Relativization in Hebrew: A Transformational Approach [Reprint 2017 ed.] 9027923914, 9789027923912


201 96 7MB

English Pages 252 [253] Year 2017

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
ACKNOWLEDGEMENT
TABLE OF CONTENTS
CHAPTER I. THE SPECIAL STATUS OF THE HEBREW LANGUAGE
CHAPTER II. RELATIVE CLAUSES WITH VERBAL PREDICATES
CHAPTER III. RELATIVE CLAUSES WITH NONVERBAL PREDICATES
CHAPTER IV. INDEPENDENT RELATIVE CLAUSES
CHAPTER V. RELATIVE MARKERS IN HEBREW
CHAPTER VI. WORD ORDER IN RELATIVE CLAUSES
CHAPTER VII. RESIDUAL PROBLEMS
APPENDIX
BIBLIOGRAPHY
Recommend Papers

Relativization in Hebrew: A Transformational Approach [Reprint 2017 ed.]
 9027923914, 9789027923912

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

JANUA LINGUARUM STUDIA

MEMORIAE

N I C O L A I VAN WIJK D E D I C A T A edenda curai C. H. VAN S C H O O N E V E L D Indiana University

Series Practica,

189

RELATIVIZATION IN HEBREW A Transformational Approach

YEHIEL HAYON

1973

MOUTON THE HAGUE • PARIS

© Copyright 1973 in The Netherlands. Mouton & Co. N.V., Publishers, The Hague. No part of this book may be translated or reproduced in any form by print, photoprint, microfilm, or any other means, without written permission from the publishers.

Printed in The Netherlands

This study is dedicated to my wife Linda and to the memory of my parents Yaffa and Jacob Hayon

A C K N O W L E D G E M E N T At the beginning of this study I would like to express only a few words of thanks to a man without whom this book would never have been published. Professor Shelomo Morag was the person who first unveiled the study of language to me, at the time when I was his undergraduate student at the University of Tel-Aviv. But Professor Morag has not only served as my teacher; he has been a guide and a friend and a light to follow. His interest in my work did not cease when I left Israel to continue my study of linguistics in the United States. He followed my work with interest, offered words of support, and finally, personally saw to it that this study was published. To Professor Morag, a true scholar and teacher, I offer my eternal gratitude. Y.H.

Columbus, Ohio

T A B L E

OF

C O N T E N T S

Chapter I.

II.

Page THE SPECIAL STATUS OF THE HEBREW LANGUAGE 1. Synchrony, Diachrony, and Modern Hebrew 2. Historical Background 3. Ancient Language in a Modern World 4. Two Problematic Solutions 5. The Character of the Present Study Notes

1 2 5 7 11 14

RELATIVE CLAUSES WITH VERBAL PREDICATES

17

1.

III.

IV.

V.

1

General Remarks Concerning the Derivation of Relative Clauses 2. Relativization Rule - Version I 3. Relativization Rule - Version II 4. The Object Marker Insertion Rule Notes

17 39 53 58 65

RELATIVE CLAUSES WITH NONVERBAL PREDICATES

74

1. Is there a Copula in Hebrew? 2. The Content of the Predicate 3. Relativization Rule - Version III 4. Relative Clause Reduction Rule 5. Possessive Attributes Notes

74 85 94 107 Ill 127

INDEPENDENT RELATIVE CLAUSES

135

1. Introductory Remarks 2. Sentences with mi/ma . kol' mi/kol ma 3. Relative Clauses with No Overt Head Noun 4. The Derivation of Some Adverbial Clauses Notes

135 137 168 171 176

RELATIVE MARKERS IN HEBREW

182

1. 'aser vs. se2. 'aser/Se- vs. haNotes

-

182 188 195

Chapter VI.

VII.

Page WORD ORDER IN RELATIVE CLAUSES

196

1. Introductory Remarks 2 . The Underlying Word Order o f Modern Hebrew 3 . Word Order and Prosody 4 . Word Order and S t r e s s 5. Word Order and Pronomlnalizatlon 6 . Word Order Inside Relative Clauses Notes

196 198 204 207 209 213 217

RESIDUAL PROBLEMS

219

1. Deletion of a P r e p o s i t i o n a l Phrase 2 . Successive vs. Stacked Relative Clauses 3. E x t r a p o s i t i o n from an NP Notes

219 222 228 230

APPENDIX

231

BIBLIOGRAPHY

236

C H A P T E R

I

THE SPECIAL STATUS OF THE HEBREW LANGUAGE

1.

Synchrony, Diachrony, end Modern Hebrew The present study is concerned with a synchronic description

of one chapter in the syntax of Modern Hebrew, i.e., the relative clause and some other related problems. It was de Saussure1 who first pointed out the distinction between synchronic and diachronic descriptions and emphasized that a linguist who fails to keep synchrony and diachrony apart ends up with falsified judgments.

Ever since it was introduced, de Saussure's claim

has been one of the fundamentals of modern linguistics. However, a large number of linguists who have been dealing with the description of Modern Hebrew did not feel quite at ease when they tried to keep the levels of evolutionary and static linguistics apart. Peretz, for example, writes: As to the new Hebrew, as it has a special fate and special conditions also its method of study is different—it is the synchronic-diachronic (panchronic) method. When Hebrew of all periods and levels is reincarnated, the present stage of our language is not understood and does not exist without the past.2

2 This uneasiness among Hebrew linguists concerning the maintenance of the strict differentiation between the two methods of studyhas a strong bearing on the special status of Modern Hebrew among other languages of the world.

Nevertheless, I believe that linguists such as

Peretz have confused two concepts:

the difficult vs. the impossible.

I admit the fact that a synchronic description of Modern Hebrew is quite elaborate and complicated, but I cannot accept the claim that it is impossible, useless, or misleading.

2.

Historical Background The history of Hebrew may be divided roughly into the follow-

ing four periods: (a)

The Biblical period:

approximately 1200-130 B.C.

(b)

The Mishnaic 3 period:

approximately 130 B.C.-600 A.D.

(c)

The Mediaeval period:

approximately 600 - the end of the 18th century

(d)

The period of new and Modern Hebrew:

the 19th century

until todayBiblical Hebrew is the language used in the books of the Old Testament.

The Biblical language represents a wide variety of linguis-

tic and stylistic characteristics:

there are distinct differences among

the poetic, prosaic, prophetic, and legal parts of the Bible on the one hand, and between the early and the late books on the other hand.

3 However, we are not interested here in these differences.

In the

Biblical period Hebrew was used both in speaking and in writing. "Mishnaic Hebrew" is an abbreviated and rather technical term for the language of the tanna'im and the 'amora'im. The tanna'im were the Hebrew scholars who created the so-called "Oral Law," which generally includes the following: Mishna

- a collection of oral laws compiled by Rabbi Judah the Prince.

The Mishna received its final shape

at around 180 A.D. Tosefta - Supplement to the Mishna. Barayta - Oral laws which Judah the Prince did not include in the Mishna. Halachaic Midrashim - Legal commentaries on the Bible. The 'amora'im were the Rabbinic sages who followed the tanna'im and interpreted their oral laws. At the beginning of the Mishnaic period Hebrew was still alive, but later on the language gradually ceased to be spoken and was replaced by Aramaic.

In other words, the tanna'im spoke Hebrew and the

'amora'im spoke Aramaic.

It should be emphasized that, although Hebrew

stopped being spoken, it was still used in prayers, in religious studies and ceremonies, and in literary writings.

Thus, in the middle of the

Mishnaic period, Hebrew was transformed into a status of a "holy language" or "language of the book." Hebrew died at this point.4

From a linguistic point of view,

4 In the third period, the mediaeval one, the language was still dead or dormant.

Scholars, poets, sages, and translators still used it,

but again not for daily speaking.

The literary work that was done in

Hebrew in the mediaeval period represents a mixture of Biblical and Mishnaic linguistic characteristics with the addition of a strong influence of Arabic.

The impact of the language of this period on Modern

Hebrew is very small, and mainly in the vocabulary, because the revival process was practically based upon the Bible and the Mishna which are the samples of classical Hebrew at the stage at which it was alive. The first step towards the revival was taken by the writers of the Enlightenment Movement (end of the 18th to the end of the 19th century). 5

But the real signs of the revival were seen at the beginning

of the 20th century when Hebrew began to be used also in speaking. The revival succeeded because of a group of devoted and enthusiastic people 6 who not only urged the people to learn and to speak Hebrew, but also opposed other enthusiastic scholars who rejected the idea of the revival. 7

On the eve of World War I, however, it was

obvious that Hebrew became a language once again with full life, in writing as well as in speaking.

Today, there are already Hebrew speakers

to whom the language has been native for three generations.

5 3.

An Ancient Language in a Modern World The revival was directed toward taking an ancient language,

which had been dead for sixteen centuries, and reactivating it at the beginning of the 20th century.

It meant taking the language of the

Biblical prophets and the sages of the Mishna and using it in a world of cars, airplanes, and rockets. The biggest problem which was faced by the first speakers of Modern Hebrew was that of a wide gap in the vocabulary.

This gap was

closed either by giving new meanings to old words, or by innovating new words. 8 As to the other parts of the grammar--i.e., phonology, morphology, and syntax--they were also revived on the pattern of classical Hebrew. 9 As the main concern of this paper is syntax, I shall concentrate on this part of the revived language. The syntax of Modern Hebrew consists of the following elements: (l.l)

(i)

The syntactic characteristics of Biblical Hebrew;

plus:

(ii)

The syntactic characteristics of Mishnaic Hebrew;

plus:

(iii)

The new syntactic developments which emerged in the last fifty to sixty years;

minus:

(iv)

The syntactic phenomena of (i) and (ii) which disappeared from Modern Hebrew.

6 These facts give Hebrew a very special status among the languages of the world.

All the other languages have different historical

strata, and each stratum differs from the former one.

It is nearly

impossible, or at least it is quite rare, for example, that a syntactic construction which disappeared from English in an early stratum (such as the 10th century) would reappear again in a later stratum.

Because

of this absence of an interaction between two stages of linguistic development, the last stratum is usually the only one which has practical importance.

This is not the case with Modern Hebrew.

This

language not only includes many characteristics of previous historical strata, but it continues borrowing from the first stages of its life. Of course, new borrowings from the previous strata are found mainly in the literature and not in daily use.

It is a well known fact that the

poetic language is always more active than the standard language. The situation described at the beginning of this section explains why some Hebrew linguists claim that synchrony and diachrony cannot be kept strictly apart in the description of Modern Hebrew (see Section 1 above).

They claim that the revival process drew past and

present together to the surface and thus the state of the language must be reflected in the method of the description of this language.

7 4.

Two Problematic Solutions

This last claim has had a direct impact on the character of linguistic research studies of Modern Hebrew.

It is almost impossible

to find a pure synchronic approach among these studies.

Most of the

linguistic studies which have been written since the revival belong to one of the following two categories:

(l.2)

(i)

Studies which are written from a so-called

"panchronic"

point of view (e.g. Peretz's miSpat). (ii)

Studies which describe a certain corpus, i.e. linguistic analyses of a certain writer or a certain literary composition (e.g. Rubinstein's hammispat; O m a n ' s Nominal ).

I should like to emphasize that the reason for this situation has not resulted from the fact that it so happened that all Hebrew linguists are interested only in these two approaches to linguistic analyses.

Rather, it happened because these linguists believe that

these two methods of description are the only methods by which one should approach Modern Hebrew

(and pure synchronic description should

wait until Modern Hebrew is no longer a sad contamination of past and present). A s to studies such as Peretz's [see (l.2)(i) above], I do not see any significant difference between them and any purely diachronic

8 study.

I should like to cite here only one typical example from

Peretz's mispat which I believe illustrates "the panchronic principle." The quotation summarizes Peretz's discussion concerning the anaphoric Pronoun in relative clauses: (1.3)

From what has been said about the location of the anaphoric Pronoun we see: 1.

In the Bible it comes after the main parts of the sentences10 or after the first of them.

2.

In the Mishnaic language its location is usually after the main parts of the sentence or after the first of them, but sometimes it precedes them, especially when the relative clause is not adjacent to the relativized IToun.

3.

In the mediaeval period the Mishnaic usage applies.

4.

In the new literature it usually follows the main parts of the sentence or the first of them.

5.

In the standard language of our time it usually precedes the main parts of the sentence. 1 1

Two questions may be raised in accordance with a description such as this: (1.4)

(i)

What makes this panchronic description different from any pure diachronic description?

(ii) Why can't we omit statements 1-4 above and keep only statement 5 which is a perfect synchronic statement concerning standard Modern Hebrew?

9 As to question (l.4)(i), de Saussure himself already discussed this matter and pointed out that "synchronic truth is so similar to diachronic truth that people confuse the two or think it superfluous to separate them" (p. 96).

This confusion which most linguists have rid

themselves of already is still a common phenomenon in Hebrew grammars, possibly because of the fact that in Modern Hebrew, more than in any other language, diachrony and synchrony are so similar. A s to (l.4)(ii), again one may find a discussion of this matter in de Saussure's Course:

(l.5)

In linguistics as in chess there are rules that outlive all events. But they are general principles existing independently of concrete facts. When we speak of particular, tangible facts, there is no panchronic viewpoint. . . . These general principles are precisely what serve as a criterion for determining what belongs to language and what does not. A concrete fact that lends itself to panchronic explanation cannot belong to language (p. 95).

To sum up this point., synchronic description of Modern Hebrew is possible even though perhaps it is complicated.

If we look at

(i)-(iv) of (l.l) above we may understand that the syntax of Modern Hebrew is subject to many more stylistic variations than other languages. These variations, of course, echo the influences of the previous periods on Modern Hebrew, but by no means should they lead to panchronic description.

For example, the Relative Marker in Modern Hebrew is

realized either as ' aser or se- or ha-.

A synchronic description should

10 only specify in which types of sentences each relative marker may occur, and it should not describe the diachronic processes which have led to the present situation.

This last task belongs to the historical

description of Hebrew. In other words, a grammar of Modern Hebrew should deal only with the items specified in (l.l)(i)-(iii) above minus (l.lj(iv), and no more.

Grammars such as Peretz's [which, incidentally, also include

in their description (l.l)(iv)] are merely grammars written with a purely diachronic viewpoint in mind. So far I have dealt with grammars of type (l.2)(i). concern myself now with studies of type (l.2)(ii).

Let me

As was mentioned

above, the main reason which led to this type of grammar was the claim that Modern Hebrew is not yet crystallized, and thus every grammar which tries to characterize Modern Hebrew ends up as an incomplete or misleading grammar.

Thus, the linguists who hold this opinion come to the

conclusion that the safest way is to choose a certain corpus and to write the grammar of this corpus. 12

I should emphasize that writing a

grammar of a corpus is an important and fruitful task (mainly for stylistic analyses) but such a grammar cannot replace a general synchronic grammar which describes the competence of a native speaker of a certain language.

11 5.

The Character of the Present Study As was mentioned at the very beginning of this chapter, this

study purports to be a synchronic one.

It is based on an analysis of my

own idiolect in Modern Hebrew, and its purpose is to illustrate my competence in this language insofar as relative clauses are concerned. It is not difficult to discover justification for such a study in Modern Hebrew.

Hebrew is my native language; I therefore see no difference

between my competence in this language and the competence of any native speaker of any other natural language.

If it so happens that some

aspects of the Biblical or Mishnaic language are reflected in my idiolect, I am not forced to describe the entire linguistic reality in these periods of the language.

Those Biblical or Mishnaic aspects which are

present in my idiolect will be synchronically analyzed.

Those Biblical

or Mishnaic aspects which are not found in my idiolect will be ignored. In other words, this study is not concerned with rare and ancient forms of relative clauses which are not found in standard Modern Hebrew, despite the fact that they may be found in the literary works of some modern Hebrew writers.

It is my claim that such phenomena do not

belong to the grammar of Modern Hebrew, but rather to stylistic studies of contemporary Hebrew literature. It was mentioned above that this study is primarily based upon my own idolect.

However, I do believe that it adequately reflects the

12 general grammatical situations which exist in standard Modern Hebrew 1 3 insofar as relative clauses are concerned. Basically, I have employed the theoretical model of Noam Chomsky in his Aspects of the Theory of Syntax (1965).

In addition,

other studies by generative grammarians (previous to or more recent than Aspects) have served as an influence upon my treatment of the topic. This study is concerned with some syntactic transformational rules in the generative grammar of Modern Hebrew.

However, some base

rules of the syntactic component are also discussed (e.g. the content of the Predicate, the nature of the Copula, etc.). In the examples which are cited in this study, the following conventions are used: (a)

Each sentence is written in two ways:

in Hebrew, in broad

phonemic transcription; and in literal translation into English. (b)

Any item appearing in parentheses in the Hebrew transcription is optional.

(c)

Any item appearing in parentheses in the English translation indicates that the item does not appear in the Hebrew and was added in English for ease of understanding.

(d)

In the English translation, any item X appearing in parentheses and preceded by an equal sign, e.g. . . . Y (=X) . . indicates that the item X is the English equivalent of Y,

which is the literal translation. hu fiosev 'al rut. 'He thinks on (=about) Ruth.'

For example

14 NOTES TO CHAPTER I

1.

de Saussure, Course, pp. 79-100.

2.

Peretz, mlspat. p. 11.

3.

The meaning of the term Mishnaic will be clarified in the course of the discussion.

4.

The language ceased to be spoken approximately at the end of the 2nd century.

5.

The Enlightenment Movement was a socio-cultural revolution which took place in Europe in the 18th century.

This general movement

had its impact on the European Jews several decades later.

The

goals of this movement among the European Jews were the following: spreading education among the Jewish people, breaking the wall between the European Jews and the peoples among whom they lived, achieving civil rights for Jews, and reviving the secular Hebrew literature. 6.

The most outstanding and active figure in this group was Eliezer Ben-Yehuda

7.

(1858-1922).

This period is known as the period of the "battle of the languages" (milAemet halla^onot).

8.

The linguists who innovated the new words did their work in accordance with the linguistic system which is represented in classical Hebrew.

That is to say, they invented words whose roots and forms

agreed with the fundamentals of the semantics and morphology of Biblical and Mishnaic Hebrew.

Obviously, there are also many

borrowed words in Modern Hebrew which do not agree with the older Hebrew linguistic system. 9.

These facts explain the great difference between the Greek child who attempts to read Homer in the original and the Israeli child who attempts to read the Old Testament in Hebrew--the former is faced with an enigma while the latter can read the text almost without difficulty.

10.

He uses the term "main parts of the sentence" to designate the Subject and the Predicate.

11.

Op. cit.. pp. 108-109.

12.

Even modern Hebrew linguists who are acquainted with the notions of "competence" vs. "performance" and whose methods of description are influenced by the theory of generative grammar cannot rid themselves of their habit of working with a corpus.

Rubinstein, for

example, chooses as his corpus a group of writings of different types (e.g. prose written by different contemporary writers, scientific and literary essays, journalistic articles, etc.). Being acquainted with Chomsky's claim that a grammar should reflect the competence of the native speaker, Rubinstein seems to use some acrobatic arguments in order to justify his method. he says:

For example,

16 In order to overcome the practical difficulty this study is based on objective data as large as possible. It is not intended to treat the examined texts as a defined In corpus whose grammar we are attempting to write. front os us lies "a big notebook" of examples which substitute the limited intuition which does not furnish us with sufficient information about the analyzed language. (hammispat, p. 45) I think that any comment such as this which argues that a big notebook of examples is a better replacement for the intuition of the native speaker is so obviously false that I will not bother with rebuttal. By "standard Modern Hebrew" I refer to the formal language used in speaking, in the newspapers, and I exclude the poetic and the vulgar language.

C H A P T E R

II

RELATIVE CLAUSES WITH VERBAL PREDICATES

1.

General Remarks Concerning the Derivation of Relative Clauses1 Jespersen2 discusses three types of relative clauses:

tive, nonrestrictive, and. continuative.

restric-

Following his discussion, I

shall illustrate these three types of clauses by giving examples from Hebrew: Restrictive: (2.1)

hafiayalim %eracu kadima hayu 'amicim. "Hie soldiers that ran forward were brave.1

Nonrestrictive: (2.2)

hafiayalim, seracu kadima, hayu 'amicim. 'The soldiers, that ran forward, were brave.'

Continuative: (2.3)

hu natan 'et3 hamixtav lapakid ^ehe'etik 'oto3. 'He gave OM 3 the letter to the clerk that copied OM + him (=it)3. '

The difference between restrictive and nonrestrictive (appositive) clauses is the following:

sentence (2.l) implies that only the

18 soldiers who ran forward were brave, while (2.2) implies that all the soldiers, all of whom were brave, ran forward.

