A CATEGORICAL APPROACH TO THE ENGLISH "DERIVED NOMINAL CONSTRUCTION": 沖縄地域学リポジトリ

Title

A CATEGORICAL APPROACH TO THE ENGLISH

_{"DERIVED NOMINAL CONSTRUCTION"}

Author(s)

Aniya, Sosei

Citation

沖縄短大論叢 = OKINAWA TANDAI RONSO, 4(1): 65-83

Issue Date

1989-03-31

URL

http://hdl.handle.net/20.500.12001/10627

A CA l'aC£0RICAL APPROACH TO THE ENGLISH

... DERIVED NOMINAL CONSTRUCTION"

Sosei Aniya 0. Introduction

In "Remarks on Nominalization ", Chomsky· (1970) reported a parallelism between the complements of vetbs·.and those of corre-sponding nouns, so-called nominals. On the strength of transforma-tional discipline, he attempted to capture what 'Wl8S· taken to be the

grammatical :relations among phrases such as destroy the city/ destruction of the city/ the city~s destruction. Following the spiirit

of Chomsky's a.nalysis, transformationalists revisited the CIDD'Struction from different angles in accordance with theoreticill developments. This article uncovers fundamental prablems in transformational analyses of the construction such as those underlined above. Representative analyses appeared in Jackendoff (1977), Stowell (19 81 ), Chomsky (1981 ), and Emonds ( 1983 ) will be examined ana-lytically and critically one by one vis-a-vis, ·thus pointing out

not only problems per se but their consequences to the nucleus

of each analysis. As a competitor to those analyses, a categorical treatment will be introduced. ·It will be shown that the competitor provides a straightforward and natural account for the issues which will be elucidated shortly.

1. Critique of transformational approach I:Jackendoff's analysis

Guided by X-bar theory, Jackendoff ( 1977) proceeds to ~he analysis of derived nominal constructions ( viz., the underlined phrases) illustrated below.

(1)

b. explain the result /explanation of the result c. establish a. claim / establishment of a claim

Following Chomsky's (1970) position that in the unmarked cas:e, verbs\ and nouns have the same complement structures, Jackendoff

assu~es that the derived nominal construtions in (1a ), (lb), and ( 1c) are derived from the underlying forms destruction the city, explanation the result, and establishment a claim, respectively

by means of an obligatory transformation inserting of. The rule

is formalized as in (2) .

(2) Of-Insertion ( Jackendoff, 1977:70 ) 1:

( N' N-(Prt)- NP-X) ---... 1- 2- of+3- 4 OBLIGATORY

Of particular interest to us here is the resultant structure ( Sb ) which would be obtained by applying Of- Insertion to ( 3a ) .

(3) Of- Insertion

a. ( N' destruction ( NP the city) ) ---... b. ( N, destruction of- ( NP the city ) )

In terms of tree diagram, ( 3b ) might be expressed as m

(4)

N'

~

N

N"

I

of

destruction the city

Jackendoff dismisses Chomsky's ( 1970) then suggestion by arguing that the inserted of in ( 3b ) cannot be the preposition of since there is no PP in the underlying structure. Thus, Jackendoff con-cludes that the element of must be a specified grammatical

39:•££:70) . To him Of ....: Insertion is a mechanism which carries out an operation of inserting a semantically dummy element of between the head noun and the· immediately. following NP so that a grammatical surface structure such as (3b ) gets to be surfaced. This scenario seems to conform to the X- bar theory and strict su bcategorization properties.

Some problems, however, arise from an unexpected··,side when Jackendoff tries to correlate a derived nominal construction with

another with the help of movement transformations. To stte this in proper perspective, we shall begin with a. list of rules relevant to the present discussion. ( Where the lOwer ·case de.no:tes, an empty node. )

(5) Agent- Postposine: ( Jackendoff, 1977:88 ) :

( NP-X-N-Y- ( by np) -Z) --- np-2-3-NP PP 4-5-1-7 OPTIONAL (6) NP- Preposing ( Jackendoff, 1977:87 ): np- Y- N-NP-Z---4-2-3-0-5 OPTIONAL

(7) Poss- Insertion ( Jackendoff, 1977:102): ( N.., N"'- Y- N") ---1 +poss -2-3

OBLIGATORY

Armed with the above rules, Jackendoff demonstrates a der-ivation of the phrase the city's destruction by the enemy. Con-sider,

(8)

. . Agent- Postposing a. the enemy destructiOn the city by

np---. . NP- Pre posing b. np destruction .the city by the

enemy·---. . Poss- Insertion

c. the city destruct1on 0 by the enemy---d. the city's destruction 0 by the enemy

In addition to the above solution, Jackendoffs system seems to be capable of generating phrases such as destruction of the city by the enemy and the enemy's destruction of the city.

following derivations.

