcategorizations may reveal underlying ~logicalv

relationships and allow the identification of Noverlappingtv

categories has been anticipated by Chomsky (1964). Chomsky

discusses the classification of lexical formatives into

hierarchies of categories, in the following terms:

1 Suppose we have a three level hierarchy. Then C1is the

2 class of all words. Let C1 =Nouns, C; =Verbs,

2 C: =Adjectives, C L =everything else. Let

3 3 1' . . . , C be subcategories of Verbs (pure .I

transitives, those with inanimate objects, etc.); . . .

(Chomsky 1964 :387)

H e then asks:

What i s t h e n a t u r a l p o i n t where cont inued refinement of t h e category h i e r a r c h y should come t o an end? This i s no t obvious. A s t h e grammatical r u l e s become more d e t a i l e d , we may f i n d t h a t grammar i s converging wi th what has been c a l l e d l o g i c a l grammar. That i s , we seem t o be s tudy ing smal l over lapping c a t e g o r i e s of fo rmat ives , where each ca tegory can be c h a r a c t e r i z e d by what we can now (g iven t h e grammar) recognize as a semantic f e a t u r e of some s o r t . . . (op. c i t . : 3 8 7 , f n . 8 )

I n t h e l i g h t of t h e program f o r grammar ske tched i n t h e

preceding e x t r a c t , we can s e e t h a t o u r a n a l y s i s has been i n t h e

r i g h t d i r e c t i o n . A "continued refinement of t h e ca tegory

h i e r a r c h y w ( i n t h e ca se of e x c i t e , e t c . ) a long semant ic l i n e s

revea led a c l a s s of verbs ( l l a l ) : a c l a s s of w c a u s a t i v e v verbs

wi th Exper iencer o b j e c t s . I n t e r e s t i n g l y , t h i s r e s u l t converged

wi th t h e r e s u l t of a s tudy of s u b c a t e g o r i z a t i o n a l

correspondences , which revea led t h e same c l a s s of verbs . The

d i scovery of such c a t e g o r i e s is what Chomsky views as a s t e p

towards f l l og i ca l " grammar.

FOOTNOTES TO CHAPTER TWO

Fol lowing t h e p r a c t i c e o f Jackendof f ( 1 9 7 7 ) , I s h a l l r e p r e s e n t

b a r s as p r imes : e. g., 21x1.

A p r e v i o u s v e r s i o n o f t h e m a t e r i a l p r e s e n t e d i n t h i s

c h a p t e r and t h e n e x t i s t o a p p e a r i n L i n g u i s t i c A n a l y s i s , under

t h e t i t l e Express ing Cross -Ca tegor ia l S e l e c t i o n a l

Correspondences : - An A l t e r n a t i v e t o t h e Syntax Approach. I wish -- t o thank J o e l Hus t , R ichard DeArmond, K a r a t t u p a r a m b i l J a y a s e e l a n

and Al f redo Hur tado f o r u s e f u l d i s c u s s i o n s .

Such a s t r u c t u r a l p a r a l l e l i s m i s n e c e s s a r y i n a t h e o r y ( l i k e

t h e s t a n d a r d t h e o r y ) where in grammat ica l r e l a t i o n s a r e d e f i n e d

by s t r u c t u r a l conf i g u r a t i o n s . Thus Jackendof f writes : "The

g e n e r a l p r i n c i p l e e n t a i l e d by t h e X 1 Convention i s t h a t i f

p a r a l l e l grammatical r e l a t i o n s e x i s t i n two d i f f e r e n t

c a t e g o r i e s , t h e c a t e g o r i e s must be s y n t a c t i c a l l y p a r a l l e l w i t h

r e s p e c t t o t h a t grammat ica l r e l a t i o n w (1977:37-38).

We assume t h a t a rguments i n t h e f u n c t i o n a l s t r u c t u r e c a r r y

t h e m a t i c i n f o r m a t i o n (Agent , Theme, Goal , e t c . ) and o t h e r

semant ic i n f o r m a t i o n ( c o n c r e t e , an imate , human, e t c . ) ; c f .

J ackendof f (1972:36-43). Thus g i v e n t h e f u n c t i o n a l s t r u c t u r e

{ ( "Znt ' Theme Argumen t

the projection rule substitutes the reading of [Nwl,

[+SubjlH'] for the Agent argument.

Observe that verbs are never strictly subcategorized for

subjects in the standard theory, but selectional features for

subjects are included in their lexical entries.

5 Thus suppose we claim that a projection rule referring to the

notion Theme can be generalized for adjectives and deadjectival

nouns. We must then countenance two types of projection rules

--"grammaticaln and "thematicw -- in Ss and NPs. But now it becomes imperative to specify the conditions under which either

type of rule applies. Suppose we stipulate that the

wgrammatical~ rule applies for restrictions within an Xtl', and

the "thematicw rule applies elsewhere. This condition holds for

the S; but it fails for the NP, since the restrictions in

deadjectival as well as deverbal nominals involve the specifier

and the head of the same Xwl. We might argue that the type of

projection rule in the NP is determined by the type of

projection rule in the S. But this tells us nothing about NPs

for which there are no corresponding Ss. Selectional

restrictions must be enforced in yesterday's weather, John's

attitude, Mary's uncle; the theory we envisage provides no hint

as to whether the rules in these NPs refer to thematic

relations, or to grammatical relations.

Observe also how such a theory would account for the

subcategorizational correspondences between Ss; cf. (i) and (ii)

below:

(i) The door opens.

(ii) The door is open.

The restrictions on NPs occupying subject position In (1) and

(ii) are identical. But the rule in (li) enforces the

restrictions of a predicate adjective, and must therefore refer

to the notion Theme, whereas the rule in (i) refers to the

"subject-ofw the intransitive verb open.

6 In later work (Chomsky 1973, 1977), Chomsky adopts the

position that the possessive NP is the "subjectw of the noun

phrase. The motivation he provides is not subcategorizational

parallelism, but the behaviour of this NP with respect to the

Specified Subject Condition. If there are alternatives to this

condition (as suggested by Brame (1977)), then even this

argument for regarding the possessive NP as a "subjectw is not

very strong.

7 Nominals like ( 9 ) do have variants like (i), where the nominal

has an - of-complement:

(i) the amusement of the children (at John's stories)

However, this should not be taken as indicative of a parallelism

in the grammatical relations of amuse and amusement, for this

of-complement canno t cooccur w i t h a f i l l e d Poss-NP l l s u b j e c t l ' - ( c f . ( 8 b ) ) , o r w i t h a ( p o s t p o s e d ) by-phrase .

The - of-NP p o s i t i o n i n t h e noun p h r a s e i s ( u n d e r t h e

t r a d i t i o n a l llrnovementv a n a l y s i s ) f i l l e d i n one o f two ways. For

' t r a n s i t i v e 1 NPs l i k e d e s t r u c t i o n -- of t h e c i t y , t h i s p o s i t i o n i s

f i l l e d i n t h e base . For ' i n t r a n s i t i v e 1 NPs l i k e - t h e prowth - of

t h e c o r n o r t h e h e i g h t of t h e b u i l d i n g , a r u l e o f NP-postposing -- - -- ( c f . Jackendof f (1977 :gOff ) ) moves a n NP from t h e Poss-NP

p o s i t i o n t o t h e - of-NP p o s i t i o n . Thus t h e o c c u r r e n c e o f a n NP i n

t h e l a t t e r p o s i t i o n does n o t ( i n i t s e l f ) i n d i c a t e a

correspondence w i t h t h e o b j e c t o f a r e l a t e d v e r b .

C o n t r a s t s l i k e ( i b ) , ( l i b ) below appear t o be a n i c e

c o n f i r m a t i o n o f t h i s a n a l y s i s , which u t i l i z e s t h e d i s t i n c t i o n

between l 1 t r a n s f o r m a t i o n a l " and u l e x i c a l v p a s s i v e s .

(1) a . The c h i l d r e n were annoyed (amused) by J o h n ' s

s t o r i e s .

b. * t h e c h i l d r e n I s annoyance (amusement) by J o h n ' s

s t o r i e s

(ii) a. The c h i l d r e n were annoyed (amused) a t J o h n ' s

s t o r i e s .

b. t h e c h i l d r e n ' s annoyance (amusement) a t J o h n ' s

s t o r i e s

If annoyed Q, amused a, are a n a l y z e d as l l t r a n s f o r m a t i o n a l l l

p a s s i v e s , b u t annoyed aJ, amused -J a t as l l l e x i c a l p a s s i v e s " (by

Wasowls Criterion 5), the unacceptability of b~ in the nominal

follows from the inability of transformationally derived

structures to participate in lexical rules (compare Wasow

1977:359, fn.15, examples (a) and (c)).

However, we must also note some recalcitrant facts. The

llsubjects" of the nominals confusion, humiliation, hurt, and

inspiration correspond to the subjects of the participles - be

confused, - be humiliated, - be -3 hurt and - be inspired (and to the

objects of the corresponding verbs). We would therefore like to

analyze these participles as wlexicallf passives, and relate the

nominals to the participles. (Additional evidence that these

participles are adjectives is provided by the "unpassivesw

unhurt, uninspired; cf. Hust (1978), Siege1 (1973, 1974).) Now

these participles select &:

(iii) He seemed confused by the questions.

(iv) John seemed humiliated by the disclosures.

(v) I was very hurt by that remark.

(vi) The poet seemed more inspired by the dinner than by the

sunset . (Notice that - un- attachment, modification by very or -' more and

the context NP seem - - -9 are diagnostics for adjectives.) The &

here is apparently lexically selected. (As Wasow notes, "there

is nothing to prohibit a lexically derived passive from taking

Q as its associated preposition, or as one of several

alternativesw (1977:349)). But even this & is unacceptable in

the nominal:

(vii) his confusion (*by the questions)

(viii) John's humiliation (*by the disclosures)

(ix) My hurt (*by that remark)

( x ) the poet's inspiration (*by the sunset)

On the other hand, if we argue that this by is the l1regularW &

and the participles are Ntransforrnationalll passives, the

nominals cannot be related to the participles.

This imperfect correspondence in the syntactic frames of

the participles and nominals under consideration suggests that

their relation may be less direct than the

Lakoff-Jackendoff-Wasow analysis argues for.

9 Anderson (1977 :371-372) seems to make the same point. However,

he (incorrectly) assumes that Wasow advocates the derivation of

all nominals from past participles. He also advances an - alternative solution, namely, to subcategorize verbs like amuse,

annoy, etc., to take idiosyncratic prepositions ("we note that

these verbs have to contain, in their lexical entries,

information concerning idiosyncratic preposition selectionu),

and to derive the nominals from wintransltivew instead of

wtransitivell verbs. This solution is unworkable, since the verbs

amuse, annoy, etc. never take idiosyncratic prepositions; it is

only the participles in the context be which must be so

subcategorized:

( i ) * T h e news had annoyed a t Mary.

( i i ) * M a r y had annoyed a t t h e news.

10 Chomsky i n i t i a l l y s u g g e s t s (1970:192) t h a t s t r u c t u r e s l i k e

John amused - t h e c h i l d r e n are t r a n s f o r m a t i o n a l l y d e r i v e d from

John c a u s e [ -- t h e c h i l d r e n - be amused], j u s t as - John grows

tomatoes might be t r a n s f o r m a t i o n a l l y d e r i v e d f rom John c a u s e

[ tomatoes p o w ] . L a t e r , he t r e a t s [+cause ] a s a l e x i c a l f e a t u r e ,

which we must " r e s t r i c t . . . w i t h r e s p e c t t o t h e f e a t u r e t h a t

d i s t i n g u i s h e s d e r i v e d nominals" (op. c i t . : 2 1 5 ) .

l1 The p a r t i c i p l e exhaus ted i s one o f t h e p r o b l e m a t i c H l e x i c a l l f

p a s s i v e s which s e l e c t 9 ( s e e f n . ( 8 ) , above) . I t s a d j e c t i v a l

s t a t u s i n ( 3 0 a ) i s shown by i t s a b i l i t y t o o c c u r i n c o n t e x t s

d i a g n o s t i c f o r a d j e c t i v e s , i.e. l e a v e NP -9 seem , and i t s

a b i l i t y t o t a k e m o d i f i e r s l i k e ve ry . The examples i n ( 3 1 ) show

t h a t when t h e NP which i s modif ied i s i n a n i m a t e (and a

non-Exper iencer) , t h i s p a r t i c i p l e does n o t o c c u r i n c o n t e x t s

d i a g n o s t i c f o r a d j e c t i v e s .

l2 For c l a r i t y of p r e s e n t a t i o n , I d i v e r g e from t h e conven t ion of

t h e s t a n d a r d t h e o r y (a rgued f o r i n Chomsky (1965), and adop ted

by Hust ( 1 9 7 8 ) ) o f g i v i n g n e g a t i v e l y marked s e l e c t i o n a l

f e a t u r e s . The f e a t u r e [ + [+Animate]] Is e q u i v a l e n t t o t h e

f e a t u r e [- [ - ~ n i m a t e ] ] of t h e s t a n d a r d t h e o r y , and

[ +- [-Animate]] i s e q u i v a l e n t t o [ - [+Animate]].

CHAPTER THREE

THE EXPLANATORY FAILURE OF GRAMMATICAL FUNCTIONS

1. Two Hypotheses -- The question now arises whether we have merely identified a

class of principled l~exceptionstl to an otherwise valid paradigm

of correspondences, ltSub,ject of the S v and "Poss-NP, and

"0bJect of the Verbt1 and "of-NPtt - (much as we might identify a

class of principled exceptions to the rule of Passive); or

whether these verbs and nouns are not exceptional at all, and it

is the paradigm which is unjustified. The theoretical interest

of this question is whether lexical redundancy rules should

still be mappings between grammatical functions, but have access

to thematic information (at a cost); or whether thematic

functions are a more appropriate basis for stating these rules

(and subcategorizational correspondences). The latter approach

has been suggested by Anderson (1977), who proposes that

recurring correspondences of thematic and grammatical functions

in given syntactic frames be abstracted and stated as general

principles of (the) language, so that individual lexical rules

need no longer explicitly relate grammatical functions.

If lexical rules are mappings between grammatical

functions, we will apparently need two rules for the

nominalizations of transitive verbs, illustrated schematically

i. Subject V [-Experiencer] I 1 Objllf

POSS-NP N (of) NP

11. Subject V [+Experiencer]

1 1 Ob;.LI ( Prepositional N Poss-NP

Object

An alternative in line with the analysis suggested by Lakoff

(1970), Jackendoff (1975), and Wasow (1977), would dispense with

rule (I), subpart (11) above; it would derive NPs with

[+Experiencer] nsubjectsv by the "regular" rule for deadjectival

nominals, which can be illustrated as (2):

Subject (be) A (PP)

Notice however that even this solution must countenance rule

(li), which is thematically restricted.

Further, the rules which relate verbs and adjectival

passives must also contain thematic information. We have seen

t h a t whether o r n o t a v e r b h a s a n a d j e c t i v a l p a s s i v e i s n o t

e n t i r e l y u n p r e d i c t a b l e ; whi le v e r b s w i t h E x p e r i e n c e r o b j e c t s

c h a r a c t e r i s t i c a l l y have a d j e c t i v a l p a s s i v e s , t h e

[ +- [-Animate]] v e r b s e x c i t e , e x h a u s t , d e p r e s s , a g i t a t e and

s a t i s f y (which t a k e non-Experiencer o b j e c t s ) do n o t have

a d j e c t i v a l p a s s i v e s . The r e l e v a n t l e x i c a l r u l e must be s o s t a t e d

a s t o exc lude t h e l a t t e r c l a s s . We may t h e r e f o r e s a y t h a t i n

t h i s t h e o r y , l e x i c a l redundancy r u l e s ( i n g e n e r a l ) must t a k e

i n t o accoun t two k i n d s o f i n f o r m a t i o n : grammat ica l and

t h e m a t i c . 1 I s h a l l r e f e r t o a t h e o r y which s t a t e s a l e x i c a l r u l e

as a n o p e r a t i o n on grammat ica l f u n c t i o n s which i s s e n s i t i v e t o

t h e m a t i c i n f o r m a t i o n as t h e ? n i x e d u v e r s i o n o f t h e lfgrarnrnatical

f u n c t i o n s v h y p o t h e s i s of l e x i c a l r u l e s . (The "puren v e r s i o n o f

t h i s t h e o r y , where in l e x i c a l r u l e s a r e d e n i e d a c c e s s t o t h e m a t i c

i n f o r m a t i o n , h a s been shown t o be d e s c r i p t i v e l y i n a d e q u a t e . )

The t h e o r y I wish t o oppose t o t h e 9nixedw v e r s i o n of t h e

vgrammat ica l f u n c t i o n s v h y p o t h e s i s i s t h e " t h e m a t i c f u n c t i o n s w

h y p o t h e s i s . I n t h i s t h e o r y , l e x i c a l r u l e s relate argument

s t r u c t u r e s , and t h e s y n t a c t i c frames t h a t p a r t i c u l a r argument

s t r u c t u r e s are "mappedff i n t o a r e i n d e p e n d e n t l y s p e c i f i e d . Thus

t o accoun t f o r t h e d a t a c o n s i d e r e d s o f a r , w e s h a l l p o s t u l a t e a

r u l e which relates a v e r b w i t h t h e argument s t r u c t u r e (Causer ,

E x p e r i e n c e r ) t o a noun o r a n a d j e c t i v e w i t h t h e argument

s t r u c t u r e ( E x p e r i e n c e r , X ) , where XZCauser, and i s p o s s i b l y

n u l l .

*

I shall show that the "grammatical functions" hypothesis,

in either its l1pureW or its flmixed" version, imposes a

limitation on lexical rules that renders them incapable of

expressing the types of generalizations about

subcategorizational correspondences that need to be expressed.

But first, I wish to consider a model for lexical rules that has

recently been proposed by Wasow (1980). Wasow1s motivation for

including thematic functions in lexical rules is somewhat

different from ours, and he advocates the "mixed" theory that I

shall argue against. However, there are some interesting points

of convergence between his theory and the analysis we have so

far developed, which I shall point out.

2. Mador and Minor Lexical Rules. - - With the possibility of capturing distributional

regularities in the lexicon in the framework of Remarks, the

issue arose of the necessity for a class of structure-preserving

transformational rules in the grammar, distinct from lexical

rules. Wasow (1977) has described why a number of linguists

argued for the elimination of structure-preserving

transformations.

The primary motivation for structure-preserving transformations has been to account for regularities in co-occurrence restrictions. More specifically, when there are two syntactic constructions with the following three properties, then their relationship can be formulated as a structure-preserving transformation: (i)

they both can be generated by the rules of the base; (11) the morphological forms that appear in one construction are predictable from those that appear in the other; and (111) the co-occurrence restrictions in one construction are predictable from those in the other. Recently, several linguists have pointed out that it is, in general, also possible to relate constructions satisfying (1)-(111) by means of lexical redundancy rules. This is accomplished by isolating a key word in each construction (generally a verb) and expressing the regularities between the constructions in terms of the lexical entry for that word, especially the contextual features in the entry for that word. Freiden (1974, 1975) and Bresnan (1976) have suggested that all structure-preserving transformations can be reformulated as lexical redundancy rules. (Wasow 1977:328)

It has been suggested that such a reanalysis offers a

natural explanation for the fact that these rules preserve

structure; other advantages, such as reduced generative capacity

and a closer approximation to psychological reality, have also

been claimed in Its favor.

Initially, Wasow, in his well-known article

vTransformations and the lexiconw (1977), argued against the

elimination of structure-preserving transformations. The major

part of his article was devoted to an examination of the passive

in English, a paradigm case of a structure-preserving rule.

Wasow showed that there were in fact two rules of passive in

English, one the traditional rule which does not alter syntactic

category (the verbal passive), the other a rule which took verbs

as input to output past participial adjectives (the adjectival

passive). He illustrated that the two rules, though both

structure-preserving, differed in productivity and in conditions

on application. (Thus we saw in Chapter One that the adjectival

passive was lllocalll, in the sense that not every immediately

post-verbal NP could be preposed by this rule.) Wasow argued

that the differences between the verbal passive and the

adjectival passive were in fact the differences we would expect

between a transformational rule and a lexical rule. Thus, he

concluded that not all structure-preserving rules were lexical

rules.

The distinctions that Wasow pointed out between the verbal

passive and the adjectival passive were indisputable. What was

perhaps not equally evident was that these differences indicated

a transformational versus lexical dichotomy in

structure-preserving rules. The maintenance of a class of

transformational rules which were structure-preserving led to

the loss of an elegant demarcation between transformational and

lexical rules, and it left the structure-preserving property

unexplained. Acknowledging these problems, Wasow (1980) accepts

the structure-preserving property as the boundary between

lexical and transformational rules. The problem he addresses is

that of retaining the distinctions between a rule-type like the

verbal passive and a rule-type like the adjectival passive in a

framework wherein both these rules are lexical, and both are

stated as operations on grammatical relations. His solution is

to adapt the suggestions of Anderson (1977), and incorporate

thematic functions into lexical rules. He therefore makes the

f o l l o w i n g p r o p o s a l :

L e t u s suppose t h a t t h e f u n c t i o n a l s t r u c t u r e i n l e x i c a l e n t r i e s i s a s p e c i f i c a t i o n of which t h e m a t i c r e l a t i o n s shou ld be a s s i g n e d t o t h e e lements mentioned i n t h e s y n t a c t i c c o n t e x t . Then we may d i s t i n g u i s h two t y p e s of l e x i c a l r u l e s : t h o s e t h a t make r e f e r e n c e t o t h e m a t i c f u n c t i o n s and t h o s e t h a t do n o t . The fo rmer would cor respond t o r u l e s t h a t my e a r l i e r p a p e r c a l l e d l e x i c a l , and thJe l a t t e r t o t h o s e t h a t I c a l l e d t ransformat ions ' .

Un l ike t h e 1977 model, t h e 1980 model p r o v i d e s d e f i n i t i o n s of

grammat ica l r e l a t i o n s i n terms of " r a t h e r s u p e r f i c i a l s t r u c t u r a l

configuration^.^^ Thus Wasow c o n t i n u e s :

t h e d i s t i n c t i o n s my e a r l i e r p a p e r t r i e d t o c a p t u r e i n terms of t h e d i f f e r e n c e between s t r u c t u r a l r u l e s and r e l a t i o n a l r u l e s i s now t o be handled i n terms of t h e d i f f e r e n c e between ( s u p e r f i c i a l ) grammat ica l r e l a t i o n s and t h e m a t i c r e l a t i o n s .