In (2.2) then, the

relative clause is in apposition to haftayalim.

Also, there is a differ-

ence between the way (2.1) and (2.2) are pronounced:

in (2.2) there is

a pause before and after the relative clause, but such a pause does not exist in (2.l); furthermore, the stress pattern of an appositive clause does not resemble that of a restrictive clause. As for the continuative relative clauses, they are obviously closer to nonrestrictive rather than to restrictive clauses.

Notice

that both continuative and nonrestrictive clauses have paraphrases with coordinate structures; (2.2) and (2.3) may then be expressed as (2.4) and (2.5), respectively, without any change in meaning: (2.4)

haftayalim hayu 'amicim, vehem racu kadima.

"The soldiers were brave, and they ran forward.' (2.5)

hu natan

'et hamixtav lapakid, V9hu he'etik

'oto.

'He gave OM the letter to the clerk, and he copied OM + him (=it).' Continuative relative clauses are more rare in Hebrew than, for example, in English, and sentences such as (2.5) are much more widespread than (2.3).4 Because of this closeness between nonrestrictive and continuative clauses, I shall deal with them as one type of sentence.

As a

19 matter of fact, continuative clauses do not constitute a separate type of sentence, but they are rather a subtype of nonrestrictive clauses. Thus, the division into types of clauses should be the following: (2.6)

(i)

Restrictive

(ii)

Nonrestrictive a.

Continuative

b.

Noncontinuative

Thus, the discussion will from this point on differentiate only between (i) and (ii) of (2.6) with the understanding that everything that is said concerning (ii) applies in exactly the same manner to both continuative and noncontinuative clauses. Because of the mentioned differences between restrictive and nonrestrictive clauses—both in meaning and pronunciation—it has been argued by generative grammarians that these two types are derived from different underlying structures.

There are slight differences between

grammarians as to the exact underlying structures from which these sentences are derived, but almost all of them agree basically on the following assumption: (2.7)

(i)

Restrictive relative clauses are embedded in the relativized NP.

(ii)

Nonrestrictive relative clauses are derived from coordinate sentences.

20 If w e want to state

(2.7) schematically, we may say that the

underlying structures of the two types of relative clause are "basically the following: 5

(2.8)

(i)

Restrictive

(ii)

Nonrestrictive

The circled NP's are identical.

This identity of Noun

Phrases is a prerequisite for the application of the relativization transformation, which will be discussed later on in this chapter. I have mentioned before that nonrestrictive clauses are derived from conjoined sentences, and I should like now to elaborate this point.

Consider the following sentences:

21 (2.9)

(i)

davicL higia*

'etmol, vadavid^ hu hamanahel hefiadas.

'David^ arrived yesterday, and David^ he (=is) the director the new ( = t h e new director).' (ii)

david -- hu hamanahel heftadas -- higia' 'etmol. 'David -- he (is) the director the new -- arrived yesterday.'

(iii)

david, £ehu hamanahel heftada£, higia' 'etmol. 'David, that he (is) the director the new, arrived yesterday.'

These three sentences are paraphrases of each other.

Sentence

(ii) is derived from (i) by moving the second conjoined sentence (which contains an NP identical to the one in the first conjoined sentence) next to the identical NP in the first conjoined sentence.

The result is

a sentence which contains a parenthetical clause with a noticeable pause before and after it [marked with dashes in sentence (ii)]. Sentence (iii) is derived from (ii) by relativizing the parenthetical clause. Thus it seems that sentence (2.9)(ii) may be derived from (2.9 ) (i) by an embedding transformation, and sentence (iii) may be derived from sentence (ii) of (2.9) by the relativization rule. Notice that after the embedding of the parenthetical clause, sentence (2.9)(ii) has an underlying structure which is similar to that of a restrictive relative clause [cf. (2.8)(i)].

We may thus assume

22

that the relativization rule may generate both restrictive and appositive relative clauses.

However, there is a problem involved in this

assumption which was discussed by Jacobs and Rosenbaum without reaching a final solution.

I shall cite their discussion and thereafter propose

a solution which appears to be a correct one, at least for Hebrew, if not universally: The fact that the relative clause transformation can apply here [on sentences such as (2.9)(ii) above—Y.H.] raises a perplexing question, the answer to which is not currently known. You will recall from the restrictive relative clause chapter that the relative clause transformation operates on a structure of the following sort: NP NP

S

However, the pronunciation of non-restrictive relative clauses suggests that such clauses are not embedded inside of noun phrases, but rather are adjoined to noun phrases as in the following diagram, which illustrates the structure of a sentence after a non-restrictive clause has been adjoined to the subject noun phrase: S NP

S

VP

Thus, it appears that the relative clause transformation must operate on two distinct environments rather than just one. It is perhaps the case that the non-restrictive clause transformation [the parenthetical clause embedding (in my terms)—Y.H.] actually does incorporate the non-restrictive clause into the noun phrase, then the relative clause transformation applies, and finally the non-restrictive clause is moved outside of the noun phrase. But this is a speculation at best, at present, and it is necessary to await the results of further grammatical research.6

23 In short, Jacobs and Rosenbaum could not reach a definite decision concerning the question of which of the following solutions is preferable:

(2.10)

(i)

The parenthetical clause is first embedded into the Noun Phrase, then relativization applies

(on one

environment), and finally the nonrestrictive clause is moved outside the Noun Phrase, (ii)

The parenthetical clause is adjoined to, rather than embedded in the Noun Phrase, and the structural index of the relativization transformation applies to two distinct environments.

As far as Hebrew is concerned, the second possibility seems to be the most plausible.

The reason for this is the following:

in

English, the only nonsemantic difference between restrictive and nonrestrictive relative clauses is the pronunciation (i.e. the pauses and the intonation); in Hebrew, on the other hand, the difference is quite often one of pronunciation as well as of structure.

Notice, for

example, the following two sentences:

(2.11)

(i)

Restrictive: hayayin ^e'al hasulfian h u miyavan. 'The wine that on the table he (=is) from Greece.'

24 (ii)

Nonrestrictive: hayayin, ¡Sehu *al hasulftan, hu miyavan. 'The wine, that he (=it) on the table, he (=is) from Greece.'

In sentence (i) the embedded identical NP is deleted, while in (ii) it is pronominalized.

It is clear that this difference must be

accounted for by the application of the relativization rule.

Conse-

quently, this rule must operate on two different structures and the structural change of the relativization rule should indicate in which cases the identical NP is deleted and in which cases it is pronominalized.

I should note, parenthetically, that in Hebrew relative clauses

whose Predicates are Verbal the difference between restrictive and nonrestrictive clauses is only in pronunciation, while in relative clauses with non-Verbal Predicates the difference is both in structure and in pronunciation.

This point will be illustrated throughout this

chapter and the next one, and I shall not pursue its discussion at this point.7 I may now state the transformational rule for embedding a parenthetical clause: (2.12)

[[ X 1 1

NP 2 2+456

Y] s 3 3

[W 4 0

NP 5 jZi

Z ]g]s 6 ¡6



25 Rule (2.12) converts a structure like (2.13)(i) into (2.13)(ii): (2.13)

(l)

David.

heftadas he (=is) the director the new

At this point I should like to discuss the derivation of restrictive relative clauses.

As I mentioned above, most grammarians

agree that in its underlying structure, the restrictive relative clause is dominated by the node of the relativized Noun, though some of them differ in the details of this analysis. crucial in many cases.

These details, however, are

26 I should like to first indicate which approach I shall adopt throughout this study, and later I shall attempt to argue against the other approaches which I did not adopt. I assume the grammar of Hebrew to have a rule such as (2.14)

NP

-

NP (S)

The optional S is the restrictive relative clause. This analysis has been used by several generative grammarians such as Ross (in Constraints), Postal (in Cross-Over), Jacobs and Rosenbaum (in English), and Emmon Bach and Stanley Peters (in personal communications). However, some other generative grammarians assume the restrictive clause to be derived from different underlying structures. Chomsky, for example, derives the restrictive relative clause from the Determiner-node of the relativized NP, in the following manner:8 (2.15)

(i)

NP

(ii) Det

-

(Det) N (S') (Pre-Art^of) Art (s")

[Where S" in (ii) is the restrictive relative clause.] Thus, according to my analysis, a sentence such as (2.l) above has the deep structure shown in (2.16), while according to Chomsky it has the deep structure shown in (2.17):

27

(2.16)

hayu 'amicim were brave hafiayalim The soldiers

haftayalim racu kadima The soldiers ran forward

(2.17)

NP

The

VP

The soldiers ran forward However, there are some difficulties involved in this

analysis.

The motivation for Chomsky's analysis is based on the fact

that not every Noun can be relativized. grammatical but not (ii):

For example, (i) of (2.18) is

28 (2.18)

(i)

ha'i£ £era'ita

'etmol haya pa'am ro% hamem&ala.

'The man that you saw yesterday was once the Prime Minister.' (ii)

* david ^era'ita 'etmol haya pa'am ro£ hamemsala. 'David that you saw yesterday was once the Prime Minister.'

Chomsky's analysis will adequately prevent (2.18)(ii) from being generated, but on the other hand, it will also prevent the following grammatical sentence from being generated:

(2.19)

hadavid se'ani msdaber 'alav lo haya ro% mem^ala. 'The David that I am talking about h i m not was (= was not) Prime Minister.'

Chomsky's remedy for this difficulty is the assumption that there is a redundancy rule in the lexicon which states that any proper Noun can be used also as a common Noun.

But such a convention will

lead to another difficulty; that is, it will allow the ungrammatical sentence (2.20) to be generated:

(2.20)

*hadavid gar po. 'The David lives here.'

One may suggest a n ad hoc solution for this, i.e. that ha('the') may precede a proper Noun only when S" [see (2.15)(ii)] also

29 occurs under the Determiner-node.

But this suggestion obscures the fact

that david by itself is also definite, e.g. it cannot occur in certain environments which require indefiniteness, such as the environment in (2.21): (2.21)

yeg

bafieder.

'There is/are

in the room.'

Thus, Nouns such as ra'as ('noise'), kelev ('dog'), bafiura ('girl') may occur in (2.2l) but not hara'ag, hakelev, habafiura, or david. A plausible solution to this problem is the following: 9

Since

proper Nouns are definite, we may have a convention such as (2.22). (2.22)

[hadavid] -* [j> david] / except before relative clauses.

With this solution, the motivation for Chomsky's analysis has no bearing, because according to (2.22) proper Nouns should be derived with a Determiner from the base component. Another difficulty which is involved in Chomsky's derivation of relative clauses arises when one deals with stacked relative clauses. Consider, for example, the following sentence: (2.23)

anafinu madabrim 'al hamafiaze Sehilhiv 'et hakahal, seliga' 'et hamavakrim. 'We speak (= are speaking) on (= about) the play that excited OM the crowd, that drove crazy OM the critics. 1

30 According to Chomsky, this sentence should have the following phrase marker:

(2.24)

excited the crowd

hamafiaze the play

siga' hamsvakrim drove the critics crazy

Schematically, then, stacked relative clauses, according to Chomsky, have the shape shown in (2.25):

33. (2.25)

NP Det

N

Ar

In addition, Chomsky's relativization rule would look somewhat like this: (2.26)

[Np[DWt # 1 2 1

0

3

4

5

6

7

0

f>

$

0

0

condition:

8 8+4536

9 9

5=8

This, however, constitutes another difficulty because the structural index of (2.26) will not be satisfied after the embedding of the first relative clause [S^ in (2.25)]. Chomsky solves this difficulty with a special convention of adjunction.

That is, he claims that the embedded Sg will not create

a new "sister"-node adjacent to the N, but rather a new N-node will be

32 created above the existing N-node.

Schematically, then, Chomsky's

adjunction would look somewhat like this: (2.27)

NP Det

A

Art

NP N

S

Det Art

N

A

N

S

In this way, the structural index of (2.26) will still be satisfied after the embedding of the first relative clause, and it will be possible to go on embedding the rest of the stacked relative clauses in the same manner. Now, however, a new problem arises:

the identity requirement

of the relativization rule will be violated because the Noun on which relativization has already applied is no longer identical to the Noun on which relativization must apply next. Chomsky absolves this difficulty by weakening the identity condition.

He allows more specified elements to be substituted for less

specified ones. All these difficulties which arise from Chomsky's analysis in (2.15)(ii) and which may be remedied only by ad hoc modifications of the theory of grammar, do not exist if one adopts the analysis suggested in (2.14).

According to (2.14), stacked relative clauses are derived

in the following manner [cf. (2.24)]:

33

(2.28)

Prep-Phrase

S^

the play excited the kahal crowd

hamafiaze the play

hamafiaze siga' hamavakrim the play drove the critics crazy

Or, schematically, stacked relative clauses in my analysis will have the following shape: (2.29)

[cf. (2.25)]

34 Thus, the above mentioned, problems (concerning the relations between the derivation of relative clauses from the Determiner-node and stacked relative clauses) all disappear when one assumes a rule such as (2.14) above which may generate constructions such as in (2.30) when applied successively: (2.30)

a.

NP [Rule (2.14)]

b.

NP~S [Rule (2.14)]

c.

NP'Y'S [Rule (2.14)]

d.

NP~S~S~S

Incidentally, rule (2.14) also enables the handling of cases of relativized coordinate Nouns in a much simpler way than (2.15)(ii). (2.31)

For example: (i) ha'i£ vaha'isa £ehayu kan hem morim. 'The man and the woman that were here they (=are) teachers.'

35

they (=are) teachers

the man

and

the woman

I should like to mention briefly two other analyses of restrictive relative clauses and to discuss some counterexamples which will demonstrate the advantage of (2.14) over them. Annear, in her Relative Clauses, derives both restrictive and nonrestrictive relative clauses from conjoined sentences.

As to the

derivation of nonrestrictive relative clauses from conjunctions, I maintained the same position [see, for example, (2.8) above].

However,

there is evidence that nonrestrictive relative clauses should not have the same underlying structures as restrictive clauses.

Consider, for

example, the following sentence: (2.32)

kel 'eftad £eroce lihyot hipi magadel se'ar 'arox. 'Every one that wants to be (a) hippie grows hair long (=long hair).'

36 According to Annear's claims, (2.32) is a paraphrase of a conjoined sentence such as (2.33), which is obviously not true: (2.33)

kpl ' eftad roce lihyot hipi, vaxol 'etfad magadel se'ar 'arox. 'Everyone wants to be (a) hippie, and everyone grows hair long (=long hair).'

Even in cases where the restrictive relative clause is indeed a paraphrase of conjoined sentences, there is still evidence contrary to Annear's analysis.

Sentence (2.34)(i) below is a paraphrase of

(2.34) (ii) but when these sentences undergo the Question transformation they are no longer paraphrases of each other [compare (i) and (ii) of (2.35)]: (2.34)

(i)

ha'i^a ^edavid 'ohev hi yafa. 'The woman that David loves she (=is) pretty. 1

(ii)

david 'ohev 'i£a , vaha'i^a hi yafa. 'David loves a woman and the woman she (=is) pretty.'

(2.35)

(i)

ha'im ha'isa ^edavid 'ohev hi yafa? 'QW 1 0 the woman that David loves she (=is) pretty '

(ii)

ha'im david 'ohev '13a vsha'im hi yafa? 'QW David loves a woman and QW the woman she (=is) pretty?'

37 Thus, the cases discussed above and other evidence (e.g. the difference in the pronunciation in restrictive vs. appositive relative clauses) support the claim in favor of assigning different underlying structures to restrictive vs. nonrestrictive clauses. Gladney 1 1 derives restrictive relative clauses in the following manner: (2.36)

NP

-

N~S

This analysis is closer to the analysis applied in this study [i.e. (2.14)].

The only difference between (2.14) and (2.36) is that

Gladney implies that the relativization transformation requires identity of Nouns, while my analysis requires identity of Noun Phrases.

Sentence

(2.3l) above, in which the relativized constituent was a coordinate Noun structure (ha'i£ vaha'iSa 'the man and the woman') supports my claim.

Additional evidence for the claim that the identity is of Noun

Phrases, rather than of Nouns, may be illustrated by the following example: (2.37)

'ani makir 'et habanim £el david sekanu kan fianut. 'I know OM the sons of David that bought here (a) store.'

habanim £el david is the identical constituent which occurs in the two sentences which underly (2.37), and the relativization transformation applies to the whole constituent.

Such a constituent is

obviously dominated by an NP-node rather than by an N-node.

38 In summary of this section, let me mention again the main points and conclusions which have been discussed thus far: (2.38)

(i)

There exist in Hebrew two types of relative clauses:

(ii)

restrictive and nonrestrictive

(appositive),

Restrictive clauses are derived from an underlying structure such as NP (NP.

Nonrestrictive clauses are derived from an underlying structure such as S

(NP)—^

1

—(NP)-

[The circled NP's are identical.] (iii)

It has been shown that in Hebrew, the difference between restrictive and nonrestrictive is not always in pronunciation only, but also in structure.

This fact leads us to the conclusion

that a nonrestrictive clause is adjoined to

39 the relativized NP rather than embedded in it, and that the relativization transformation operates on distinct structures, (iv)

A rule for embedding nonrestrictive

(parenthe-

tical) clauses was stated, and evidence was given to support the derivation adopted here for restrictive relative clauses over other proposed analyses. We may now continue on to the next sections in which the discussion will be less general and theoretical than in the present section.

Sepcifics will be dealt with--details which are characteristic

of Hebrew relative clauses.

These details, plus the general themes

which have been discussed thus far and which have just been summarized in (2.38) will enable us to deal with the exact formulation of the Hebrew relativization transformation.

2.

Relativization Rule - Version I Throughout this chapter, I will deal only with relative

clauses whose Predicates contain a Verb.

I do so only for the sake of

the clarity of the discussion, because (as will be demonstrated in this chapter and in the next one) there are some basic differences between the characteristics of clauses with Verbal Predicates vs. clauses with non-Verbal Predicates.

40 However, the final shape of the relativization rule Is intended, to apply to both types of clauses. To make things more explicit, this chapter will deal with relative clauses whose Predicate-Phrase develops in the following manner:12

(2.39)

(i)

Predicate- Phrase

(ii)

VP

V



Aux VP

(place)(Time)

(NP)(Prep-Phrase)(Prep-Phrase)(Manner)

Before going into a systematic treatment of this type of sentences, I should like to mention again the following fact:

in Verbal

relative clauses, the difference which exists between restrictive and appositive clauses on the surface structure is always a difference in pronunciation only [cf. (2.l) and (2.2) above].

This fact should be

borne in mind, because in Chapter III we shall notice that in relative clauses with non-Verbal Predicates the differences between these two types of clauses lie

in structure as well as in pronunciation.

Let us look now at some examples in more detail:

(2.40)

(i)

ha ' isa ¡Sekanta po dira hi 'omanit. "The woman that bought here

(an) apartment

she (=is)(an) artist.' (ii)

gveret tal, sekanta po dira, hi

'omanit.

'Miss Tal, that bought here (an) apartment, she (=is)(an) artist.'

41 (2.41)

(i)

hu sone 'et ha ' iìsa gekanta po dira. 'He hates OM the woman that bought here (an) apartment. 1

(ii )

'ani sone 'et 'adon levi, £ekol boker make 'et 'iSto. 'I hate OM Mr. Levy, that every morning hits OM his wife.'

(2.42)

(i)

kol layla 'ani ttolem 'al ha'isa £ekanta po dira. 'Every night I dream on (=about ) the woman that bought here (an) apartment. 1

(ii)

hu fiolem 'al 'elizabet teylor, gehitfiatna 'im ri(5ard barton. 'He dreams on (=about) Elizabeth Taylor, that got married with Richard Burton. 1

Each of the sentences numbered (i) in (2.40) - (2.42) above contains a restrictive relative clause, while each of the sentences numbered (ii) contains a nonrestrictive relative clause. Two facts must be mentioned now, before dealing with the grammatical rules by which these sentences are generated. (2.43)

(i)

In all three sentences the relativized NP functions as the Subject of the relative clause. However, the same NP has different grammatical

42 functions in the main clause:

In (2.40) it is

the Subject; in (2.4l) it is the direct Object; and in (2.42) it is the indirect Object, (ii)

In all three examples, the N P is deleted from the relative clause.

I mentioned these two facts because later I shall illustrate the following phenomena:

(2.44)

(i)

The grammatical function of the N P in the relative clause has an influence on the relativization rule, but this rule is not influenced by the grammatical function of the N P in the main clause,

(ii)

When the N P of the relative clause is not the Subject, it is not always deleted (cf. Section 3 below).