(9)

Observe the

· b Agent- Postposing

a. the enemy destruction the c1ty y np---+

d · h · b Of- Insertion

b. 0 estructlon t e c1ty y the enemy---+ c 0 destruction of the city by the enemy

(10)

. . Of- Insertion

a. the enemy destructiOn the c1ty ---.

. . Poss- Insertion

b. the enemy destruction of the.c1ty ···----. c. the enemy's destruction of the city

Thus, Jackendoff's system appears to work perfectly ·well.

Recall now that we already saw how Jac.kendoff derived the phrase destruction of the city from its underlying form destruc-tion the city by means of Of- Insertion. With this in mind, we raise the following question: How does Jackendoff account for the phrase the city's destruction? A moment's reflection is likely to show that the phrase in question . ·would be derived from the un-derlying form destruction the city by applying two rules succes-sively, NP- Preposing and Poss- Insertion. This idea, however, does not work since NP- Preposing requires an empty NP in subject position. Apparently, a wrinkle is needed here. Jackendoff ( 1977:88 ) suggests that the correct underlying form is rather np

destruction the city with an empty NP in the subject position. If we follow this line of thought, the following derivation may be obtained.

(11)

. . NP- Preposing

---. . Poss- Insertion b. the city destructiOn 0 ---+

e .. the city's deskuciton 0

The above solution rather prompts a further question with respect to NP- Preposing. As stated in (6). NP- Preposing Is an optional transformation. Thus, we ask: What happens if we do not apply the rule in the first place ? In that case, the opera-tion followed by The- Inseropera-tion (See (12) ~low,) may. take place. This gives rise to the derivation pictured in (13).

(12) The- Insertion ( Jackendoff, 1977:102 ): ( N"' np..,.Y-N") ---+the-2-3

OBLIGATORY

(13)

. . Of- Insertion

a. np destructiOn the city---'-+

. . The- Insertion

b. np destructiOn of the city·---+ c. the destruction of the city

In essence the role of The- Insertion here is to fill an empty NP, thus preventing it frQm surfacing. But this means that the article the as a whole is equated lwith an :NP isuch as the enemy, the city, etc. This seems to be a far- fetched assumption. More-over, the following phrase without the determiner the at the ini-tial position is by all means a well- formed string. 3

(14} destruction of the city

At this point one might pose a question concerning the Agent-Postposing. The question can: ·be worded as follows: What prevents an ungrammatical phrase such as the enemy's destruction of t~ city by from surfacing? As stated in (5), the rule in question is an optional transformation whose structural description contains an empty np.

2. Critique of transformational approach

n:

Stowell's analysis

Within the framework of Stowell's ( 1981 ) Cass assignment theory, Case may be assigned by categories such as rp, V, and tensed clause. On the other hand, (

+

N ) categories such as N, and A may not assigh Case. It is conspicuous .that the first three Case.;.., assigning categories. do not f"Orm a natural class since tensed clauses, unlike P and V, bear the feature ( +N) Although this is not a trivial matter, we do not pursue this issue here because it is outside of our objective at hand. Instead, we proceed to more stimulating topic. It deals with NP arguments appear-ing at D- structure in a position which is not adjacent to a (- N) head. Representatives of such case are derived nominal construc-tions where Case assignment is done in relation to of- insertion. The existence of dummy preposition of, however, seems to raise

a technical problem with respect to the categorial status of d as exemplified in fi~.

fi~ ( N, the destruction ( ct of ( NP the city) ) )

The spirit of Chomsky's ( 1970) theory of the lexicon is to relate certain verbs and their corresponding nouns. Thus, for example, destruction is included in the lexical entry for destroy.

Therefore, the theory predicts that both destroy and destruction

subcategorize for NP complements. This is the reason why the underlying form for fi~ is set up as destruction the city .. Now

by the Projection Principle, representations projected from the lexicon must observe the subcategorization properties of lexical items at each syntactic level, i.e., D- structure, S- structure, and LF. (See. Chomsky, 1981). Hence, if destruction subcategorizes

for NP complement , it must select NP at S- structure. also. It

count for subcategorization.