Wasow t h u s p o s t u l a t e s two t y p e s of l e x i c a l r u l e s , which he

c a l l s 9 n a j o r n and "minoru l e x i c a l r u l e s . A t y p i c a l l e x i c a l r u l e

i s g i v e n t h e form ( n < - - - m ) . q i n d i c a t e s t h e morpho log ica l PU

e f f e c t , - P and Q a r e t h e grammat ica l c a t e g o r i e s o f i n p u t and

o u t p u t ( r e s p e c t i v e l y ) , and - n<---m - symbol izes t h e change from t h e

grammat ica l r e l a t i o n - m t o t h e grammat ica l r e l a t i o n - n. Major and

minor r u l e s a r e d i s t i n g u i s h e d i n t h e f o l l o w i n g ways:

a. Minor r u l e s s t i p u l a t e t h e t h e m a t i c r e l a t i o n of g, major r u l e s do n o t . b. I n major r u l e s , P and Q must be i d e n t i c a l . c . Only minor r u l e s may have i d i o s y n c r a t i c a l l y marked e x c e p t i o n s . (op. c i t . )

The verbal passive Is now a major lexical rule, which is stated

as follows: -en (I<---2). The adjectival passive is now a minor vv lexical rule, which is stated as follows: If 2=t(heme),

Note the reference to Theme in the adjectival passive rule.

The operations of the adjectival and the verbal passive are

otherwise here identical; both turn direct objects into

subjects. It is the thematic condition that precludes "indirect1'

objects (which are Goals) and wraisedll objects (which have no

thematic status) from undergoing the adjectival passive rule,

even when they occupy direct object (1.e. immediately

post-verbal) position. This is the wlocalnessll property that (as

we saw) distinguishes the adjectival from the verbal passive,

and the explanation here is essentially that suggested by

Anderson (1977).

Wasow1s main motivation for incorporating thematic

functions into (minor) lexical rules is thus a theoretical one;

it rests on the assumption that all structure-preserving rules

are lexical rules. If this assumption is denied, his case for

thematic functions is correspondingly weakened. What is needed

is empirical evidence to support the incorporation of thematic

functions into lexical rules. Our data provide this evidence. We

saw that the animate Experiencer objects of excite, exhaust,

agitate, depress and satisfy undergo the adjectival passive, but

that the inanimate obdects do not. If we tried to account for

this differentiation in terms of grammatical relations (as in

the 1977 model), we would be forced into the counterintuitive

position that the Experiencer objects of these verbs are direct

objects, but that the non-Experiencer objects are not.

To put the same point somewhat differently, there was some

initial plausibility to the view that the immediately

post-verbal NP in double object constructions, accusative-

infinitive constructions, and idiomatic constructions was not

the verb's direct object in the lexicon. The double object

construction V NP1 NP2 alternates with another construction V

NP2 to/for NP1, where the I1first objectw is a prepositional

object; the accusative-infinitive construction allows NPs like

there to follow the verb, showing that this Nobjectll has its

origin in the subject position of the complement clause; and the

idiom [[take]$advantage] of],+s possibly present as a unit in N the lexicon. The failure of such post-verbal NPs to undergo the

adjectival passive rule could therefore be conceivably due to

their non-direct object status. In the case of verbs like

excite, however, there is no such evidence for assigning

non-direct object status to inanimate objects at any stage. The

explanation here must appeal to thematic functions, irrespective

of whether the verbal passive is a transformational rule that

refers solely to structural properties, or a lexical rule that

refers to grammatical relations.

Notice that the verbal and adjectival passive again differ

in productivity. Verbal passives of the inanimate objects of

excite, etc., are possible:

( 3 ) a. The protons were excited by gamma rays.

b. The lever is depressed by the air pressure, and

water flows.

c. The soap solution is agitated by the machine's

rotary action, so that it lathers.

d. The non-distinctness requirement is satisfied by the

lexical entry, and lexical insertion goes through.

e. The country's resources were exhausted by

indiscriminate mining, resulting in dependence on

foreign oil.

Compare also (4)-(7), where only Experiencer objects appear as

subJects of an adjectival passive construction.

(4) a. John moved (touched) the stone.

b. *The stone seemed moved (touched).

(5) a. The story moved (touched) John.

b. John seemed (un)moved (touched) by the story.

(6) a. The authorities relaxed the regulations.

b. *The regulations seem relaxed.

(7) a. A shower and a shampoo relaxed John.

be John seems relaxed,

Moreover, our data show that reference to thematic

functions is a general property of minor lexical rules. The

characteristics of a minor lexical rule according to Wasow

(1980) are that it can alter syntactic category, and that it may

have idiosyncratic exceptions. On both these counts, the

nominalization rule is a minor lexical rule (it was its

idiosyncratic character that motivated the Lexicalist Hypothesis

in the first place). Thus it is significant that the

nominalization rule and the adjectival passive rule show a

common sensitivity to the Experiencer-non Experiencer

distinction, which the verbal passive rule is immune to. The

distinction between major and minor lexical rules is thus shown

to be consistent, and one that must be incorporated into a

framework where there are no structure-preserving

transformational rules. Wasowts model is unique in this respect

(as he notes, "Bresnants theory, as she has presented it so far,

says nothing about the major-minor rule distinctlon").2

Wasowls optimism concerning a satisfactory resolution of

the problem of identifying the thematic functions relevant to

lexical rules also turns out to be justified. He suggests lla

line of research (which) holds some promise," which is very

close to the method we pursued in Chapter Two. Noting that a

lack of rigorous justification for proposed thematic assignments

undermines the force of any explanation of wexceptionaln

behaviour, he observes that

(t)he ideal answer to this sort of charge would be a characterization of the semantic content of the thematic relations which Is precise enough to make their assignments as transparent to the intuitions of the native speaker as are control relations. However, given the failure of substantial previous efforts (e.g., Gruber (1965), Jackendoff (1976)) to achieve the requisite level of precision (see Hust and Brame (1976)), there is little cause for optimism on this score. An alternative means of lending substance to the thematic assignments could be provided by considering a wider range of minor rules. If I am correct in claiming that all such rules are sensitive to thematic relations, then each assignment will make predictions with respect to the operation of a number of different rules. These assignments would then predict a clustering of properties, and hence could not be said to be arbitrary.

It is precisely such a wclustering of propertiesv that was

seen above, that motivated the thematic function Experiencer.

Causative verbs with Experiencer objects were shown to

consistently have both an adjectival passive, and a "skewedn

nominalization subcategorizationally parallel to this adjectival

passive. This evidence was corroborated by evidence from "finerw

subcategorization, along the lines suggested by Chomsky (1964).

3. A Cross-Morphological Regularity - -

The question posed by the data in Chapter Two (repeated

here as (8), with the [ + - [+Animate]] subcategories of verbs

like excite included) was whether causative verbs with

Experiencer objects are "regularly exceptionaln in exhibiting a

llskewed" nominalization, subcategorizationally parallel to a

past participial adjective.

(8) a. agitate, amaze, amuse, annoy, astonish, bore,

confuse, delight, depress, disappoint, disgust,

dismay, distress, elate, embarrass, excite, exhaust,

fascinate, frustrate, humiliate, hurt, inspire,

interest, irritate, (dis )please, puzzle, relieve,

satisfy, surprise, vex

b. agitated (at, over), amazed (at), amused (at),

annoyed (at), astonished (at), bored (with),

confused (by), delighted (at), depressed (at),

disappointed (with), disgusted (at), dismayed (at),

distressed (at), elated (at), embarrassed (at),

excited (at), exhausted (by), fascinated (with),

frustrated (at), humiliated (by), hurt (by),

inspired (by), interested (in), irritated (at),

(d1s)pleased (at), puzzled (at), relieved (at),

satisfied (with), surprised (at), vexed (by)

c. agitation (at, over), amazement (at), amusement

(at), annoyance (at), astonishment (at), boredom

(with), confusion (*by), delight (at), depression

(at), disappointment (with), disgust (at), dismay

(at), distress (at), elation (at), embarrassment

(at), excitement (at), exhaustion (*by), fascination

(with), frustration (at), humiliation (*by), hurt

(*by), inspiration (*by), interest (in), irritation

(at), (d1s)pleasure (at), puzzlement (?at), relief

(at), satisfaction (with), surprise (at), vexation

(*by)

The theoretical interest of this question, we said, was whether

to adopt the "mixedw version of the llgrammatical functionsv

hypothesis of lexical rules, or the lfthematic functionsf1

hypothesis of lexical rules. A related question was whether the

nominalizations of these verbs should be related to the past

participial adjectives instead of to the verbs themselves, as

suggested by Lakoff (1970), Jackendoff (1975), and Wasow (1977).

I undertook to show that the verbs, adjectives and nouns under

consideration are not "irregularw in any way, and that it is the

requirement that lexical rules refer to grammatical functions

which prevents us from seeing the underlying regularity in the

observed pattern of subcategorizational correspondences.

In order to see the limitations of the wgrammatical

functions" hypothesis, let us briefly advert to the simplest

(and strongest) form of this hypothesis: the model that

Jackendoff (1977) and Wasow (1977) have in mind (call it the

"pure grammatical functionsw model). The prediction that this

model makes is that the subcategorization of a nominal N will

parallel the subcategorization of a predicate P - from which it is

morphologically derived. This may be schematized as in (91,

where the arrow indicates morphological derivation and the

dotted line indicates subcategorizational parallelism. The

predicate P is a verb in (gal, an adjective in (gb).

Examples (10a,b) and (lla,b) illustrate the schema above.

(10) a. John criticized the book.

b. John's criticism of the book

(11) a. The book is readable.

b. the book's readability

Now there may be other predicates to which the nominal N is

morphologically related, but from which it is not

morphologically derived. Consider for example the predicates

(be) critical and read. The subjects of these predicates (which

are not the morphological wbases't of criticism and readability)

may, or may not, correspond to the "subjectn of the nominal.

Thus we have (12)-(13):

(12) a. John was critical of the book.

b. John's criticism of the book.

(13) a. John read the book.

b. the book's readability

Observe that the subject (and of-complement) of be critical are - - parallel to the wsubjectll (and - of -complement) of criticism,

whereas it is the object of read that corresponds to the

"subjectw of. readability. But these correspondences do not

validate or invalidate (respectively) the model we are

discussing, for it makes no direct predictions in these cases.

Rather, we must first ascertain the morphological (and hence

subcategorizational) relation of - be critical to criticize and of

read to readable, since the latter predicates are the - morphological "basesI1 of the nouns. Thus, since the subject of

be critical is also the subject of criticize, it corresponds (as - expected) to the llsubjectll of criticism. But since it is the

object of - read that corresponds to the subject of readable, the

llsubjectsll of the verb and the,deadjectival noun are not

expected to be parallel; as indeed they are not.

Given a range of Ss and an NP with morphologically related

predicates, theref ore, the "pure grammatical functions l1 model

predicts a parallelism between only one of these Ss, and the NP.

The S that the NP1s subcategorizational frame is predicted to be

parallel to is that S whose predicate is the morphological

I1basew of the head noun. 3 Obviously, any claim that the NP1s

subcategorizational frame should parallel those of the whole

range of Ss with morphologically related predicates is

immediately refutable, since the Ss do not show such a

parallelism among themselves. The link with morphological

derivation is therefore a crucial one for this model.

This crucial dependence on the links provided by

morphological derivation is the reason why the problem of the

examples (8a-c) is not solved, for this model, by relating the

nouns in (8c) to the participles in (8b). Observe first that the

verbs in (8a) each have not one, but two adjectival

(participial) derivatives: present participial adjectives, and

past participial adjectives. The present participles take as

subject the subject of the verb, but the past participles take

as subject the object of the verb:

(14) a. The stories amused the children.

b. The stories were amusing.

c. The children were amused.

Given a noun amusement which is morphologically derived from

neither amusing nor amused, it is - a priori impossible to predict whether its "subjectw will correspond to the subject of - be

amusinq or of - be amused. This correspondence can only be

inferred from the subcategorizational correspondence of

amusement to (its morphological base) amuse. The prediction of

the "pure grammatical functionsw model is that the wsubjectw of

amusement will correspond to the subject of amuse, and

( t h e r e f o r e ) t o t h e s u b J e c t o f be amusing. The c a s e o f - amuse-amusement-(be) - amusing s h o u l d t h u s be p a r a l l e l t o t h a t of

c r i t i c i z e - c r i t i c i s m - ( b e ) - c r i t i c a l . I t i s p r e c i s e l y because

amusement l l v i o l a t e s n t h i s p r e d i c t i o n t h a t a p a r a l l e l i s m i s

observed i n t h e l l s u b j e c t s l l o f t h e nominal and t h e p a s t

p a r t i c i p i a l a d j e c t i v e ; t h i s p a r a l l e l i s m a r i s e s by v i r t u e of t h e

f a c t t h a t bo th t h e s e l l s u b j e c t s l l co r respond t o t h e v e r b ' s o b j e c t .

The a n a l y s i s which r e l a t e s amusement t o amuse$ a p p e a r s t o

accord w i t h t h e "pure1' v e r s i o n o f t h e l lgrammatical f u n c t i o n s v

h y p o t h e s i s , by r e s t o r i n g a p a r a l l e l i s m i n t h e l l s u b j e c t s l l of one

S and NP p a i r i n a paradigm o f m o r p h o l o g i c a l l y r e l a t e d

p r e d i c a t e s . But t h e p a r a l l e l i s m which i s t h u s r e i n s t a t e d i s

mere ly a n o b s e r v a t i o n , n o t a p r e d i c t i o n .

The p r o p e r t y of b e i n g a b l e t o e x p r e s s o n l y p a i r w i s e

r e l a t i o n s between l e x i c a l i t ems l1l inkedt1 by morpho log ica l

d e r i v a t i o n i s i n f a c t a g e n e r a l p r o p e r t y of l e x i c a l r u l e s a s

c u r r e n t l y f o r m u l a t e d . Thus t h e a c t i v e - p a s s i v e r u l e relates o n l y

a p a i r of S s , such a s ( 1 5 ) and ( 1 6 ) :

( 1 5 ) Someone r e a d t h e book.

( 1 6 ) The book was read.

I t does n o t ( d i r e c t l y ) r e l a t e (16) t o ( 1 7 ) o r (18) :

( 1 7 ) The book i s r e a d a b l e .

(18) The book reads well.

The choice of which pair of Ss to relate is obviously determined

by a (postulated) morphological derivation, which derives - be

read from transitive read and not from readable or intransitive -3

read.

Observe that the rules of the "mixed" version of the

"grammatical functionsH hypothesis are also subject to the same

restrictions. Such rules (cf. (1-2) above) may apply to finer

subcategories, by virtue of their containing thematic

information; but they may still express only pairwise relations,

the choice of the pair being determined by morphological

derivation. As a further illustration of this property, consider

again the formalism for minor lexical rules given by Wasow

(1980) : "if m=(some thematic function)&, then l@pQ(n<---m). l1 Such

a rule specifies a subcategorizational correspondence between - m and - n (e.g., and vsubjectll) in conjunction with a

morphological operation(P; it relates the subcategorization of a

pair of predicates such that one is morphologically derived from

the other. On the other hand, rules which relate only thematic

structures may predict a subcategorizational pattern in a range

of Ss and an NP. Further, they may apply to lexical items

exhibiting diverse patterns of morphological relationships.

(These points are illustrated below.)

I shall now show that there are generalizations about

subcategorizational correspondences which cut across different

morphological paradigms, and which cannot be expressed by

pairwise comparisons. Consider the four different morphological

paradigms schematically illustrated In (lga-d). The arrows

indicate morphological derivation, and the dotted lines indicate

observed subcategorlzational parallelism. Recall that the "pure

grammatical functionsft model predicts that the dotted line will

follow the arrow.

For the "pure grammatical functionsN model, the paradigms

represented by (lga) and (19d) are "irregular. Further, even

the vmixedv version of the wgrammatical functionsw hypothesis

cannot say anything about the relation indicated by the dotted

line in (lga) and (lgd). In order to express this relation, it

needs the presence of an "arrow," 1.e. a morphological

derivation. However, when the four sets of data exemplifying the

four schemata of (19) are represented in terms of argument

structures (thematic functions) alone -- ignoring morphological

derivation --, we obtain (as we shall see) the single general schema (20):

v (Causer, Experiencer)

Subject Object

N

(Experiencer, X )

POSS-NP PP

A

(Experiencer, X )

Subject PP

The generalization represented by (20) should be apparent.

Schema (19a) represents the relation of amuse to amusement

and amused (and the other examples in (8a-c)). It should be

clear that these data conform to the representation in (20).

Schemata (lgb-c) are exemplified in (21a-d).4

(21) a. The news saddened (gladdened, cheered, tired) Bill.

b. *The news1 sadness (gladness, cheerfulness,

tiredness) of Bill

c. Bill was sad (glad, cheerful, tired).

d. Bill's sadness (gladness, cheerfulness, tiredness)

In (21a) and (21d), as in (8a-c), the object of the verb

corresponds to the vsubJectw of the nominal. However, here we do

not readily see an instance of the "skewed1' paradigm, since the

nouns in (21d) are plainly deadjectival (as are the verbs sadden

and pladden). The ''pure grammatical functionsv model therefore

predicts the parallelism in (21c-d); the wsubjectw-object

correspondence in (21d), (21a) is to be inferred from the

relationship of the verbs in (21a) to the adjectives in (21~).

But observe that the same pattern of subcategorizational

correspondences is predicted by our hypothesis about

Experiencers. The verbs in (21a) are causative verbs with

Experiencer objects. The objects of the verb correspond to the

"subjectslt of the nominal and the subjects of the adjectival

predicates with Experiencer arguments, as expected. The

''thematic functions~ hypothesis thus allows us to treat (8a-c)

and (21a-d) as instances of the same pattern of

subcategorizational correspondences, namely (20).

The case of tire tire and tirednes is a particularly *' illuminating example of the problems created by morphological

derivation for the "pure grammatical functionsw model. The

adjective tired is morphologically a past participle, just like

the adjective exhausted (and the other adjectives in (8b)).

However, the noun tiredness is formed by -ness - attachment to the adjective, while the noun exhaustion (like the other nouns in

(8c)) is formed from the verb (suffixes like -ment -9 -(at)ion --'

characteristically attach to verbs, to form nouns). The ''pure

grammatical functionsu model therefore predicts that *the - walk's exhaustion - of Mary will be grammatical, but that *the - walk's

tiredness - of Mary will be ungrammatical. The wthematic

functions11 hypothesis, however, correctly predicts that the

semantically similar verbs tire and exhaust (strictly, the

subcategory of the latter with the feature [ + [+Animate]] 1,

and their adjectival and nominal derivatives, will have the same

pattern of subcategorizational correspondences.

Consider now the hypothetical situation wherein the

nominals in (8c) are subcategorizationally parallel to the verb,

1.e. where John's amusement -- of the children is grammatical. Such

a hypothetical paradigm would violate the generalization about

Experiencers. Our hypothesis thus not only accounts for (8a-c),

and generalizes it with (21a-d); it predicts that the language

would be more complex if the facts were otherwise.5 The llpure

grammatical functionsl1 model, on the other hand, predicts that

(8a-c) and (21a-d) should be non-parallel. Further, when this

prediction fails, this model is still unable to see (8a-c) and

(21a-d) as instances of the same subcategorizational pattern.

The morphology tells us that in (8a-c), there is a relation

between an S with a verbal predicate, and an NP; while in

(21a-d), there is a relation between an S with an adjectival

predicate, and an NP. The only way to reconcile the two

paradigms is to ignore the morphology, postulate a lexical llgapl'

where the nominalizations of the verbs (8a) should be, and

consider both the nouns in (8c) and (21d) as deadjectival.

Recall that this is the solution proposed by Lakoff, Jackendoff,

and ~ a s o w . 6

Consider next t h e schema ( l g d ) , i l l u s t r a t e d i n (22a-d).

( 2 2 ) a . The news t e r r i f i e d ( h o r r i f i e d ) Mary.

b. * the t e r r o r ( h o r r o r ) of Mary by t h e news

c. Mary was t e r r i f i e d ( h o r r i f i e d ) a t t h e s i g h t of

blood.

d. Mary ' s t e r r o r ( h o r r o r ) a t t h e s i g h t of blood

I n (22a) we have verbs which a r e morphological ly der ived from

t h e nouns i n (22d1.7 The same arguments t h a t mot ivated t h e '!pure

grammatical f u n c t i o n s H model f o r verbs and deve rba l nouns ought

t o apply I n t h e c a s e of nouns and denominal verbs . Observe t h e

p a r a l l e l i s m s i n (23) - (24) :8

(23 ) a. J o h n ' s apology t o B i l l

b. John apologized t o B i l l .

(24) a. J o h n ' s s c r u t i n y (summary) o f t h e a r t i c l e

b. John s c r u t i n i z e d (summarized) t h e a r t i c l e .

I n (22a-d), however, t h i s p a r a l l e l i s m does n o t ob t a in ; t h e

vsubjec t f l of t h e nominal corresponds t o t h e o b j e c t of t h e verb.

This i s a problem f o r t h e "pure grammatical f u n c t i o n s w

hypothes i s . If (22d) i s r e l a t e d t o (22a) on t h e s t r e n g t h of t h e

morphological evidence, t h e grammatical f u n c t i o n s of t h e S and

t h e NP do no t co inc ide . If (22d) i s r e l a t e d t o (22c ) , t h e

subcategorizational relation conflicts with the morphological

relation.9 Observe that once again, the thematic function

Experiencer is crucially involved. The object in (22a), and the

"subjectw in (22c-d), are Experiencers. The paradigm (22a-d)

thus generalizes with (8a-c) and (21a-d).