I now come to the point where I may formulate the first version of the relativization rule.

This version is but tentative and is

meant to generate adequately only sentences such as (2.40)- (2.42) above. As this study progresses, further types of sentences will be introduced, and the first version of the relativization rule w i l l be rechecked as to its capacity to correctly generate the new sentences.

If the first

version appears incapable of doing so, corrections will be introduced irj the rule.

This process will be repeated again and again, until a

43 f i n a l v e r s i o n o f t h e r u l e i s found which w i l l a d e q u a t e l y g e n e r a t e t h e v a r i e t i e s o f s e n t e n c e s with r e l a t i v e

all

clauses.

As a f i r s t approximation, I propose t h e f o l l o w i n g

relativiza-

tion rule: (2.45)

[X 1 1

NP # [ W NP Z ] # Y L 2 3 4 5 6 7 8 2 0

KM+4 J0

Conditions:

6

0 8

(i)

2=5

=4>

( i i ) fNP?

S e v e r a l comments should be made concerning t h i s r u l e . symbol "EM" stands f o r " R e l a t i v e M a r k e r . "

In examples ( 2 . 4 0 ) -

The (2.42)

I used t h e R e l a t i v e Marker s e - which i s t h e most common one in spoken Modern Hebrew.

However, t h e r e e x i s t o t h e r R e l a t i v e Markers i n Hebrew.

The R e l a t i v e Markers a r e not always i n f r e e v a r i a t i o n s , but r a t h e r t h e r e a r e c e r t a i n environments i n which each R e l a t i v e Marker may o c c u r . T h e r e f o r e , I use f o r t h e moment t h e element "RM" i n t h e

relativization

r u l e , and l a t e r i n Chapter V ( R e l a t i v e Markers i n Hebrew) I s h a l l

state

t h e c o n d i t i o n under which each R e l a t i v e Marker may o c c u r . Condition ( i ) in ( 2 . 4 5 ) means t h a t t h e two NP's a r e both i n t h e i r s t r u c t u r e and i n t h e i r r e f e r e n c e . i s a d i f f i c u l t and complicated one.

identical

The t o p i c o f r e f e r e n c e

C o r e f e r e n t i a l i t y i s not merely a

grammatical d e v i c e but r a t h e r a g e n e r a l p h i l o s o p h i c a l q u e s t i o n which has

44 direct bearing on the theory of grammar.

I shall not enter into the

general question of coreferentiality here as this topic deserves a wide and deep investigation which is far beyond the scope of this s t u d y . 1 3 However, there is one aspect of this question which is related to the framework of the present study, i.e. how coreferentiality should be formally represented in underlying structures. following convention:

Chomsky suggested the

"Suppose that certain lexical items are designated

'referential' and that by a general convention, each occurence of a referential item is assigned a marker, say, an integer, as a feature." 1 4 Bach, on the other hand, proposed that underlying primitive Noun Phrases are purely indexical and that identity of these indices alone is relevant.15 In the present study I shall adopt the following notational device:

All identical lexical items having identical subscripts are

assumed to be coreferential.

This convention does not necessarily mean

that I prefer Chomsky's analysis over Bach's--in fact the opposite is true.

I adopt this notation only for the sake of simplification of the

presentation. What is the purpose of condition (ii) in (2.45)?

Recall that

the relativization rule was meant to apply to two distinct structures: one for appositive clauses and one for restrictive clauses.

In other

words, it was intended that rule (2.45) apply to structures such as (2.46)(i) and to structures such as (2.46)(ii) as well:

45 (2.46)

(i)

Underlying structure for an appositive clause [after the application of the parenthetical clause embedding, cf. rule (2.12) above]: S

(ii)

Underlying structure for a restrictive clause: S

In (2.46) (i), the appositive clause is adjoined to the relativized N P and thus is dominated by the same S-node that dominates the relativized NP. In (2.46)(ii), the restrictive clause is embedded in the relativized N P and is dominated b y it. Condition (ii) in (2.45) states that this rule applies whether the relative clause is dominated by the S-node (for appositive clauses) o r by the NP-node (for restrictive clauses).

The notation in this

condition is adopted from Bach's "Have" and "be".

46 Thus, rule (2.45) generates the same output of constituents for appositive and restrictive relative clauses. however, are not branched in the same manner.

These constituents,

When the output of this

rule becomes the input for the phonological rules, the differences in branching will enable the handling of the differences between restrictive and appositive clauses insofar as the pausal and intonational patterns are concerned. At this point, before pursuing the treatment of the various types of Hebrew relative clauses, I should like to discuss the special characteristics of the Hebrew relative clause which differentiate it from relative clauses in some other languages. Although it may have already been understood from the relativization rule and from the examples given above, I should like to make the following fact very explicit:

Relativization in Hebrew takes place

not by means of a relative Pronoun but rather by means of a Relative Marker.

This fact actually means that the relativization process in

Hebrew is not involved with any movement of elements inside the relative clause.

This is not the case with languages such as English in which

the identical N P in the embedded sentence is moved to the beginning of the relative clause: i.e. relativization is a reordering transformation in English but not in Hebrew. Thus, many general syntactic constraints on movement of elements are consequently imposed on English relative clauses, but not on Hebrew relative clauses.

47

In his most exhaustive study, Constraints, Ross dealt with the constraints on reordering transformations in English, and I would like to discuss some of these constraints (obviously those which are concerned with relativization) in relation with the syntactic reality in Modern Hebrew. On page 16, Ross states that "elements of relative clauses may not be questioned or relativized."

This restriction, however, is

not valid for Hebrew relative clauses.

Thus, sentences such as raftel

makira 'et ha'jg ¡Serima 'et haftalban ('Rachel knows OM the man that cheated OM the milkman') may be embedded as a relative clause in an NP whose head Noun is haftalban ('the milkman'), e.g. (2.47)

ra'iti 'et haftalban Seraftel makira 'et ha'is serima 'oto. 'I saw OM the milkman that Rachel knows OM the man that cheated OM + him.' 1 6

Notice, incidentally, that the above mentioned restriction does hold in Hebrew insofar as questioning is concerned, because the Question transformation in Hebrew, as in English, is involved with element movement. (2.48)

Thus (2.48) is also ungrammatical in Hebrew:

*('et) mi raftel makira 'et ha'iS serima? ' (OM) who Rachel knows OM the man that cheated?'

48 On page 17, Ross brings up another case in which relativization cannot take place in English, and again this restriction is not applicable in Hebrew for the same reasons mentioned above.

He states

the fact that in English, "elements of sentences in apposition to such sentential nouns as fact, idea, doubt, question, etc., cannot be questioned or relativized."

However, the following sentence, which is

ungrammatical in English, is grammatical in Modern Hebrew: (2.49)

hine hablkini seha'uvda serut lavsa 'oto hirgiza 'et 'ilana. 'Here is the bikini that the fact that Ruth wore OM + h i m 1 6 (=it) annoyed OM Ilana.'

Again, the last restriction is valid in Hebrew as far as the Question transformation is concerned; therefore, sentence (2.50) is ungrammatical in Hebrew also: (2.50)

*('et) ma ha'uvda serut lavsa hirgiza 'et 'ilana? ' (OM) what the fact that Ruth wore annoyed OM Ilana?'

Furthermore, the constraint imposed on relativization in coordinate NP structures does not hold in Modern Hebrew. words (p. 22):

In Ross'

"An NP which is a conjunct in a coordinate NP structure

cannot be questioned or relativized."

But in Modern Hebrew, in which

no movement of elements is involved with relativization, sentences such

as walas sone 'et harepubllkanlm va'et hademokratim ('Wallace hates OM the Republicans and OM

17

the Democrats.') may be embedded as a relative

clause into an NP such as harepublikanim, e.g.: (2.51)

'anafinu madabrim ' al harepublikanim sewalas sone 'otam ve'et hademokratim. 'We talk (=are talking) on (=about) the Republicans that Wallace hates OM + them and O M 1 7 the Democrats.'

The discussion thus far was not intended to claim or disclaim the universality of Ross' constraints on variables (e.g. "the Complex NP Constraints" or "the Coordinate Structure Constraint"), which are directly related to the examples discussed above. 1 8

The sole purpose of

this discussion has been to emphasize the differences which exist between relative clauses in Hebrew and in other languages such as English.

These differences stem from the fact that in English, the

identical NP in the embedded sentence is moved to the beginning of the relative clause, joins with the Wh-, and yields the relative Pronoun. In Hebrew, on the other hand, the identical NP in the embedded sentence is deleted or pronominalized (without being moved), and the embedding is done by the Relative Marker.

This process enables Hebrew to have

relative clauses in which any number of sentences may intervene between the relativized NP and the embedded sentence in which the identical NP occurs.

Schematically, then, relativization may take place in structure

50 such as (2.53) in Hebrew but not in languages in which relativization involves movement of the identical NP: (2.53)

Notice that our relativization rule (2.45) accounts for this fact because the variable W in the structural index of this rule may contain S. Thus, rule (2.45) adequately applies to sentences such as (2.47) above whose deep structure is the following:

51 (2.54)

haftalban the milkmanif NP

rima ftalban # cheated milkman Notice that the two pairs of identical NP's in (2.54)(NP NP.) do not agree in their Determiners: J

and

the NP in the main clause is

definite, while the NP in the relative clause is indefinite.

The rela-

tionship between Determiners and relative clauses has been discussed byseveral generative grammarians. three points were emphasized:

In early treatments of this question,

52 (a) There is a distinction between definite, indefinite, and generic Determiners; (b) the main difference between generic and definite Determiners is that the latter has an anaphoric nature which relates its occurrence to the presence of a relative clause; (c) the restrictions on the occurrence of the different Determiners may be stated in syntactic terms (e.g. depending on the occurrence of relative clauses after a Noun). Later studies proved that this last point cannot always be justified. 2 0

In other words, it was shown that the quality of definite-

ness is not always predicted by the occurrence of a restrictive relative clause, but rather the relativized NP may be either definite or indefinite, depending on its semantic interpretation.

Compare, for

example, these two sentences: (2.55)

kaniti sefer se'eHkol katav C'oto). 'I bought (a) book that Eshkol wrote (OM + him (=it) ).'

From:

kaniti sefer. 1 'e'gkol katav 'et hasefer^

(2.56)

kaniti 'et hasefer %e'e£kol katav ('oto). 'I bought OM the book that Eshkol wrote (OM + him (=it) ).'

From:

'e^kol katav sefer. kaniti 'et hasefer.

53 Sentence (2.55) contains what Annear calls a "descriptive" clause and (2.56) contains what she calls a "restrictive" clause. These sentences illustrate then that the quality of the Determiners cannot be predicted on syntactic grounds.

Rather, the

choice of different Determiners in the deep structure may later play an important role when the semantic projection rules give different interpretations to sentences such as (2.55) and (2.56).

Furthermore, as a

consequence of this analysis, we must also assume that the identity condition does not require that the two NP's agree in their definiteness.

3.

Relativization Rule - Version II Let us now consider some relative clauses in which the iden-

tical NP does not function as the Subject of the relative clause: (2.57)

(i)

habafiura sera'ita 'otah 3 'etmol bamasiba hi rakdanit. 'The girl that you saw OM + her 3 yesterday in the party she (=is) (a) dancer. 1

(ii)

'ilana, Sera'ita 'otah 3 'etmol bamssiba, hi rakdanit. 'Ilana, that you saw OM + her 3 yesterday in the party, she (=is) (a) dancer.'

54 (2.58)

(i)

habafiura £edan 'ohev hi 'afiot. 'The girl that Dan loves she (=is)(a) nurse.'

(li)

*'liana, £edan 'ohev, hi 'afiot. 'liana, that Dan loves, she (=is)(a) nurse.1

(2.59)

(i)

habafiura sehu fiolem 'aleyha21kol layla hi dugmanit. "The girl that he dreams on (=about ) hei*21 every night she (=is ) (a) model.'

(ii)

rut, 'Sehu fiolem 'aleyha21kol layla, hi dugmanit. 'Ruth, that he dreams on (=about) hei^1 every night, she (=is)(a) model.'

(2.60)

(i)

*habafiura £ehu fiolem 'al kol layla hi dugmanit. 'The girl that he dreams about every night she (=is)(a) model.'

(ii)

*rut, 'Sehu fiolem 'al kol layla, hi dugmanit. 'Ruth, that he dreams about every night, she (=is ) (a ) model. '

In (2.57) - (2.6O), the sentences marked (i) contain restrictive relative clauses and those marked (ii) contain nonrestrictive clauses. The conclusions that may be drawn from these examples are the following:

55 (2.6l)

(i)

If the identical N P functions in a restrictive relative clause as a direct Object it is either pronominalized [see (2.57)(i)] or deleted [see (2.58) (i) ].

(ii)

If the identical N P functions in a nonrestrictive clause as a direct Object, it must be pronominalized [see (2.57)(ii)].

Its deletion leads

to a n ungrammatical sentence [see (iii)

(2.58)(ii)].

If the identical N P functions as an indirect Object in either a restrictive or nonrestrictive relative clause, it must be pronominalized [cf. (2.59) and

(2.60)].

A t this point I should like to mention another fact which actually should have been included in the previous section in which I discussed relative clauses in which the identical N P functions as the Subject.

However, I intentionally delayed the treatment of this

phenomenon u p to this point because it may now be handled more effectively after knowing the facts just mentioned in (2.6l) above. A s it may be recalled, I stated in the previous section that in the process of relativization, the identical N P is deleted from the relative clause, providing that it functions as the Subject of the relative clause.

This is indeed true, but only when this N P is not part

of a coordinate N P structure. 2 2

However, if it is dominated by a

56 coordinate NP-node, it is pronominalized in the relativization process, no matter what function it performs in the relative clause.

This

phenomenon is illustrated in the following examples [cf. also (2.5l) above]: (2.62)

(i )

zehu hasafikan^ sehu^ 'o ricard barton yasafiaku baseret hayisra'eli hefiada£. 'This is the actor, that he. or Richard Burton 1 1 will play in the movie the Israeli the new (=in the new Israeli movie).'

(ii)

zot 'ilana.,J Sehi. varut ma'ohavot bifrank i 1 sinatra. 'This is liana., that she. and Ruth are in love 1' 1 in (=with) Frank Sinatra. 1

With these remarks I conclude my discussion concerning the conditions under which the NP inside a relative clause (with a Verbal Predicate) is deleted or pronominalized. It is obvious that the first version of the relativization rule (2.45) is not adequate now for generating the sentences that have been discussed in the present section.

According to the way in which

(2.45) is stated, the identical NP in the relative clause is always deleted.

However, the sentences which have been introduced in this

section require a change in the rule which will enable it to delete or to pronominalize the embedded NP in the correct environments.

57 I therefore propose the following new version of the relativizatlon rule which will cover all the sentences which have been introduced thus far (i.e. all the varieties of relative clauses with Verbal Predicates):

(2.63)

SD 2 3 :

SO:

[X NP 1 2

# 3

1 2

j

NP 5

tw 4

Z L # 6 s 7

Cn H + P m RM+4 | ^ j L 7 ° J j 6

0

Y 8

L

s

8

SC (a) - either (i) if 4 contains V, or (ii) if 5 is dominated by a coordinate NP-node SC (b) - elsewhere Where (i) 2=5

(ii)fs 1 |NPj

=>4...6

This new version of the rule implies that the embedded NP is pronominalized when it is either a direct or an indirect Object (both, of course, are preceded by a Verb), or if the embedded NP is a part of a structure such as NP

This new version of the rule is able to adequately generate all the sentences which have been introduced thus far except one:

Rule

58 (2.63) may generate sentences such as (2.57)(i) but not (2.58)(i).

I

should like to emphasize the fact that these two sentences are pure paraphrases of each other, and they do not even differ in any degree of emphasis.

The only difference between them is a stylistic one, and

thus, I believe that the deletion of the Pronoun in (2.58)(i) is a matter which should not be dealt with o n the grammatical level but rather on the level of performance.

I therefore consider rule

(2.63)

above completely adequate for generating all the grammatical sentences that have been introduced thus far.

4.

The Object Marker Insertion Rule

In the phrase marker of the sentences which have been introduced so far, I have not included the Object Marker [see, for example, (2.54) above].

The reason for this is simply the fact that this

element may be introduced by a rather simple transformational rule. In this section I shall deal with the formulation of the Object Marker insertion rule.

As mentioned in note 3 of this chapter,

'et precedes a direct Object in cases when the latter is one of the following:

(2.64)

(i)

A Proper Noun

(ii)

A Noun preceded by the definite Article ha-.

(iii)

A Pronoun

59 (iv)

A Noun modified by a possessive Pronoun, e.g. bayit ('house' p P o s s ^ ' a n i

('I') => beyti ('my

house') I should like to further elaborate case (iv) which is related to the question of smixut ('construct state') in Hebrew.

Consider the

following examples :

(2.65)

(i)

hu kana bayit. 'He bought (a) house.'

(ii)

hu kana beyt-kolnoa'. 'He bought (a) house-movie (=movie theater).'

(lii) hu kana 'et beyt-hakolnoa'. 'He bought OM house-the movie (=the movie theater).' (iv)

hu kana 'et beyt-dan. 'He bought OM house-dan (=Dan's house).'

(v)

hu kana 'et beyti. 'He bought OM my house. '

Any Noun in Hebrew may be either in the independent state [as in (i) of (2.65)] or in the construct state [as in (2.65)(ii)-(v)]. A Noun is in the construct state when it is followed by another Noun and is related to it in the manner N^of^N, where the word of is implied both by the construction N ^ N and by the form of the first word (the

60 construct state form).

It should be noted that the independent and

construct forms of a Noun are not always distinct.

Thus, a word like

'voice' is kol both in the independent and the construct state, but its syntactic function may be predicted by the environments in which it occurs, e.g. hakol fiazak ('The voice (is) s t r o n g ' ) — i n d e p e n d e n t ; kol-hara'am fiazak ('The voice of the thunder (is) strong' )--construct state. W h e n we have a smixut construction in which the second Noun is a Pronoun, the first word is inflected.

In other words, the first word

is in the construct state, and a pronominal suffix is added to it [cf. (2 .65) (v) ]. Notice that if such a construction is definite, the Article follows the second Noun [sentence

(iii) in (2.65)].

Notice furthermore that if a smixut construction functions as a direct Object in a sentence, it is preceded by the Object Marker only if it is definite; that is, if the second Noun is preceded by ha-, or if it is a proper Noun, or if it is [+Pro] [cf. (ii) with (iii) - (v) in

(2.65)]. Let us assume the following underlying structure for the

smixut constructions:

61 (2.66)

NP.

Eventually, a rule such as (2.67) below would be required, in the grammar of Hebrew: (2.67)

[ [X^NP^Poss] 2 3 1 0

f)

f>

4 4+312

This rule would convert (ha ^kafeToss^bayit into bayit"Poss~ (ha Fkafe.

A morphophonemic rule will convert bayit Poss into beyt.

Should the second Noun be [+Pro], the first Noun would be inflected. Notice again that the Object Marker precedes a smixut construction only if the node Ng in (2.66) is either [+Pro] or a Proper Noun, or if the Determiner-node dominates a definite Article.

This

situation is quite similar to the situation described in (i)-(iii) in (2.65) above. Before arriving at the formulation of the Object Marker insertion rule, I should like to introduce a proposal made by Bach concerning the Determiner.

He proposes that "the definite, indefinite

62

and generic distinctions are treated as features of nouns rather than as independent morphemes in the base." 24

In a note, Bach adds:

Among the arguments for treating the definite and indefinite determiners as features are the following: pronouns and proper names act syntactically like definite nouns in some respects: the definite article is used with modified proper names, otherwise some proper names have it, some do not; many of the rules involving N are simpler if we can omit the determiner until rather late in the transformations. Semantically, proper nouns and definite noun phrases are alike. The advantages of this analysis in other languages are obvious, where we sometimes have suffixes, sometimes proclitics, sometimes both, sometimes only positional characteristics representing definiteness. p. 464 The description given above concerning the regularity which underlies the Object Marker insertion supports Bach's proposal and arguments.

I therefore adopt Bach's proposal and I suggest the conven-

tion that from this point on, any NP node which contains the definite Article, a proper Noun, or a Pronoun is marked as [+Definite]. Thus the Object Marker insertion rule may be stated simply as (2.68)

[ X V 1 2 1

NP 3 2

Y ] 4

OM+3

Condition:

4

3=[+Definite]

This modification not only results in economy of feature specification but also emphasizes the common characteristics which tie together definite Nouns, proper Nouns, and Pronouns.