The question concerning the status of

a

in (1$ is closely connected with the theory of Case assignment. If the operation of of- insertion saves the NP complement from being ·caught in the net of Cass Filter,4 then the dummy preposition of is taken

to be a Case- assigner. This followS' from Case Assignment under Government ( CAUG

)~

The dummy preposition of in (1$ is ad-jacent to its object. Th.is satisfies CAUG. Thus , we might put forward the following assumption: By definition, of in (15) is the head of .a·~ Mo!reover; ( - N) categories such as P and V may assign Case as discussed earlier. Therefore of in (1$ is a prep-osition hence

a

must be PP. This outcome is, however, in direct . conflict with the aforementioned conclusion drawn from the strict subcategorization and Projection Principle.

To break the deadlock, Stowell adopts Chomsky's ( 1981 ) sug-gestion in which the following two assumpticms are· crucial: Firstly, the strict subcategorization must be always ·satisfied at D-structure. Secondly, the Projection Principle is insensitive to categorial distinctions between phrases. What this means is that an:y phrase XP may appear as a complement of N at S- structure. The contradiction is now resolved. Thus, the above - mentioned conclusion that the status of

a

is PP would be.· a sufficient one for Stowell.

There is yet another possible solution proposed by Chomsky

( 1981:51): The of- insertion rule forms a neutralized NP-PP of the form ( -

V )

which shares properties of both NP and PP. This does not seem to be a plausible solution. If tHis line of thought is followed then there seems to be a possibility where an un-grammatical surface form such as

*

of the city destruction might

be generated in conjunction with Move~

a .

Suppose that the from ( -

V

J

is granted, the categorial status of the inserted of should still be spelled out at some point since P is a case-assigner whereas N is not in this particular case.

Let us examine another problem which involves Case and 1J -role assignment, and an NP- internaL. application of Move-

a.

Stowell expresses difficulty in dealing with the problem pictured in fiG).

U6)

a. ( the destruction of- ( the city

J J

b. ( ( the city's

J .

destruction ( e

J . J

. c.

* ( (

the city's

J .

destruction

of~

( e

J . J

1 1

In both cases ( 16a ) and ( 16b), the condition on IJ -role as-signment 6 is satisfied, hence IJ -

rGl~

assignment can be proceeded without any problem. Stowell thinks that the ungrammaticality of ( 16c) may be due to the violation of Empty Category Principll possibly with some involvement of 'Case clash'. The scenario assumed here by Stowell would be as follows: First, the of-inser-tion rule applies at D- structure creating the result illustrated in

( 16a) . Second, Move-

a

comes into play and pre poses the NP

the city into the prenominal position. Finally, genitive Case is

assigned to the NP the city at S - structure.. Thus, structure

( 16c) would surface creating the most undesirable situation as noted by Stowell.

There seems to be a non- trivial point overlooked in the above chain of processes. After the NP complement was moved into the J>renominal position by Move-

a,

a s-trange thing hap-pened. The determiner modifying destruation disappeared. Needless to say, this requires an explanation.

3. Critique of transformational approach

m:

Chomsky';s analysis In Chomsky's(1981)flra:mework, the undel:'lying :form destruction the city may undergo either one of the following wo opera· tions: On the one hand, we may apply of- insertion rule and obtain the result as in ( 16a ) . On the other hand, we may apply Move-a first and then the genitive - s assignment. This gives rise to the structure as in ( 16b) . In either: case, the Case Filter is satisfied.

A problem which might be posed here has to do with the formula of of- insertion rule. Chomsky ( 1981:50) formalizes the rule of of- insertion as follows.

U'll

NP--- ( of) NP in env.: (

+

N) p

Chomsky leaves open the exact formulation: of the rule. If the rule adjoins of to-NP forming ( NP of NP), then. the undesirable result

* ( (

of the city's) . destruction ( e) . ) might be generated

gant-1 1

ing that Move- a and genitive-s· assigJ!I:ment were in force. (But the ungrammatical phrase might be accouted for if it is seen as a case of "Case conflict "). ·

The rule

U'll

may apply to "transitive' adjectival "·forms such as proud John and produces proud of John .as the result. In s1,1ch cases, however, Move-a will not allow the construction to escape the Case Filter because genitive Case is not assinged in pre- adjectival position. This necessitates the obligatory appli-cation of the of- insertion. (See Chomsky 1981:49;140 fn 26 ). It seems rather natural to take the idiomatic phrase be proud of as a single constituent, hence the of at issue is nof the product of the of- insertion.