But now the "pure grammatical functions" model has not only

made a consistently wrong prediction in two different

morphological paradigms where Experiencers have appeared; in the

case of (22a-d), we have a strong intuition that the paradigm in

fact is regular. It seems to me that this intuition has its

roots in two facts: first, the subcategorization of the noun is

"given," since it is morphologically basic; second, verbs ending

in -= are causative verbs. In general, the subjects of intransitive (non-causative) constructions appear as the objects

of transitive (which Includes causative) constructions (compare

the Theme-Rule of Anderson (1977) ). The paradigm (22a-d) reveals

an interesting fact: when the relevant semantic factors are

sufficiently overt as to be available to unanalyzed intuition,

our expectations about the subcategorizational correspondences

in Ss and NPs depend on perceived semantic roles, and not on a

simple correspondence in grammatical functions.

We have thus far examined four morphological paradigms

where Experiencers are involved, and the same pattern of

subcategorizational correspondences has been seen to hold in all

four cases. The underlying unity in the subcategorizational

p a t t e r n emerged o n l y when t h e s y n t a c t i c frames of t h r e e

p r e d i c a t e s were compared a t once: a v e r b a l , a n a d j e c t i v a l and a

nominal p r e d i c a t e . The p a i r w i s e comparison o f S s and NPs canno t

r e v e a l such r e g u l a r i t i e s .

But we have s e e n t h a t such a p a i r w i s e comparison w i l l be a n

e s s e n t i a l f e a t u r e of any v e r s i o n o f t h e vgrammat ica l f u n c t i o n s H

h y p o t h e s i s . Thus c o n s i d e r how t h e wmixedu v e r s i o n of t h i s

h y p o t h e s i s w i l l d e a l w i t h t h e t h r e e morpho log ica l paradigms we

have d i s c u s s e d , To d e a l w i t h d e v e r b a l nomlnals s o as t o

d i s t i n g u i s h t h e d a t a o f (8a-c) ( i l l u s t r a t i n g t h e schema ( 1 9 a ) )

from t h e %ormalV c a s e s , it w i l l p o s t u l a t e t h e two "marked1!

r u l e s ( l i , ii) ( r e p e a t e d h e r e a s ( 2 5 ) ) :

( 2 5 ) 1- S u b j e c t V [ " f z i e n c e r ]

POSS-NP N ( o f ) NP

ii. S u b j e c t V Objec t 1 I [+Exper iencer ]

( P r e p o s i t i o n N PO&-NP -a1 O b j e c t )

I n t h e c a s e o f (21a-d) (illustrating t h e schemata ( l g b - c ) ) , i t

w i l l have no c a u s e t o p o s t u l a t e a Itmarked" r u l e men t ion ing t h e

t h e m a t i c f u n c t i o n E x p e r i e n c e r , s i n c e - a l l s u b j e c t s o f a d j e c t i v a l

p r e d i c a t e s co r respond t o t h e l l s u b j e c t s w of d e a d j e c t i v a l

nominals.10 The required rule is (2) (repeated here as ( 2 6 ) ) :

(26) Subject

In the case of denominal verbs, in order to distinguish the

"skewedN correspondence exhibited by (22a-d) from the vnormaln

correspondence in (23) and (24), the theory will need two

"markedI1 rules, which we may represent as (27) :

Subject V

Object V

It should be apparent that these rules do not express the

generalization we have noticed; nor do they constitute any kind

of explanation of the phenomena under consideration.

Let us conclude the discussion with a rather nice bit of

evidence for the Experiencer hypothesis (as against the pairwise

comparison approach). Among the examples in (8a-c) is the

verb-adjective-nominal triplet delight, delighted (at), - delight

(at) - :

(28) a. The news delighted Mary.

b. *the news1 delight of Mary

c. Mary was delighted at the news.

d. Mary's delight at the news

From (28a-d) we must conclude that delight Is one of the nouns

exhibiting the "skewedw pattern, being subcategorizationally

parallel to the participial adjective. However, there is an

intransitive verb delight, and this verb has a corresponding

nominalization:

(29) a. Mary delights in tormenting small insects.

b. Mary's delight in tormenting small insects

The wsubjectw of the noun delight Is thus at the same time

parallel and non-parallel to the subjects of the verbs delight.

The facts, however, are straightforward. The "subject of the

noun is an Experiencer; it corresponds to the subjects of those

Ss in which Experiencers appear as subjects.11

4. The Thematic-Syntactic Mapping --

There is thus a variety of evidence to show that the

parallelism in the grammatical functions of the S and the NP

postulated by Jackendoff is unjustified, and the proposal for

generalizing the projection rules of verbs and deverbal nouns

fails. One task which remains is to specify a general method of

enforcing the shared selectional restrictions of lexically

related words.12 The other task is to specify the formalism for

lexical redundancy rules under the "thematic functionsv1

hypothesis.

Let us consider in some detail the proposals of Anderson

(l977), to which I have made brief references earlier. Anderson

observes that there are recurrent regularities within a

language, and perhaps across languages, in the syntactic

positions in which certain thematic functions are

characteristically realized. He suggests that the grammar

provide for a statement of these regularities in the

thematic-syntactic mapping. Two such regularities he notes are

the preference of Agents for subject position, and the

occurrence of Themes as subjects of intransitive verbs or

objects of transitive verbs; and he informally designates these

regularities as the "Theme-Rulen and the "Agent-Rulew. Note that

such "mappingw rules would extract a core of regularities from

thematic-syntactic correspondences which, in existing

frameworks, are specified for individual lexical entries. Thus

Jackendoff (1972, section 2.4) assumes an an indexing procedure

for correlating syntactic positions and thematic functions, in

his illustrative examples of lexical entries. Similarly, Wasow

(1980) gives equations for thematic-syntactic correspondences,

e.g., "l( 1. e. subject )=a(gent)." The rules suggested by Anderson

do not therefore result In any ad hoc extension of the grammar. -- On the contrary, they simplify lexical entries, by allowing

thematic-syntactic equations or indices to be omitted from the

entry if the information contained in them is entirely regular

and predictable. 13

Assuming, then, that the grammar specifies a Theme-Rule and

an Agent-Rule, Anderson illustrates how this affects the minor

lexical rule relating transitive and intransitive verb pairs.

Jackendoff (1975) proposes the rule (30) :

Anderson points out that

+NP1

NP1 W A

. . .this rule . . . explicitly establishes the fact that the NP in Direct Object position In a clause with break(tr.) corresponds semantically to the NP in Subject m i o n In a clause with break(intr.). In light of the Theme-Rule, however, we can see that this association is exactly the natural one; and that it need not be stated as part of the lexical relation at all. In fact, neither verb need contain any explicit association of particular NP in its syntactic environment with particular positions in the semantic representation, for these will follow directly from the Theme-Rule and the Agent-Rule. (Anderson 1977 :368)

~-----t

NP2 CAUSE (NP1 W)

In this model, we may represent the relation between

transitive and intransitive verbs as follows:

We may now state the selectional restrictions of intransitive

' tv

tNP NP

(Agent, Theme)

+NP

(Theme)

and transitive verbs on the thematic functions Agent and Theme,

and the syntactic positions corresponding to these functions

will be automatically restricted by the Agent-Rule and the

Theme-Rule, which (thus) function as projection rules.14 Let us

-

illustrate this with the verbs break:

/brAk/

+v +NP

(Theme 1

+breakable

( Agent, Theme )

+concrete +breakable

If such a thematic-syntactic mapping can also be specified for

NPs, the same mechanism which enforces shared selectional

restrictions in the domain of the S, for Ss with lexically

related predicates, can be extended to account for the

subcategorizatlonal correspondences In Ss and NPs. As our

recognition of thematic functions is sharpened, we may postulate

more such functions; and there may be corresponding

modifications in rules like the Theme-Rule. We have already

noted that the notion "Themew is too gross for the statement of

lexical redundancies and subcategorizational correspondences.

The thematic-syntactic mapping (we expect) will typically be a

many-to-one mapping, since the number of syntactic positions

available will (likely) be much smaller than the number of

thematic functions.

We may conceptualize the mapping from argument structures

to syntactic frames in the following way. Let there be a

hierarchy of grammatical functions in the S and the NP, and a

hierarchy of thematic functions, approximately as in (33):15

(33) Thematic functions Grammatical functions

Causer 1 Subject/Poss-NP

Theme, Experiencer 2 Object/of-NP - X 3 Prepositional phrase

(X#Causer or Theme/Experiencer)

Let us hypothesize that the highest available thematic function

is mapped on to the highest available grammatical function. This

means that if there is a Causer in the argument structure,

Experiencer/Theme has the rank 2; if there is no Causer, it has

the rank 1. Thus given an argument structure (Causer,

Experiencer/Theme), and a syntactic frame Subject/Poss

NP-Object, Causer will occupy Subject or Poss-NP position, and

Experiencer/Theme will occupy Object or or-NP position. Given an

argument structure (Experiencer, X) or (Theme, X), (where there

is no Causer), and a frame Subject/Poss NP-PP, the Experiencer

or Theme will occupy the Subject or Poss-NP position. Note that

in the NP, the Poss-NP position is optional. In this case, if we

have the syntactic positions of-NP and PP, and an argument - structure (Experiencer,X) or (Theme,X), the Experiencer or Theme

will occupy the (higher) - of-NP position, and - X will go to the PP

pos i t ion.

We can now give a lexical entry the following schematic

The mapping specification correlates thematic functions with

syntactic positions. We assume that where this mapping is

predictable from the hierarchy in (33), it need not be stated as

part of the lexical entry.

representation. (We assume that each lexical

specified, as in the framework of Jackendoff

Given such a hierarchy, the lexical rules relating verbs,

nouns and adjectives can be viewed simply as rules relating

argument structures, for the subcategorizational correspondences

they produce will be predictable. In my attempts to formulate

(34) - 1. /phonological representation/

2. lexical category

3. subcategorization frame

4. semantic representation, 1. e

argument structure in terms of

thematic functions

5. mapping specification -

t h e r u l e s i n t h e s e terms, I s h a l l be assuming t h a t t h e r e a r e

s e p a r a t e morpho log ica l and semant ic redundancy r u l e s , and t h a t

t h e same semant ic r u l e can be p a i r e d w i t h more t h a n one a f f i x . I

s h a l l j u s t i f y t h i s s e p a r a t i o n i n t h e n e x t c h a p t e r , and show how

t h e s e m a n t i c r u l e s may be a c c e s s e d by r u l e s of a f f i x a t i o n .

L e t u s suppose t h a t i n t h e s i m p l e s t c a s e , t h e r e a r e r u l e s

l i k e t h e f o l l o w i n g :

( 3 5 ) For any v e r b w i t h a c e r t a i n argument s t r u c t u r e , form a

noun w i t h t h e same argument s t r u c t u r e .

(e .g. c r i t i c i z e - c r i t i c i s m , e x c i t e - e x c i t a t i o n )

( 3 6 ) For any a d j e c t i v e w i t h a c e r t a i n argument s t r u c t u r e ,

form a noun w i t h t h e same argument s t r u c t u r e .

(e .g .sad-sadness , - p u r e - p u r l t y , c a l l o u s - c a l l o u s n e s s )

(37) For any noun w i t h a c e r t a i n argument s t r u c t u r e , form a n

a d J e c t i v e w i t h t h e same argument s t r u c t u r e .

( e . g. f u r y - f u r i o u s , g l o r y - g l o r i o u s )

L e t us c a l l r u l e s l i k e ( 3 5 ) - ( 3 7 1 , r u l e s of I n h e r i t a n c e . L e t us

f u r t h e r h y p o t h e s i z e t h a t t h e o u t p u t of t h e r u l e s r e l a t i n g

argument s t r u c t u r e s i s s u b j e c t t o f i l t e r s , which e x p r e s s t h e

n o t i o n w p o s s i b l e argument s t r u c t u r e w f o r t h e o u t p u t c a t e g o r y ;

and t h a t t h e r e i s a ( p o s s i b l y language s p e c i f i c ) f i l t e r t h a t

r u l e s o u t nouns w i t h t h e argument s t r u c t u r e (Causer ,

E x p e r i e n c e r ) :

( 3 8 ) * N , (Cause r , E x p e r i e n c e r )

T h i s w i l l f i l t e r o u t t h e o u t p u t o f r u l e (35) f o r v e r b s wi th t h i s

argument s t r u c t u r e . Now t h e r e i s a r u l e of D e c a u s a t i v i z a t i o n ,

which we need f o r r e l a t i n g c a u s a t i v e v e r b s t o t h e i r p a s t

p a r t i c i p i a l a d j e c t i v e s . Suppose we s a y t h i s r u l e is more

g e n e r a l , and t h a t i t r e l a t e s v e r b s t o nouns as w e l l a s t o

a d j e c t i v e s . We can f o r m a l i z e t h i s g e n e r a l i z e d r u l e i n terms of

t h e f e a t u r e sys tem of Chomsky (1970) , where v e r b s have a f e a t u r e

[-N], and a d j e c t i v e s and nouns s h a r e a f e a t u r e [+N]. The

r e q u i r e d r u l e i s ( 3 9 ) :

( 3 9 ) [-N (1.e. V)

L ( ~ a u s e r , E x p e r i e n c e r ) ( E x p e r i e n c e r , (X) ) ]

T h i s r u l e w i l l p roduce a d j e c t i v e s and nouns l i k e amused and

amusement from amuse.

I n a d d i t i o n , we know t h a t t h e r e i s a r u l e of

C a u s a t i v i z a t i o n t h a t produces v e r b s from nouns and a d j e c t i v e s

( h o r r o r - h o r r i f y , &-sadden).

L(~x~eriencer, (X)

tNP NP

(Causer, Experiencer) 1 Notice now that rules (40) and (39) are inverses of each other,

and that rules (36) and (37) would be inverses of each other.

Moreover, corresponding to rule (35), there is an inverse rule

which retains the argument structure of the noun for the verb

(summary-summarize, scrutiny-scrutinize). In a theory wherein

the same rule specifies morphological and semantic information,

such inverse rules must be kept separate, in order to preserve

information about the "directionw of morphological derivation. I

have indicated, however, that I shall argue for a distinct body

of affixation rules. These affixation rule8 will be

unidirectional, and will supply the necessary information "is

derived from." This allows us to give the semantic rules simply

as bidirectional rules, expressing the notion "1s related tow.

That is, the language appears to have rules which relate the

argument structures of nouns, verbs, and adJectives, which are

neutral with respect to the direction of morphological

derivation.

We may therefore give the following semantic rules. I

indicate below each rule the lexical items related by it.

(41) Causativization-Decausativization

+ N ( i . e . N o r A ) -N ( 1 . e . V)

( P P ) 1 <----> kNP NP

( E x p e r i e n c e r , ( X ) ) (Cause r , E x p e r i e n c e r ) 1 amusement, amused amuse h o r r o r , h o r r i f i e d h o r r i f y t i r e d t i r e s a d sadden

( 4 2 ) I n h e r i t a n c e 1

! NP 1 <---- > IN +NP NP

(Argument, Argument) Argument)

c r i t i c i z e e x c i t e summarize

c r i t i c i s m e x c i t a t i o n summary

( 4 3 ) I n h e r i t a n c e 2

! I I

A

+NP ( P P )

(Argument, (Argument) ) (Argument, (Argument) )

s a d t i r e d f u r i o u s

s a d n e s s t i r e d n e s s f u r y

P o s t s c r i p t : Thematic Func t ions

The s t r a t e g y I have pursued f o r i d e n t i f y i n g a t h e m a t i c

f u n c t i o n E x p e r i e n c e r o f r e l e v a n c e t o l e x i c a l r u l e s h a s been t o

b e g i n w i t h a n examina t ion of s u b c a t e g o r i z a t i o n a l

co r respondences , r a t h e r t h a n w i t h a t h e o r y of t h e m a t i c

f u n c t i o n s . C o n s i d e r i n g t h e p r e s e n t l a c k of a c o n s t r a i n e d t h e o r y

i n t h i s area, t h i s i s pe rhaps t h e r i g h t s t r a t e g y . However, I

s h a l l i n t h i s s e c t i o n make a n a t t e m p t t o i n t e g r a t e t h i s t h e m a t i c

f u n c t i o n i n t o t h e f a m i l i a r s y s t e m of Gruber (1965) and

Jackendoff (1972,1976) , and p o i n t o u t some problems t o be

r e s o l v e d i n t h i s a r e a .

Jackendof f ( 1 9 7 6 ) , f o l l o w i n g Gruber ( 1 9 6 5 ) , h y p o t h e s i z e s

t h r e e b a s i c c l a s s e s of p r e d i c a t e s : Mot ional , P u n c t u a l and

D u r a t i o n a l . T h e i r schemat ic r e p r e s e n t a t i o n s are g i v e n below.

( 4 4 ) a. G O ( x , y , z ) ( M o t i o n a l )

b. BE ( x , y ) ( P u n c t u a l )

c. STAY ( x , y ) ( D u r a t i o n a l )

I n a d d i t i o n , t h e n o t i o n of c a u s a t i o n i s e x p r e s s e d by t h e

f u n c t i o n s i n ( 4 5 ) :

( 4 5 ) a. CAUSE ( x , e ( v e n t ) )

b. LET ( x , e ( v e n t ) )

We thus a r r i v e a t t h e fo l lowing familiar d e f i n i t i o n s .

( 4 6 ) a - Causer: 1 6 ~ i r s t argument of CAUSE o r LET

b. Theme: F i r s t argument of GO, BE o r STAY

c. Source: Second argument of GO

d. Goal: Third argument of GO

e . Location:17 Second argument of BE o r STAY

Our f i r s t t a s k i s t o r e f i n e t h e s e f u n c t i o n s , f o r we have

s t r e s s e d t h a t t h e system i n (46) does no t p rov ide enough

in format ion f o r l e x i c a l r u l e s . For i n s t a n c e , t h e sen tences - t h e

a i r - p r e s s u r e depressed - t h e l e v e r and -- t h e - news depressed Mary

would i n t h i s system have i d e n t i c a l semant ic r e p r e s e n t a t i o n s :

( 4 7 ) CAUSE(the a i r p r e s s u r e , GO(the l e v e r , BE(the l e v e r , not

dep re s sed ) , BE(the l e v e r , d e p r e s s e d ) ) )

1. e. dep re s s (Causer, Theme)

(48) CAUSE( t h e news, GO(Mary, BE(Mary, n o t dep re s sed ) ,

BE(Mary, d e p r e s s e d ) ) )

i. e. dep re s s (Causer, Theme )

We might e f f e c t t h e requi red ref inement by t a k i n g i n t o

account t h e ~ l o c a t l o n a l mode" t o which t h e f u n c t i o n s GO and BE

must be r e s t r i c t e d i n (47) and (48 ) . Jackendoff (1976:102)

p o s t u l a t e s u l o c a t i o n a l modesff such a s P o s i t i o n a l , Posses s iona l

and I d e n t i f i c a t i o n a l , which can be s p e c i f i e d as r e s t r i c t i v e

markers on GO, BE and STAY :

The marker Positional affixed to a semantic function . . . indicates that the Location or Source or Goal of that function specify claims about where the Theme is; the marker Possessional indicates that they specify claims about whose the Theme is. . . . A parameter Identificational . . . indicates that the Location or Source or Goal of the function to which it is affixed specify claims about what the Theme is.

The attractiveness of this proposal lies in the claim that

it is not accidental that a verb which can locate entities in

the physical domain can also locate entities in the abstract

domains of possession or identification: "in the simplest case,

the verb stays fundamentally the same, changing only the

restrictive modifier from one locational mode to anotherv (op.

cit.:103 -104). Thus Jackendoff suggests that the verbs turn in

the - coach turned into - driveway and the - coach turned into a -- pumpkin are both specified for a semantic function GO, but that

the locational modes differ (Positional in the former case,

Identificational in the latter). Similarly, keep in keep - the

book on the shelf and keep in keep the book are treated as --- -- differing in the modes Positional and Possessional, while

sharing a function STAY.

Since we have noticed a similar generalization in the

semantic fields of verbs like depress, a consideration of the

relevant locational modes would appear to be an appropriate

starting point for integrating the notion Experiencer into this

system. We may f i r s t r e v i s e t h e r e p r e s e n t a t i o n (47) t o ( 4 9 ) (we

w r i t e G O ( x , BE(x,not y ) , BE(x,y)) as BECOME y ( x ) , t o d i s t i n g u i s h

these two-place p r e d i c a t e s f rom p r e d i c a t e s l i k e ~ i v e ) :

( 49) CAUSE( t h e a i r p r e s su re , BECOME-posit DEPRESSED( t h e

l e v e r ) )

To s i m i l a r l y r e v i s e ( 4 8 ) , we must determine t h e a p p r o p r i a t e

l o c a t i o n a l mode. But t h e r e l e v a n t l o c a t i o n a l mode aga in appears

t o be P o s i t i o n a l . I n s ay ing t h e news depressed Mary, we a r e no t -- say ing - who Mary belongs t o o r - what she is; r a t h e r , we a r e aga in

s p e c i f y i n g h e r p o s i t i o n , but t h i s time i n terms of a s c a l e of

emotions: i n terms of t h e world wi th in .

Let us then p o s t u l a t e two new l o c a t i o n a l modes, Exte rna l

and I n t e r n a l , and add t h e s e as r e s t r i c t i v e markers t o GO-posit:

GO-posit. e x t ( e r n a l ) , and ~ o - ~ o s i t . i n t ( e r n a l ) . 18 We now r e v i s e

(49) t o (501, and (48 ) t o ( 5 1 ) .

(50) CAUSE ( t h e a i r p r e s s u r e , BECOME-posit.ext DEPRESSED(the

l e v e r ) )

(51 ) CAUSE ( t h e news, BECOME-posit.int DEPRESSED(Mary))

We may now d e f i n e what we have been c a l l i n g Exper iencer as t h e

f i r s t argument of GO-posit. i n t . , BE-posit. i n t , o r

STAY-posit. i n t .

The a n a l y s i s s o f a r sugges t s t h a t themat ic f u n c t i o n s may be

d e f i n e d a t two l e v e l s of s p e c i f i c i t y . On a f i r s t , b r o a d e r ,

l e v e l , they may be d e f i n e d as i n ( 4 6 ) ) w i t h r e f e r e n c e t o on ly

one of t h e f i v e semant ic f u n c t i o n s . A t t h i s l e v e l , t h e f u n c t i o n

Exper iencer was i d e n t i f i e d mere ly a s a Theme. On a second,

f i n e r , l e v e l , they may be d e f i n e d w i t h r e f e r e n c e t o t h e semant ic

f u n c t i o n s , as w e l l as t o t h e l o c a t i o n a l mode t o which t h e s e

f u n c t i o n s are r e s t r i c t e d . T h i s i s t h e l e v e l which d e f i n e s

n o t i o n s l i k e P o s s e s s o r o r E x p e r i e n c e r .