63 Finally, I want to discuss the relative order of application of the Object Marker insertion rule and the relativization rule.

In

the first approximation, it appears that the relativization rule is always one cycle up from the OM-insertion rule. evidence which makes this assumption doubtful.

However, there is Let us consider the

following sentence and its underlying structure: (2.69)

(i)

kaniti 'et hasefer sedan katav 'oto. 'I bought OM the book that Dan wrote OM+him (=it).'

(ii)

Sn

hasefer the book

dan katav sefer Dan wrote (a) book

It is clear, that in the first cycle (i.e. in Sg) the OMinsertion rule will not apply because the direct Object is not [+Definite].

The relativization rule (2.63) then applies and the

embedded NP is pronominallzed. insertion rule applies.

In the second cycle (S^), the OM-

Such a derivation will yield this ungrammatical

sentence instead of (2.69):

64 (2. 70)

*kaniti 'et hasefer "Sedan katav hu.

The only way in which this ungrammaticality may be avoided is by assuming the OM-insertion rule to be last or post cyclic.

In so

doing, the final output of the cyclic rule would be kaniti hasefer sedan katav hu.

Then the OM-insertion rule will apply and 'et will be

inserted both in front of hasefer and hu.

65

NOTES TO CHAPTER II

Many of the arguments in this section are not properly mine.

In

the writing of this section I was greatly influenced by the discussions in Stanley Peters' course on "Problems in Transformational Grammars."

The course was offered in 1967-68 at the University of

Texas. The Philosophy, pp. 112-113. The word 'et is an Object Marker ("OM") which precedes the direct Object when the latter falls into one of the following categories: (a) When it is preceded by a definite Article: hu roce 'et hasefer. 'He wants OM the book.' (b) When it is a proper Noun: hu 'ohev 'et rut. •He likes OM Ruth.' (c) When it is a Pronoun: hu 'ohev 'otah. 'He likes OM + her.' (d) When it is a Noun accompanied by a possessive Pronoun: hu lo makir 'et dodi. 'He (does) not know OM my uncle.'

66 Otherwise there is no Object Marker, e.g. hu 'ohev bafturot ('he likes girls'). The Object Marker is an inflected word and obviously it takes the inflected forms when the direct Object is a Pronoun. Thus, 'et + hi

'otah [cf. (c) above], ' et + hu =»'oto [as in

example (2.3) in the text]. The problem of the Object Marker will be discussed in greater detail in Section 4 of this chapter. 4.

Teachers of Hebrew grammar urge their students to avoid the use of continuative relative clauses and to use conjoined sentences instead.

See, for example, Peretz's mi^pat, who claims that the

use of continuative relative clauses "may blur the content of the sentence and even distort it" (p. 150).

Warnings such as this do

not help, of course, and continuative relative clauses are found in Modern Hebrew, especially in the language of the radio and the newspapers in which syntactic borrowings from English are quite common.

However, as was mentioned, the usage of continuative

relative clauses in Modern Hebrew is still much more infrequent than it is in English, for example. 5.

The tree diagrams throughout this study are branched in different degrees of detail;

in some cases even different nodes in the same

diagram are branched in different degrees of precision. only for the sake of practicality:

I do so

I include the details only when

they are directly relevant to the discussion at hand.

67 The subscripts (e.g. S^, S^, etc.) have no systematic significance, and are used merely as expository devices. Jacobs and Rosenbaum, English, pp. 261-2. One may wonder why I compare the situation in Hebrew to that in English since this study is not a contrastive study of EnglishHebrew syntaxes.

My motivation in doing so is the following:

different generative grammars are written for different languages; however, generative grammarians have been vitally interested in the question of linguistic universals.

These universals may be found

only through the study of as many languages as possible.

Such

studies may lead the linguist to some conclusions concerning which phenomena are more or less "natural" in languages.

Such conclu-

sions have a direct impact on the formulation of the base rules of the grammar because these rules are intended to be universal. However, even phenomena which are not directly related to the structure of the base rules must also be studied from a contrastive point of view.

I am referring especially to some major transforma-

tional rules, e.g. relativization, which are found in most (if not all) natural languages.

It may be the case that such major trans-

formations apply differently in different languages, but it may be realized that the conditions under which such transformations apply are universal.

That is, it may be the case that the structural

index for a rule such as relativization is universal while the

68

output (the structural change) of this rule differs from language to language.

The reason for my references to English is the fact

that more transformational studies have been done in this than in any other language. If future studies prove that "universal structural indices for major transformations" have any realistic grounds, and if evidence from English alone cannot supply the answer to which possibility—(i) or (ii) of (2.10)—is more desirable, then the evidence given here from Hebrew and evidence from some other languages would assist in making thé choice between (i) and (ii) of (2.10). 8.

Aspects, pp. 107, 128-129.

9.

This solution was suggested by Postal (cf. On so-called 'Pronouns').

10.

QW = Question Word.

11.

On Relative Clauses in Russian, p. 1.

12.

Cf. Chomsky, Aspects, p. 107.

13.

For a discussion of some of the problems involved in the question of coreferential identity, see Ross' Pronouns and Reference and Counterparts.

14.

Op. cit., p. 145.

15.

See Bach's Nouns.

He writes:

"I have argued on the basis of many

pieces of evidence that it is reasonable to suppose that all nouns come from relative clauses based on the predicate nominal

69 constituent.

Further, I have proposed that the referential indices

assumed to occur with nouns in current theory be replaced by a system of operators and variables much like those used in logic but clearly different in detail, and that these elements rather than actual pronouns or the like be used to tie together the sentences underlying a single complex utterance" (p. 12l). 16.

As to this anaphoric Pronoun, see

above and especially

Section 3 below where I discuss relative clauses in which the NP in the relative clause does not function as the Subject. 17.

In coordinate NP's which function as direct Objects or as parts of Prepositional Phrases, the Object Marker or the Preposition precede each conjunct of the coordinate NP. (a)

Thus:

'ani roce 'et ha'iparon va'et ha'et. 'I want OM the pencil and OM the pen.'

(b)

hu fiosev rak 'al kesef vs'al min. 'He thinks only on (=about) money and on (=about) sex.1

18.

The fact that the Hebrew Question transformation obeys Ross 1 constraints does in fact suggest the possibility that some (or all) of these constraints have some universal character.

19.

Notice that in the surface structure this Pronoun does not occur [cf. sentence (2.47)].

In Hebrew, the Person of an inflected Verb

in the Past and Future tense is redundant because it may be

predicted by the different inflectional affixes which correspond to the different Persons.

Some examples will illustrate this

paradigmatic regularity. Past

Future (same Pronouns)

samarti

'I guarded'

'e^mor

sakalti

'I weighed'

'eskol

samarta

'you (m.s.) guarded'

tismor

£akalta

'you (m.s.) weighed'

tiSkol

samart

'you (f.s.) guarded'

ti£ma ri

sakalt;

'you (f.s.) weighed'

tiSksli

Marnar (¿)

'he guarded'

yi^mor

sakal

'he weighed'

yi^kol

samra

'she guarded'

tismor

Sakla

'she weighed'

tiSkol

In the Present tense, however, the Person cannot be predicted by the form of the Verb and therefore it is not deleted. 'ani

Somer

'I (m.s.) guard'

'ata

somer

'you (m.s.) guard'

hu

Corner

'he guards'

'ani

someret

'I (f.s.) guard'

'at

someret

'you (f.s.) guard 1

hi

someret

'she guards'

Thus:

71 'anafinu

Somrim

we (m.p.) guard'

'atem

'Somrim

you (m.p.) guard'

hem

^omrim

they (m.p.) guard'

' a nafinu

s omr ot

we (f.p.) guard'

'aten

¡Somrot

you (f.p.) guard'

hen

somrot

they (f.p.) guard'

The conclusion from this description is quite obvious:

the Pro-

nouns should appear in the deep structure; then, a general rule of agreement would add the correct affixes to the stems (e.g. 'ata~ P a s t e s - k - 1 =» 'ata lakalta).

And then, a general rule of

Pronoun deletion would specify the circumstances under which the Pronouns should be deleted (i.e. 1st and 2nd Person which correspond to a Verb in the Past or Future tense). This claim was made in studies written independently by Annear (English and Mandarin Chinese) and Kuroda (Generative).

The former

based her arguments on semantic grounds and the latter on syntactic ones. For a good survey of the early approaches to the relationship between Determiners and relative clauses see the first chapter of Annear's paper. I basically adopt the conclusions reached by Annear and Kuroda concerning this question and absolve myself from restating

72 here all the arguments which have been discussed by them and which by and large apply to Hebrew as well. 21.

Prepositions are inflected in Hebrew in the same w a y as the Object Marker

(cf. note 3 above), thus:

la- ('to') + -i^ (lst person,

singular) = li; la- + -xa (2nd person, masculine singular) = laxa; la- + - a x (2nd person, feminine singular) = lax; la- + -o = lo ('to him');

la- + - a h = lah ('to her'), etc.

Some Prepositions take the plural suffixes rather than the singular ones which were mentioned above. Prepositions is 'al ('on, about') in (2.59). of this Pronoun is ' alay ('on me'), 'alevxa

One of these Thus, the inflection ('on you, masculine

singular'), 'alayix ('on you - f.s.'), 'alav ('on h i m 1 ) ,

'aleyha

('on her' ). It is clear, however, that these inflections will be handled by adequate morphophonemic rules.

In the deep structure,

a word like lo would appear as la- ( ' t o ' ) ^ h u 22.

('he').

This term is used in accordance with Ross' definition in Constraints which reads:

I propose to define the notion coordinate structure conforming to the schematic diagram in (4.85). (4.85)

A

73 Of course, since (4.85) is intended, to be a universal definition, it must be understood as containing not the English morphemes and and or, but rather a more abstract, language-independent representation of these terms.

23.

SD = Structural Description SC = Structural Change

24.

"Have and "be", p. 464.

C H A P T E R

III

RELATIVE CLAUSES WITH NONVERBAL PREDICATES 1.

Is there a Copula in Hebrew? Before dealing with relative clauses whose Predicates are

non-Verbal, I would like to discuss several general questions which are related to non-Verbal Predicates. I shall begin with a discussion concerning the Copula in Hebrew.

First, I shall introduce a list of examples, followed by a

discussion of these sentences, and finally I shall draw some conclusions.

All the examples below are simple affirmative sentences bear-

ing no special emphasis on either one of their constituents. (3.1)

(i)

david hu haganav. •David he (=is) the thief.1

(ii)

hamaSaret hu haganav. 'The butler he (=is) the thief.1

(iii)

'ata hu haganav. 'you he (=are) the thief.'

(iv)

*david haganav. 'David (is) the thief.'

(v)

-*haros saret haganav. 'The butler (is) the thief.'

(vi)

'ata haganav. •You (are) the thief.'

75 (vii)

'onasis--hamilyoner hamsfursam. 'Onassis (is) the millionaire the famous.1

(viii) hahatkafa--hahagana hatova beyoter. 'The attack (is) the defense the good most (= the best defense). (3.2)

(i)

david hu ganav. •David he (=is) (a) thief.'

(ii)

hams'fearet hu ganav. 'The butler he (=is) (a) thief.'

(iii) *'ata hu ganav. 'You he (=are) (a) thief.' (iv)

david ganav. 'David (is) (a) thief.'

(v)

hama'Saret ganav. 'The butler (is) (a) thief.'

(vi)

'ata ganav. 'You (are) (a) thief.'

(3.3)

(i)

matl§evim hem hamila ha'afiarona bamada*. 'Computers they (=are) the word the last in the sciences.'

(ii)

cvi'ut hi hatxuna habsisit £el ha'adam. 'Hypocrisy she (=is) the trait the basic of the man.'

76 (iii)

*znut hamikcoa' ha'atik beyoter. 'Prostitution (is) the profession the old most (= the oldest profession).'

(iv)

?znut--hamikcoa' ha'atik bayoter. [Translation the same as (iii)3

(3.4)

(i)

znut hi mikcoa* 'atik. 'Prostitution she (=is) (a) profession old.'

(ii)

cayarim hem 'omanim. 'Painters they (=are) artists.'

(iii)

*znut mikcoa' 'atik. 'Prostitution (is) (a) profession old.'

(iv)

znut--mikeoa' 'atik. [Translation same as (iii)]

(3.5)

(i)

david hu toy. 'David he (=is) good.'

(ii)

hasefer hu tov. 'The book he (=is) good (=The book is a good one).'

(iii)

*'ata hu tov. 'You he (=are) good.'

(iv)

david tov. •David (is) good.'

(v)

hasefer tov. "The book (is) good.'

77 (vi)

'ata tov. 'You (are) good.1

(3.6)

(i)

*'ecim hem yerukim. •Trees they (=are) green.'

(ii)

^uJftan hu lif'amim 'agol. '(A) table he (=is) sometimes round.'

(iii)

*ecim ysrukim. 'Trees (are) green.'

(iv)

-xluljian lif'amim 'agol. '(A) table (is) sometimes round.'

The examples given above obviously do not exhaust the many possible varieties of sentences with non-Verbal Predicates.

Neverthe-

less they may serve as a good basis for a general discussion concerning the so-called Copula in Hebrew. As may have been noticed, the sentences have been classified in a way which will facilitate the discussion. Group (5.1) contains sentences whose Predicates are Nominal and whose Subjects and Predicates both are definite. Nouns, Det + N, Pronouns.)

(i.e., Proper

The general characteristic of this group

is that its sentences do require a Copula.

The absence of the Copula

will yield appositive constructions1 rather than sentences.

Thus,

(3.1) (iv), for example, means David the thief, rather than David is

78 the thief.

The Copula may be absent in sentences whose Subject is a

Pronoun [see (3.1)(vi)].

Notice also, that if the Copula is absent

from other sentences in this group, there must be a lengthy pause between the Subject and the Predicate in order for the sentence to be grammatical [compare (iv) (v) with (vii) (viii) of (3.1)]. Group (5.2) contains sentences whose Predicates are Nominal, whose Subject is definite and whose Predicate is indefinite.

In this

group, the Copula may or may not be present, except in sentences in which the Subject is a Pronoun; in the latter case, the Copula cannot appear [see (3.2) (iii)].

I should emphasize, however, that sentences

of this group which lack the Copula are much more widespread than those with the Copula. Group (5.5) contains sentences whose Predicates are Nominal, whose Subject is indefinite, and whose Predicate is definite.

Before

even dealing with this type of sentence, I should like to mention the fact that the occurrence of such sentences in normal discourse is extremely rare.

In reality one usually tells something "unknown" (Predi-

cate) about something "known" (Subject).

This type of sentence contra-

dicts this tendency. Notice that the grammatical sentences in this group are all "definition sentences."

Furthermore, the Predicate contains an em-

bedded sentence (relative clause).

I could not discover any grammatical

sentence in which the Predicate does not contain an embedded sentence.

79 Later in this study it will be demonstrated that the existence of the copulative Pronouns in such sentences depend on the existence of the embedded sentence [cf. (4.24) and (4.28) in Chapter IV]. Group (5.4) contains sentences whose Predicates are Nominal, and whose Subject and Predicate both are indefinite.

In this group

the Copula is mandatory, unless a lengthy pause exists between the Subject and the Predicate.

Here again, as in Group (3.1), the absence

of the Copula will yield appositive constructions1 rather than complete sentences; that is, the lack of a Copula in a sentence such as (3.4) (ii) would yield the constituent Painters-artists (=artistic painters) rather than painters are artists. Group (5.5) contains Adjectival Predicates and definite Subjects . The general tendency in this group is to omit the Copula, even though there are stylistic paraphrases which do contain Copulas [compare (i) and (iv) of (3.5)].

Sometimes, however, the presence of the

Copula leads to a slightly different meaning [compare (ii) and (v) of (3.5)].2 Group (5.6) contains Adjectival Predicates and indefinite Subjects.

In this group, the Copula is obligatory because its absence

would yield constructions of a Noun followed by its modifier instead of complete sentences.3 than trees are green.

That is, (3.6)(iii) means green trees rather Notice, incidentally, that sentences of this

type are often ungrammatical even if the Copula is present.

In natural

80 discourse, sentences such as (3.6)(i) would usually "be uttered with the Article ha- preceding the Subject.

The reason for this was discussed

above in connection with Group (3.3). Ignoring minor details, the situation described above may b e summed up in the following chart which illustrates general tendencies rather than the complete reality: 4

Subject

Predicate

Copula

(i)

[+Def]

[+Def]

+

(ii)

[+Def]

[-Def]

-

(iii)

[-Def]

[+Def]

+

(iv)

[-Def]

[-Def]

+

(v)

[+Def]

Adjective

-

(vi)

[-Def]

Adjective

+

A t first glance, (3.7) m a y give the misleading impression that the presence of Copulas in Hebrew non-Verbal sentences is more dominant than its absence.

But a deeper look into the matter leads

to a n opposite conclusion; that is, that the absence of the Copula is basic and its presence is secondary.

W e have already noticed that

cases such as (iii) and (vi) of (3.7) are very rare in regular discourse.

This diminishes the number of pluses in the Copula column by

two, and thus w e remain with 2 pluses vs. 2 minuses.

Let m e emphasize

again that in cases (i) and (iv) and (vi) of (3.7) the Copula is

81 obligatory only for one p u r p o s e — t o avoid the misinterpretation of nonVerbal sentences as constituents of a Noun and its modifier (whether this modifier is a Noun or an Adjective), i.e.: (3.8)

(i)

ca.yar hu 'oman. '(A) painter he (=is) (an) artist.'

vs.

cayar 'oman. 'Painter-artist (=artistic painter).'

(ii)

politika'im hem •armumiyim. 'Politicians they (=are) crafty.'

vs.

politika'im *armumiyim. 'Politicians crafty (=crafty politicians).'

The question is then the following:

should the Copula be

introduced by the base rules or rather by transformations?

The infor-

mation I have given above shows that at least as far as the native speaker's intuition is concerned, the Copula is not considered to be fundamental in non-Verbal sentences, but rather is thought to be an element which is introduced "to avoid misunderstandings." But this is not the only motivation for not having the element "Copula" in the base.

I would like, therefore, to introduce fur-

ther evidence in support of this claim. Notice that the so-called Copula must always occur in the Past and the Future tenses, e.g.

82 (3.9)

(i)

ha'uga mecuyenet. •The cake (is) excellent.'

(il)

ha'uga hayta mscuyenet. 'The cake was excellent.'

(iii) ha'uga tihye macuyenet. 'The cake will "be excellent.' When the Tense of the sentence is Past or Future, the Copula is simply the Verb h.y.y. ('be') in the Past or Future tense.

However,

when the Tense of the sentence is Present, the Copula is one of the third person Pronouns.

This fact has lead some linguists to claim that

the Copulas in the Present tense are anaphoric Pronouns rather than linking elements.

See, for example, Segal's dig dug:

This Pronominal unit is a linking element between the Subject and the Predicate only from the point of view of linguistic usage and habit. From a grammatical point of view the Pronoun is just a second Subject which replaces the main Subject that remains out of the sentence boundaries. Thus in the sentence 'ani hu hatame ( 1 1 he (=am) the impure') we have a double Subject: 'ani ('I'), hu ('he'). The sentence actually has two parts, hu hatame ('he the impure') and the main Subject is out of the sentence boundaries. p. 181 Historically, the so-called Copula in Hebrew indeed had the nature of an anaphoric Pronoun.

I do not believe, however, that this

is the way it should be explained from a synchronic point of view. However, the use of copulative Pronouns in Hebrew can in no way be paralleled to the use of the Verb be in English, for example.

83 From the discussion above, the following conclusions may be drawn:

In some cases, the presence of the Copula is mandatory; in some

cases the absence of the Copula is mandatory; and in the remaining cases the presence or absence of the Copula depends on stylistic variations.

The first two instances must be handled by the grammar of Hebrew

while the latter should be dealt with on the stylistic level. It is obvious here that I am interested only in the impact that the existence or absence of the Copula has on the grammatically of sentences rather than the impact on stylistic variations. The question again is--should the base rules generate all non-Verbal sentences with Copulas and then transformational rules delete the Copulas where their absence is mandatory?

Or, should the base

rules generate sentences without Copulas and then transformational rules insert the Copulas where their presence is mandatory? I prefer the second alternative, and I shall introduce another argument supporting my choice. Consider the following sentences: (3.10)

(i)

haganav hu 'aharoni. 'The thief he (=is) Aharoni.'

(ii)

haganav--'aharoni. 'The thief (is) Ahroni.'