4. Critique of transformational approach W: Emonds' analysis Quite a different solution for the derived nominal

construe-tion is proposed by Einonds"( 1983 ). Within his framework, Weak-ened Sisterhood Restriction on Su bcategorization (See (1.8)) induces certain deep structure empty nodes such as in

fift.

QB) Weakened Sisterhood Restriction on Subcategorization ( WS

R ):

...-.,

Given X,

+

C D ···, all lexical (non- empty) material under. sisters of X must be dominated by ( that is, can be

,..._

factored into ) a sequence C D · ···

( Emonds: 1983 ) X

---X pp

lex~cal

- - - -_p

_D

I

I

0 lexical

Thus, for example, a deep structure subcategoi:ization such as destruction, N,

+ _

NP induces the empty deep structure P node analogous to Q~ above by the WSR.

The induced empty node in Q~ is said to be justified by the insertion of grammatical formative of, hence the phrase destruction of the city may be obtained for instance. But one might pose the question: What happens if the constituent fol-lowing the induced empty node moves away ? In that case the induced empty node is deleted by Node Elimination.

(20) Node Elimination:

Suppose X, D 'induces an empty node C in the structure ( B C ·•· D· · · ) , D the leftmost liermjnal in B. If D becomes an empty, non- variable D. co indexed with a local

ante-1

J'.

C is obligatorily delt;!ted when the coiri.de:xing· rule applies. ( Emonds, 1983 ) Granting the Node Elimination, one might expect tthat the empty node Pas in

(Lit

w.ould disappeJlr ,when. the: 'NP 'O)mplement

the city moves into the prenominal position and coindexed with

its trace at the terminal D. Emonds' system . seems :to be quite capable of handling derived nominal constructions such as in C!U ·

provided that the relevant transformational rules were applied to

the deep structure (2~. ( See Emonds, 1976 ) .

a. the city's destruction by the enemy b. the destruction of the city by the enemy

~ _NP ----~

=::::::::::::::----NP N pp PP

thl

I

/ \ .

./\P

destruction P NP enemy

I

I

I

I

t,

_{: : ~...:>.} Ah _{t e c1ty} • b _{y ~...:>.} A

!

! ___ --- -- ---~--- ---_j

i

Agent NP- Postposing , '---·- --- --- ---l Complement NP- Preposipg

Although Emonds' proposal seems to gain some advantage over the past analyses, it is ·not free from problems. First , of- insertion ( which is not formalized in Emonds ( 1976; 1983 ) ) should be constrained in such a way so that it ·does not operate in the derivation of ( 21a ) . Otherwise, the Node Elimination would be blocked. Second, the two movement rules, the Agent NP-Postposing and the Complement NP-Preposing as pictured in

(2~

would lead to 8 -criterion violations. 8 Finally

~

one might inquire into the origin of the determiner the appeared in the

ini-tial position in ( 21b) .

So far we have witnessed the details of the past analyses with respect to the derived nominal construction. To summarize and recapitulate we give the following list of crucial points.

t.!~

a. The status of the optional Agent- Postposing is questionable. ( Jackendoff, 1977 )

b. The strange disappearance of the determiner the takes place

in connection with the complement NP- Preposing. ( Stowell, 1981)

c. The formularizational problem is found with respect to the of- insertion rule. ( Chomsky, 1981; Emonds, 1976)

d. The interaction of rules leads to fJ -criterion violations. ( Emonds, 1976 )

As we have seen,: each of the past analyses is different in treating the derived· nominal construction, but they all meet on common ground: In the unmarked case, verbs and nouns have the same complement structures. Fox example, destroy and de-struction both subcategorize for NP complement. And the key

for accounting derived nominal phrases is the transformational rules ( together with Filtering devices ) . In the following sec-tion, we provide a competitor to the past analyses.

5. A categorical approach

The theory adopted hereafter for analyzing the issue is called categorical grammar originated by Brame ( 1984; 1985 ) . In oppo-sition to the doctorine of transformational grammar, the theory rejects phrase structure rules, tree- structures, and transformational rules of any kind. In categorical grammar, intrinsic properties of lexical items are essential and they are thought of as the locus

of grammatical relations. . Withi_n this model, the composition of suffixes is maJierializedl by Suffixation. So called words, phrases, and sentences are generated by a device called Word Induction, a combiner by which lexical items are connected in an explicit and mechanical fashion. ( See Brame ( 1984; 1985 ))for details ) . Let us now show how the derived nominal construction un-der consiun-deration is accounted for within our theory. To begin with, consider the following lexical specifications for the relevant lexical items.