We have s e e n t h a t t h i s f i n e r l e v e l i s e s s e n t i a l f o r t h e

s t a t e m e n t of l e x i c a l r u l e s and s e l e c t i o n a l r e s t r i c t i o n s . The

q u e s t i o n t h e n a r i s e s i f t h e b r o a d e r l e v e l i s n e c e s s a r y a t a l l .

One m o t i v a t i o n f o r r e t a i n i n g t h i s l e v e l would be i f i t s u f f i c e s

f o r t h e t h e m a t i c - s y n t a c t i c mapping, and t h e r e i s some e v i d e n c e

t h a t t h i s might be t h e c a s e . Thus w e see t h a t w i t h r e s p e c t t o

t h i s mapping, E x p e r i e n c e r s a p p e a r t o behave l i k e Themes. They

occupy t h e o b j e c t p o s i t i o n of c a u s a t i v e v e r b s , s h a r i n g t h i s

p o s i t i o n w i t h (non-Exper iencer) Themes. Again, t h e r e a r e a few

c a s e s e x h i b i t i n g a cor respondence between a n E x p e r i e n c e r - o b j e c t

and a n E x p e r i e n c e r - s u b j e c t , which a p p e a r t o be s u b c a s e s of

Anderson 's Theme-Rule:

(52) a . Tormenting s m a l l i n s e c t s d e l i g h t s Mary.

b. Mary d e l i g h t s i n t o r m e n t i n g small i n s e c t s .

(53) a . H e w o r r i e s h i s mother.

b. H i s mother w o r r i e s ( a b o u t h im) .

We may anticipate that as our recognition of thematic functions

is sharpened, there will be a number of cases of lfcoincidencev

in the thematic-syntactic mapping. Now if we find that some set

of refined thematic functions tl, . . . tn which behave similarly with respect to this mapping can also be subsumed

(from the semantic point of view) under one broad thematic

function, the coincidences can be explained, and the statement

of the mapping rules simplified, if the broad level is retained.

This must, however, remain a very tentative proposal. For I

must also note some counterevidence against the inclusion of

Experiencer under Theme. Consider (54b):

(54) a. John touched the table.

b. The story touched John.

According to Gruber (1965:37) and Jackendoff (1972:43-44), the

verb touch takes Theme as subject, and Location as object. But

we have seen that the object of touch in (54b) is an

Experiencer. The behaviour of touch with respect to the

adjectival passive rule is entirely parallel to that of move:

(55) a. *The table seemed touched by John.

b. John seemed touched by the story.

(56) a. John moved the stone.

b. The story moved John.

c . *The s t o n e seemed moved by John.

d. John seemed moved by t h e s t o r y .

But -3 move u n l i k e t o u c h , i s c o n s i d e r e d t o have a Theme as i t s

o b j e c t . Thus we a r e f o r c e d t o say t h a t move h a s a n

Experiencer/Therne o b j e c t , whi le t o u c h h a s a n

Exper iencer /Loca t ion o b j e c t . 1 9

There i s a d e e p e r q u e s t i o n h e r e , namely whether t h e

t h e m a t i c s t r u c t u r e s of p r e d i c a t e s of emotion, c o g n i t i o n ,

p o s s e s s i o n , and s o on , can be reduced t o , and r e s t a t e d i n te rms

o f , t h e t h e m a t i c s t r u c t u r e s of p o s i t i o n a l p r e d i c a t e s . If t h e

t h e o r y m a i n t a i n s t h a t t h i s can be accompl ished, i t must o f f e r us

s t r i c t g u i d e l i n e s f o r t h e m e t a p h o r i c a l i n t e r p r e t a t i o n o f

n o n - p o s i t i o n a l p r e d i c a t e s i n terms of a w l o c a l i s t i c w system. I n

t h e absence o f such g u i d e l i n e s , t h e sys tem i s open t o abuse .

As an i l l u s t r a t i o n of t h e problem, c o n s i d e r t h e t r e a t m e n t

of p r e d i c a t e s of p o s s e s s i o n i n t h i s t h e o r y . Jackendof f

(1976:101) s tar ts o u t w i t h a sys tem where in t h e p o s s e s s e d i t e m

i s a lways Theme, and t h e p o s s e s s o r i s L o c a t i o n ( f o r s t a t i c

p r e d i c a t e s ) , o r Source o r Goal ( f o r dynamic p r e d i c a t e s ) . Thus we

have t h e t h e m a t i c a s s ignments i n ( 5 7 ) :

(57) a.

The book belonged t o t he l i b r a r y .

possessed

THEME

b.

pos ses so r

LOCATION

Max owned an iguana.

possessor possessed

LOCATION THEME

B i l l had no money.

possessor possessed

LOCATION THEME

However, Jackendoff ( fo l lowing up an obse rva t ion of G r u b e r l s )

n o t e s t h a t l l there i s a s o r t of converse of Possess ive l o c a t i o n

i n t h e fo l lowing exp res s ionsw (op. c i t . : 1 3 4 ) :

(58) a. Nelson r a n o u t of money.

b. A r i i s i n t h e money.

c. Fred came i n t o a l o t of money.

Reca l l t h a t Theme i s in formal ly c h a r a c t e r i z e d as the e n t i t y t h a t

moves, o r whose l o c a t i o n i s a s s e r t e d . By t h i s c r i t e r i o n , t he

Themes i n (58) seem t o be Nelson, - Ari and - Fred. Jackendoff

t h e r e f o r e p o s t u l a t e s a new l o c a t i o n a l mode - Poss l . I n t h i s

l o c a t i o n a l mode, t h e p o s s e s s o r i s Theme, and t h e possessed i t em

is Location (or Source or Goal).

Now in cases like (58a) and (58c), the flmovement" of the

subject away from or towards the money is perhaps fairly

obvious, and the inverse locational mode Possf is perhaps

identifiable.20 But the undesirable consequences of postulating

inverse locational modes are seen in the following analyses of

Ostler (1979).

Ostler attempts to develop a system of "linking rules"

(essentially our mapping rules) for argument structures and

syntactic frames. He notes that for a verb with the argument

structure (Possessed, Possessor), two syntactic realizations are

possible. In some frames, the possessed woutranksv the possessor

(the -- book belongs to the library). In other frames, the -- possessor "outranksw the possessed (Mary has/owns/possesses the

book). - A similar alternation of ranking is seen in the Dative

alternation: - John pave -- a book(possessed) - to Mary (possessor),

John gave Mary (possessor) 5 book (possessed). The solution - - Ostler proposes is the following. We know from Jackendoff's

analysis that there are two locational modes, Poss and Possl. We

may now say that belon4 - to is a Poss verb, while

have/own/possess are Posst verbs; and always require Theme to

outrank Location/Source/Goal. We thus arrive at the following

assignment of thematic functions, which may be compared with

those in (57) above. C

(59) a.

The book belonged to the library.

possessed

THEME

b.

possessor

LOCATION/GOAL

Max owned an iguana.

possessor possessed

THEME LOCATION/GOAL

C.

Bill had no money.

possessor possessed

THEME LOCATION/GOAL

In this system, Dative alternation verbs are simply specified

for both Poss and Possl modes. For example:

(60)

a.

John gave Mary a book.

possessor possessed

THEME LOCATION/QOAL

John gave a book to Mary.

possessed possessor

THEME LOCATION/GOAL

But this solution robs thematic functions of any semantic

content,21 without gaining anything by way of explaining the

facts. 22

In view of the latitude that the theory offers with regard

to the semantic content of notions like Theme, such notions are

at present of little value for formalizing lexical rules. In

conclusion, I must stress that the nature of thematic relations

is a very ill-understood area, and it has not been my intention

here to present a coherent system of thematic functions; I

believe such an attempt would be premature. My intention rather

has been to motivate one thematic function, and on the basis of

this, to suggest that this is the most appropriate level for

lexical rules. Hopefully, further investigation of lexical rules

will allow the development of a satisfactory system of thematic

functions. Whether the resulting system will be integrable into

the system we now have remains to be seen. In the meanwhile,

there is little option but to proceed along the lines suggested

by Chomsky (1965:75):

A linguist with a serious interest in semantics will presumably attempt to deepen and extend syntactic analysis to the point where it can provide the

information concerning subcategorization, instead of relegating this to unanalyzed semantic intuition, there being, for the moment, no other available proposal as to a semantic basis for making the necessary distinctions.

FOOTNOTES - TO CHAPTER THREE

There is some evidence that the rules far -- and -ive - adjectives must also refer to thematic information. Consider

( I-vi ) :

(I) The walk exhausted Mary.

(ii)*The walk was exhaustive.

(111) The walk was exhausting.

(iv) This list exhausts the possibilities.

(v) This list is exhaustive.

(vi) *This list is exhausting.

The walk in (I) is a "Causer," the list in (iv) is not. Compare -- -- also John/This example suggests - the following analysis, - this

example - is suggestive, *John - is suggestive.

See also DeArmond (1980).

30r vice-versa; cf. (23)-(24) below.

4~ince we are here interested only in the fact that in both

(lgb) and (19c) the noun is deadjectival, I have combined the

two paradigms.

This prediction has some consequences for the identification

of wexceptionsn in the lexicon. Thus although the majority of

causative verbs with Experiencer objects do not have a

nominalization with the Experiencer as llobject," nominals like

(i)-(iii) below are apparently acceptable for some speakers:

(I) John's disappointment of his audience

(ii) The teacher's inspiration of the students

(iii) John's embarrassment of Mary

Example (i) is cited by Anderson (1977:372); (11)-(iii) are

cited by Newmeyer (1979), who also cites Tomls disappointment - of

Sue. From the point of view of the Experiencer hypothesis, these - noun phrases must be regarded as exceptions. For the "pure

grammatical functionsv hypothesis, however, these nominals

represent the regular case. Apart from the statistical

difference in the number of "exceptions" under the two

hypotheses, note that the alleged correspondence in the frames

of the verb and the noun is subject to restrictions. Thus, - the

news1/ results1/ performance's disappointment -- of the audience

(equivalently, - the disappointment -- of the audience 3 - the news/

the results/ the performance), the sunset's inspiration of the - - - -- poet (equivalently, the poet's inspiration 3 - the sunset), and

the disclosurels embarrassment of Mary (equivalently, the - - embarrassment - of Mary Q - the disclosure) seem to me to be

totally unacceptable, although - the news/ - the results/ - the

performance disappointed - the audience, - the sunset inspired the

poet, and - the disclosure embarrassed Mary are perfectly

a c c e p t a b l e .

A more d i f f i c u l t counterexample t o t h e Exper iencer

h y p o t h e s i s i s provided by ( 1 v ) - ( v i ) :

( i v ) Mary charmed t h e hos t e s s .

( v ) The h o s t e s s was charmed.

( v i ) Mary 's charm

Assuming t h a t - t h e h o s t e s s i s an Exper iencer i n ( i v ) , we do not

get t h e expected Exper iencer s u b j e c t i n ( v i ) . I have no

e x p l a n a t i o n f o r t h i s . Notice t h a t ( v i i ) i s n e v e r t h e l e s s

ungrammatical :

( v i i ) * M a r y l s charm of t h e h o s t e s s

Advocates of t h e Jackendoff-Wasow a n a l y s i s might p o i n t ou t

that t h e theory of word based morphology proposed by Aronoff

(1976) appears t o a l low t h e morphological d e r i v a t i o n of

amusement, annoyance, i r r i t a t i o n , e t c . , from t h e corresponding

past p a r t i c i p l e s . I n t h i s theory we might p o s t u l a t e WFRs

p r o d u c i n g [ [ [ X I ed] ment] e t c . , and l l t runca te t l t h e p a r t i c i p i a l V A N

i n f l e c t i o n . This a n a l y s i s f a c e s a h o s t of morphological

p r o b l e m s , however. There a r e well-motivated s u f f i x e s -ment -9

-ance and - ( a t ) i o n -- which a t t a c h t o verbs . We must p o s t u l a t e

homonymous s u f f i x e s which a t t a c h t o [Xed] t h e s e cannot be t h e A '

' l s a m e N as t h e deve rba l s u f f i x e s , f o r a WFR ope ra t e s on a u n i t a r y

s y n t a c t i c o s e m a n t i c base (op. c i t . : 4 8 ) . Apart from t h e l a c k of

I n d e p e n d e n t evidence f o r "dead j e c t i v a l l 1 -ment -' -ance - and

- ( a t ) i o n -- -- t h e y do n o t a t t a c h t o any o t h e r a d j e c t i v a l bases

such as [Xable] [Xive] [XinglA-- t h e r e i s e v i d e n c e a g a i n s t a A' A'

p a s t p a r t i c i p i a l b a s e f o r t h e s e s u f f i x e s . Cons ide r f i r s t

morpho log ica l r e s t r i c t i o n s on b a s e - s u f f i x combina t ions . Given

t h e h y p o t h e t i c a l base [Xed] which i s f r e e t o combine w i t h -ment, A

-ance and - ( a t ) i o n , - - - we must e x p l a i n why from [ i r r i t a t e d ] w e ge t A

[ i r r i t a t i ~ n ] ~ and n o t * i r r i t ( a t e ) m e n t , * i r r i t ( a t e ) a n c e . T h i s

f a c t i s e x p l a i n e d i f w e t a k e i n t o accoun t t h e morpheme -ate - of

i r r i t a t e d , i g n o r i n g t h e -9; Xate is a t y p i c a l b a s e f o r Xat ion

(Marchand 1969:259). Again, Aronoff (op. c i t . : 5 6 , fn .10) n o t e s

t h a t X c i t e may t a k e e i t h e r -ment - ( i n c i t e m e n t ) o r - a t i o n

( c i t a t i o n ) . P r e c i s e l y t h i s v a r i a t i o n i s s e e n e x c i t e , e x c i t e m e n t ,

e x i t a t i o n ; i f [ e x c i t e d ] i s t h e b a s e of e x c i t e m e n t , t h e -ed must A -

once a g a i n be i g n o r e d i f t h e v a r i a t i o n i s t o be e x p l a i n e d . But

i f t h e p a r t i c i p i a l i n f l e c t i o n is i g n o r e d by b a s e - s u f f i x

combinatory p r i n c i p l e s , and t r u n c a t e d b e f o r e i t r e a c h e s t h e

s u r f a c e , t h e r e i s no e v i d e n c e f o r i t s p r e s e n c e a t any

t h e d e r i v a t i o n , i. e . no ev idence f o r a b a s e [Xed] . A

Secondly , t h e r e i s a " l e v e l o r d e r i n g w problem i n

h y p o t h e t i c a l d e r i v a t i o n . S i e g e 1 (1974) p o s t u l a t e d two

a f f i x e s : + boundary (Class I o r Level I ) a f f i x e s , and

s t a g e of

t h i s

c l a s s e s o f

# boundary

(Class I1 o r Leve l 11) a f f i x e s . Aronoff r e t a i n s t h i s d i s t i n c t i o n

( s e e a l s o A l l e n ( 1 9 7 8 ) , P e s e t s k y ( 1 9 7 9 ) , f o r j u s t i f i c a t i o n of

t h e two l e v e l s ) . I n word based morphology, t r u n c a t i o n o p e r a t e s

on ly b e f o r e + boundary s u f f i x e s . Thus i f t h e -ed - i s t o t r u n c a t e ,

the suffixes which attach to it must have t boundaries. Since t

boundary suffixes do not normally appear outside # boundary

suffixes, the -ed - must itself therefore be a + boundary suffix.

But Allen (op. cit.:38ff.) presents evidence that -ed - is a #

boundary suffix: (i) nominals and negations of - Xed are always

formed with Level I1 suffixes, and not with Level I suffixes,

e.g. tired#ness, *tired+ity; un#hurried, *inhurried;

un#announced,*in+announced; (11)-s is stress-neutral and

stressless, a typical characteristic of # boundary suffixes;

(iii) underlying non-syllabic r-, which appears in syllabic form

before # boundaries and non-syllabic form before + boundaries,

is syllabic before -ed - (sober, sobriety, sobered). Allen observes that this evidence is consistent with the fact that

"only Level I1 derivational suffixes exhibit the phenomenon of

lcopyingl the shape of inflectional affixes."

Thirdly, the impossibility of *unXment, *unXance, and

*unXation is further evidence against an adjectival base for

these nouns, since "un- - appears on nouns . . . only if these nouns have been derived from adJectivesn (Siege1 1973:303);

e.g., untruth, unkindness.

Interestingly, Aronoff motivates morphologically "abstractw

derivations like those considered above, partly on the basis of

semantic facts. We shall return to an examination of word based

morphology in Chapter Four.

7 Marchand (1969:300) notes that -3 has formed desubstantival and deadjectival derivatives in English from the sixteenth

century onwards. If the nouns in (22d) were formed from the

verbs in (22a), we would expect *terrification,*horrification;

cf. glory-glorify-glorification,

identity-identify-identification.

The suffix -ize - f orms desubstantival (winterize, lionize, dramatize) and deadjectival (peneralize, liberalize, popularize)

derivatives. Nouns formed from Vize end in -ism (criticism, - - plagiarism) or -ation (generalization).

I

Notice the shared preposition - at in the complements of the

adjective and the noun in (22c-d). (Compare also the parallelism

below in the - of-NP complements to the noun and the adjective:

Mary's terror - of growing old, Mary - Is terrified - of prowing old.)

Shared prepositions were one argument for relating the nouns in

(8c) to the past participles in (8b), although the nouns were

morphologically deverbal. By the same token, we must now relate

the morphologically basic nouns in (22d) to the past participles

of denominal verbs in (22c), instead of to the denominal verbs

(22a) themselves. It appears that once again a generalization is

being missed. The shared preposition should be accounted for not

by positing various arbitrary relations in the lexicon, but in

terms of shared thematic functions signalled by particular

prepositions.

lo Wasow (1980) does not consider minor lexical rules involving

adjectives. Since the primary function of thematic functions in

his framework is to encode the llexceptionalityll of lexical rules

for verbs arising out of their lllocalness,ll It is unclear to me

if he would require a reference to thematic functions in

apparently exceptionless cases. It could be argued that the

additional level of thematic functions should be accessed only

when necessary, in the interests of theoretical parsimony. If,

however, reference to thematic functions is obligatory for all

minor rules, rule (26) could be modified as follows:

Subject (be) A (PP)

11 There are also transitive verbs (and verbs which take other

types of complements) which have Experiencer subjects, and which

have nomlnalizations exhibiting the llnormallf correspondence; cf.

hate, hope, fear, love, re~ret, pity, repent, admire.

l2 This task (in fact) reveals the fundamental inadequacy of the

approach to this problem suggested In XI Syntax. The

cross-categorial generalization of grammatical relations

obviously cannot help In capturing the subcategorlzational

correspondences of a derivationally related pair like

read-readable. In fact, in the majority of cases of

derivationally related words, we must postulate lexical

redundancy rules that alter grammatical relations. However (for

some unclear reason) X I Syntax supposes that the generalization

strategy will be fruitful in the case of the nominal derivatives

of verbs and adjectives. Accordingly, the proposal of X1 Syntax

is confined to this small subdomain of derivational morphology.

But the task that the Lexicalist Hypothesis imposes on the

grammar is more general: it is to capture regularities in the

subcategorizational frames of - all lexically related words. The

"neutraln lexical entry, which economizes on the statement of

shared selectional restrictions, is one such general device.

What is needed is a device of similar generality which

economizes on the enforcement of shared selectional

restrictions, regardless of whether the syntactic positions

which must be thus restricted represent the same grammatical

functions or not.

13 The necessity for such rules was observed as early as

Jackendoff (1972); he notes that one mechanism that must be

built into the lexicon is "a set of rules describing

generalities among the lexical correlations of thematic and

grammatical relations," and suggests that the solution is to

"state a redundancy rule in the lexicon that would make lexical

items contain less independent information if they conform to

such generalizations. The redundancy could be expressed in terms

of not needing to specify superscripts in the lexical entryv

(op. cit. :42).

DeArmond (1980) attempts to develop such rules, in the

context of a discussion of the lexical entry for the verb open.

He introduces the "modal features" [+Transitive], [-Transitive],

and [+Instrumental] for process verbs, and gives rules for

predicting semantic functions from these features, as also the

syntactic positions occupied by these functions. See DeArmond

(op. cit.) for details. See also Ostler (1979).

14 Such an enforcement of selectional restrictions removes

Jackendoffts main argument for generalizing the grammatical

relations of the S to the NP. (Hornstein (1977) makes the same

point: if selectional restrictions are not placed on grammatical

functions, "the motivation for a generalized notion lsubject-ofl

evaporatesn (op. cit:141)).

Prof. DeArmond has pointed out to me that an argument could

still be made for a wsubjectw in the NP, based on facts like the

following:

(I) the tendency for John to forget his keys

(ii)Johnls tendency to forget his keys

Example (ii) appears to be the analogue, in the NP, of Raising

to Subject. However, it also appears to be a unique case. Verbs

like seem, appear, happen have no nominalizations; adjectives

like certain and likely, which have nominalizations, do not

permit raising in the NP (examples (111)-(v) are from Chomsky

(1970 ~188-189) ) :

(i1i)John is certain (likely) to win the prize.

(iv)*Johnts certainty (likelihood) to win the prize

(v) John's certainty that Bill will win the prize

I therefore wish to leave open the question of whether a

"subjectv in the NP can be justified on grounds other than

selectional parallelism.

l5 Cf. Jackendoff (1972) for an early attempt to set up a

hierarchy of thematic functions. Cf. also Hust and Brame (1976)

for some critical comments.

I use the term Causer rather than Agent to Indicate that the

first argument of CAUSE or LET need not be animate or exercise

volition.

17 Ostler (1979) suggests that Location is a static Goal.

Jackendoff (1978) suggests that Source and Goal be collapsed

into the function Path.

18 Jackendoff (op. cit. : 110) envisages the postulation of

additional locational modes as one way of enriching his system.

The occurrence of more than one locational mode as a restrictive

modifier has also been anticipated (op. cit.:138-139).