(3.11)

(i)

david hu 'arxitekt. 'David he (=is) (an) architect.'

84 (ii)

david 'arxitekt. •David (is) (an) architect.1

Notice that the deletion of the Copula in (3.10) must result in a heavy pause between the Subject and the Predicate, which is not the case in (3.11). If we choose the first alternative (i.e., Copula in the base), (3.10) and (3.11) will have the shape of (3.10)(i) and (3.1l)(i) at the end of the syntactic component. If we choose the second alternative (i.e., no Copula in the base), (3.10) and (3.11) will have the shape of (3.10)(i) and (3.11) (ii) at the end of the syntactic component.5 The second output is preferable over the first one not only because (3.11)(il) is more common in speech than (3.11)(i), but also because of the fact that with the second output as the basis of the stylistic rules it will be much simpler to handle the pausal differences between (3.10)(ii) and (3.11)(ii).

That is, if one chooses not to

generate the Copula by the base rules, and the final output of the syntactic component does contain a Copula (introduced by transformation, of course), and this Copula is to be deleted for stylistic reasonsits deletion must result in a pause. In other words, the first alternative (i.e., Copula in the base) is not able to show the difference between Copula deletion which results in a pause and a deletion which does not result in a pause.

85 The last argument which I would like to point out here in favor of base rules which do not contain Copula over the opposite alternative is that "by so doing the grammar of Hebrew demonstrates the great difference between non-Verbal sentences in Past and Future tenses vs. sentences whose Tense is Present.

In sentences whose Tense is

Past or Future, the so-called Copula is merely a Tense-Marker which depends only on the Tense-node in the phrase marker.

In sentences

whose Tense is Present, the so-called Copula is a kind of interpretive element which is sometimes needed and sometimes not needed.6

Thus, in

my analysis, the introduction of the so-called Copula either depends upon the content of the Tense-node (Past or Future) or upon the application of insertion transformations (in Present tense). This analysis not only represents the intuition of the Hebrew speaker concerning the Copula, but also, I believe, simplifies the grammar of Hebrew. Later when I deal with the differences between restrictive vs. nonrestrlctive non-Verbal relative clauses, it will be shown once again how the absence of the Copula in the base is advantageous for the analysis of these types of sentences.

2.

The Content of the Predicate The discussion and the conclusions above were highly influ-

enced by Bach's paper "Have" and "be."

In this paper, the author asks:

86 In many languages, for instance Russian, Arabic, Malay, Tagalog, there exist equational sentences consisting of a subject and an immediately juxtaposed nounphrase, adjective or locational phrase with no overt verb corresponding to the form be derived from Copula in the grammar of English. In setting up base rules for such languages, or in interpreting rules such as Chomsky's as candidates for a universal base component we are immediately faced with the following question: Should the element Copula be retained in the base and deleted transformationally in the appropriate places ? p. 462 Following this question Bach suggests (on the basis of languages with equational sentences and on the basis of evidence from English itself) that the Copula should not be introduced by the base rules, but rather should be inserted by transformational rules.

Hence,

to my previous arguments now is added Bach's argument which is based on the grounds that the absence of the Copula from the base may lead to a more universal character of the base rules. Following his own proposal, Bach thus claims that Chomsky's rule (3.12) below would be better stated as (3.13):7 (3.12) VP

' Copula

Predicate

(NP) (Prep-Phrase)(Prep-Phrase)(Manner) S' Predicate (3.13) VP - (V (Manner))

/'Predicate 4 (NP)(PP)(PP) I

S'

87 Bach also states the transformational rules which insert have and be in the adequate constructions.

I shall not repeat these rules

here as they differ from those which are found in Modern Hebrew grammar. I would like, however, to emphasize at this point that these transformational rules are much simpler in Modern Hebrew than in English as they refer only to the Verb h.y.y.8 ('be') and not to have.

English

sentences such as I have the book are expressed in Modern Hebrew in a quite different syntactic structure (see Section 5 below).9

Further-

more, in Modern Hebrew there are no "perfect" tenses which are found in languages such as English and French, and which are formed by have and avoir. respectively. I therefore believe that Bach's proposed rule (3.13) may simplify the grammar of Modern Hebrew not only because of the arguments which were brought out in the previous section but also because of the fact that by adopting this rule the analysis of non-Verbal relative clauses may be simplified. On the basis of the facts given above [see (3.9)] concerning the difference between non-Verbal sentences in the Present tense vs. non-Verbal sentences whose Tense is Past or Future, in my analysis the Verb h.y.y. ('be') will be adjoined to the Auxiliary and be considered as a Tense-Signifier while the copulative Pronouns will be inserted by another transformation and will be dominated directly by the S node. Hence, the Tense-Signifiers of sentences such as (ii) and (iii) of (3.9) will be handled by a rule such as

88

(3.14)

[ X

NP

Tense 10

Predicate Y ] s

1

2

3

4

5

1

2

5+h.y.y.

4

5

=»>

Obviously, l a t e r morphophonemic rules w i l l convert 5+h.y.y. t o i t s f i n a l i n f l e c t e d form depending on the Tense and on the Person, Gender, and Number of the Subject, (3.15)

(i)

e.g.

hayeled Past^h.y.y.^yafe =*> hayeled haya y a f e . 'The boy was (m.) n i c e . 1

(ii)

ha uga Future h . y . y . y a f a . => ha'uga t i h y e y a f a . •The cake w i l l be ( f . )

pretty.'

On the other hand, the copulative Pronouns w i l l be inserted i n t o non-Verbal sentences ( i n case t h e i r presence i s o b l i g a t o r y ) by a transformational r u l e which may look roughly l i k e :

(3.16)

[gx

1 1

[ ^Det) N

2

]Np

Y

• r

3 2

3+Pro Condition:

(

D e t

)

N

Adj '4

]np

Pred S

4 3 does not contain Tense. 1 0

89 Rule (3.16) basically describes the regularity which underlies the mandatory occurrences of copulative Pronouns in non-Verbal sentences whose Tense is unmarked (i.e., Present tense).

This rule

roughly covers the cases described in the previous section, and it is by no means claimed that the shape of this rule is final (see note 4). Rule (3.16) states that the copulative Pronouns are inserted in a nonVerbal sentence whose Tense is Present under the following conditions: either both the Subject and the Nominal Predicate agree in their definiteness (i.e., both are either definite or indefinite); or the Subject is indefinite and the Predicate is an Adjective.

A later rule will

state that the inserted copulative Pronoun should agree with the Subject in Gender and Number. Thus, according to my analysis, the following two sentences in (3.17) will differ in their deep structures in the manner described in (3.18): (3.17)

(i)

hasefer yakar. 'The book (is) expensive.'

(ii)

hasefer haya yakar. 'The book was expensive.'

90 (3.18)

(i)

The book

yakar expensive

[Rule (3.16) does not apply.] (ii)

NP

Ape Tense

VP Adj

hasefer

Past yakar

[Rule (5.14) appli es, and the output is hasefer Past h.y.y. yakar.] At this point I would like to concentrate the discussion on a different t o p i c — t h a t is, the -way in which the Predicate is developed. Bach suggests the following r u l e : 1 1 (3.19)

'(Degree) Adj (NP) ' Predicate

1

Time Place

I believe that this rule should be somewhat different for Modern Hebrew grammar.

I therefore suggest the following modification:

91 (3.20) Predicate ->•

The only difference between (3.19) and (3.20) is that the latter also contains a Prepositional-Phrase.

This change in the rule

was made in order to account for sentences with possessive attributes (see Section 5 below) and sentences such as: (3.21)

'ani makir 'et ha'i§a bama'il hakaftol. 'I know 0M the woman in the coat the blue.1

(3.22) habafiura *im hamitriya hi 'aftot. 'The girl with the umbrella she (=is) (a) nurse.1 The deep structure of (3.21) cannot have the shape sketched in (3.23) below even though (3.23) may be the adequate deep structure for an English sentence which is the equivalent of (3.21):

92

a coat

the coat is blue

Such a deep structure is not adequate for Hebrew because this language lacks the Verb have.

Furthermore, and even more impor-

tant, English and French sentences such as I have a book or J'ai un livre are expressed in Hebrew by a sentence whose Predicate is a Prepositional Phrase.

I do not want to enter into the details concerning

such sentences in Hebrew at this point since this topic will be dealt with at length in Section 5 below where I discuss reduced relative clauses which yield possessive attributes. Thus, according to my analysis, (3.21) has the following deep structure:

93

the woman ha'i^a the woman

Predicate Prep-Phrase

hama'il kafiol the coat (is) blue Up to this point the discussion was centered around questions which are concerned with the nature of Hebrew non-Verbal sentences in general.

I shall now pursue the analysis of non-Verbal sentences which

are embedded as relative clauses.

3.

R e l a t i v i z a t i o n Rule - Version I I I Close examination of non-Verbal r e l a t i v e clauses i n d i c a t e s

t h a t there e x i s t s a c l e a r d i s t i n c t i o n between the s t r u c t u r e of r e s t r i c t i v e vs. n o n r e s t r i c t i v e (appositive)

clauses.

I s h a l l f i r s t i l l u s t r a t e t h i s d i s t i n c t i o n by a l i s t of examples.

L a t e r , I s h a l l discuss the impact t h a t these sentences have

upon the r e l a t i v i z a t i o n r u l e

(2.63).

The sentences presented below a r e arranged in t h r e e s e t s : Set A contains sentences with appositive r e l a t i v e c l a u s e s j set B cont a i n s sentences with r e s t r i c t i v e r e l a t i v e c l a u s e s ; Set C contains sentences with a t t r i b u t e s which a r e derived from reduced r e l a t i v e clauses A. I .

( 3 . 2 5 ) hana'Sim, ~§ehen 'armumiyot. gormot c a r o t . 'The women, t h a t they (are)cunning, cause troubles ( 3 . 2 6 ) hanasi, "Seboker hu. gar bateksas. 'The p r e s i d e n t , t h a t ( a ) cowboy he ( i s ) , l i v e s in Texas. 1 ( 3 . 2 7 ) harahitim, "Sehem miSvedya, yafim vailazakim. •The pieces of f u r n i t u r e , t h a t they ( a r e ) from Sweden, ( a r e ) p r e t t y and s t r o n g . 1 ( 3 . 2 8 ) ha'Slatim, 'Sehem 'im zarfian. manacnacim. 'The s i g n s , t h a t they ( a r e ) with phosphorus, shine

95 A.II.

(3.25) hanaSim, "Sehayu 'armumiyot, garmu carot. 'The women, that were cunning, caused troubles. 1 (3.26) hanasi, ^ehaya boker. gar bateksas. 'The President, that was (a) cowboy, lived in Texas.' (3.27) harahitim, "Sehayu miSved.ya, hayu yafim vafiazakim. 'The pieces of furniture, that were from Sweden, were pretty and strong.' (3.28) ha'Slatim,

ehayu *im zarftan, nicnacu.

'The signs, that were with phosphorus, shone.' B. I.

(3.25) *hana'§im 'Se*armumiyot gormot carot. 'The women that cunning cause troubles.' (3.26) »hanasi "Seboker gar bateksas. 'The president that cowboy lives in Texas.' (3.27) harahitim 'Semi'Svedya yafim vaftazakim. 'The pieces of furniture that from Sweden (are) pretty and strong.' (3.28) ha'Slatim ~§e'im zarftan manacnacim. 'The signs that with phosphorus shine.'

B.II.

(3.25) hanasim

ehayu 'armumiyot garmu carot.

(3.26) hanasi "Sehaya boker gar bateksas. (3.27) harahitim ^ehayu miSvedya hayu yafim vaftazakim. (3.28) ha'Slatim

ehayu 'im zarfian nicnacu.

(Translation as in A.II., except for the punctuation: commas are needed in B.II.)

no

96 C.

(3.25) hanaSim ha'armumiyot gormot carot. 'The women the cunning (=the cunning women) cause troubles. 1 (3.26) hanasi haboker gar bsteksas. 'The President the cowboy lives in Texas. 1 (3.27) harahitim miévedya yafim vafiazakim. 'The pieces of furniture from Sweden (are) pretty and strong.' (3.28) ha'Slatim * im zarhan menacnacim. 'The signs with phosphorus shine.'

The following conclusions may b e drawn from the sentences given above:

(3.29) Group A.I. (i)

This group contains appositive relative clauses in which the Tense is Present (the unmarked Tense).

(ii)

In this type of sentences, Pronouns occur in the relative clause.

(Iil)

In sentence (3.26), the Pronoun is preceded b y the Predicate of the relative clause. This change in the word order has a stylistic significance rather than a grammatical

97 (3.30) Group A.II. (i)

This group contains appositive relative clauses in which the Tense is non-Present.

(ii)

The Tense-Signifiers

Past Future

h-y-y-

must occur, as was described in Section 2 above [cf. rule (3.14)]. (3.31) Group B.I. (i)

This group contains restrictive relative clauses in which the Tense is Present (unmarked ).

(ii)

This group differs from group A.I. not only in meaning "but also in the pausal and stress patterns.

In the sentences of A.I. there

is a pause before and after the relative clause (marked here "by commas) and also the stress pattern of restrictive clauses is different from that of appositive clauses. (iii)

Furthermore, the sentences in B.T. appear to be divided into two different types:

if

the Predicate of the relative clause is an Adverbial, or a Prepositional Phrase [see (3.27) and (3.28)], the Pronoun is absent

and the sentence is grammatical.

On the

other hand, if the Predicate is an NP or an Adjective, the sentence is ungrammatical unless it undergoes the reducing transformation [cf. (3.25) (3.26) of B.I. and of C]. (3.32) Group B.II. (i)

This group contains restrictive relative clauses in which the Tense is non-Present.

(ii)

The differences between the sentences of this group and group A.II. is in the pausal and stress patterns [cf. (3.31)(ii) above].

(3.33) Group C. (i)

This group contains reduced restrictive relative clauses.

That is, this group is

derived from B.I. by deleting the relative marker. (ii)

Notice that if the Predicate of the relative clause is an NP or an Adjective, it must agree with the NP in the main sentence insofar as the Determiner is concerned.

Let us see now how all these facts may be accounted for in the rules of Modern Hebrew grammar.

But before so doing let me mention

99 a g a i n t h e f a c t t h a t has a l r e a d y been d i s c u s s e d i n Chapter I I ,

i.e.,

t h a t t h e s e n t e n c e s of group A a r e d e r i v e d from t h e s t r u c t u r e given i n ( 3 . 3 4 ) w h i l e t h e s e n t e n c e s of groups B and C a r e d e r i v e d from t h e s t r u c t u r e given in

(3.35):

(3.34)

(3.35) NP NP i

S • NP.J-

The deep s t r u c t u r e of s e n t e n c e s such a s ( 3 . 2 5 ) of A . I . and A . I I . , f o r example, h a v e t h e n t h e f o l l o w i n g s h a p e :

100 (3.36) (i)

Deep structure of (3.25) of A.I. S-,

hanaSim The women

gormot carot cause troubles

(li)

hana'Sim

hana§lm the women

Adj \ armumyot cunning

Deep structure of (3.25) of A.II.:

garmu carot

Predicate I Adj I 'armumiyot

Before trying to apply our rules to these deep structures and before attempting to see if the output is correct, we should answer a general question which is concerned with group A.I.

101 As was noticed, before, the relative clauses of this group contain Pronouns. is.

One may wonder what the origin of these Pronouns

In other words, are they copulative Pronouns or are they anaphoric

Pronouns?

For example, is the Pronoun hen in (3.25) derived in the

manner described in (i) or (ii) in (3.37) below? (3.37) (i)

a.

hana'Sim-i gormot carot

b.

hana'Sim^ hen 'armumiyot. •

hana'Sim, "Sehen 'armumiyot, gormot carot. (The identical NP in the relative clause is deleted, and hen is a copulative Pronoun in the deep structure.) (ii)

a.

hana'Sim^ gormot carot.

b.

hana'S im^ * armumi yot. •

hana&imj , laehen-; 'armumiyot, gormot carot. (The identical NP in the relative clause is pronominalized into hen and there is no copulative Pronoun in the deep structure.) Sentences such as (3.25) of A.I. cannot give the answer to this question, but other sentences can. following sentences:

Consider, for example, the

102 (3.38) (i)

'ani ga'on. 'I (am) (a) genius.1

(ii)

'ani, %ega'on 'ani. 'exbo'S 'et ha'olam. 'I, that genius I, will conquer the world.1

(iii) *'ani, 'Sega'on hu. * exbo£ 'et ha*olam. "I, that genius he (=is, am), will conquer the world.'13 These sentences illustrate the point that the Pronoun in the relative clause is anaphoric rather than copulative. supported by the following facts:

This argument is

if the Subject of a non-Verbal sen-

tence is a Pronoun, the copulative Pronoun is always in the third person; thus 'ani hu hamaragel ('I he (=am) the spy').

Now, if the Pro-

nouns in embedded nonrestrictive non-Verbal clauses were copulative, sentence (3.38)(iii) would be grammatical, which is not the case. Sentence (3.38)(ii) illustrates then that the Pronoun in the relative clause is anaphoric to the relativized NP (in this case the relativized MP is a Pronoun and the anaphoric Pronoun is identical to it). We thus conclude that the deep structure in (3.36)(i) (without a copulative Pronoun) is the adequate one. Let us check now to ascertain whether the rules that have been introduced so far are able to generate nonrestrictive clauses with non-Verbal Predicates [e.g., (3.25) of A.I. and A.II].

103 I shall begin with (3.25) of A.I. whose deep structure is given in (3.36)(i) above.

The copulative Pronoun insertion rule (3.16)

does not apply to S^ in (3.36)(i) because the structural index is not met.

Then the parenthetical clause embedding rule (2.12) applies and

Sj is embedded into S^ and thus (3.36)(i) is converted into the following structure:

(3.39)

hana^im

Predicate Adj \ * armumiyot

Now we apply the relativization rule (2.63) to the structure in (3.39) and the result is the ungrammatical sentence [cf. (3.25) in group A.I.]:

(3.40) ~*hana§im, 'Se'armumiyot, gormot carot. 'The women, that cunning, cause troubles.'

In order to remedy this ungrammaticality, I must introduce a change in the relativization rule (2.63).

That is, the rule must be

104 remodified so that it will also pronominalize the embedded NP in cases when the Predicate of the appositive relative clause is non-Verbal and the Tense of this clause is Present. Let us now see if our rules are adequate for generating nonrestrictive non-Verbal clauses whose Tense is non-Present (i.e., group A.II.).

Let us apply our previous rules to (3.25) of A.II. whose deep

structure is given in (3.36)(ii) above.

The Tense-Signifiers insertion

rule (3.14) applies to S^ of (3.36)(ii) and h.y.y. is adjoined to the Tense-node.

Then the parenthetical clause embedding rule (2.12) applies

and S3 is embedded into Sg yielding the following structure: (3

'

'armumiyot Finally the relativization rule (2.63) applies to (3.41) and yields the correct sentence (3.25) of A.II: garmu carot.

hana'Sim. ~§ehayu •armumiyot,

Thus, our relativization rule need not undergo any changes

insofar as generating nonrestrlctive non-Verbal relative clauses in the Past tense is concerned.

105 Our relativization rule will also generate all the sentences in B.I. and B.II.

The ungrammatical sentences (3.25) and (3.26) in

B.I. may be handled very easily: in Hebrew must be introduced.

a rule for reducing relative clauses

In the conditions under which this rule

applies it should be indicated that the application of this rule is obligatory if the Predicate is an NP or an Adjective [of. (3.25) and (3.26) with (3.27) and (3.28) in groups B.I. and C],

This rule for

reducing relative clauses will be dealt with in the next section of this chapter. The conclusion to be drawn from the entire previous discussion then is that the only change that must be made in the relativization rule (2.63) is to enable it to pronominalize the embedded NP under one more condition--when the relative clause is nonrestrictive, non-Verbal, and in the Present tense.

This is the main argument to which I re-

ferred in Section 1 of the previous chapter, when I claimed that the relativization rule should operate on two distinct environments.

With-

out doing so it would be impossible to show the differences between restrictive and nonrestrlctive non-Verbal clauses whose Tense is Present insofar as the pronominalization or the deletion of the embedded NP is concerned. I shall now introduce the modified relativization rule.