{24) Primitive Nullary Words a. <enemy, Nl

b.< city, Nl c. <destruction, Nl Primitive Unary Words d. <the, Dl N>

e. <destroy, VID> Primitive Binary Words

-P f. <N

I

of, P g

I

D

>

' -P

>

g. <N I s, S g I N ag.l D' h.<NI by, P ?

The initial component .of the formulae in the· first row is taken to be phonetic or orthographic words. The second component is its intrinsic category. The items in the first row lack their argument categories, hence nothing comes after the vertical line. Such words are called the primitive nullary word. Each lexical item in the second row carries a single argument category, i.e., the third component in the formulae. Thus, the term primitive unary word is utilized. The items in the third row take two argument categories, one on the left and the other on the · right.

Lexical items which take two argument categories ue: named the primitive binary word.

A brief explanation for the sym bois in rM> should be also in order. The uppercase Latin N is thought of as noun, "and D as determiner. The P with superscript· g in ( 24f) symbolizes prep-osition with genitive function. The symbol S gin ( 24g) denotes suffix with genitive function. The superscript +--· p in ( ?4£~ and

( 24g) indicates the direction of prominence relation. Finally, P with superscript ag in ( 24h ) signifies preposition with agentive function.

Our next move is to demonstrate ·how words are put together. The following devices are provided in order to formaHy state

9

the linking procedure.

125) Word Induction: a. If L =<x (/J

I

'1fr --·

,1/r

>

e

LEX and i ' 1' n L =<y,Y j ' 1

alo ···

1' ' 8

>

e LEX, then < x- y,

cp

1/{,

a

I

~,

· · ·

m ,om,~····,

1/r

>

e

LEX. n

b. If Li=<x, 1/r1

al

81 , · · · , 8

m>

e

LEX and Li=<1/r0 , ••• ,

Yr.

I

y,

cp

I

e , · · • , e

>

e

LEX and L = < z, T

e

I

fi ,

1 I k k 1 I

··· fi _{' j}

> e

LEX then <x-y-z _{' '} ·'~ _{'Y 1}

am

T

e

I

8 ••• 8

T 1 1 ' ' m~

Vr

2 , ••• ,

Vr

0, /i1 , ••• , /ii ,

e

2 , •• • ,

e

k

> e

LEX.

By ( 25a ) , for example, ( 24d) may be combined with ( 24b ) . By similar procedure, ( ·24d ) and I( 24a) may be joined. These proc· esses can be pictured as in (26).

(26)

a. <the, Dl N> (<city, Nl) =<the -city, DNI

b. <the,. DIN> (<enemy, Nl )=<the-enemy, DNI

The result of ( 26a ) may be connected with ( 24e ) analogously as shown below.

(27) <destroy,VID>( <the- city, DNI)=<destroy- the- city,VDNI From the above exposition , it should be clear. now that the coupling of words may take place if and only if the argument category of one lexical item is the same type .as·lllhe head of in-trinsic category of another lexical item.

Let us now proceed to the word induction involving primi-tive binary words. Lexical it~ms ( 24c ) , ( 24h ) and ( 26a ) may be concatenated by Word Induction (25b). Thtis; we obtain the following result.

Cl~ ( <destruction, Nl ) <Niby, P8_{giD> ( <the-enemy, DNI )}

= <destruction- by- the- enemy, NP ag DN I

Given ( 24 b ) , the induced word in ( 28 ) , and ( 24g) , we. may put all these items together at ·once py ( 25b) . This process is shown below.

~

(<city, Nl' ) <-NI t s,

s:l

N> (<destruction'-

by-the-en-emy, NP ag DNI ) =<city's -destruction- by- the-enemy ,

Ns-:

NP ag DNI

To this result we may attach(24d). Thus, we obtain the following.

+-P

00 <the, D,l N > ( <city's- destruction- by- the- enemy, NS g NP ag DNI ) = <the-city's-destruction- by-the-enemy, DNs : NP ag DNI

By analogous procedure, other examples can be also accounted for. The reader is invited to verify that by himself. Some rep-resentative results of concatenation procedure are given below.