We might note that the predicates we have identified as

taking Experiencer arguments, and which we now suggest are a

realization of the locational mode Internal, coincide with a

class of predicates decomposed in terms of a semantic primitive

FEEL in the system of Wierzbicka (undated). She gives the

following as examples: sad, upset, glad, Joyful, regrett,

(un)happy, (un)pleasant, (dis)pleased, (dis)satisfied,

disappointed, surprised, amazed, angry, indignant, irritated,

afraid, worried, repentant, remorse, ashamed, embarrassed,

humiliated, proud, contempt, admiration, envious, jealous, pity,

compassion, grateful, and vengeful.

l9 We can of couree resolve this problem by saying that the two

senses of touch have different broad thematic assignments. This

is not implausible, considering that if one touches a table, the

table does not undergo a change of state, whereas if one is

touched by a story, one undergoes a change from an unsympathetic

to a sympathetic state. However, this would mean that the

generalization of the meaning of a verb involves more than a

simple switch in the locational mode.

Prof. DeArmond has pointed out to me that if Experiencers

are Themes, the sentences below must have Themes in subject

position, although the verbs are transitive and allow passivi-

zat ion.

(1) John enjoyed the party.

(ii) John p i t i e s Mary. f e a r s admi res r e s e n t s e n v i e s

T h i s c o n t r a d i c t s t h e h y p o t h e s i s of Jackendof f (1972:44) t h a t

s e n t e n c e s w i t h Themes as s u b j e c t s do n o t p a s s i v i z e . (See a l s o

Wasow ( 1 9 8 0 ) , DeArmond (1980) ).

However, t h e s e are n o t t h e o n l y counterexamples t o t h i s

h y p o t h e s i s . Hust and Brame (1975 :249) c i t e t h e f o l l o w i n g ( = t h e i r

(111) Washington i s bordered by Oregon, Montana and Canada.

( i v ) The k i n g ' s c a r r i a g e was f l a n k e d by two pha langes of the

r o y a l g u a r d .

( v ) The column was capped by a n o r n a t e s c r o l l .

( v i ) The cowboy was o n l y touched by t h e ambusher ' s b u l l e t .

( v i i ) T h i s s e n t e n c e i s preceded by t h e s e n t e n c e numbered

( 1 7 d ) and i t i s fo l lowed by t h e rest o f t h e paper .

I have a l r e a d y a rgued t h a t t h e v e r b a l p a s s i v e (whether l e x i c a l

o r t r a n s f o r m a t i o n a l ) i s n o t s e n s i t i v e t o t h e m a t i c i n f o r m a t i o n .

20 Not ice , however, t h a t -- run o u t can be used i n t r a n s i t i v e l y , and

t h e o b j e c t of t h e t r a n s i t i v e v e r b t h e n a p p e a r s as s u b j e c t :

( i) Nelson r a n o u t of money.

(ii) The money r a n o u t .

If t h i s i s a n i n s t a n c e o f t h e Theme-Rule, t h e n Nelson i n ( i)

cannot be t h e Theme.

P

21 The problem of giving content to these notions is seen again

in the thematic assignments of the verbs strike and regard in

Jackendoff (1972:45) and Wasow (1980): for Jackendoff, they have

inverse thematic assignments, while for Wasow, they have the

same thematic assignments, differing only in control.

22 Cf. Oerhle (1976) for an insightful discussion of the Dative

alternation.

CHAPTER FOUR

MORPHOLOGICAL AND SEMANTIC ASPECTS OF WORD FORMATION

1. Preliminaries. - In the previous chapter I hypothesized that semantic

lexical rules do not specify the morphological relationships

between the lexical items they relate. The exclusion of

morphological Information allowed us to isolate a set of

semantic relations for semantically similar verbs, nouns, and

adjectives, which occurred in paradigms wherein the lldirectionll

of the morphological derivation varied as shown below (the

arrows indicate morphological derivation):

amuse

N A N-A amusement amused tiredness tired

sadden horrify

sadness sad horror horrified

I argued that the independence of the "thematic functions~

hypothesis from morphological derivations was its crucial

advantage over the llmixedll version of the ugrammatical

functionsn hypothesis.

In this chapter1 I present evidence that the separation of

morphological and semantic redundancy rules in the lexicon is

necessary independently of the "thematic functionsu hypothesis.

I show in section 2 that even when the wdirectionw of

morphological derivation is held constant from a lexical

category - X to a lexical category - Y (as under the llgrammatical functionsu hypothesis), a single semantic rule may intersect

various morphological rules, and a single morphological rule may

intersect various semantic rules. The resulting

cross-classification of morphological and semantic redundancy

(which was first noticed by Jackendoff (1975)) cannot be

captured by rules required to specify both kinds of information.

In particular, the model of word formation rules (WFRs)

developed by Aronoff (1976) is seen to be inappropriate for

lexical redundancy rules, and to pose problems for the semantics

as well as the morphology of word formation. I argue for a model

of the lexicon wherein affixation rules do not specify semantic

operations, but access a body of independently existing semantic

operations. I show that such a conception of word formation can

be reconciled with a current theory of morphological structure.

2. The Cross Classification of Morphology and Semantics -- - -- That the lexicon must contain separate morphological and

semantic rules is suggested by Jackendoff (1975).2 Consider his

rule for noun-verb pairs like decide-decision (given below), and

his comments on it:

ABSTRACT RESULT 1 OF ACT OF NPl Is I Z-ING NP2 - - J

tNP1 (P) NP2

NP1 Z NP2 -

. . , redundancy rule (3) (is) a rule relating lexical items both at the morphological and semantic levels. In fact, this formulation will not do. It claims that there is a particular meaning, ABSTRACT RESULT OF ACT OF V-INQ, associated with the ending -ion. However, there are several different semantic rela-ns obtaining between -ion - nominals and their related verbs, and there are several nominalizing endings which can express the same range of meanings . . . The picture that emerges is of a family of nominalizing affixes and an associated family of noun-verb semantic relationships.

Jackendoff goes on to show that the three deverbal nominalizing

suffixes -ment -' -ion - and -a1 - (e.g., refuse-refusal) can each

express any of the three semantic relations "abstract result of

act of V-ing," "group that Vts," or "act or process of V-ing."

He s u g g e s t s t h a t t h e t h r e e morpho log ica l and t h e t h r e e semant ic

r u l e s be i n d e p e n d e n t l y a v a i l a b l e , such t h a t t h e c h o i c e o f one

r u l e from e a c h s e t y i e l d s a p a i r o f morpho log ica l and semant ic

r u l e s t o r e l a t e a p a i r o f words. T h i s is i l l u s t r a t e d i n t h e

t a b l e below (examples from J a c k e n d o f f ) , where each noun i s

r e l a t e d t o t h e c o r r e s p o n d i n g v e r b by t h e morpho log ica l r u l e a t

t h e head of i t s column, and t h e semant ic r u l e a t t h e l e f t of i t s

row.

Semantic r u l e s

S 1 ABSTRACT RESULT

OF ACT OF V-ING

S2 GROUP THAT V 1 s

S 3 ACT OR PROCESS

OF V-ING

Morphologica l r u l e s

M1 M2

- i o n -ment

d i s c u s s i o n argument

congrega t Ion government

c o p u l a t i o n e s t a b l i s h

-ment

M3

-a1

r e b u t t a l

r e f u s a l

The i s s u e t h a t Jackendof f l e a v e s u n r e s o l v e d , however, i s

whether t h e t h r e e r e a d i n g s he g i v e s d e v e r b a l nouns are indeed

p r o d u c t s of t h r e e s e p a r a t e semant ic r u l e s , o r o f one ambiguous

semant ic r u l e ; o r even whether t h e r e i s a p r i n c i p l e d way of

differentiating the two kinds of cases. This issue surfaces when

we notice that some words in the table above can occur in more

than one meaning slot. For example, refusal means "abstract

result of act of refusing11 (cf.Sl) in - a letter of refusal, and discussion means '!act or process of discussingw (cf. S3) in - our

discussion lasted - all night. If the semantic rules S1 and S3 are

distinct, the claim would be that there are two homonymous

lexical items refusal (related to refuse by (M3,S3) and (M3,Sl)

respectively) and two homonymous lexical items discussion

(related to discuss by (M1,Sl) and (Ml,S3) respectively). But if

S1 and S3 are the same rule, no such differentiation of lexical

items would be implied.

We see this problem when we compare the semantic

interpretation for nouns of the form Xousness (e.g.,

callousness) given by Aronoff (1976:38). Here too, three

readings are possible:

( 4 ) a. 'the fact that Y is Xous' --- - His callousness surprised me.= The fact that he was

callous surprised me.

b. - 'the extent - to which - - Y is Xousv

His callousness surprised me.= The extent to which

he was callous surprised me.

c. - 'the guality - or state - of being Xous'

Callousness is not a virtue.= The quality or state

of being callous is not a virtue.

The question is whether there are three discrete semantic rules

associated with -ness -, or only one ambiguous one. Jackendoff1s

system would appear to choose the former alternative. Aronoff,

however, adopts the latter hypothesis.3 He proposes a system of

WFRs which perform an operation that is "both syntacticosemantic

and morphophonological~ (op. cit.:85), and are governed by the

"one suffix, one rulev principle. In this theory, each affix

(strictly, the phonological operation of the WFR) is associated

with a syntacticosemantic operation, and the question of

cross-classification does not arise. We may conceptualize the

WFR as follows:

(5)

Base Word

Morphological rule -- WFR -- Semantic rule

~erivid Word

The issue of whether an affix can be associated with more

than one semantic rule is, as Aronoff observes, a difficult one

to resolve on semantic grounds alone. But we need not confine

ourselves to semantic intuition. We can instead examine whether

a semantic operation has characteristic syntactic consequences

which allow us to identify it, and whether the same consequences

consistently follow from the same semantic rule when it applies

across a variety of affixes. We can also examine if more than

one such identifiable semantic operation is paired with the same

affix. The results of such a procedure reveal that Jackendoff's

system is superior to that of Aronoff.

Let us, to begin with, consider again the list of verbs

with Experiencer objects. The nominals of these verbs are formed

with a variety of suffixes, as the sample in (6) shows:

(6) a. V, V+ment: amuse-amusement, excite-excitement

b. V, V+(at)ion: irritate-irritation,

inspire-inspiration

c. V, V+ance: annoy-annoyance

d. V, V+0: delight-delight, surprise-surprise

At least four morphological rules are needed to relate causative

verbs with Experiencer objects to their deverbal nominal

derivatives. We have seen, however, that the syntactic and

semantic relation between the verb and the noun is constant. The

noun's subcategorization is regularly the vinveraew of the

verb's (Subject corresponds to PP-complement, ObJect to

Poss-NP); and the noun's semantic interpretation is regularly

paraphrasable as "state of being V-ed." This semantic relation,

which relates the nouns to a semantically coherent class of

verbs, is crucial to any explanation of the noun's

subcategorization. The semantic rule is thus clearly

identifiable (from its subcategorizational consequences), and

can be readily distinguished from other semantic rules for

deverbal nouns (which do not have the same subcategorizational

consequences).

The semantic rule for verbs with Experiencer objects and

their nominal derivatives must thus generalize across the

morphological rules of -ment -9 -(at)ion - 9 -ance -9 and 0

nominalization. The "thematic functionsw hypothesis is now seen

to carry such a generalization of a semantic rule to its logical

conclusion. It proposes to isolate a single semantic relation,

with its characteristic subcategorizational correlates, across

not only paradigms wherein the fldirectionn of morphological

derivation is held constant while the affixes are allowed to

vary, but across paradigms wherein the l'directionw of derivation

itself varies.

Consider, on the other hand, how the facts in (6a-c) might

be handled in a theory wherein lexical rules are modelled on

WFRs, and the syntacticosemantic relation between the verb and

the noun is inseparable from the morphological relation. Four

rules will suffix -ment J -(at)ion --, -ance -9 and 0, to the verbs,

and output nouns. Each of these rules will specify that the

verb's object corresponds to the Poss-NP specifier of the noun,

and its subject to an optional PP-complement; and that the noun

has the interpretation "state of being V-ed." Thus even the

weaker generalization that these verbs have virregular,u

wskewedw n o m i n a l i z a t i o n s ( c o m p a ~ e d t o c r i t i c i z e - c r i t i c i s m ) w i l l

be s c a t t e r e d a c r o s s t h e f o u r r u l e s demanded by t h e morpho log ica l

r e l a t i o n s .

T h i s i s on ly one problem f o r l e x i c a l r u l e s which i m i t a t e

WFRs. For j u s t a s t h e same semant ic r u l e may g e n e r a l i z e a c r o s s

v a r i o u s morphologica l r u l e s , t h e same morpho log ica l r u l e may

g e n e r a l i z e a c r o s s v a r i o u s semant ic r u l e s ( c f . a n n e x a t i o n " t h e

a c t of annexing," v e x a t i o n " t h e s t a t e of b e i n g vexedw) . When

( f o r example) t h e same n o m i n a l i z i n g s u f f i x i s a t t a c h e d t o

homonymous v e r b p a i r s , t h e r e s u l t i s homonymous noun p a i r s which

a r e r e l a t e d t o t h e i r v e r b b a s e s by d i f f e r e n t s e m a n t i c r u l e s . We

have a l r e a d y encoun te red such c a s e s o f homonymity. R e c a l l t h a t

v e r b s , l i k e d e p r e s s , e x h a u s t , s a t i s f y and a g i t a t e have two

argument s t r u c t u r e s , w i t h two c o r r e s p o n d i n g p a t t e r n s o f

n o m i n a l i z a t i o n . The semant ic r u l e s f o r t h e two k i n d s of ve rbs

a r e h e r e d i f f e r e n t : " s t a t e of b e i n g V-edm i n one c a s e , " a c t ( i o n )

of V-ingfl i n a n o t h e r . However, t h e morpho log ica l r u l e t h a t bo th

t y p e s of ve rbs undergo i s t h e same:

I n f a c t , when we examine t h e n o m i n a l i z a t i o n s of

Exper iencer -ob jec t v e r b s more c l o s e l y , we f i n d t h a t t h e s e

nominals e x h i b i t a l s o a d i f f e r e n t t y p e o f homonymity: they o c c u r

i n [-Count] and [+Count] p a i r s . When t h e s e nominals mean Ifstate

of be ing V-edn (and a r e s u b c a t e g o r i z a t i o n a l l y p a r a l l e l t o a p a s t

participial adjective) , they must have a syntactic feature

[-Count]. Evidence for this is that the nouns cannot have a

plural form in the relevant subcategorizational frame.

( 8 ) *the children's amusements

' the stories

annoyances

disappointments embarrassments excitements surprises

interests In J

The non-pluralizability of the noun is here predictable from its

stative interpretation. In general, nouns denoting states or

qualities do not pluralize: compare John's sincerity

(*sincerities), deafness (*es), dumbness (*es) .4 We can formally

express this property of the nouns in (8) by giving them a

feature [-Count], as we do for a noun like sincerity.

There are, however, other contexts in which such nouns are

apparently specified for a feature [+Count], for here plural

forms are possible, as is the indefinite article - a(n): -

(9) a. The many amusements available to a tourist in

Jamaica include surfing and skin-diving.

b. We missed the plane and lost our baggage, but these

were minor annoyances (irritations).

c. John is an embarrassment and a disappointment to his

family .

d. The chocolate chip cookies were a rare delight.

The contexts for the [+Count] nouns do not show the regular

correspondences that we have noted, with the contexts for the

related verbs and adjectives. Thus non-animate possessive NP

specifiers are here possible, although non-animate NPs cannot

occur as the objects of the verbs or the subjects of the

adjectival predicates.

(10) a. The eveningfs chief amusement (excitement) was a

strip-tease by a male dancer.

(Cf. '*Something amused the evening; *the evening

was amused )

b. The day's surprises were not yet at an end.

(Cf. f*Something surprised the day; *the day was

surprisedf.)

Also, the [+Count] nouns, unlike the [-Count] nouns, cannot take

the verb's subject as an optional PP-complement:

(11) a. *Maryfs interests in the stories

b. Mary's interests include stamp-collecting and

head-hunting.

The specification of strict subcategorizational and selectional

features for the noun thus differs according to whether it is

[-count] or [+Count]. We must therefore postulate two lexical

Observe now that the [+Count] nouns are paraphrasable not as

"state of being V-ed," but as "events, activities, objects,

etc., that V.ll The countable and uncountable nouns must

therefore be produced from (or related to) the verb by distinct

semantic rules. Thus the difference in meaning correlates with

the differentiation of lexical items based on syntactic

considerations, proving again our point that it is possible to

motivate distinct semantic rules on grounds other than purely

semantic or intuitive ones.5 Now, the semantic rule ltevents,

activities, objects, etc., that Vw generalizes across nouns with

a variety of affixes (amusements, irritations -9 annoyances,

surprise0s). But more interestingly, the morphological rules of

ment (at)ion, ance and 0 are also seen to cut across the two -3 - - - semantic rules "state of being V-edw and "events . . . that V." The morphological relation between the verb and the [-Count]

noun is in every case the same as that between the verb and the

[+count] noun; only the semantic relation differs.6

items in such cases in order to ensure correct lexical

insertion, e.g. (12) : - amusement

[+Count]

idiosyncratic features

-

(12) - - amusement

[-Count]

+ (PP)

[+ - [+Abstract]]

.(=V1s subject) -

Consider ano the r c a s e of homonymity: t h e noun assurance. I n

( 1 3 ) , a s su rance i s ambiguous between t h e two read ings "ac t ion ,

o r r e s u l t of a c t i o n , of Viogtl and " s t a t e of being Ved,n shown i n

(131) and (1311) r e s p e c t i v e l y .

(13) The h a r e ' s assurance t h a t he would win t h e r ace

( I ) f a i l e d t o comfort t h e t o r t o i s e .

(11) was matched only by h i s l a z i n e s s .

But t h e noun i s disambiguated when i t occu r s wi th a - to-NP

complement, o r wi th a p l u r a l morpheme. Compare (14) - (15) :

(14) The h a r e ' s assurance t o t h e t o r t o i s e

( I ) was g iven i n t h e presence of a l l t h e s p e c t a t o r s .

(ii)*was matched only by h i s l a z i n e s s .

(15) The h a r e ' s a s su rances

(I) f a i l e d t o comfort t h e t o r t o i s e .

( i i ) * w e r e matched only by h i s l a z i n e s s ( e s ) . I n (15) and ( 1 4 ) , only t h e non-s ta t ive read ing of assurance is

poss ib l e . The ungramrnaticality of ( 1 4 1 1 ) shows t h a t t h e noun

wi th t h e s t a t i v e meaning "conf idence, sureness1' does no t

subca t ego r i ze a wGoalN argument, an e n t i t y t o whom t h e assurance

i s given.7 The ungrammatical i ty of (1511) shows t h a t ( l i k e o t h e r

s t a t i v e nouns) assurance on t h i s read ing cannot be [+Count].

There must t h e r e f o r e be two l e x i c a l i tems assurance , r e l a t e d t o

the verb by the same morphological rule but by different

semantic rules.8

We have now identified at least three semantic rules for

nominalizations, which must intersect four morphological rules.

This is illustrated in the table below. I have introduced, in

parentheses, lexical items which intuitively meet the

specifications

completeness.

Semantic Rules

Morphological Rules

state of events. . . act or result being V-ed that V of V-ing

amusement amusements (acknowledgements )

irritation irritations (derivations)

annoyance annoyances assurances assurance

surprise surprises (displays)

This is precisely the cross-classification of morphological and

semantic relationships noticed by Jackendoff. We have, however,

succeeded In tying the semantics to observable consequences,

justifying the separation of semantic rules. If we now require

the same lexical rule to specify morphological and semantic

information, we will need tweLve rules instead of seven, to

relate the nouns in (16) to their verbs. If we identify yet

another semantic rule, we will need sixteen rules instead of

eight; and so on. The loss of generalization is apparent,

Observe that the data above cannot be dismissed as

idiosyncratic (I. e., as products of vsemantic driftf1). The

semantic regularity of the nouns in the first column of (16) is

not in question. For the nouns in the second column, we were

able to give a paraphrase with a variable, This is the

traditional method of expressing regular semantic relationships,

and one which Aronoff informally adopts. We might note that it

is the readings "action of V-ingw and "event or state of being

V-edw that Aronoff (op. cit.:33) himself suggests as the

expected reading of deverbal nouns, and these are precisely the

two readings of assurance that we have identified.

This suggests that the inappropriateness of the WFR model

for this data arises from the inadequacy of the model itself.

For the justification for separating a theory of word formation

from a theory of word analysis is that existing words tend to be

peculiar, and acquire idiosyncracies. When such idiosyncracies

are removed and only regularities are taken into account, we

must expect a "matchw between the two theories; as Aronoff

observes, it is "possible, but highly unlikely," that the rules

for analyzing existing words are radically different from the

rules for making up new words.

Let us outline more explicitly how the WFR model will

handle facts like those in (16), to see at what points the model

is inappropriate. Aronoff assumes each word in the lexicon to be

a fully specified entity (as we have); there are no partially

specified lexical entries. The two subcategories of (e.g.)

amusement must therefore be given two fully specified lexical

entries. Further, the syntacticosemantic operation of a WFR

vspecifies the semantics of the output as a compositional

function of the meaning of the base, and assigns the output to a

specific major lexical category - - in a specific sub categorization^

(op. cit.:85, emphasis added). If so, the same

syntacticosemantic operation cannot have produced (e.g.) both

the [-Count] and [+Count] nouns amusement, which have different

subcategorizations. This again accords with our analysis, and

confirms that the WFR which performs the morphological operation

of -ment attachment does not perform a unitary

syntacticosemantic operation. We must rather speak of a WFR

which consists of the -ment attachment rule, and of (at least) - two syntactico- semantic operations: "state of being V-ed," and

"events . . . that V." The meaning and the subcategorization of

the derived noun amusement depends on which of these two

operations has applied.

But here there is a problem. We have hypothesized diverging

paths from the morphological to the semantic part of a WFR, and

bifurcated the latter part:

(17) WFR #ment - Morphology: [A] ----> [[A] #merit]

V V N Syntax and Semantics:

"state of being V-edw "events . . . that Vv Noun, [-Count] Noun, [+Count]

However, the postulation of such bifurcations is not a mere

matter of filling in the details of a WFRts semantic operation;

it violates the "very strict 'one suffix, one rule1 basisw (op.

cit. :89) of the theory. To preserve this principle, we must

postulate at least two WFRs of -ment for the case above, with - one morphological and one syntacticosemantic operation each.