This

rule (plus the rule for reducing relative clauses which will be stated later) will adequately generate all the relative clauses that have been

106 introduced thus f a r - - r e s t r i c t i v e and n o n r e s t r i c t i v e , with Verbal and non-Verbal Predicates : ( 3 . 4 2 ) SD:

[X 1

SÇ:

1

NP # [ 2

3 2

W

NP

4

5

RM+4

0

Z] g #

Y]g

6 7 8

i(a)[+Pro]Î6 loo 0 f

0

8

SC (a) - e i t h e r ( i ) i f 4 contains V, or ( i i ) i f 5 i s dominated by a coordinate NP-node, or ( i i i ) i f 4 . . . 6 i s dominated by S and does not contain Tense. SC (b) - elsewhere Conditions:

(i)

2=5

(ii)

S

NP?

The only d i f f e r e n c e between t h i s version of the r e l a t i v i z a t i o n r u l e and the previous one [ r u l e ( 2 . 6 3 ) ] i s the t h i r d condition under which s t r u c t u r a l change (a) a p p l i e s .

This condition means t h a t the

embedded NP i s pronominalized i f the r e l a t i v e clause i s n o n r e s t r i c t i v e (and not dominated by the node of t h e r e l a t i v i z e d NP, but r a t h e r adjoined t o i t ) , and i f i t s Tense i s Present.

Obviously, t h i s change

enables t h e r u l e t o generate the grammatical sentences of group A . I . instead of sentences such as * ( 3 . 4 0 ) above.

107 4.

Relative Clause Reduction Rule I should like to initiate this section with the following

information: duced.14

in Hebrew, only non-Verbal relative clauses may be re-

Therefore, the following grammatical English sentences are

ungrammatical in Hebrew: (3.43) *ha'i5 ra'ita ('oto) hu more. 'The man you saw (OM+him) he (=is) (a) teacher.1 (3.44) »ra'iti 'et ha'is ddbarta 'ito 'etmol. 'I saw 0M the man you talked with him yesterday.1 Consequently, our relative clause reduction rule should be limited to non-Verbal sentences only.

In the previous section I illus-

trated the way in which restrictive relative clauses are reduced (in group C).

I indicated there that: (3.45) (i)

Restrictive clauses whose Predicate is an NP or an Adjective must be reduced [cf. (3.31)(xii)].

(ii)

After the reduction, if the resulting attribute is an NP or an Adjective it must agree with the modified Noun Phrase insofar as the Determiner is concerned.

108 Let us examine now the situation in which appositive relative clauses are reduced: (3.46) (i)

david levi. ~§ehu ma^orer umafianex, kibel pras. 'David Levy, that he (is) (a) poet and (an) educator, received (an) award.1

(ii)

david levi, ma'Sorer umafianex, kibel pras. '.David Levy, (a) poet and (an) educator, received (an) award.'

(3.47) (i)

hayayin, ~feehu 'al ha^ulfian, hu miyavan. 'The wine, that he (=it) (is) on the table, he (=ls) from Greece.'

(ii)

»hayayin, 'al ha'feulfian, hu miyavan. 'The wine, on the table, he (=is) from Greece.'

(3.48) (i)

'ilana, "Sehi pihilit, hitfiatna kvar. 'liana, that she (is) smart, got married already.'

(ii)

*'ilana, pikftit, hitftatna kvar. 'Ilana, smart, got married already.'

These sentences lead to the following conclusions: (3.49) (i)

In appositive clauses, unlike restrictive clauses, the reduction is optional

[both

109 (i) and. (ii) of (3.46) are grammatical]. (11)

In appositive clauses, reduction is permissible only if the Predicate of the relative clause is an NP.

(iii) After the reduction, the pauses before and after the appositive clause remain, and the result is an apposition. (iv)

This apposition does not have to agree with the relativized

NP insofar as definiteness

is concerned [cf. (3.46)(ii) above].

Thus

we may have sentences such as miryam makeba, zameret kuSit, hofi'a kan ' etmol ('Miriam Makeba, (a) singer negro (=a Negro singer) appeared here yesterday') along with miryam makeba, hazameret hakuSit, hofi'a kan 'etmol ('Miriam Makeba, the singer the Negro (=the Negro singer) appeared here yesterday'). I may now state the relative clause reduction rule and the rule for the agreement of definiteness between a Noun and its modifier which was derived from a reduced restrictive relative clause:

110 ( 3 . 5 0 ) R e l a t i v e clause reduction X

NP RM Y

1 1 2

2

Aux

Predicate

Z =>

3

4

5

6

7

0

0

0

6

7

Conditions:

(i)

5 does not contain Tense;

(ii)

obligatory i f 6 dominates NP or Adj and NP

2 . .

.7;

( i i i ) i f S => 2 . . . . 7 , applies optiona l l y only i f 6 dominates NP. Condition ( i ) means t h a t a non-Verbal r e l a t i v e clause may be reduced only i f i t s Tense i s Present (the unmarked Tense).

Thus,

the following ungrammatical sentences a r e prevented from being generated: (3.51) ( i )

*hayalda hayta yafa halxa. "The g i r l was p r e t t y went.'

(ii)

*david, haya more, n i f t a r . 'David, was (a) t e a c h e r , passed away. 1

Condition ( i i ) means t h a t i n case the r e l a t i v e c l a u s e i s r e s t r i c t i v e and i t s P r e d i c a t e i s an NP or an A d j e c t i v e , the reduction i s obligatory [ c f .

( 3 . 2 5 ) and ( 3 . 2 6 ) of groups B . I . and C above].

Condition ( i i i ) enables n o n r e s t r i c t i v e r e l a t i v e clauses t o be reduced only i f t h e i r Predicate i s nominal [ c f .

( 3 . 4 6 ) - ( 3 . 4 8 ) above].

Ill And finally, we have the following rule: (3.52) Agreement of definiteness +N +Def 1

2

1

2

At this point, the modified relativi zation rule and the new rules that have been introduced in this section adequately generate all the sentences that have been introduced thus far.

5.

Possessive Attributes The use of the possessive has undergone several developments

through the history of the Hebrew language.

I shall review these

developments, and then illustrate how the current usage of the possessive in Modern Hebrew may be best explained by positing the Biblical usage as the underlying structure from which the modern usages are derived. I should emphasize that by so doing I am not attempting to apply the so-called "panchronic approach" which was discussed and rejected in the first chapter of this study.

I reached the conclusion

that the Biblical usage should be posited as the underlying structure for Modern Hebrew by a purely synchronic analysis.

That is, I reviewed

112 the syntactic reality found in Modern Hebrew--and Modern Hebrew a l o n e — and I endeavored to find the simplest and most plausible way for describing this reality.

The findings of my study have led me to posit

certain underlying structures on which certain transformational rules should apply.

At the end of this analysis it was found that those under-

lying structures, which were arrived at on independent grounds, correspond to the Biblical usage, and furthermore, the intermediate stages of derivation correspond to historical stages in the usage of the possessive in Hebrew. tive grammars:

Such a phenomenon is not unusual in writing genera-

quite often a succession of transformational rules,

which are written for a synchronic analysis, reflect historical developments . 1 5 In the Bible, 1 6 the following constructions are used for expressing possessive attributes: (a)

A cluster of Nouns in which the first Noun (nomen

regens) is in the construct state, and the second (nomen rectum) is in the independent form, e.g.: (3.53) ' elohey ha'Samayim velohey ha'arec (Genesis 24,3) 'The God of heaven and the God of the earth1 (3.54) za'akat sedom (Genesis 18,20) 'The cry of Sodom' [The independent form of 'elohey ('God of') is 'elohim ('God'); and of za'akat ('The cry of 1 ) is za'aka ('cry').]

113 If the nomen rectum is a Pronoun, the nomen regens is inflected in agreement with the Person, Gender, and Number of this Pronoun. In other words, the Pronoun is a suffix added to the first Noun, e.g.:

(3.55) ma'asehu ('his treatment'), in Judges 4,9, is derived from ma'ase ('treatment') plus the suffix -hu

('his').

The derivation of this type of construction (smixut) was discussed above, in Section 4 of Chapter II.

(b)

At times, this cluster of Nouns is "broken" by the

Preposition la- ('to') which is inserted in front of the second Noun: «

A

N I9- N.

In this case, both of the Nouns have the independent

form.

Thus: (3.56) hacofim laSa'ul (First Samuel 14,16) 'The watchmen to (=of) Saul' (3.57) ben layi'Say (First Samuel 16,18) '(A) son to (=of) Jesse'

If the second Noun is a Pronoun, the Preposition la- is inflected by a pronominal suffix.

(c)

There are cases in which the Biblical Relative Marker

'a^er precedes the la- mentioned in (b):

Thus:

114 (3.58) "Sir ha^irim 'a§er liSlomo (The Song of Songs 1,1) •The song of the songs that to Solomon (=the song of songs which is Solomon's)' Again, if the nomen rectum is a Pronoun, the Preposition 18is inflected. In the last books of the Bible, the Relative Marker "Se- begins to appear as a substitute for the common Biblical Relative Marker ' a'S er, thus : (3.59) mi tat o 'Seli'Slomo (The Song of Songs 3,7) 'The litter of Solomon' The following changes in the usage of the possessive took place in the Mishnaic period: (a)

The usage of the construct state (smixut) which is

widespread in the Bible, is very limited in the Mishna. (b)

The usage of the construction N'ls-^N disappeared in

the Mishnaic period. (c)

The construction N RM^ls-^N became the common usage

in expressing the possessive in the Mishnaic period. usage underwent certain changes.

However, this

At the beginning of the Mishnaic

period, the usage was identical to that in (3.59).

Later, the two

morphemes "Se- and Is- were considered one free morpheme %el ('of'). This morpheme was treated as a Preposition and thus it was inflected.

115 A simple chart will demonstrate the structural change: (3.60)

Late Biblical and Early Mishnalc period (i)

'§e+l8+&8lomo =4.

#Sel#

fSeliSlomol17

#S8lomo#

('of Solomon')

(ii) £e + I9+ ha + zavattim #§elazavattLm # 1 B (iii) Se + 19 + i

Late Mishnaic Period

#§el#

#haz3vafiim#

('of the sacrifices'

#Seli#

#Sel + i#

#^eli#

('of me')

In Modern Hebrew, the following situations exist: (a)

In spoken Hebrew the construct state is used mainly in

idiomatic clusters, e.g., beyt-sefer ('house of book' = 'school'), ba'al-bayit ('owner of house' = 'landlord').

On higher stylistic lev-

els, it is also used in nonidiomatlc constructions, e.g., delet-hamisrad ('the office's door'), mafialat- ha'is

('the man's sickness').

(b) The construction N^ia-^N is generally not in current use, but may be found occasionally in highly poetic writings. / \

^

rv

(o) The construction N RM 18- N

[in the way it was used in

the Mishna, c.f. (3.60) above] is the most common usage today.

Thus,

both (3.61)(l) and (ii) below are found in Modern Hebrew, but (ii) is much more widespread.

116 (i) beyt-david

(ii) habayit £el david

'David's house' •§iri

ha'Sir 'Seli

'my song' Before dealing with the r'ules which generate the possessive attributes, I should like to illustrate here the Hebrew sentences which correspond to those English sentences such as I have a book. In the Bible, such a sentence is also expressed by the construction

[cf. (3.56), (3.57) above].

In other words, both

'their mouth' and 'they have a mouth' are expressed in the same construction; thus: (3.62) pe lahem valo yadaberu, 'eynayim lahem valo yir'u. (Psalms 115,5) '(A) mouth to them and not they will speak, eyes to them and not they will see (= they have mouths, but they speak not; eyes have they, but they see not).' This structural similarity of two units which are different semantically was probably the reason for creating the new construction lAwfla-~N.

117 Sentences such as (3.62) are occasionally found in the Bible with the word ye'S ('existence'); e.g.:

(3.63) ye£ li ray. (Genesis 33,9) 'Exists to me much (=1 have enough).' Sentences such as (3.63) are more common in Modern Hebrew than (3.62).

Notice, however, that ye£ may occur only in the Present

Tense, because in the Past and Future tense both sentences such as (3.62) and (3.63) have the same structure, i.e., in both of them the structure is N Tense h.y.y. la- N.

For example, (3.65) below is the

Past/Future equivalent of (3.64)(i) and (ii): (3.64) (i)

sefer li. '(A) book to me (=1 have a book).1

(ii)

sefer ye& li. '(A) book exists to me (=1 have a book).'

(3.65) sefer haya/yihye li. '(A) book was/will be to me (=1 had/will have a book).' These examples suggest that ye& is inserted by a transformational rule, and that the underlying structure of (3.64) and (3.65) is as follows:

118 (3.66) (i)

For (3.64): S NP"

"VP

/

\

sef er (A) book

Predicate Prep-Phrase Prep

NP

Ato

A am

The underlying structure (3.66)(i) will generate sentences such as (3.64)(i)j sentences such as (3.64)(ii) will "be generated by applying the ye'S insertion transformational rule which I do not intend to state here. Basically then, the ye'S insertion is similar to the insertion of the copulative Pronouns [cf. (3.16) above] which are also inserted transformationally when the Tense of a non-Verbal sentence is unmarked.20

119 The underlying structure in (3.66)(ii) will adequately generate sentences such as (3.65).

The Tense-Signifiers insertion rule

[see (3.14) above], which applies to non-Verbal sentences in general, applies here also and inserts h.y.y. Notice, incidentally, that this type of sentences was one of the primary reasons for my claim that rule (3.20) above must contain a Prepositional Phrase. Let us now summarize the examples that have been presented and discussed thus far, add some more variations, and see how they are all derived: (3.67) (i)

beyt-fiolim 'House of sick people (=hospital)'

(ii)

bat hafialban •The milkman's daughter'

(Hi)

'afii 'my brother'

(3.68) (i)

kesef ray la'i§ haze. 'Money much to the man the this (=this man has a lot of money).'

(ii)

kesef ye'S la'i§. 'Money exists to the man (=the man has money).

(iii)

carot yihyu latalmid. 'Troubles will be to the student (=the student will have troubles).'

120 (3.69) (i)

habaftura 'Sel dan yafa. •The girl of dan (is) pretty.'

(ii)

habaftura 'Seye'S la dan (hi) yafa. 'The girl that exists to dan (she (=is)) pretty (=»the girl that Dan has is pretty).'

(iii) habaftura ^ehayta ladan hayta yafa. 'The girl that was to Dan was pretty (=the girl that Dan had was pretty).' (iv)

»habaftura 'Se'Sel dan (hi) yafa.

(v)

hamafiberet, "éehi 'Sel dan, 'avda. 'The notebook, that she (=it) of Dan, got lost.1

The derivation of sentences such as (3.67)(i)-(iii) was discussed above, in Section 4 of Chapter II, and therefore will not be discussed here. The underlying structures of sentences such as (3.68) (i)(iii) is basically the following:

121 (3.70)

S

Sentence (3.68)(i) is derived directly from (3.70), providing that the Tense-node is absent; Sentence (3.68)(ii) is derived from (3.70) and by the application of the ye§ insertion rule; and sentence (3.68)(iii) is derived from (3.70) and by the application of the TenseSignifiers insertion rule (3.14) (the Tense-node dominates Future). As for the derivation of the sentences in (3.69), I propose to derive them from relative clauses, in the following manner:

122

(5.71) Sentence (3.69)(i)

bafiura (a) girl

Predicate | Prep-Phrase Prep lato

yafa pretty

NP dan Dan

After the application of the relativization rule on (3.71), the following sentence is generated: (3.72) *habaKura "Seladan yafa. Notice that this sentence is exactly the same as the usage of possessives in the early Mishnalc period [cf. chart (3.60) above]. However, today the constituent 'Seladan has separated and become 'Sel dan. Notice incidentally, that the separation of this constituent must apply before the rule that converts la- + ha- into la- (cf. note 17).

This rule applies within word boundaries.

See, for example,

123 (ii) of (3.60):

'Selazsvaftim vs. 'Sel hazavaftlm. The structure ]|e +

la - + hazavaKim underlies both words, except that the second one is derived after the reorganization of the word boundaries takes place. Thus, we assume that a rule such as (3.73) below exists in the grammar of Hebrew: (3.73) SD:

|e+ la - + 1 2

MP 3

SC: (i) 1

2

(ii)l

2

§

3 3

#

SC (ii) if 3 = [+Pro] Structural change (i) converts ^e + la- + dan into #Sel# #dan#. Structural change (ii) converts 'Se- + la- + i_ into (instead of

#§el#

#i#).

One may claim that Sg in (3.71) might be better stated as follows:

NP

VP

bafiura

Predicate Prep-Phrase

Sel

dan

124 This analysis is based on the assumption that IS el is an independent lexical entry, and that it functions as do all the other Prepositions.

This analysis should be rejected on the following

grounds: (a)

It will generate ungrammatical sentences such as (iv)

(b)

It fails to show the inherent relations between the

in (3.69).

sentences in (3.68) and those in (3.69) above. So far, I have touched upon sentences (i) and *(iv) of (3.69). However, all the remaining sentences in (3.69) are also adequately generated by the analyses which have been applied throughout this section.

Sentence (ii) of (3.69) has the same underlying structure as

(3.69)(i) [see (3.71) above] except that sentence (3.69)(ii) is generated by both the application of the relativization rule and the ye& insertion rule. Sentence (v) of (3.69) has an appositive relative clause, and thus the anaphoric Fronoun hi is generated by the relativization rule (3.42). ing:

The underlying structure of this sentence is the follow-

125

Predicate

mafiberet

Prep-Phrase Prep dan

18-

The d e r i v a t i o n of ( 3 . 6 9 ) ( v ) was a r r i v e d at the i n the f o l l o w i n g manner: (3.76) ( i )

[

[ hamafrberetf [ [ hamafrberet 1 s/NP NP S 2 L NP KP

[

[ mafiberet ladan] ] ] VP S3 S3 VP S2

['avda]

] VP S i

The r e l a t i v i z a t i o n r u l e embeds S j i n t o Sg and y i e l d s :

126 (ii)

[ l

[ hamaftberetl [ hamafiberet ie la dan! S 1 l NP NP S 2 ['avda1J 1J VP S 1 The relativization rule embeds Sg into Si (hamafib er et in S^ is pronominal!zed because this is a nonrestrictive clause):

(iii)

[ fhamaftberetl [ 'Se hi £e la dan] Si NP S 2 S2 f'avda1 ] VP Si Later rules, which will organize the word boundaries, will yield the expected sentence

(iv)

hamafiberet, 'Sehi 'Sel dan, 'avda.

In summary, I have demonstrated that only by positing the Biblical structure given in (3.62) as the underlying structure can we adequately explain both the stylistic variations given in (3.68) and (3.69) and the close relation between sentences such as hasefer 1ieli ('my book') and sefer (ye'S) li ('I have a book.').

127 NOTES TO CHAPTER III I am referring to constructions such as profesor ftaver (Professor friend = Associate Professor), 'i§a 'almana ('woman widow*), ma's orer 'Oman ('poet artist = artistic poet'), hasar dayan ('the minister Dayan'), etc.

The second Noun in such constructions is

obviously derived from a relative clause with a Predicate Nominal. Cf. Rubinstein's hammiSpat. pp. 80-83. In Hebrew, the Adjective always follows the Noun which it modifies and agrees with it in definiteness, in Gender, and in Number.

For

example: (a) A nice boy - yeled nefimad (b) The nice boy - hayeled haneftmad (c) Nice boys _ yeladlm nefimadim (d)

The nice girls - hayaladot haneftmadot

I have dealt with quite simple non-Verbal sentences and I have shown that the occurrence of the Copula depends primarily upon the quality of definiteness of the Subject and the Predicate.

This is

indeed the main factor which dictates the presence or the absence of the Copula. There are, however, some other factors which influence the insertion of the Copula. dlc.

Some of these factors are purely proso-

For example, if the Subject or the Predicate is lengthy

(e.g., if they contain embedded sentences), the Copula is inserted

128 no matter what the quality of definiteness of the Subject of the Predicate.

(See also Chapter IV.)

Regretfully, there have not been any systematic studies which cover all the varieties of this type of sentence.

Recent

studies on the character of the Hebrew nominal sentence (e.g., Rubinstelnfe;hammiSpat) avoided the problem by claiming that the presence or absence of the Copula is purely stylistic. This topic is extremely broad and intricate, and its full treatment is beyond the scope of the present study.

The main goal

of any study on this topic should aim to define explicitly the cases in which the presence or the absence of the Copula results in different degrees of grammatlcalness or stylistic varieties. If the second alternative is chosen, there must be a transformational rule that inserts the Copula in cases when the Subject and the Predicate both have the same quality of definiteness. Notice, incidentally, that the Verb be (h.y.y.) does have Present tense forms.