(iJ)

a. (<destruction, Nl) <Niof, i>:ID>( <the-city, DNI) = <aestruction-of-the-city, NP: DNI

+-p

I

I

agl

b. (<destruction-of-the-city, NP gDN )<N by, P D > (<the-enemy, DNI ) =<destruction-of-the-city-by-the

+-P ag

I

enemy, NP g DNP. DN

c. <the, DIN> (<destruction-of-the-city-by-the-enemy,

+-P ag

I

NP g DNP DN ) =<the- destruction- of- the- city-by- the

enemy, DNP ~ DNP ag DNI

d. (<enemy, Nl )<NI 's, s~IN> (<destruction-of-the-city, NP

~

DNI ) = <enemy's- destruction- of- the- city, Ns

~

NP: DNI

e. <the, DIN> ( enemy's-destruction-of-the-city,

NS~

N"P:

DNI )=<the-enemy's-destruction-of-the-city, DNs

~NP~

DN.I'

6. Concluding remark

As explicated above, our theory provides the straightforward and natural account for the examples given in this article. More-over, the problems listed in ( 23 ) do not even arise within the framework of categorical grammar. This follows from the fact that the competing grammar does not hinge on transformational rules and filtering devices. One might ask at this point: What about the transformationalists' claim with respect to the sub-categorization properties such as destroy, V,

+ _

NP and destruction,

N,

+ _

NP? This does not seem to gain wide acceptance. Consider the following examples where N's lack complement NP's yet they are in fact well- formed strings.

a. the great destruction b. the clear- cut explanation c. the public establishment

The above examples would lead to Projection Principle violations if the nouns are specified once for all as destruction, N,

+ _

NP,

Within our framework, destruction and destroy for example are lexically specified respectively as ( 24c) and- { 24eU. The former is a nullary word, hence it does not· take an argument category. To the contrary, the latter is a unary word, hence it takes an a~­

gument category, in this case D. As exemplified in (27), ( 28), ( 29 ) , ( 30) , and ( 31 ) , grammatical relations are inherent, so to speak, in the relevant lexical items. Thus; all we need to do is to connect the words together in accordance with the spirit of Word Induction

Notes

1. The meaning of of+3 in the structural change is not clear . Judging from the footnote (See Jackendoff, 1977:70 fn 13 ), however, the of+ NP does not seem to form a single constituent.

2. See Jackendoff ( 1977:70 fn 13;108 fn 3 ) .

3. This might not be a problem at all if two underlying forms are assumed: The one with an empty np in subject position as in np

destruction the city, and the other without the empty np · as in de-struction the city.

4. The Case Filter is stated as follows:

"*

N, where N has no Case.

"

(See Stowell, 1981:111; Chomsky, 1981:49).

5. The CAUG is formulated as follows: In the configuration(a.B···) or ( ··· ,9

a ) , a

Case- marks ,9, where

( I l

a

governs

.a

and (jj)

a

is adjacent to ,9, and

(jjj)

a

is ( -N)

( See Stowell, 1981:113 ) 6. The condition on fJ - role assignment IS worded as follows:"

fJ-role m:ay only be assigned to A-positions which are associated with PRO or Case. " ( See Stowell, 1981:111 ) .

7. The Empty Category Principle is expressed as follows: " ( NP e ) must be properly governed." (See Stowell, 1981:133; Chomsky, 1981: 250).

8. In Chomsky ( 1981:36 ) it is stated that each argument bears one and only one fJ - role, and each fJ -role is assigned to one and only one argument.

9. The basis of the following rules is Brame's lectures delivered at University of Washington in 1986.

References

Brame, M.1984. " Recursive categorical syntax and morphology." Linguistic Analysis 14, 4:265-287 .

. 1985. "Recursive categorical syntax H: n-arity and variable continuation." Linguistic Anlaysis 15, 2-3: 137-176.

Chomsky, N. 1970. "Remarks on Nominalization." in R. Jacobs and P. Rosenbaum eds. ( 1970 ) .

. 1980. "On the Representation of Form and Function." Paper presented at the CNRS Conference at Royaumont, France, June 1980 .

. 1981. Lectures on Government and Binding. Foris, Dordrecht

Emonds, J. 1976. A Transformational Approach to English Syntax . .

Academic Press, New York .

. 1983. "Subcategorization and Induced Empty Nodes: An

Ex-*h

,

planation for Case Resistance, Control, and t at- trace. Class handout in LING 561.

Jackendoff, J. 1977. X Syntax: A Study of Phrase Structure. Cam-bridge: MIT.