The "one suffix, one ruleH principle is more clearly seen

in the unitary base hypothesis. This hypothesis states that "the

syntacticosemantic specification of the base, though it may be

more or less complex, is always uniquew (op. cit.:48). Under the

provision for wcomplexity,w a WFR may refer to the "category,

subcategory, selection and lexically governed entailment and

presupposition" of its base. What the unitary base hypothesis

prohibits is a disjunction in the base, with semantic operations

and derived subcategorizations sensitive to this disjunction;

for this would again bifurcate the WFR. Consider for example the

affix #able -9 which (Aronoff notes) attaches to verbs

(acceptable) and, less productively, to nouns (obJectionable,

marketable, profitable, knowledgeable, sizeable, saleable,

f a s h i o n a b l e ) , t o form a d j e c t i v e s . The d e v e r b a l a d j e c t i v e s have a

r e a d i n g l lcapable of b e i n g - Xed (where - X is t h e b a s e ) , " w h i l e t h e

denomlnal a d j e c t i v e s have a r e a d i n g ' ' c h a r a c t e r i z e d by - X (where - X

1s t h e base ) - "10 S i n c e t h e form o f t h e s u f f i x , i t s Level ( o r

boundary) s p e c i f i c a t i o n , and t h e c a t e g o r y of t h e o u t p u t , is t h e

same i n bo th c a s e s , w e might t r y t o combine d e v e r b a l and

denominal - a b l e - i n t o one r u l e . T h i s h y p o t h e t i c a l r u l e would

o p e r a t e on e i t h e r a v e r b - o r a noun, w i t h t h e semant ic o p e r a t i o n

v a r y i n g w i t h t h e l e x i c a l c a t e g o r y of t h e b a s e :

( 1 8 ) ( h y p o t h e t i c a l ) WFR # a b l e - CAIx----> C [ AIX#ableJA

where X = N o r V

Syn-sem op. 1 Syn-sem op.2

Base :X=N Base : X=V

' c h a r a c t e r i z e d by N 1 ' c a p a b l e of b e i n g V-ed

The u n i t a r y base h y p o t h e s i s p r o h i b i t s t h e d i s j u n c t i o n i n ( 1 8 ) ,

and p r e c l u d e s t h e p o s s i b i l i t y of a s i n g l e # a b l e - WFR. But c o n s i d e r now how t h e b a s e of t h e n o m i n a l i z a t i o n r u l e s

of -ment -3 - ( a t ) i o n - -3 -ance - and 0 must be s p e c i f i e d . We see t h a t

t h e s e s u f f i x e s s u r f a c e on b o t h l l r e g u l a r l l and llskewedll

n o m i n a l i z a t i o n s :

( 1 9 ) a. i. J o h n ' s acknowledgement o f t h e l e t t e r

11. the children's amusement (at his antics) - b, I. the annexation of Oudh by the British

ii. the children's irritation (at his antics) - c. I. John's acceptance of Mary's resignation -

ii. the children's annoyance (at his antics) - d. i. John's display of courage

ii, the children's surprise (at his antics)

We have shown that the pattern of nominalization, and its

semantics, is dependent on a disjunction in the

syntacticosemantic specification of the base. The disjunction is

not in the category, but in the subcategory. Let us illustrate

this with the rule of #ment. -

( 20) [A]~---- > CCAIV#mentlN Syn-sem Base 1 : Syn-sem Base 2 :

V with Experiencer Object V with non-Experiencer Object

Semantic operation: Semantic operation:

state of being V-ed action of V-ing

Subcategorization: Subcategorization:

inverse of V1s parallel to V18

There is no way to collapse the bases or the operations in

(20) without loss of descriptive adequacy. The unitary base

hypothesis thus forces us to postulate two sets of affixes

-ment, -(at)ion, -ance and 0, for the data in (19).

The result of the "one suffix, one rule" principle,

therefore, is the proliferation of homonymous affixes: if there

are - n different semantic operations associated with an affix, we must treat the affix as n different affixes.11 This is clearly - unsatisfactory, and the "one suffix, one rulev principle (from

which the unitary base hypothesis follows) must be given up.12

But it is apparent that the maintenance of the "one suffix,

one rule" principle is essential for maintaining the integrity

of the WFR as a rule relating a base word to a derived word at

two levels simultaneously: the morphological, and the

(syntactico)semantic. If this principle held, we should have

been able to state the relation between two words A and B simply

as "A is derived from B by the WFR of Wev Everything about the

morphological, the syntactic, and the semantic relation of A to

B would follow from this simple statement. Let us illustrate

this by assuming, for the sake of argument, that the WFR of +ee - has a unitary base [V, +transitive, +animate object] (as

suggested by Aronoff), and performs a unitary semantic

operation, (roughly) "recipient of the action of V." Now if we

said that employee was derived from employ by the WFR of +=, it would follow that employ was a transitive verb with an animate

object; the meaning of the derived word would also follow from

the rule, in a straightforward way. In contrast, to say now that

agitation or derivation is derived from agitate or derive by the

-(at)ion -- WPR is to say very little; for the WFR no longer has a

unique base, and no longer uniquely determines the semantic and

the subcategorizational relationship between the base and the

output. All that the statement now tells us for certain is the

morphological relationship. To determine the syntactic and

semantic relationship, we must probe again into the

syntacticosemantic subclass of the base ("is it an

Experiencer-obJect verb or not?"), and ascertain which semantic

operation has applied to this base.

In vertically bifurcating the semantic part of a WFR, then,

we have also bifurcated the WFR laterally, separating the

morphological process of word formation from the

syntacticosemantic process. In section 5 I suggest that this

separation has consequences for the morphology that are not

undesirable.

3. A Model - -

Our data suggest that the rules of afflxation which output

the morphological forms of words are not themselves specified

with semantic operations, but that they are able to access a set

of independently existing semantic operations in the lexicon.

Such a model would explain the "sharingw of semantic rules by a

range of affixes, and eliminate the redundant statement of the

same semantic rule as part of a number of affixation rules. As

regards the semantic operations, we may hypothesize that there

a r e b a s i c a l l y two k i n d s o f o p e r a t i o n s a v a i l a b l e . One k ind would

r e l a t e argument s t r u c t u r e s ( s p e c i f i e d i n t e rms of t h e m a t i c

f u n c t i o n s ) , and r e s u l t i n t h e obse rved r e g u l a r i t i e s i n t h e

s u b c a t e g o r i z a t i o n a l frames of r e l a t e d words. I n s t a n c e s of t h i s

k ind would be t h e r u l e s we gave i n Chap te r Three as redundancy

r u l e s , r e p e a t e d h e r e a s ( 2 1 ) - ( 2 3 ) .

( 2 1 ) S1: Causativization-Decausativization

( i . e . N o r A )

(PP)

( E x p e r i e n c e r , ( X I ) (Cause r , E x p e r i e n c e r )

S2: I n h e r i t a n c e 1

F i l t e r : *N, (Cause r , E x p e r i e n c e r )

L( Argument, Argument) ( Argument , Argument ) 1

(23) S3: Inheritance 2

A N

+ NP (PP) * +NP-(PP) (Argument, (Argument)) 1 I (Argument, (Argument)

The other kind of operation (we may hypothesize) does not

relate argument structures, and (therefore) does not give rise

to subcategorizational regularities. The words produced by the

latter kind of operation would have more idiosyncratic meanings

and syntactic features. Among operations of this kind, have

identified one we may call Concretization - * which produces [+Count] nouns like amusement:

(24) S4 : Concretization

Noun, +Count

CAUSER OF STATE

Let us now hypothesize that an affixation rule specifies

the following information. It specifies the category (or

categories) of its input, the category of its output, the

boundary, shape and position of the affix; and a list of

morphologically permitted bases. This much is essentially the

morphological part of the Aronovian WFR. For the semantics, let

us hypothesize that the rule contains a reference to a list of

semantic rules. Thus the #ment - affixation rule may be given as in (25):

t

[ A] ,--- > [ [ A ] #merit] N

List of morphological bases:

beX (bewilder)

enX (enjoy)

Xcite (excite)

Xuse (amuse)

etc.

Semantics:

Rule S1 (Causativization-Decausativization)

Rule S2 (Inheritance)

Rule S4 (Concretization)

etc.

Suppose now the lexicon contains the following lexical entry.

amuse

(Causer, Experiencer)

mapping: (redundant) -

Rule (25) says that given such a verb Xuse, we can suffix #ment -

t o i t t o form a noun. For t h e semant ics of t h e noun, i t r e f e r s

us t o r u l e s ( 2 1 ) - ( 2 4 ) above. The l e x i c o n t e l l s us t h a t amuse

has t h e argument s t r u c t u r e (Causer, Expe r i ence r ) . Rule ( 2 1 ) s ays

t h a t f o r verbs wi th t h i s argument s t r u c t u r e , t h e r e may be nouns

o r a d j e c t i v e s wi th t h e argument s t r u c t u r e (Exper iencer , ( X ) ) .

Thus we a s s i g n amusement t h i s argument s t r u c t u r e , and e n t e r t h e

word i n t h e lex icon . (We assume t h a t t h e f e a t u r e [-Count] f o r

t h e noun i s p r e d i c t a b l e from i t s semant ic i n t e r p r e t a t i o n ; c f .

fn .3 above.) Next we t a k e t h e same morphological form amusement,

and apply t h e r u l e (22) above; bu t t h e ou tpu t of t h i s r u l e i s

f i l t e r e d out . Running down t h e l i s t , we f i n d t h a t Conc re t i za t ion

i s a p p l i c a b l e . Thus t o t h e same noun-form amusement we a s s i g n a

reading "causer of s t a t e u and a f e a t u r e [+Count], d e r i v i n g

another noun amusement, which we e n t e r i n t h e l ex i con .

We t h u s have t h e fo l lowing r e l a t i o n s h i p s between amuse and

i t s d e r i v a t i v e s :

amuse

/phon. repr. /

v + NP NP

(Causer, Experiencer)

mapping: (redundant)

amusement

/phon. repr . / 1 N, +Count c

other features

amusement

Fphon. repr. /

(Experiencer,(X))

I mapping: (redundant) P=a t -L

This entry expresses the fact that the two nouns amusement are

derived from amuse by the same morphological rule, but by

different semantic rules. The correlation between the argument

structure and the syntactic frame is provided by the

thematic-syntactic mapping specified in the previous chapter. In

(27), the preposition - at selected by the noun may be regarded as

wcostlessfl if an appropriate characterization of the thematic

function - X can be given.

Let us now specify another affixation rule, which derives

adjectives from verbs and nouns.

(28) M2: #ed -

Y=V or N

List of morphological bases:

.

. Semantics:

Rule S1 (Causativization-Decausativization)

etc.

Notice that this affixation rule accesses the same semantic rule

S1 as the Ument rule did. Following a procedure similar to that

outlined above, we derive an adjective amused from amuse. We now

have the following morphological and semantic relationships

between amuse and its nominal and adjectival derivatives:

amuse

/phon. repr. /

v +NP - NP

(Causer, Experiencer)

mapping: (redundant)

amusement

/phon. repr. / by S4 [ N , +count 1 +-

other features

amused

I mapping : (redundant )

iphon. repr. /

N , -Count

+ NP (PP)

(Experiencer,(X))

mapping : (redundant )

P=at

The network above shows that while the two nominal

derivatives share a morphological relation with amuse but have

different semantic relationships, the [-Count] noun and the

addective have the same semantic relationship with amuse,

differing only in the morphological relationship. The

subcategorizational correspondences in the frames of amuse,

amused and amusement [-Count] now follow from this semantic

relationship, and the thematic-syntactic mapping. The fact that

the adjective and the [-Count] noun both select the same

preposition at may be regarded as non-coincidental if the thematic function - X in their argument structures is taken into account, and the selection of the preposition is dependent on

this thematic function.

Below I give additional samples of affixation rules, and of

the lexical entries related by these rules. Notice again the

recurrence of the semantic rules S1, S2 and S3 above In the

specifications of these affixation rules.

CAI,----> [ [ A +(at)ion] t N

List of morphological bases:

Xate (irritate)

Xify (glorify)

Xcite (excite)

etc.

Semantics:

Rule S1 (Causativization-Decausativization)

Rule S2 ( ~nheritance )

Rule S4 (Concretization)

etc.

(31) illustrates how this rule relates the two senses of agitate

to the two senses of agitation and the adjective agitated. (I

use the term "Ext(erna1) Themew to denote the first argument of

GO-posit.ext. We may speculate that the two senses of agitate

are themselves related by a redundancy rule like the following:

a g i t a t e

/phon. r e p r . / 1

tNP NP

(Causer , Ext .Theme)

mapping : ( redundant )I

a g i t a t e

/phon. r e p r e /

v

tNP NP

(Causer , Exper iencer )

a g i t a t i o n

C/phon. repr . /

1 (Causer, Ext .Theme)

mapping : ( redundant ) 1

a g i t a t i o n

/phon. repr . / 1

mapping: ( redundant ) I . P=at , over A

a g i t a t e d

/phon. repr . /

A

tNP (PP)

(Expe r i ence r , (X) )

mapping : ( redundant : . P=at , over

List of morphological bases :

Xor (terror)

etc.

Semantics :

Rule S1 (Causativization-Decausatlvizatlon)

etc.

List of morphological bases:

.

.

Semantics:

Rule S 1 (Causativization-Decausatlvlzatlon)

etc.

List of morphological bases:

. Semantics:

Rule S3 ( Inheritance)

etc.

by M4 I horr i fy

horror

/phone repr. /

N

t N P ( P P )

(Experiencer, ( X ) )

mapping: (redundant:

P=at -

horr i f l e d

/phon. repr. /

v tNP NP

(Causer, Experiencer)

mapping: (redundant)

/phone repr. / 1

(Experiencer , ( X ) ) I mapping: (redundant ) I

t i r e

Fphon. repr. / 1

I (Causer, Experlencer) I mapping : ( redundant )j

t i r e d t i r ednes s

/phon. repr. / -

repr. /

N

t N P

(Experlencer)

mapping: ( r e d ~ n d a n t ) ~ - mapping : ( redundant )-

- by M6

by S 3 -

sadness

/phon. repr . / N

+ NP

(Experiencer)

mapping: (redundant)

sad

jphon. repr. /

A

+ NP

(Experiencer )

-mapping: (redundant:

sadden

(Causer, Experiencer ) 1 mapping: (redundant) 1

4. Implications for Productivity and Semantic Coherence - - - Our conception of the process of word formation, which

allows us to examine the morphological and the semantic process

each in their own right, gives us a slightly different

perspective on notions like Nproductivity" and usemantic

coherence. " Consider first productivity. Aronoff shows that the morphological productivity of an

affixation rule varies with the morphology of the base. Thus

given the deadjectival nominalizing suffixes #ness and +=,

#ness - is in general the more productive suffix; but for adjectives of the form -9 Xile += is more productive than #ness - (servile, servility, *servileness). That is, there are more

Xility words than Xileness words in the language. Similarly,

#ment is very productive with - enX and - beX verbs (Aronoff op.

cit.:53, Marchand 1969:332). Aronoff therefore suggests that a

productivity index be associated with each morphological form of

the base of a WFR. We may indicate this as in (38):

(38) affix #ment - Forms of the base: beX, p. index 1

enX, p. index 1

Xcite, p. index 3

etc.

Given the assumption that the semantic base and the semantic

operation of a WFR are unique, this is as much as can be said

about the productivity of a WFR, in Aronoffls theory.

It appears, however, that there is another dimension to the

phenomenon of productivity, which our view of word formation

enables us to capture. We have said that an affixation rule may

have access to more than one semantic operation. The same rule

can therefore produce words belonging to different sense-groups.

It transpires that such a rule can exhibit a preference for one

of the semantic operations available to it, and produce more

words of that sense-group than of any other. Thus consider

Marchand1s comments (op. cit.:332) on the suffix Iment. - Marchand first notes that Xrnent nouns can have one of four

readings: (1) "act or fact of -", e.g. appointment, development;

(2) "something concrete or material connected with -", e.g.advertisement, equipment; ( 3 ) "the state of being -edn,

"chiefly from verbs denoting mental or emotional states," e.g.

astonishment, embarrassment; and (4) "the place connected with

-", e.g. encampment, settlement.

Marchandts sense group ( 3 ) is of course the class we have

identified as the nominalizations of Experiencer-object verbs,

derived by our rule S1 (see (21) above). Of this sense group,

Marchand observes that it

has been especially productive for the last 150 years. The following words, all with the basic meaning wembarrassment, bewildermentn or the like, were coined in the 19th century : astoundment , bedevilment, bemuddlement, bepuzzlement, besetment, betanglement, bewilderment, bewitchment, dazement, dazzlement,

disillusionment dispiritment, displeasurement f lusterment ,Mdlement, muddlement , perplexmeAt. More recent are perturbment 1901, puzzlement 1922.

The earliest words in this sense group that Marchand cites are

astonishment and amazement, in the 16th century. Considering

that "by 1300, -ment - was obviously a derivative suffix," the semantic rule S1 appears to have become available to the #ment

suffixation rule relatively late, and to have reached its peak

of productivity three centuries later.

Suppose now that not only the morphological forms of the

base of #ment -3 but also the semantic operations it permits, are

each assigned an index of productivity. We may then

conceptualize the history of the semantics of #ment - in the following way.

(39) 14th century: Inheritance

16th century: Inheritance (higher productivity)

Decausativization (lower productivity)

19th century: Decausativizatlon (higher productivity)

Inheritance (lower productivity)

Assigning an index of productivity to permitted semantic

operations might also explain why it is excitement, rather than

excitation, which is related to the Experiencer-object verb

excite. We may speculate that while the morphological base Xcite

is equally productive with the suffixes #ment - and +(at)ion, -- the

Decausativization rule has a higher productivity with #ment than

with +(at )ion. -- Consider next the phenomenon of wsemantic coherence." A

derived word is said to be semantically coherent, in Aronoffls

theory, to the extent that its semantics is predictable from the

semantics of its base and the (supposedly unitary) semantic

operation of the WFR. Aronoff associates semantic coherence with

productivity, and productivity (as'we have seen) with the

morphology of the base (op. cit. :62-63). Thus in this theory,

the coherence of a derived word may vary with the morphology of

the base. For example, +a derivatives from the base - 9 Xile

which (we have seen) is a highly productive base for this

suffix, would be expected to be more coherent than +a derivatives from the less productive base - Xous. Similarly, since

#ment - attaches most productively to - enX and beX verbs, we would - expect enXment and beXment nouns to be more coherent than other

Xment nouns. Moreover, a WFR which has few morphological

restrictions on its base is hypothesized to be likely to be more

productive, and therefore produce more coherent words, than a

WFR with many such restrictions.

However, Aronoff notes that although wproductivity goes

hand in hand with semantic coherence, . . . we have no real evidence as to which of these is primary, or even whether they

are really distinct matters. " In this context, it is striking that the [-Count] nominalizations of Experiencer-object verbs

are always semantically coherent, regardless of the particular

a f f i x on t h e noun, o r t h e morphology of t h e base ; whereas t h e

[+Count] d e r i v a t i v e s a r e n o t s o s e m a n t i c a l l y c o h e r e n t . T h i s

s u g g e s t s t h a t t h e coherence i s a p r o p e r t y of i n d i v i d u a l s e m a n t i c

o p e r a t i o n s , i - e . , t h a t o u r r u l e S1, which relates argument '

s t r u c t u r e s , is more c o h e r e n t t h a n o u r r u l e S4. Moreover, t h e

coherence of a n a f f i x a t i o n r u l e as such a p p e a r s t o be a f u n c t i o n

of t h e s e m a n t i c o p e r a t i o n s i t p e r m i t s o r p r e f e r s . We may

i l l u s t r a t e t h i s by comparing A r o n o f f t s a c c o u n t of t h e coherence

of d e a d j e c t i v a l nouns i n +a and #ness w i t h t h a t i n o u r -9

system.

R e c a l l t h a t nouns of t h e form Xousness have t h r e e r e a d i n g s :

( a ) t h e f a c t t h a t - Y i s Xous ( b ) t h e e x t e n t t o which Y is Xous -9 - -9

and ( c ) t h e q u a l i t y o r s t a t e of be ing Xous. Aronoff o b s e r v e s

t h a t #ness d e r i v a t i v e s have o n l y t h e s e t h r e e r e a d i n g s , whi le

+a d e r i v a t i v e s have t h e s e r e a d i n g s p l u s I fo the r r e a d i n g s :

t e c h n i c a l s e n s e s , c o n c r e t e nouns, count nounsw ( o p e c i t . : 3 8 ) . He

d e s i g n a t e s d e a d j e c t i v a l nouns w i t h t h e t h r e e r e a d i n g s above a s

s e m a n t i c a l l y c o h e r e n t , and c o r r e l a t e s t h e g r e a t e r p r o d u c t i v i t y

of #ness - w i t h i t s g r e a t e r coherence .

We now f i n d t h a t t h e r e are [+Count] nouns nouns i n #ness as - wel l . We f i n d a l s o t h a t t h e s e m a n t i c s of t h e += nouns i s

coheren t o n l y when they are [-Count], and t h a t t h e s e m a n t i c s of

t h e #ness - nouns i s n o t c o h e r e n t when t h e y are [+Count]. Thus t h e

semant ics of t h e nouns ( i n ) c r e d u l i t y ( * i e s ) , g a r r u l i t y ( * i e s )

tenacity(*ies), - and vivacity(*ies), derived from the (less

productive) base -3 Xous and fragility(*ies), servility(*ies),

derived from the (more productive) base -9 Xile is transparent.

The semantics of the [+count) +ity derivatives in (40) Is not so

transparent (examples f rom Aronoff ) :

(40) a. How many varieties of fish are there in the pond?

b. All the town's notables and notorieties were there.

c. They admired his dress, but only as a curiosity.

d. What a monstrosity!

e. There are many discontinuities in your story.

Similarly, the semantics of the [+Count] #ness - derivatives in (41) is not so transparent.