However, these forms are not used as the copulative

elements in non-Verbal sentences whose Tense is Present (as was pointed out, Pronouns are used instead). See Chomsky, Aspects, p. 107 and Bach, "Have" and "be." p. 463. On page 473 Bach introduces another version of this rule, i.e., VP

-* (V (Prep-Phrase)(Manner))

129 Bach's motivation for introducing this new rule is based primarily on his claim that "there seem to be no reasonable results if we use our VP rule to choose Prep-Phrase but no V."

I do not adopt

this modification of the rule because, as I shall later illustrate, there are Hebrew sentences in which a Prep-Phrase (not derived from Place or Time) is the Predicate of a non-Verbal sentence.

I there-

fore still consider rule (3.13) the basis for my discussion. 8.

h.y.y. are the three consonants of the Verb-root, which is a discontinuous morpheme.

9.

Cf. Bach's observation:

"when we look at some other languages, we

see that the use of a special form like have is a peculiarity of English" (op.cit. p. 468). 10.

The existence of the item Tense implies that the sentence is either in the Past or in the Future.

In other words, in this analysis, I

am adopting Bach's following proposal: Suppose the tense in which no copula appears is the present tense (as is commonly the case), and suppose further that the present tense for such languages is the unmarked tensej that is, Tense is optional and the result of choosing no tense is the present. (op.cit. pp. 478-479) 11.

Op.cit. p. 473.

12.

Word order in Hebrew is much looser than in many other languages, e.g., English.

In some cases, a change in the word order results

in a greater degree of emphasis [this is the case with sentence

130 (3.26) of A.I., for example], and in some cases it has no influence on the meaning or on the degree of emphasis.

I shall deal with the

question of word order in Hebrew in greater detail in Chapter VI. 13.

As to the word order in the relative clauses, see note 12 above.

14.

This statement needs further clarification.

In fact, there are

instances in which the Relative Marker may be deleted from a relative clause whose Predicate is Verbal.

Consider, for example, the

following three paraphrases: (a) habafiura "Sedibarnu 'aleyha 'etmol hayta na'arat gogo. "The girl that we talked on (=about) her yesterday was (a) go-go girl.' (b) habafiura 'Se'aleyha dibarnu 'etmol hayta na'arat gogo. 'The girl that on (=about) her we talked was (a) go-go girl.' (c) habafiura 'aleyha dibarnu 'etmol hayta na'arat gogo. 'The girl on (=about) her we talked yesterday was (a) go-go girl. ' The Relative Marker may be optionally deleted only if the NP embedded in the relative clause is a direct or an indirect Object, and if this Object is moved to the beginning of the relative clause. It is obvious that the Relative Marker deletion depends on a certain movement transformation and follows it.

I believe, how-

ever, that this movement transformation should be handled on the

stylistic level rather than on the syntactic level.

If this is

the case, then the Relative Marker reduction, which should follow the movement transformation, also belongs to the stylistic level. However, this matter is part of the broad and intricate question of general word order in Hebrew.

I deal with this ques-

tion in Chapter VI of this study. 15.

This point was also illustrated in Chomsky and Halle's The Sound Pattern.

While dealing with a synchronic description of English

phonology, they were led to "postulate underlying forms which are quite abstract.

Furthermore, these abstract underlying represen-

tations are, in general, very close to conventional orthography" (p. 48).

The explanation they give obviously applies to the case

with the Hebrew possessives as well:

"It is a widely confirmed

empirical fact that underlying representations are fairly resistant to historical change" (p. 49). 16.

Most of the Biblical examples are gleaned from Gesenius1 Hebrew Grammar.

In reviewing the developments which took place in the

Mishnaic period, I was aided by Segal's diqduq.

The nonliteral

translation for Biblical quotations is taken from the English translation of the Holy Scriptures, The Jewish Publications Society of America, Philadelphia, 1955. 17.

la- + "SoHebrew.

l i a c c o r d i n g to a general phonological rule in

132 18.

la + ha-

la- according to a general phonological rule in Hebrew.

19.

Is + 'ani

20.

Further support for this claim may "be found in the following exam-

li .

See note 21 in Chapter II.

ples : (a)

milfiama bamizrafi hatixon. '(A) war (is) in the east the middle (= in the Middle East).1

(b)

milftama ye§ bamizrafi hatixon. 1

(c)

(A) war exists in the east the middle.'

sefer ye§ ' al ha^uljian. '(A) book exists on the table.1

(d)

hasefer (hu) 'al IraSulfian. 'The book (he (=is) ) on the table.1

(e)

hasefer ye'Sno ' al ha^ulftan. 'The book exists on the table.'

These sentences show that ye& is inserted not only in those Hebrew sentences which are equivalent to English sentences which contain the verb have, but also in non-Verbal sentences whose Predicate is Place.

Notice that if the Subject is [+Definite], the word ye'S

is inflected [sentence (e)]. Further evidence that ye& is a copulative element similar to the copulative Pronouns may be demonstrated by the fact that the negative of ye'S is ' eyn, and that ' eyn is also a negation word for any sentence in the Present tense.

133 I do not delve Into all the details of sentences

containing

ye§, because everything concerning ye& which is directly relevant to the question of relative clauses has been mentioned already. However, one of the most interesting questions is the question of word order in sentences containing ye'S.

I shall only hint

at this matter. Notice that, generally speaking, Hebrew sentences begin with the Subject

(providing, of course, that no element in the sentences

bears emphasis). ye'S.

This is not the case with sentences containing

Presumably, because of the influence of non-Semitic

languages

(mainly English), the Subject of ye§-sentences is often understood as a direct Object and therefore moved to the end of the sentence. Therefore, a sentence such as ye& li sefer

('exists to me

(a) book (=1 have a b o o k ) 1 ) is the nonemphatic usage while the usage sefer ye& li usually implies that sefer bears emphasis (that is, I have a book and not something else).

Needless to say, the word

order in the second sentence is the word order of the deep structure. The influence of non-Semitic languages

(in which a book is

a direct Object in sentences such as I have a book) is so strong in the formation of ye&-sentences

that one of the most widespread

grammatical mistakes in Modern Hebrew is the following:

134 *ye§ 11 'et hasefer •Exists to me OM the book.1 That is, the Object Marker is inserted in front of the Subject in a sentence which obviously does not contain any Verb or any direct Object.

C H A P T E R

IV

INDEPENDENT RELATIVE CLAUSES

I.

Introductory Remarks The following phenomenon is quite familiar in many languages:

at times, Nouns are deleted from constructions of the type Noun + Modifier, and consequently, the Modifier stands for both itself and the deleted Noun.

Thus we have pairs such as in (4.1) with no basic

semantic differences between them: (4.l)

(i)

' f i a x a m fto^ev pa'amayim. '(a) man wise thinks twice.' fiaxam holey pa'amayim.

(ii)

'ani sone 'anasim kamcanim. 'I hate men (=people) stingy.' 'ani sone kamcanim.

(iii)

ha'il hagavoah yaca rison. 'The man the tall left first.' hagavoah yaca rigon.

The modifiers in (4.l) are obviously derived from reduced relative clauses.

136 In Hebrew, Nouns may be deleted in the same manner, even if they are followed by a true (nonreduced) relative clause; for example: (4.2)

(i)

'ani mafizir 'et ' aser lakaftti. 'I return OM that I took.'

(ii)

'ata hu seracafita 'et harakdanit. 'You he (=are) that murdered OM the dancer.'

In sentences (i) and (ii) of (4.2) then, we have independent relative clauses; that is, they are not preceded by a relativized NP (at least insofar as the surface structure is concerned)

and they stand

for both the relativized NP's and the relative clauses that modify them. However, in this section, I shall discuss not only sentences such as those in (4.2) but also sentences in which the -relativized element is mi ('who'), ma ('what 1 ), and kol ('all, every'), e.g.: (4.3)

(i)

'ani yodea' ma £ehu roce. 'I know what that he wants.'

(ii)

mi sekatav 'et hasefer haze hu mufra'. 'Who that wrote OM the book the this (=this book) he (=is ) disturbed.'

(4.4)

hu natan lah kol ma sehi racta. 'He gave to her all what that she wanted.'

137 (4.5)

kol 'a£er 'ani mavakeS hu hamlaca ktana. 'All that I ask he (=is) recommendation small (= a small recommendation).' The presentation of the discussion will benefit if I pursue

the following order:

I shall first deal with relative clauses whose

head Nouns are ma and mi [cf. (4.3) above]; I shall next deal with sentences in which kol ma and kol mi are the relativized items [cf. (4.4) above]; finally I shall deal with sentences containing relative clauses with no head Nouns [cf. (4.2) above] and with relative clauses whose head is kol [cf. (4.5) above].

2.

Sentences with mi/ma, kol mi/kol ma Before dealing with mi and ma as the head of relative clauses,

I should like to discuss the general character of these two words and the character of Hebrew Question sentences. The two words, ma and mi, are usually translated as 'what.' and 'who' respectively, but this translation is quite misleading if one is dealing with a transformational analysis of Hebrew. The origin of 'what' and 'who' (whether they occur in interrogative sentences or in relative clauses) is quite different from the origin of ma and mi.

'What' and 'who' consist of two elements:

wh +

Pronoun, while ma and mi are pure Pronouns designating unspecified Nouns.

In other words, mi/ma may be characterized as follows:

138 (4.7) ml

ma

+N

+N

+Pro

+Pro

+Human

-Human

-Definite

-Definite

As a result of the fact that mi and ma are pure Pronouns, transformations which apply to Nouns in general also apply to mi/ma. This is the case at least insofar as the relativization transformation is concerned. mi/ma.

Therefore we have relative clauses whose head Nouns are

For example:

(4.6)

(i)

mi £eba lamasiba hit'axzev. 'Anyone that came to the party was disappointed.'

(ii)

ten li ma ^ekanita. 'Give to me the thing that you bought.'

In Question sentences, ma and mi are best translated as 'what' and 'who', but it must be borne in mind that in questions as well as in declarative sentences mi and ma are still Pronouns.

In interrogative

sentences mi and ma are understood as Question elements only because the Question transformation has moved them to the beginning of the sentence and has changed the intonation of the sentence.

That is, Hebrew

interrogative sentences containing ma and mi are derived in a manner

139 quite different from English sentences containing 'what1 or 'who'.

In

English questions, 'what' and 'who' are the result of two operations: (a) the questioned Pronoun is moved to the beginning of the sentence; (b) it is adjoined to the element VJh- yielding what or who(m). 1

In

Hebrew, on the other hand, only the first operation takes place because mi and ma exist as such in the underlying structure of the interrogative sentence. Thus, the underlying structure of a sentence such as (4.8)(i) would be that shown in (4.8)(ii): (4.8)

(i)

ma rafiel kanta basupermarket? 'What Rachel bought (=did Rachel buy) in the supermarket?' S

(ii) Quest

NP

VP

rafie 1

Prep ba-

NP hasupermarket

140 The Question transformation moves ma to the beginning of the sentence and in the phonological component this sentence will be assigned the intonation of an informative question. Incidentally, ma and mi once meant 'something' and 'someone' in any type of sentence, but today they are used in this sense primarily as heads of relative clauses [cf. (4.6) above].

However, a new form

has been created for designating 'something' and 'someone' in simple sentences.

The reason for this is quite obvious:

if mi and ma were

used as 'someone' and 'something' in simple sentences, the only difference that would exist between 'who came?' and 'someone came' would be intonation:

(4.9)

(i)

mi ba?

(ii)

mi ba•

Thus, the following new forms were created:

mi ¡Sehu (or

misehu) for 'someone' and ma %ehu (or ma^ehu) for 'something.'

However,

these new forms still do not change the situation described above.

This

is because mi sehu (='whoever he is') and ma sehu (='whatever it is') themselves are derived from mi and ma followed by a relative clause:

141 (4.10)

NP NP

S

I

\

hu

Predicate

I

NP

I

ml/ma To sum up:

today mi is used in simple sentences mainly if

they undergo the Question transformation.

In declarative sentences, ma

and mi are usually followed by a relative clause. 2

In any case, mi and

ma must be considered pure Pronouns. Let us look now in more detail at some relative clauses whose head Nouns are mi or ma. Group A: (4.1l)

(i)

mi ^elamad kibel ciyun toy. 'Anyone that studied received grade

good (= a

good grade.' (ii)

ma ^ehictaber bsme^ex sanim parac lafeta*. "The thing that accumulated during years exploded suddenly.'

142 (4.12)

(i)

ml Seberaxti ('oto) 'ax^av katav maftaze moderni. 'The one that I greeted (OM + him) now wrote (a) play modern•'

(ii) ma ^e'aslta yaharos 'et hakaryera £elxa. 'The thing that you did will destroy OM the career of you.' (4.13)

(i) mi Sedibarti 'ito katav mafiaze moderni. 'The one that I spoke with him wrote (a) play modern.' (ii) ma Me'ata mazalzel bo hayom hir'i^ 'et ha'olam lifney £natayim. 'The thing that you disrespect in him (=it) today excited OM the world before two years (=two years ago).'

Group B: (4.14)

(i) fiaxam hu mi ISelomed min hanisayon. 'Wise he (=is) the one that learns from the experience.' (ii) ze hu ma seharas 'et hakol. 'This he (=is) the thing that destroyed OM everything.'

143 (4.15)

(i)

yaron h u ml £epagasti ('oto) bs'eropa. 'Yaron he (=is) the one that I met (OM + him) in Europe.'

(ii)

ze h u ma £emacati po. 'This he (=is) the thing that I found here.'

(4.16)

(i)

yadid h u mi se'ata fio^ev 'alav kol hazman. '(A) friend he (=is) the one that you think on h i m all the time.'

(ii)

ze h u ma %e'ani macape lo tamid. "This is the thing that I hope for h i m

(=it)

always.'

Group C: (4.17)

(i)

h u her'a li 'et mi sekatav 'et hamafiaze. 'He showed to me OM the one that wrote OM the play.'

(ii)

h u her'a li ('et) ma £enafal min ha'ec. 'He showed to me (OM) the thing that fell from the tree.'

(4.18)

(1)

h u her'a li 'et mi ^ehe'elavta. 'He showed to me OM the one that you insulted.'

(ii)

h u her'a li ('et) ma £ekatavta. 'He showed to me (OM) the thing that you wrote.'

144 (4.19)

(i)

hu her'a li 'et mi seravta 'ito. 'He showed to me 0M the one that you quarreled with him.'

(ii)

hu her'a li ('et) ma sesifiakta bo. 'He showed to me (0M) the thing that you played with him (=it). '

Group D: (4.20)

(i)

hu siper li 'al mi

seganav 'et hakesef.

'He told to me on (=about) the one that stole the money.' (ii)

hu siper li 'al ma Sepocec 'et haya^iva. 'He told to me on the thing that exploded 0M the meeting.'

(4.21)

(i)

hu siper li 'al mi sepga^tem bamasiba. 'He told me on the one that you met in the party.'

(ii)

hu siper li 'al ma se'asitem bamasiba. 'He told me on the thing that you did in the party.'

(4.22)

(i)

'ani madaber 'al mi seca'akta 'alav. 'I speak on the one that you yelled on him.'

145 (ii)

hu siper li 'al ma Se'ata 'osek bo. 'He told to me on the thing that you deal in him (=it).'

The independent relative clauses in Group A function as the Subject of the sentence; in Group B they function as the Predicate; in Group C they function as the direct Object; and in Group D they act as the indirect Object.

Furthermore, in each group there are three pairs

of sentences—in sentence (i) mi is the relativized morpheme in each pair, and in (ii) ma is the relativized morpheme.

The first pair in

each group has mi/ma as the Subject within the relative clause; the second pair has mi/ma as the direct Object within the relative clause; and the third pair has mi/ma as the indirect Object within the relative clause. Let us now look at each group of sentences separately.

Group A Let us assume the following underlying structure for sentences such as (4.11) (i ):

146 (4.23)

klbel ciyun toy received grade good

mi.

studied

W h e n the relativization rule (3.42) applies to (4.23), the correct sentence (4.1l)(i) is generated.

Sentence

(4.1l)(ii) is also

adequately generated in a similar manner. A s for the sentences in (4.12), notice that in sentence

(i),

the embedded mi may be either pronominalized or deleted, exactly as was noted above in Chapter II.

However, in sentences such as (ii), the

embedded ma is deleted (there is no 'oto in this sentence).

Further-

more, in sentences such as (i), it is more common to delete than to pronominalize. The sentences in (4.13) are adequately generated by the relativization rule, i.e. the embedded mi/ma are pronominalized.

147 Group B All the sentences in this group are ultimately non-Verbal because their Predicates are mi/ma, and these morphemes are obviously dominated by an NP-node. Notice that all the sentences in this group contain the copulative Pronoun.

This phenomenon is not only true for Predicative

independent clauses whose head Noun is mi/ma; rather, it is a general phenomenon which may be stated as follows: (4.24)

If a relative clause is dominated by the Subjectnode or by the Predicate node of a non-Verbal sentence, this sentence must contain a copulative Pronoun.

It should be emphasized that this statement applies to any type of relative clauses, independent and nonindependent as well.

Fol-

lowing are some examples: (4.25)

hasapar ^eracafi 'et 'i£to hu mstoraf. 'The barber that murdered OM his wife he (=is) insane.'

(4.26)

'ilana hi baftura ^elomedet harbe. 'liana she (=is) (a) girl that studies a lot.'

(4.27)

david hu seracafi 'et hafiasfanit. 'David he (=is) that killed OM the stripper.'

148 I shall therefore introduce the following modification in the copulative Pronoun insertion rule [cf. (3.16) above]: (4.28)

[X 1

NP d

1

2

Y o

[

{NP, 4

3+Pro

Conditions:

4

Adj} b

]predicate]s

5

(i)

3 does not contain Tense.

(ii)

either 2 or 4 contains S or 2 and 4 are [o SVO + SO + SO) Even i f word permutations take place in the f i r s t of the conjoined sentences, the d i r e c t i o n of Gapping i s s t i l l forward. ample: (6.23)

s e f e r kana david. vsdan hafiberet. ' ( A ) book bought David, and Dan ( a ) notebook. 1

For ex-

203 (6.24)

kana david sefer, vadan mafiberet. 'Bought David (a) book, and Dan (a) notebook. 1

Ross suggests the following two hypotheses concerning Gapping: 6

(6.25)

(i)

The order in which Gapping operates depends on the order of elements at the time that the rule applies: if the identical elements are on left branches, Gapping operates forward; if they are on right branches, it operates backward,

(ii)

Gapping is a n "anywhere rule."

The fact that in Hebrew there exists only forward Gapping suggests that the Verb cannot b e on right branches.

On the other hand,

hypothesis (ii) in (6.25) appears to me to b e invalid considering sentences such as (6.23)-(6.24).

This hypothesis was made b y Ross on

the basis of Russian sentences in which changes in the order of elements cause a change in the direction of Gapping.

Sentences

(6.23)-

(6.24) illustrate that in Hebrew, the direction of Gapping depends on the order of the elements in the deep structure [providing, of course, that hypothesis (6.25)(l) is valid]. In conclusion, if w e assume that both the Verb and the Object develop from the VP-node, the following orders are possible:

(6.26)

(i)

S V 0

(ii)

S~0~V

J

NP VP

204 (iii) V 0 S (iv)

0 V~S

(orders such as

VP NP

and o""s V are automatically eliminated

because they imply that the Verb and the Object are not dominated by the VP-node.) The orders S'tTv and V~0~S are either ungrammatical or unacceptable. The order 0 V S was proven to be a noninherent order by sentences in which both the Subject and the Object are indefinite. The only remaining possible order then is m

,

which is

thus the inherent word order of Modern Hebrew.

3.

Word Order and Prosody There is evidence that at times, changes in word order depend

on prosodic factors.

The permutations which take place under these

circumstances are sometimes a tendency only and sometimes they take place obligatorily. The following examples will illustrate this claim: (6.27)

(i)

hu natan ladan 'et hamafiberet TSedavid hevi lo 'etmol. 'He gave to Dan OM the notebook that David brought to him yesterday.'

(ii)

? hu natan 'et hamafiberet ^edavid hevi lo 'etmol ladan.

205 (6.28)

(i)

hu natan li 'et hasefer. 'He gave to me CM the book.'

(ii)

* hu natan 'et hasefer li.

Sentences (!) and (ii) of (6.27) contain the same elements in different orders:

in (i) the order is Subject Verb indirect Object di-

rect Object, while in (ii) the direct Object precedes the indirect Object.

Usually the relative order of the direct and indirect Objects

is rather loose; thus the following two sentences do not differ from each other in any way (including emphasis): (6.29)

(i)

hu natan 'et hasefer ladan. 'He gave 0M the book to Dan.'