(41) a. I remembered his many kindnesses.

b. The awkwardnesses of the evening were soon

forgotten.

c. There are many weaknesses In your analysis.

In (41), Xness does not mean "the fact, quality, state or extent

of being X . " Kindnesses, weaknesses and awkwardnesses mean

winstances of kindness, weakness or awkwardness," just as

discontinuities means winstances of disc~ntinuity.~

The interesting fact is that #ness appears to produce fewer

[+Count] derivatives than +a. NOW, we have hypothesized that

there are two kinds of semantic operations accessed by

a f f i x a t i o n r u l e s ; one r e l a t i n g argument s t r u c t u r e s (and

a s s ign ing coherent r e a d i n g s ) , and ano the r kind which i s more

i d i o s y n c r a t i c . The +a r u l e appears t o a l l ow both kinds of

ope ra t ions , whi le t h e #ness r u l e appears t o f a v o r t h e former. - This "pre fe rencen i s what we have sugges ted be formal ized i n

terms of i n d i c e s of p r o d u c t i v i t y f o r pe rmi t t ed semant ic

ope ra t ions . The g r e a t e r semantic coherence of #ness nouns t h a t - Aronoff n o t e s i s a r e f l e x of t h i s p re fe rence .

5. Residual I s sues . - The c e n t r a l i n s i g h t of t h e theory of word based morphology

is t h a t t h e semant ics of word format ion must be based on t h e

word, and no t on t h e morpheme. There a r e morphemes without

meaning; t h e r e a r e morphological ly complex words whose meanings

a r e no t t h e sum of t h e meanings of t h e i r morphemes; t h e r e a r e

words which s h a r e morphemes without s h a r i n g any meaning ( s t a n d ,

unders tand, wi ths tand; t ake , under take) . The meaning of a

der ived word, i n s o f a r as i t is p r e d i c t a b l e , is p r e d i c t a b l e from

t h e meaning of a - word from which i t i s der ived . "The s i g n

g r a v i t a t e s t o t h e word."

This i s undoubtedly a va luab le i n s i g h t . Thus t h e semantic

r e l a t i o n s h i p s between t h e words i n each column of (42) a r e

t r a n s p a r e n t t o t h e i n t u i t i o n s of t h e n a t i v e speaker , while such

semantic r e l a t i o n s h i p s as may e x i s t between t h e words i n each

column of (43) a r e n o t :

(42)

a. d u r a b l e

d u r a b i l i t y

( 4 3 ) a . d u r a b l e - endure - d u r a t i o n -

b. endure c . e v a c u a t e d. vacan t

e n d u r a b l e evacuee vacancy

endurance e v a c u a t i o n

b. - v a c a n t

v a c a t i o n - vacuum - e v a c u a t e

A t h e o r y of t h e s e m a n t i c s of p r o d u c t i v e word f o r m a t i o n must

t h e r e f o r e b e g i n a t t h e l e v e l o f t h e word, and n o t of t h e

morpheme.

The p remise o f word based morphology, however, i s t h a t t h e

t h e o r y of t h e s t r u c t u r e of words must a l s o b e g i n w i t h t h e l e v e l

of t h e word. Here w e r u n i n t o problems, f o r t h e a f f i x does n o t

v g r a v i t a t e w t o t h e word: t h e r e a r e " r e g u l a r l y d e r i v e d words,

s e m a n t i c a l l y t r a n s p a r e n t , formed w i t h a f f i x e s which we know t o

be a l i v e and w e l l i n t h e i r o p e r a t i o n , which on t h e s u r f a c e do

n o t a p p e a r t o have been d e r i v e d from wordsn (Aronoff op.

c i t . : 8 8 ) . L e t u s t a k e A r o n o f f l s example, t h e s u f f i x +ee, which

forms nouns whose meaning i s a f u n c t i o n of t h e meaning o f

r e l a t e d t r a n s i t i v e v e r b s w i t h an imate o b j e c t s . T h i s s u f f i x

a p p e a r s sometimes on t h e ve rb , and sometimes t h e r o o t of

t h a t v e r b , which can be o b t a i n e d by d e l e t i n g i t s last morpheme1'

( l o c . c i t . ) :

(44)

a. pay-payee b. nominate-nominee

employ-employee evacuate-evacuee

present-presentee

Typically, verbs of the form - Xate have derivatives of the form

Xee instead of the expected *Xatee. The point of attachment of - the suffix is thus predictable from the morphology of the base.

A possible solution, then, would appear to be to allow the

affixation rule to vary with the morphology of the base: i.e.,

to say that each morphological form of the base defines a stem

for the suffix to attach to, which may or may not be identical

to the word.

But in a theory built on the "one suffix, one rulew

principle, where a WFR is a monolithic rule that outputs the

form of a derived word as well as its meaning and

subcategorization, this choice is not available. If the

morphological base of the +ee - WFR is evacu- or nomin-, while the semantic base is evacuate or nominate, there is no unitary base.

If the attachment rule for the suffix is sometimes word-based

and sometimes stem-based, there is a bifurcation in the

attachment rule; it is no longer the "same rulen (cf. Aronoff,

op. cit.:89-94). In this theory there is only one option, and

that is to let the suffix regularly attach to the word, and then

to "adjustv the shape of the incorrect output to the "vagaries

of reality" by rules of truncation.13

Below I discuss some problems with truncation rules. The

point of this discussion is that an examination of the

morphological process of word formation brings us to the same

conclusion that the examination of the syntacticosemantic

process did: the "one suffix, one rule" principle cannot be

maintained. There is no single rule of word formation that

outputs the form as well as the meaning of a derived word.

The obvious question about truncation is why it invariably

operates only before + boundary (Level I) suffixes, and never

before # boundary (Level 11) suffixes. These two classes of

affixes were first postulated by Siegel (1974),14 and are

retained by Aronoff (op. cit. :79ff. ) Given two classes of

affixes, Siegel showed that there is an interesting correlation

between the boundary and the manner of attachment of the suffix.

The + boundary suffixes may attach to stems as well as to words; the # boundary suffixes attach only to words. Truncation

recaptures this distinction at the surface; but in a theory

wherein both types of suffixes attach only to words, the

correlation between truncation and the + boundary appears

accidental.

A more important question is what exactly truncation rules

delete. They are presented as rules which "delete a morpheme

which is internal to an affix, in the following general manner:

[ [ root t A Ix+ B lY

1 2 3 ---- > I 0 3

where X and Y are major lexical categorie~.~

(op. cit. :88)

However, it appears that truncation rules must also (sometimes)

erase internal labelled bracketing.

In order to see why this is so, we must remember that this

theory of word formation purports to provide a motivated

labelled bracketing for words entering the phonological cycle.

As Aronoff observes, such labelled bracketing had previously

been sometimes determined in an "arbitrary and high-handedN

manner. In view of this, Brame (1974) proposed the Natural

Bracketing Hypothesis as a constraint on the assignment of

internal bracketings. His constraint was based on the

observation that the string constituting the domain of

application of the cycle of rules "itself shows up elsewhere as

an independent phonetic word sequencev (op. cit.:55). To express

this relation between a bracketed substring and its occurrence

as an independent word, Brame proposed the following definition,

(45) Definition

Two strings in phonological representations are said to

be equipotent if they are identical and at least one of

the two is not represented as a proper substring in

phonetic representations.

and the following constraint:

Natural Bracketinq Hypothesis

For a substring W to be bracketed, it must be equipotent to a string 6.

This constraint rules out bracketings like [[ortho[dox]]y] or

[[aristo[crat]]y], since - dox and - crat do not occur as words.

However, it still permits a bracketing [CfilJter] for a word

like filter, given the existence of a word - fill. To avoid the

possibility of this bracketing, Brame suggested a stronger

version of the Natural Bracketing Hypothesis:

Strong Natural Bracketing Hypothesis

For a substring Y of a string $J to be bracketed, )U must

be equipotent to a string 0, and the meaning of 9 must be a compositional function of the meaning of 0 and @ -

( @ minus VI 1.

(Brame 1974 5 6 - 5 8 ]

As Aronoff observes, it would be desirable for such a

constraint on natural bracketing to follow from a theory of

morphology. He remarks:

The question now naturally arises whether a constraint like that imposed by the Strong Hypothesis is a basic theoretical entity, or whether it falls out from more general principles. There obviously is some device which assigns these natural bracketings, and this device should have some other motivation than the mere fact

t h a t a s s i g n s n a t u r a l b racke t ings . . . . Within t h e theory of morphology o u t l i n e d above, a new word i s always formed by performing some phonological ope ra t ion on an a l r e a d y e x i s t i n g one. . . . The meaning of t h e new word w i l l a l s o be a composi t ional f u n c t i o n of t h e meaning of t h e word i t c o n t a i n s . S ince members of major l e x i c a l c a t e g o r i e s a r e always l a b e l l e d (N, V, Adj, Adv), s i n c e a l l r e g u l a r WFRs o p e r a t e on such l a b e l l e d words, and s i n c e t h e r e is no reason t o assume t h a t t h e s e l a b e l s a r e e rased i n t h e cou r se of t h e a p p l i c a t i o n of a WFR, WFRs w i l l , u n l e s s o the rwi se cons t r a ined , produce l a b e l l e d b racke t ings i n t h e i r output . I t i s c l e a r t h a t a l l t h e c o n s t r a i n t s imposed on int raword b racke t ings by t h e Natura l Bracke t ing Hypothesis a r e d i r e c t consequences of t h i s theory . I n f a c t , g iven t h i s theory , no o t h e r b r a c k e t i n g Is poss ib le . (op. c i t . :25)

With t h i s i n mind, consider t h e fo l lowing d e r i v a t i o n s

(suggested by Aronof f ) , assuming f i rs t t h a t t r u n c a t i o n l eaves

i n t e r n a l b racke t ings i n t a c t .

( 46 ) a. Base [evacu+ate$ b. [ i n c i s ( e ) + i o n k

WFRs [ [evacuta te t [ [ i n c i s ( e ) + i o n t t i v e l Truncat ion 1 2 3 1 2 3

Output [[evacu 1 p e e l N [ [ i n c i s ( e ) 1 fq+ive lA

The ou tpu t s of (46) a r e t h e i npu t s t o t h e phonologica l cyc le .

Here, t h e r e are l a b e l l e d bracke ts [ 1 around evacu-, and [ IN v around inc i se - . The Natura l Bracketing Hypothesis i s v i o l a t e d :

n e i t h e r evacu nor i n c i s e shows up as a word.

Consider a l s o t h e fo l lowing d e r i v a t i o n s of communicative.

I n (47) , t h e r e s u l t of t r u n c a t i n g t h e f i r s t - A t - i s shownl5; i n

( 4 8 ) , t h e r e s u l t of t r u n c a t i n g t h e second - A t i s shown. -

a. Base [communic+ate

WFRs +

[[corn~nunic+atel + A t +ion$ v [[[communic+ate] +At + ion v +i'"

1st At- t runcat ion 1 2 3 4 5

1 P' 3 4 5

- ion t r u n c a t i o n 1 2 3 4

1 2 B 4

Output [[[communic 1 A t v 1 ivelA

N

Once aga in , t h e ou tpu t makes t h e f a l s e c la im t h a t communic shows

up a s a verb, and communicate as a noun.

( 48)

b. Base [communictate] v WFRs [[communictate] +At +ionIN v

[[[communi~+ate]~+At +ion] +iveIA N

2nd -At truncation 1 2 3 4 5

1 2 b 4 5

-ion truncation 1 2 3 4

1 2 P 4

Output [[[communicate] v 1 tive] N A

Here there is an extra noun cycle on communicate, again

violating the Natural Bracketing Hypothesis.

These violations can perhaps be rectified if we assume that

truncation rules delete the labelled bracketings woutsidelt the

truncating morpheme. This will erase the Internal verb brackets

in evacuee, the internal noun brackets in incisive, and

(assuming that the second -At - truncates in communicative) the internal noun brackets in communicative.

Notice, however, that truncation must not erase the outer

(noun) bracketing in the following derivation (assuming again

that the second -& truncates):

Here the output is appropriately bracketed only if truncation

leaves bracketings intact. This means that truncation must

trigger erasure of bracketing only in some cases; it is not

clear that such a restriction can be formalized. Moreover,

truncation rules apply not cyclically, but in a block after all

WFRs (op. cit.:91). They must thus have access to internal

bracketing at any depth; they are a very powerful type of rule,

which can undo the work of any WFR.

The only motivatlon for retaining truncation in the

morphology is the "one suffix, one rulev principle. Since we

have shown (from an examination of the semantics of word

formation) that this principle must in any case be given up,

there remains very little justification for truncation rules.

But without truncation rules, the notion of a WFR falls apart.

Recently, Lieber (1980) has argued, from a purely

morphological point of view, for autonomous components of

nlexical structuret1 and vlexical semantics."16 She shows that in

languages like German and Latin, the sorts of stems for nouns,

verbs and adjectives which form bases for derivation and

compounding are also the sorts of stems which form bases for

i n f l e c t i o n a l p r o c e s s e s . T h i s i s s t r o n g s u p p o r t f o r t h e view t h a t

t h e l e x i c o n c o n t a i n s n o t o n l y words, but r e l a t e d s tems. L i e b e r

hypo thes izes t h a t each major c a t e g o r y N , V , A i n t h e l e x i c o n i s

subdivided i n t o n l e x i c a l c l a s s e s , 1 1 c o n s i s t i n g of words of t h a t

ca tegory type and r e l a t e d s tems. The i t e m s i n a l e x i c a l c l a s s

a r e r e l a t e d by a vmorpho lex ica l r u l e . " Thus f o r E n g l i s h s h e

g i v e s i l l u s t r a t i v e l e x i c a l c l a s s e s of v e r b s l i k e t h e f o l l o w i n g :

(50) ( = h e r ( 3 0 ) ) Class a : Xduce ' Xduct

( ( p r o d u c e , p r o d u c t ) , (conduce, c o n d u c t ) , . . . I

Class b: X s c r i b e ' X s c r i p t

{ ( p r e s c r i b e , p r e s c r i p t ) , ( i n s c r i b e , i n s c r i p t ) ,

( d e s c r i b e , d e s c r i p t ) , . . . I

Class c : X m i t X m i s

{ ( p e r m i t , p e r m i s ) , (commit, commis),

( t r a n s m i t , t r a n s m i s ) , ... )

( 5 1 ) ( = h e r ( 3 1 ) ) C l a s s d : X ' Xate

( (form, f o r m a t e ) , ( r e p r e s e n t , r e p r e s e n t a t e ) ,

( p r o c r a s t i n , p r o c r a s t i n a t e ) , ( e v a p o r , e v a p o r a t e ) , . . - 1

I n t h i s t h e o r y , bo th members of t h e l e x i c a l c l a s s a r e a v a i l a b l e

t o t h e (morpho log ica l ) r u l e s of word f o r m a t i o n . All al lomorphy

i s conf ined t o t h e s tem. The i n f o r m a t i o n a b o u t which member of

t h e l e x i c a l c l a s s i s t h e word, and which t h e stem, is provided

by subcategorizational frames in the lexicon: the stem is given

a frame which requires a following affix.

A detailed consideration of the morphological process of

word formation is outside the scope of this thesis. However, we

might note that in separating the morphological rules of

affixation from the semantic rules of word formation, we may

have made It possible to reconcile a theory of "lexical

structurev such as Lieberls, with Aronoff's view of ulexical

~ernantics.~ We may now weaken the hypothesis of word based

morphology to apply only to the semantics of word formation, and

say that the semantic base of (e.g.) evacuee is the word

evacuate, although the morphological base is evacu-. As Lieber

observes, the relation between the morphological base and the

semantic base provided by the morpholexical rules. Thus words

like stimulate, stimulant and stimulable can be related without

morphologically deriving the latter two forms from stimulate,

for

(v)erbs in -ate - . . . belong to a lexical class which is defined by the morpholexical rule X-Xate. Thus, every individual form Xate Is related by morpholexical rule to the correspondingwhich - is the base for affixation of -ant - and -able. (op. cit.:223)

These observations suggest that the (morphological part of) the

affixation rules given in the previous section can be refined in

the following way. First, the base may be given as a variable,

which may be either a lexical category or a stem. Secondly, the

list of morphological bases may be given in terms of

morpholexical classes. The morphological part of the tee - affixation rule may thus be as in ( 5 2 ) :

X= Stem or Verb

Forms of the base: XVy (pay, employ)

X(-Xate) (evacu-evacuate,

nomin-nominate)

etc.

Observe that with such a revision of affixation rules, we can

allow morphologically complex but semantically non-compositional

words (e.g. prob+able, rect+ify, gustatetion) to be formed by

the same affixation rules that apply in productive word

formation. Such words will have a morphological base, but no

semantic base. Thus when a noun pustation is formed from the

stem gustate, the semantic rule will look for a verb *pust or

*gustate, from which to derive its meaning. When this search

fails, no rule-governed meaning will be assigned to the word,

and an idiosyncratic meaning for this word will have to be

specified in the lexicon.

6. Concluding Remarks - I have argued that there is a type of lexical rule which

relates the argument structures (given in terms of thematic

functions) of derivationally related words, and suggested that

the domain of this rule type coincides with what Wasow (1980)

has identified as the "minorw lexical rule. Lexical items

related by this kind of rule display regular patterns of

subcategorizational correspondences; such correspondences are

here viewed as arising out of the principles governing the

assignment of thematic functions to syntactic frames. The

semantic rules relating lexical items were seen to be

independent of the affixation rules that output the

morphological forms of words. The separation of these rules in

the lexicon was supported by evidence from word formation.

A number of issues remain for further research. First, a

detailed examination of a wider range of such lexical rules is

necessary if we are to give content to the theory of thematic

functions, and to the thematic-syntactic mapping. Second, given

that we have postulated separate morphological and semantic

rules of word formation, questions arise as to the constraints

on each type of rule, and on the principles governing the

accessibility of semantic rules to affixation rules. Much is

known concerning the notion "possible morphological rulew; a

similar characterization of the notion llpossible semantic rulen

is lacking. It is to be hoped that we can isolate a small number

of "basic" semantic operations such as Causativization,

Decausativization, Inheritance, and so on, as the set of

possible relationships between argument structures. Such a

characterization of possible relationships might also allow us

to formulate general principles governing the assignment of

semantic operations to affixes. Thus we may ask if verbs and

adjectives are ever related by a rule of Inheritance; or if

nouns and adjectives are ever related by anything - but a rule of

Inheritance.

Again, we have suggested that deverbal nominalizing and

adjectivalizing affixation rules may share semantic operations,

as may denominal and deadjectival verbalizing suffixes. This

situation, where diverse affixes access a single semantic rule,

contrasts with the situation of suffixes like adjective-forming

-ed - or -able. - Here a single suffix (we have argued) may attach to either a noun or a verb; but the semantic operations accessed

by the suffix vary, depending on the category of the input. We

need to develop a theory which tells us why this state of

affairs obtains, rather than its opposite. That is, why do nouns

and adjectives function together in semantic rules, rather than

nouns and verbs? A possible answer is that only those lexical

categories which can be abbreviated in terms of a system of

syntactic features can function together in semantic rules. Thus

an investigation of the semantics of word formation may also

give us insights into the system of features for the

decomposition of lexical categories.

FOOTNOTES - TO CHAPTER FOUR

Some of the material in this chapter has been presented at the

Western Conference on Linguistics (1979) and the annual meeting

of the Canadian Linguistic Association (1980).

See also Newmeyer (1979).

He remarks: "It is not clear that we are dealing with

eparate readings rather than one tripartite or ambiguo~

three

1s one. I

lean towards the latter, but due to the present state of the art

of semantics, and perhaps to my own incompetence, I will leave

this very interesting question openw (op. cit.:38, fn.5).

For arguments that - n-ways ambiguous words should be treated as - n lexical items, see McCawley (1968).

'I Chomsky (1965 :215) suggests that "if the traditional view that

syntactic categorization is in part determined semantically can

be substantiated in any serious way, it can be expressed by a

redundancy rule determining syntactic features in terms of

semantic ones." The consistent correlation between the stativity

of a noun and the feature [-Count] would appear to be a good

candidate for such a rule: [+stative] ----- > [-Count].

5 Hust (1978) stresses this point.

6 Newmeyer (1979) makes the same point. The case of excitement,

excitation is unique among our examples in exhibiting a

morphological variation which apparently coincides with semantic

variation. However, we have noted that the conditioning factor

here is not the semantic rule, but the purely morphological fact

that the base Xcite is equally productive with the suffixes

-ment and (at)ion. -- We might note that another test for homonymity (1.e. words

produced by the same morphological operation but different

semantic operations) is to examine further derivational

processes that apply to words. Thus given a "regular" noun

satisfaction and a lfskewedll noun satisfaction derived from

satisfy [+- [+Animate]], we find that the prefix - dis- attaches

only to the latter:

(i) Mary's dissatisfaction with the dress

(ii)*the dissatisfaction of the requirements of lexical

insertion by the lexical entry

The adjective satisfied also undergoes - dis- prefixation.

(iii) Mary seems dissatisfied.

This test in fact leads Aronoff to recapture a distinction

between a noun government llbody that Vfsl1 and a noun government

"act or process of V-ing," that is provided by Jackendofffs

system in (3) above. Aronoff (op. cit.:54) first attempts to

place a negative restriction on the denominal adJectiva1 suffix

-al: - -a1 - "does not attach to the class of nouns of the form X ment (i.e. the class of nouns of the form Xrnent, where X is an -V-- independently occurring verb)," This fails to account for (iv):

(iv) The funds were used for purely governmental purposes.

Aronoff then notes that governmental in (iv) corresponds

semantically to the I1extended substantivieation" in (v), rather

than the "directly deverbaln sense of government in (vi):

(v) His government was defeated by a wide margin.

(vi) His government of the country has been roundly

criticized.

He proposes that "the difference between the two senses of

government can be represented in purely structural terms as

being that between Xrnent and - XVmentS -' governmental is clearly

derived from the former. If, therefore, we state the constraint

(on -a1 - attachment, R.A.) on X ment, then governmental is no II-

longer an exception" (loc. cit.). In other words, there are two

nouns government.

Notice that -%, when it attaches to constitution or

institution, again means "pertaining to the constitution or

institutionfi rather than "pertaining to the act or process of

constituting or institutingn (constitutional amendment,

institutional reform). This suggests that the relevant

restriction Involves the semantics of the nominal base rather

than its internal structure. As Aronoff (op. cit.:120) observes,

the suffix +(ic)al -- generally attaches to nouns "which denote

inherently definite things," e.g. &lobe, region, dialect,

continent, excrement, etc. An appropriate characterization of

its semantics might avoid the following two problems raised by

Aronoffls solution. First, it is not clear how the internal

bracketing is lost for one sense of government. The only

mechanism for loss of internal bracketing in this theory is

"semantic drift." But the meaning of government "body that V's"

is nowhere near as idiosyncratic as that of words like

transmission (of a car) and Prohibition (a period in U.S.

history), which are considered to have "drifted" and lost

internal structure; and there are other deverbal nouns with the

same meaning, e.g., administration, association, congregation,

organization.

Secondly, the negative restriction on the base violates the

Adjacency Condition of Siege1 (1977) and Allen(1978), to which

there are no other known counterexamples:

(i) No WFR can involve X and Y, unless Y is uniquely contained

in the cycle adjacent to X.

In a theory of morphology governed by this condition,

it becomes impossible for a WFR to refer to any conceivable property of the base at any possible cyclic depth. Rules which crucially involve the notions denominal, deverbal and dead ectival are not allowed

b t e s that a suffix X . . . For example, a ru e w - may attach only to denominal adjectives cannot be formulated, . . . since such a rule relates two items which are not in adjacent cycles; e.g. :

If the -a1 - affixation rule is to be restricted to non-deverbal nouns Xment, it must "look inside" the outer noun brackets.

Interestingly, Pesetsky (1979:36) cites an adjectival

suffix -isk - in Russian which does not attach to "abstract nouns1' which "lack semantically well-formed plurals," (e.g., nouns in

-ost - or -istv), and observes: -

the verb u ravitl 'to administer1 yields a nominaliza -R- on upravilenije, which has both meanings of the English word administration , 1. e. a body of people who administer and the process of administering . . . -isk - can apply to this word, but the meaning of the result is compositional only of the former, non-abstract

meaning.

7 Notice that we need to distinguish the optionality of a - to-NP

complement to assurance in its non-stative reading, from the

absence of a - to-NP complement to assurance in its stative

reading. Compare the discussion of optional elements in verb

subcategorization in Chapter One, fn.1.

8 The possibility of two nominalizations for assure suggests

that like depress, etc., this verb takes both Experiencer and

non-Experiencer objects. As we would expect if the object of

assure on one reading were an Experiencer, there is a related

adjective assured:

(i) The h a r e seems a s s u r e d t h a t he w i l l win.

The [-Count] noun and t h e a d j e c t i v e b o t h undergo

s e l f - p r e f i x a t i o n . - (ii) The h a r e seems ve ry s e l f - a s s u r e d .

(iii) The h a r e ' s s e l f - a s s u r a n c e was remarkable .

However, I have been unab le t o f u l l y u n r a v e l t h e complexi ty

i n t h e semant ics of a s s u r e . Thus t h e v e r b and t h e a d j e c t i v e have

a n o t h e r r e a d i n g wherein t h e v e r b i s n o t a "verb o f s a y i n g Y f ' and

n o t h i n g i s claimed about t h e menta l s t a te o f t h e v e r b ' s o b j e c t

o r t h e a d j e c t i v e ' s s u b j e c t :

( i v ) H i s slow and s t e a d y pace a s s u r e d t h e t o r t o i s e o f

v i c t o r y .

( v ) Vic to ry now seems a s s u r e d t o t h e t o r t o i s e . ( s a i d by

a s p e c t a t o r )

( v i ) The t o r t o i s e now seems a s s u r e d o f v i c t o r y . ( s a i d by

a s p e c t a t o r )

There i s no noun a s s u r a n c e on t h i s r e a d i n g .

Rardin (1975) n o t e s t h e e x i s t e n c e of homonymous [-Count]

and [+Count] nouns, where " t h e f e a t u r e [+Count] a p p e a r s t o be

a s s o c i a t e d w i t h i n t e r p r e t a t i o n s of one t y p e , [-Count] w i t h

a n o t h e r f 1 ( o p . c i t . : 3 6 ) . Using t h e many/much - t e s t , he shows t h a t

d e c i s i o n i s [+Count], but i n d e c i s i o n i s [-Count] (examples from

Rard in ) :

( v i i ) Did Martha make any d e c i s i o n s ? Some, bu t n o t many.

( v i i i ) Did Martha show any i n d e c i s i o n ? Some, b u t n o t much.

However, since indecision is stative and synonymous with

indecisiveness, he concludes that it is deadjectival. He gives a

similar argument for a dual, deadjectival as well as deverbal,

source for determination. Compare the ambiguity of (ix) (example

from Rardin) :

(ix) The committee's determination(s) angered Bill.

[-Count] reading: its determined state

[+Count] reading: what it determined

Notice that there is an adjective determined which appears

to be derived from the intransitive verb determine:

(x) Goodenough determined to win the election.

(xi)Goodenough is determined to win the election.

If the subject of (x) is an Experiencer, this might be an

argument that the Causer is optional in the input to the rule

deriving the adjectival passive. However, such a modification

does not appear to account for (xi1)-(xiii):

(xii) John opposes the changes.

(xiii) John is opposed to any change.

An explanation of these facts must await further research.

9 Thus it may be specified that the +ee - WFR requires verbs which are transitive and take animate objects, or that the prefix - re#

attaches only to verbs that entail a change of state (~ohn

punched Bill, *John repunched Bill; John punched - the holes - in

the paper, John repunched the holes in the paper). - - --

10 There is in fact a third class of able adjectives. These are

the ones whose meanings are related to intransitive verbs. (Hust

(1978) notes this fact, and credits the observation to Chapin

(1967)

(i) The weather is changeable.

(Cf. The weather changes. )

(ii) John is adaptable.

(Cf. John adapts easily. )

(iii) Foodstuffs are perishable.

(Cf. Foodstuffs perish. )

(iv) The weather is variable.

(Cf. The weather varies from day to day.)

These adjectives can be given the generalized reading "capable

of - X-ing (where - X is the base)." So our hypothetical WFR (18)

(below) strictly needs three branches, not two; but we shall

ignore this point in the discussion.

Compare Aronoff Is comment (op. cit. :48) on the treatment of

the ambiguity of fashionable ("in fashion," or "capable of being

fashionedw) and sizeable ("having great size," or "capable of

being sizedn): "Such a consistent correlation of homophony and

ambiguity can only be accounted for on the hypothesis that we

are dealing here with two different affixes, each with its own /

meaning and each with its own base."

Conversely, if - n different affixes take part in a semantic operation, we need - n different semantic operations.

12 We shall thus postulate homonymous affixes only in the cases

where (i) the Level specifications of the affixes are not the

same, or (ii) where the lexical categories output by the affixes

are not the same. (Examples of the former kind are the

adjectivalizing suffixes +able and #able -3 discussed by Aronoff

(op. cit.). Examples of the latter kind would be (i) the -en - in

redden, which is a verbalizing suffix, and the -en - in broken, which is a variant of the adjectivalizing suffix -ed* -3 (ii)

noun-forming and adjective-forming -a1 - (approve, approval, &lobe, global); and (iii) verb-forming and adjective-forming

un- (untie, unexplored).) A variation in the "semantics of the -

suffix" will not be taken as evidence for homonymous suffixes.

Allen (1978) makes a similar point, that the identification

of a suffix should be effected independently of its semantics.

Compare her comment on the adjectivalizing suffix -ed: -

. . . it was demonstrated that the -ed suffix is a Level 11, word-boundary suffix, #ed. Data presented in support of this argument included examples of both denominal and deverbal -ed. In all cases, denominal and deverbal -ed - showed idexical types of interactions with other morphological affixes. I will therefore assume that -ed - is a single adjectivalizing suffix which attaches to both nouns and verbs. (op. cit.:289, fn.21).

One argument given by Aronoff (op. cit. :48) for homonymous

suffixes in the case of denominal #able and deverbal #able is - that "the denominal adjectives always take the nominal ending

#ness and never + a (fashionableness, *fashionability;

sizeableness, *sizeability), while the deverbal adjectives show

no real preference (acceptableness, acceptability; movableness,

m~vability).~~ This argument is, however, defective. It must be

remembered that Aronoff postulates a total of not two, but three

-able suffixes: two deverbal (table and #able), and one - - denominal (#able). - What is at issue is not the distinction between the +able - and (either of) the #able suffixes; the issue - is whether there is a distinction between the two #able - suffixes. Now tity and #ness suffixation can serve as a test for

table versus #able. - This is because, under the hypothesis of a llLevel-orderedll morphology, t boundary suffixation precedes #

boundary suffixation. (Thus the fact that movable takes tity as

a suffix shows that the -able in that form is table and not -' #able.) Note (however) that ta and #ness suffixation cannot

distinguish between deverbal #able - and denominal #able. -

l3 Aronoff also invokes truncation for dealing with two other

problems, namely (i) the problem of productive word formation

from a stem which never shows up as a word, and (ii) the problem

of derived words sharing a meaning which is absent from the base

word. The first problem is illustrated by instances of *z, Xion,

Xive i n E n g l i s h ( e . g . , * i n c i s e , i n c i s i o n , i n c i s i v e ; c f . Aronoff

(op. c i t . : 2 9 ) f o r documentat ion of t h i s d a t a ) . To a c c o u n t f o r

t h i s , Aronoff s u g g e s t s t h a t ( a l l ) Xive/Xory words a r e d e r i v e d

from Xion and n o t X. T h i s s o l u t i o n i n v o l v e s t r u n c a t i o n . I n -' - s u p p o r t o f t h i s a n a l y s i s , Aronoff p r e s e n t s a h i s t o r i c a l argument

( " t h e - ion - form e n t e r e d t h e language b e f o r e t h e -* formq1), a

distributional argument ( l1the t o t a l number of words of t h e form

Xion fa r outnumbers t h e t o t a l number o f words i n a l l o t h e r

s u f f i x e s combined1@), and a t h i r d argument , which goes as

f o l l o w s :

when X does o c c u r a s an independent ve rb , and t h e s e m a n ~ i c s of X and Xion do n o t co r respond e x a c t l y , t h e meaning o f -9 ~ i v e e t c . , a lways cor responds t o t h a t of Xion. -

None of t h e s e arguments Is ve ry f o r c e f u l . From t h e p o i n t of view

of o u r p r e v i o u s d i s c u s s i o n , what t h e s e d a t a seem t o s u g g e s t i s

t h a t t h e a f f i x a t i o n r u l e s f o r d e v e r b a l nouns and a d j e c t i v e s

o f t e n s h a r e t h e same semant ic r u l e . Tha t i s , I n s t e a d of a

two-step d e r i v a t i o n where ( a ) t h e - ion - r u l e adds o r s u b t r a c t s a

meaning from i t s b a s e X, and ( b ) t h e - i v e / - o u - r u l e s o p e r a t e on

t h e a b s t r a c t base Xion t o r e t a i n t h i s ( m o d i f i e d ) meaning, we can

e n v i s a g e a d e r i v a t i o n where a r u l e which m o d i f i e s t h e meaning of

X i s s h a r e d by b o t h - ion and - i v e / - 9 . T h i s might p r o v i d e a - - - c l u e t o t h e paradigm *$, Xion, Xive/Xory.

1 4 S i e g e 1 argued t h a t Level I a f f i x a t i o n r u l e s preceded Level I1

a f f i x a t i o n r u l e s , w i t h t h e c y c l i c stress r u l e s i n t e r v e n i n g . She

showed t h a t such a n o r g a n i z a t i o n of t h e morphology e x p l a i n s why

( i ) Level I a f f i x e s a r e s t r e s s - d e t e r m i n i n g and t a k e s t r e s s ,

w h i l e Leve l I1 a f f i x e s a r e s t r e s s - n e u t r a l and stressless; ( i i )

t h e p h o n o l o g i c a l c o n d i t i o n s on t h e base of t h e a f f i x may i n c l u d e

i n f o r m a t i o n a b o u t s t r e s s f o r Level I1 a f f i x e s b u t n o t f o r Level

I a f f i x e s ; and (iii) Level I1 a f f i x e s a r e g e n e r a l l y found

"outs ide l l Level I a f f i x e s .

For f u r t h e r e v i d e n c e i n f a v o r o f a n l e v e l - o r d e r e d l l

morphology, c f . A l l e n (1978) and Pese t sky (1979) ( f o r ev idence

from R u s s i a n ) .

1 5 Aronoff l e a v e s open t h e q u e s t i o n of which - A t t r u n c a t e s i n

Xate+Ation. H i s r u l e ( 1 8 ) (op . c i t . : 9 5 ) d e l e t e s t h e f i r s t - A t .

But t r u n c a t i o n of t h e second - A t a p p e a r s t o b e n e c e s s a r y t o

accoun t f o r some stress f a c t s noted by Brame (1972) , as Aronoff

(op. c i t . : l O l , 114) o b s e r v e s . Brame shows t h a t t h e s t r e s s /

d i f f e r e n c e i n Xatory words l i k e d i v i n a t o r y ( p r e s u f f i x a l stress) /

and a s s i m i l a t o r y ( p r e - p r e s u f f i x a l s tress) i s p r e d i c t a b l e from

whether t h e morpheme - a t e - i s p a r t of t h e b a s e v e r b o r p a r t of

t h e s u f f i x . I n Bramefs a n a l y s i s , t h e s e words must have t h e

b r a c k e t i n g s shown below when they e n t e r t h e p h o n o l o g i c a l

component:

[ [ d i v i n ] A t +Or +y] [ [ ass imi l+At ] O r +y] V A V A

I n A r o n o f f l s t h e o r y , a s s i m i l a t o r y w i l l have t h e f o l l o w i n g

derivation prior to truncation (ignoring the brackets):

assimilate

assimilate + At +ion

assimilate + At +ion +ory

Two rules of truncation, -At truncation and -ion truncation,

must apply to this output. Now if the first -At truncates, the

-At in assimilator^ will appear outside the verb brackets.

On the other hand, if the second -At truncates, the -At

truncation rule here cannot be the same rule as that in evacuate

(as Aronoff observes). Notice also that truncating part of the

suffix is equivalent to saying that the form of the suffix which

attahes to Xate is -ion and not -ation. - -

l6 She writes:

. . . there is no more reason to believe that semantics should be part of the formal mechanics of word formation, than there is to suppose that semantics is part of the formal mechanisms of sentence syntax (i.e., phrase structure, transformations). It has long been a basic tenet of generative syntax that the syntax and semantics constitute autonomous components of the grammar. The claim will be made below that the "syntacticw or structural aspects of word formation should also be autonomous form lexical semantics. (op. cit. : 109)

BIBLIOGRAPHY

Allen, M.R. (1978). Morphological Investigations. Doctoral dissertation, The University o- Storrs, Connecticut.

Amritavalli, R. (1979) The representation of transitivity in the lexicon. Linguistic Analysis 5:1, 71-92.

. (forthcoming). Expressing cross-categorial selectiona1 correspondences: an alternative to the X 1 Syntax approach. Linguistic Analysis.

Anderson, S.R. (1977). Comments on the paper by Wasow. In P.W. Culicover, T. Wasow, and A . Akmajian, eds., Formal Syntax. Academic Press: New York.

Aronoff, M. (1976). Word Formation in Generative Grammar. Linguistic Inqui-onograph Nox. M.I.T. Press: Cambridge, Mass .

. (1978). Lexical representations. In D. Farkas et al., eds., Papers from the Parassession on the Lexicon. -- -- Chicago Linguistic Society, Chicago, Illinois.

Brarne, M.K. (1972). The segmental cycle. In M.K. Brarne, ed., Contributions - to Generative Phonologx. Univ. of Texas Press: Austin.

(1974). The cycle in phonology: stress in Palestinian Maltese and Spanish. Linguistic Inquiry. 5, 39-60.

8 (1977). Alternatives to the Tensed-S and Specified

Subject Conditions. Linguistics and Philosophy 1:381-411. -

Bresnan, J. (1978). A realistic transformational grammar. In M. Halle, J. Bresnan and G. Miller, eds., Lin uistic Theory and Psychological Reality. M. I .T. Press : C a b a s s . -

. (1980). Polyadicity: Part I of a theory of lexical rules. Unpublished ms., M.I.T.

Chapin, P. (1967). On the S ntax of Word-Derivation in English. Doctoral disser~t~ijii,ibr;T.

-

Chomsky, N. (1964). Degrees of grammaticalness. In J.A. Fodor and J.J. Katz, eds., - The Structure of Language. Prentice-Hall: Englewood Cliffs, N.X

. (1965). As ects of the Theory of Syntax. M.I.T. Press : Cambridge* - - -

. (1970). Remarks on nominalization. In R. Jacobs and P. Rosenbaum, eds., Readings English Transformational Grammar. Ginn and Co. : Waltham, Mass.

. (1973). Conditions on transformations. In S. Anderson and P. Kiparsky, eds., A Festschrift - for Morris Halle. Holt, Rinehart and winston: New York.

. (1977). On wh-movement. In P. Culicover et al., eds., Formal Syntax. Academic Press : New York.

DeArmond, R.C. (1969). The concept of word derivation. Lingua 22 ~329-361.

. (1980). On the lexical representation of o en in English. In R.C. DeArmond, ed., Essays in Memory -- of h Hust. Linguistic Research Inc.: Edmonton. -

Emonds, J.E. (1970). -- Root and Structure-Preserving

Transformations. Doctoral dissertation, M.I.T.

. (1976) A Transformational Approach - to English Syntax. Academic Press: mew York.

Frieden, R. (1975). The analysis of passives. Language 51 : 384-405.

Gruber, J. (1965). Studies in Lexical Relations. Doctoral dissertation, M.I.T. ~esoduced by Indiana University Linguistics Club.

. (1976). Lexical Structures - in Syntax - and Semantics. North-Holland: Amsterdam.

Halle, M. (1973). Prologomena to a theory of word formation. Linguistic Inquiry 4:3-16.

Hornstein, N. (1977). S and X 1 Convention. Linguistic Analysis 3~137-176.

Hust, J.R. (1977). The syntax of the unpassive construction in English. Linguistic Analysis 3:31-63.

. (1978). Lexical redundancy rules and the unpassive construction. Linguistic Analysis 4:61-89.

Hust. J.R. and M.K. Brame. (1976). Jackendoff on Interpretive semantics: a review of semantic Interpretation in Generative Grammar by R. Jackendof f . Linguistic Analysis 2543-277

Jackendoff, R.S. (1972). Semantic Interpretation - in Generative Grammar. M.I.T. Press: Cambridge, Mass.

(1974a). Introduction to the X' Conventiofi* Mimeograph, Indiana University Linguistics Club, Bloomington, Indiana.

. (1974b). A deep structure projection rule. Linguistic Inquiry 5:481-506.

. (1975). Morphological and semantic regularities in the lexicon. Language 51:639-671.

. (1976). Towards an explanatory semantic representation. Linguistic Inquiry 7:89-150.

. (1977). X-bar S ntax A Stud of Phrase Structure. Lingusiticrniry onograp 0.2. lifZ7K Press : Cambridge, Mass.

+ - T i #

. (1978). Grammar as evidence for conceptual structure. In M. Halle et ale, eds., Lin uistic Theor and Psychological Reality. M.I.T. Press: 6 -

Katz, J.A. and J.J. Fodor. (1964) The structure of a semantic theory. In J.A. Fodor and J.J. Katz, eds., The Structure of - - Language. Prentice-Hall, Inc., Englewood Cliffs, N.J.

Lakoff, G. (1970). Irregularitx - in Syntax. Holt, Rinehart and Winston: New York.

Lieber, R. (1980). On the Organization of the Lexicon. Doctoral -- -- dissertation, M.I.T.

Marchand, H. (1969). The Categories and Types of Present-Day - English word-~ormxon. Beck: M u z h .

Martin, S. (1972). -ive and other - ion based suffixes. Unpublished papeFM.1.~.

Matthews, P.H. (1974). Mor holo An Introduction to the Theory of Word Structure. M n ~ e r s i t y Press, TSnbridge. --

McCawley, J.D. (1968). The role of semantics in a grammar. In E. Bach and R.T. Harms, eds., Universals in Linguistic Theory. -

Holt, Rinehart and Winston: New York.

Newmeyer, F.J. (1979). nPsychological predicatesu and the structure of the lexicon. Paper presented at the Western Conference on Linguistics, Simon Fraser University.

Oehrle, R. (1975).The Grammatical Status of the English Dative Alternation. Doctoral dissertation,Mx.C

Ostler, N. (1979). Case Linkin a theor of case and verb diathesis a p p l i Z T € o d a ~ d i 7 Z EZEoFZI - dissertation, M.1.T

Pesetsky, D. (1979). Russian morphology and lexical theory. Unpublished ms., M.I.T.

Rardin, R.B. (1975). Studies in Derivational Morphology. Doctoral dissertation, Pf.ET.

Roeper, T. and M. Siegel. (1978). A lexical transformation for verbal compounds. Linguistic Inquiry 9:199-260.

Siegel, D. (1973). Nonsources of unpassives. In J. Klmball, ed., Syntax - and Semantics. Seminar Press: New York.

. (1974). To ics in English Morphology. Doctoral dissertation, M+ -

. (1977). The adjacency condition and the theory of morphology. In Proceedings --- of NELS VIII, Amherst, Mass.

StrauSs, S.L. (1979). Against boundary distinctions in morphology. Linguistic Analysis 5:4, 387-419.

Walker, J. (1936). Walker's Rh min Dictionar . Revised and enlarged by L.H. Dawson. - d k & : d

Wasow, T. (1977) . T r a n s f o r m a t i o n s and t h e l e x i c o n . I n P . C u l i c o v e r e t a l . , e d s . , Formal Syntax . Academic P r e s s : New York.

. (1980) . Major and minor r u l e s i n l e x i c a l grammar. I n T. Hoeks t ra e t a l . , e d s . , L e x i c a l Grammar. F o r i s P u b l i c a t i o n s : Dordrech t , The Nether lands .

Wierzbicka , A . ( u n d a t e d ) . Semantic P r i m i t i v e s . ( T r a n s l a t e d by A . Wierzbicka and J. Besemeres.) Athenaum V e r l a g : F r a n k f u r t .