(ii)

hu natan Isdan 'et hasefer.

However, in sentences such as (6.27) above the general tendency is to have the order given in (i) and not in (ii). The reason for this is simply the length of the elements.

The direct Object in

(6.27) is lengthier than the indirect Object, and therefore there is a tendency to utter the short element first and the long element last. It would be too harsh to claim that (6.27)(ii) is completely ungrammatical, but such a sentence is absolutely unacceptable.

Thus the sen-

tences in (6.27) illustrate a general stylistic tendency in Modern Hebrew.

206 The sentences in (6.28) illustrate basically the same phenomenon as in (6.27) except that in this type of sentence the above mentioned tendency (i.e., to delay the lengthy elements) becomes a rule. The sentences in (6,28) demonstrate that if a sentence contains two or more Objects (direct as well as indirect), and if one of these Objects is a Pronoun, this Pronoun will always precede the other Object(s).

This phenomenon is similar to the one found in English in

sentences such as the following: (6.29)

(i)

He looked up the word.

(ii)

He looked the word up.

(iii)

He looked it up.

(iv)

* He looked up it.

Consequently we have the following situations existing in Hebrew: (6.30)

(i)

dan natan sefer ladavid. 'Dan gave (a) book to David.'

(ii)

dan natan ladavid sefer.

(iii)

dan natan 'oto lsdavid. 'Dan gave OM + him (=it) to David.

(iv)

* dan natan ladavid 'oto.

(v)

dan natan lo 'et hasefer. 'Dan gave to him OM the book.1

207 (vi)

* dan natan 'et hasefer lo.

(vii) dan natan lo 'oto. •Dan gave to him 0M + it. 1 (viii) * dan natan 'oto lo. Sentences (vii)-(viii) in (6.30) demonstrate that if the two Objects are pronominalized, the shortest Object comes first.7 All these examples emphasize again the fact that word order sometimes has bearing only upon prosodic factors rather than upon syntactic ones. This fact supports Chomsky's claim that word reorderlngs belong to the level of performance rather than to the grammatical level. However, as it will be later demonstrated, this is not always the case. Before departing to the next section I should like to point out the following fact: Sentences such as (6„28)(ii), (6.30)(vi) and (viii) maybe grammatical and acceptable only if the indirect Object bears the heaviest stress in the sentence.

In this case, sentences such as (6.30)

(viii) imply that Dan gave it to him (and not to anybody else).

4.

Word Order and Stress It has been noticed above (in Section 2) that if the inherent

word order S V^O is changed there must be a change in the stress pattern in the following manner:

208 (6.31) (i)

(ii) (iii)

2 3 1 S T O 13 2

oTs

13 2 0~S~V 13 2

(iv)

V~S~0

[cf. (6.21) above.]

Let us look now at some permutations in non-Verbal sentences: (6.32)

2 (i)

3

1

david haya melex. 'David was (a) king.'

(ii) (iii)

3 2 1 haya melex. 1 david 3 2 melex haya david.

(iv)

2 1 3 david melex haya.

These sentences illustrate that when permutations occur in non-Verbal sentences, each constituent bears the original degree of stress.

This leads to the conclusion that insofar as non-Verbal sen-

tences are concerned the permutations must follow the phonological rules which assign the original stress pattern. Notice, however, that Verbal sentences do not cause such difficulty, because in this type of sentences, the stress pattern is always 1-3-2 in sentences in which stylistic inversions have taken

209

place.

Thus, the s t y l i s t i c inversions of sentences with Verbal Predi-

cates may be handled, f o r example, by introducing the feature [+Emphasis].

The existence of t h i s f e a t u r e in the f i r s t constituent of the

sentence w i l l enable the phonological rules t o assign the correct stress pattern t o the sentence. So f a r we have come t o the f o l l o w i n g conclusions:

( a ) even

though word order in Hebrew i s quite loose, there i s an inherent order which i s £TV\); ( b ) at times permutations are obligatory and at times they are optional; furthermore, the reason f o r the permutations may depend on either prosodic or emphatic f a c t o r s ;

( c ) the permutations

that have been discussed so f a r seem t o support Chomsky's claim that they belong t o the l e v e l of performance rather than t o the grammatical level;

( d ) at t h i s point i t i s not at a l l clear i f the rules of the

s t y l i s t i c l e v e l should operate immediately a f t e r the application of the transformational rules or a f t e r the application of the phonological rules.

5.

Word Order and Fronominalization There i s evidence that claim ( c ) above i s not always v a l i d

i n Hebrew.

In other words, certain s t y l i s t i c permutations appear t o

have d i r e c t bearing on the theory of grammar. Such i s the case in sentences in which pronominalization had applied and then s t y l i s t i c permutations took place.

210 In Modern Hebrew, as in English, pronominalization may always apply from left to right; it may also apply from right to left (roughly speaking) only if the NP on the left is dominated by every S-node that dominates the NP on the right, and by some S-node that does not dominate the right NP.

Thus NP" can be pronominalized in constructions

(i)(ii) and (iii) of (6.33) below, while NP' can be pronominalized only in (iii): (6.33)

(i)

S S

S

(ii)

S s

ray

(iii)

s s

NPi" NP. 1 l

Thus, pronominalization can take place in sentences such as the following:

211 (6. 34)

danj maxar ' et hama'll "Sehu^ tafar. 'Darii sold OM the coat that he^ sewed.'

(6.35)

hasefer "Sehu^ katav pirsem 'et dan^. 'The "book that he^ wrote made famous OM Dan^.'

In sentence (6.34) pronominalization applied from left to right, and in sentence (6.35) from right to left (the left NP is in a subordinate clause).

Notice, however, that hu and dan may also have

different referential indices (i.e., hu is not a result of pronominalization).

Therefore, sentences such as (6.34) and (6.35) are ambiguous. However, sentences such as (6.36)-(6.37) below are not ambig-

uous : (6.36)

hu^ maca 'etmol 'et hasefer 'Sedanj mafizik 'axSav. 'He-j_ found yesterday OM the book that Danj holds now. '

(6.37)

'ani msdaber 'ito- 'axSav 'al hasefer "SedanJ katav. J

'I speak with him. now on (=about) the book that Dan. i J wrote.' In these two sentences, the pronominalization transformation could not have applied since hu and 'ito are not in a subordinate clause.

Thus, the only interpretation of these sentences is that hu

and ('it)o are not anaphoric Pronouns of dan. Let us assume that for stylistic purposes we move the Objects of (6.36)-(6.37) to initial positions. then result:

The following sentences would

212 (6.38)

'et hasefer 'Sedan mafizik ' ax§av hu maca 'etmol,

(6.39)

'al hasefer "Sedan katav 'ani ma dab er 'ito 'axSav.

The nonambiguous sentences (6.36)-(6.37) are now ambiguous. In sentences (6.38)-(6.39), dan is now on the left and hu and ('it)o may be interpreted as anaphoric Pronouns of Dan. The permutations which converted sentences (6. 36"}-(6. 37 ) into (6.38)-(6.39) and the semantic results of these permutations force us to draw one of the following conclusions: (6.40)

(i)

The reordering transformations should precede the pronominalization transformation.

If this is the

case, the so-called "stylistic permutations" are indeed grammatical transformations, (il)

The convention concerning the direction of pronominalization is valid not only for the arrangement of the constituents in the deep structure, but also for their reordered shape in the surface structure.

Whichever conclusion we accept, we must agree that the socalled "stylistic inversions" can no longer be ignored on the basis that they "have no apparent bearing, for the moment, on the theory of grammar."8

213 6.

Word Order inside Relative Clauses The word order within Hebrew relative clauses is even more

loose than it is in main clauses.

For example, the order v ' V S which

is unacceptable (if not ungrammatical) in main clauses is quite widespread within relative clauses.

Hence, the unacceptable sentence (i)

in (6.41) below is perfectly acceptable when embedded as a relative clause in £L sentence such ELS (6.41)(ii); (6.41)

(i)

* he'eric ('et ha-) more dan. (V~cTs) 'Admired (OM the) teacher Dan.'

(ii)

hamore 'Sehe'ric 'oto dan, met. 'The teacher that admired OM + him Dan, died.•

Furthermore, I believe that sentence (6.41)(ii) is more acceptable than sentence (6.41)(iii) below in whose relative clause the prevailing order is V^S^O: (6.41)

(iii)

? hamore 'Sehe'eric dan 'oto, met.

In other words, insofar as word order is concerned, the degrees of grammaticalness vary in different types of sentences.

I be-

lieve that the topic of word order is the most unsettled one in Modern Hebrew.

It is not only difficult to draw the line between grammatical

orders and acceptable ones, but it is also difficult to distinguish between what is more grammatical or more acceptable.

The various

214 reactions of native speakers to questions concerned with word order are so different from each other that one is helpless to find any consistent regularity. Every statement that has been made throughout this chapter is based on my own idiolect, and I must confess that even some of these statements were made with many reservations.

This is the reason why I

am avoiding stating any rules which are related to permutations, as each such rule may be proven false by any native speaker of Modern Hebrew.

I do believe, however, that the inherent word order of Modern

Hebrew is indeed S^AMD, but insofar as other orders are concerned I am able to discuss vague tendencies only. Word order throughout the history of Hebrew was rather loose and this situation is still reflected in Modern Hebrew.

In our era,

however, two new phenomena are playing a role in the development of the language:

(a) there is a tendency towards systematization of pat-

terns in the language; (b) there is a strong influence of non-Semitic languages on the structure of Hebrew. The first phenomenon is obviously leading to a simpler grammar of Hebrew.

However, changing a grammar of a language is a long and

difficult process, and word order will probably be one of the last patterns to submit to systematization. The influence of non-Semitic languages is felt in several areas of Modern Hebrew grammar and also in the area of word order.

215 Obviously this influence is towards a relatively fixed word order, as it is in the influential non-Semitic languages (predominantly English). However, there is a type of tension in Modern Hebrew which has been created by two powers:

on the one hand there is the power

which pulls towards fixed patterns of word order; on the other hand there exists another power (mainly of literary writers) which keeps gleaning structures from previous periods of the Hebrew language in which word order was looser than it is today. Obviously, at a certain time in the future a vector which will be a compromise of the two powers, will be created.

Such a vec-

tor has already been created in other areas of the Hebrew grammar, but insofar as word order is concerned, the situation remains almost chaotic. The situation would have been a bit less confusing if all the questions of word order in Modern Hebrew could have been considered stylistic inversions.

We then could claim that they are not a part of

the grammar of Hebrew and should be handled by stylistic studies.

How-

ever, I have demonstrated in Section 5 above that sometimes the problem of permutations does have a direct bearing on the theory of grammar (e.g., pronominalization).9 The discussion in this chapter was not intended to solve the questions of word order, but rather to point out the fact that this question is too complicated to be solved by simply claiming that it does not belong to the theory of grammar.

216 There is one question of word order which has a direct relation to the question of relative clauses.

If the identical embedded

NP is a direct or indirect Object of the relative clause, and this Object is permuted to the beginning of the relative clause, the Relative Marker may be optionally deleted, e.g.: (6.42)

(i)

ha'iS 'Sedibarnu 'alav hu 'anarxist. 'The man that we spoke on (=about) him he (=is) (an) anarchist.1

(ii)

ha'iS "Se*alav dibarnu hu 'anarxist.

(iii) ha'i§ 'alav dibarnu hu 'anarxist. Sentences (i)-(iii) in (6.42) are paraphrases of each other. It'is obvious, however, that the optional deletion of the Relative Marker in sentences such as (iii) cannot be separated from the question of word permutations which took place in this sentence. In this case, however, I believe that both the word permutations and the Relative Marker deletion are purely stylistic variations, and I do not intend to deal with them here.

In other words, my relativi-

zation rule will generate Sentences such as (i)(il) in (6.42) but not (iii).

The treatment of sentences such as (6.42)(iii) should be post-

poned until the general question of word order in Hebrew receives thorough study and perhaps some workable solutions are arrived at.

217 NOTES TO CHAPTER VI 1.

Aspects, pp. 123-127.

2.

Chomsky refers here to "syntactically related structures with a single network of grammatical relations . . . , each member is directly related to the underlying abstract representation, and there is no internal organization—that is, no order of derivation-within the set of structures."

3.

S=Subject, V=Verb, 0=0bject.

4.

The orders S^O^V and V^O^S are not found in my idiolect. however, be found in poetic writings.

They may,

If they are found in spoken

discourse, they are usually the result of a forethought.

At any

rate, these orderings are at least unacceptable if not ungrammatical. 5.

Op. cit., pp. 126-127.

6.

Ross, Gapping.

7.

In this case the indirect Object only has one syllable while the direct Object has two.

In opposite cases, the direct Object will

come first; for example: (a)

hu zarak 'oto 'aleynu. 'He threw OM+him (=it) on us.'

(b)

* hu zarak 'aleynu 'oto.

Because ' aleynu has three syllables and 'oto has only two. 8.

Chomsky, Aspects. p. 127.

218 9.

It may still b e the case that even permutations which have connection with grammatical transformations m a y b e separated from the syntactic component and dealt with on the so-called "stylistic level."

The problem is, of course, the following:

no one has

ever dealt with the concrete character of the stylistic component and its rules, and thus, no one has any idea whether problems such as pronominalization may be handled b y the stylistic apparatus.

C H A P T E R

VII

RESIDUAL PROBLEMS

1.

Deletion of a Prepositional Phrase In Chapters II and. IV I emphasized that if the embedded NP

functions as a part of a Prepositional Phrase it must be pronominalized rather than deleted. However, there are a few cases in which such an NP is deleted. It would, in fact, be more true to say that in these cases the whole Prepositional Phrase is deleted. This phenomenon usually occurs optionally when the relativized NP is a part of a Prepositional Phrase designating Time or Place or when the VP's of the main clause and the relative clause are identical.1 Some examples will demonstrate this phenomenon: (7.1)

hu nosea' la'ir Senoladti (bah). 'He travels to the town that I was born (in her (=it)).'

(7.2) hu gar bakfar £e'ani gar (bo). 'He lives in the village that I live (in him (=lt)).' (7.3) hu holex lamakom MehakoL holxim ('elav). 'He goes to the place that everybody goes (to him (=it)) .'

220 (7.4)

hem 'amdu Mam mizman ^ehaMemeS zarfia (?bo). 'They stood there from the time that the sun shone (in him (=it)). '

(7.5)

milftemet ha'olam parca bayom £ehu nolad (bo). 'The World War broke out in the day that he was born (in him (=it)).'

(7.6)

'ani madaber *al ha^anim Selo haya lanu 'oxel (bahen). 'I speak on the years that there was not to us (=that we did not have) food (in them).'

(7.7)

'ani madaber 'al hasefer Se'ata madaber ('alav). 'I speak on the book that you speak (on him (=it)).'

(7.8)

kibalti mixtav meha'università Se'ata kibalta

(mimena).

'I received (a) letter from the university that you received (from her (=it ) ). ' Sentences (7.l)-(7.3) contain head Nouns designating Place. Sentence (7.l) is more commonly uttered with the embedded Prepositional Phrase, at least in my own idiolect.

Sentences (7.2 )- (7.3) are usually

uttered without the embedded Prepositional Phrase because notwithstanding the fact that the embedded NP designates Place, the VP's of the main clauses are identical (i.e. both the Verb and the Prepositional Phrase are identical).

221 In sentences (7.4)-(7.6) the relativized NP designates Time. In this type of sentences, the embedded Prepositional Phrase is usuallydeleted.

The reason for this is the following:

in this type of sen-

tences, the action that is expressed in the main clause usually takes place at the same time that the action in the relative clause occurs. Therefore, the Preposition in the relative clause is always ba- ('in, at').

This fact makes the content of the embedded Prepositional Phrase

redundant, and it is deleted.

Notice that in sentence (7.6) the word

lanim ('years') is part of a Prepositional Phrase which is not dominated by a Time-node, but the embedded Prepositional Phrase may still be optionally deleted. Sentences (7.7)-(7.8) have identical VP's in the main clause and in the relative clause, and thus the embedded Prepositional Phrase may be optionally deleted. I would like to sum up the information which has been provided thus far.

A Prepositional Phrase which is part of a relative clause may

be deleted under one of the following circumstances: (7.9)

(i)

The relativized NP designates Time [sentences (7.4)- (7.6)].

(ii)

The whole VP of the main clause is identical to the VP of the relative clause [sentences (7.2)(7.3), (7.7 )- (7 .8)].

222

The question is, of course, whether this phenomenon should be part of the grammar or whether it should be dealt with on the stylistic level.

I myself believe that the latter possibility is preferable, and

I therefore do not state here any grammatical rule for handling these cases.

2.

Successive vs. Stacked Relative Clauses In Section 1 of Chapter II above I dealt with the derivation

of stacked relative clauses and I claimed that according to my analysis, stacked relative clauses are derived in the following manner, as described in (2.29) above: (2.29)

NP NP NP

S. 1 S, 2

However, stacked relative clauses are not always derived in this manner.

223 Before illustrating the other derivation of stacked relative clauses, I should explain the difference between successive and stacked relative clauses. By stacked relative clauses we mean that several clauses modify the same NP; by successive relative clauses we refer to a sentence in which the first relative clause modifies a certain NP in the main clause, the second relative clause modified another NP in the first relative clause, the third relative clause modifies a different NP in the second relative clause, and so on. Two examples will clarify this difference: Stacked (7.10)

haya lah mabat ïetiher 'et ni^mati, Se'orer 'et yscaray, sehldlik 'et lavavi. 'There was to her (=she had) (a) look (m.s.) that purified (m.s.) 0M my soul (f.s.), that woke (m.s.) 0M my desires (m.p.), that lit (m.s.) 0M my heart.'

Successive (7.11)

zehu hakelev £ena%ax 'et heftatul, sebala' 'et ha'axbar, £e'axal 'et hagvina. 'This is the dog that bit 0M the cat, that swallowed 0M the mouse, that ate 0M the cheese.1

224 In sentence (7.10) all three relative clauses modify the Noun ma bat ('look').

In sentence (7.1l) the first relative clause modifies

the Noun hakelev ('the dog'), the second relative clause modifies the Noun hefiatul ('the cat'), and the last relative clause modifies ha'axbar ( 'the mouse'). Obviously, sentences such as (7.10) are derived from underlying structures such as (2.29) above, while sentences such as (7.1l) are derived from underlying structures such as (7.12) below: (7.12)

S

NP. 0

S NP

NP

Furthermore, every stacked relative clause is usually preceded by a pause longer than the one which precedes a successive relative clause. 2

225 In sentences such as (7.10) there is no way to misinterpret the relative clauses as being successive, because of two reasons: (a) the existence of a pause before each relative clause; (b) the fact that the Verbs in the relative clauses agree in Gender and Number only with the first relativized NP (in the main clause) and not with any other NP inside the stacked relative clauses. 3 Notice, however, that this last criterion (plus the one mentioned in note 3) is not valid for sentences such as (7.1l), because here the Verbs in the relative clauses agree in Gender and Number with any Noun in this sentence.

Selectional restrictions cannot solve the

problem either, because the dog may be able to do all the actions mentioned in this sentence (i.e. the dog may not only bite the cat, but may also swallow the mouse and eat the cheese).

Nevertheless, I believe

that every one would interpret this sentence as containing successive rather than stacked relative clauses (even though the last interpretation is not at all impossible).

The reason for interpreting the rela-

tive clauses of (7.1l) as successive probably is related to several conceptions that we all have about the world.

We usually associate dogs

with cats, cats with mice, and mice with cheese. However, there are some cases in which the distinction between stacked and successive relative clauses is almost invisible, 4 if the former are derived in the manner described above in (2.29). for example, the following sentence:

Consider,

226

(7.13)

ha'i£ Seracafi 'et haftalban leganav 'et hayahalomim hu po£ea* yad.ua* . 'The man that murdered OM the milkman that stole OM the diamonds he (=is) criminal known (= a known criminal).'

This sentence is obviously ambiguous. 5

One reading is (a) that

the man both murdered the milkman and stole the diamonds, and the other reading is (b) that the man murdered the milkman, and the milkman was the culprit who stole the diamonds. Eventually, if the reading is (a), there will be a lengthier pause before the word Seganav ('that stole'). Usually, however, a sentence such as (7.13) is interpreted as containing successive rather than stacked relative clauses.

In such

cases where the only distinction between successive and stacked relative clauses lies in the difference of the pausal length, the Hebrew speaker prefers to use stacked relative clauses, which are derived from the following underlying structure [rather than from the underlying structure given in (2.29) above]:

227 (7.14)

In other words, if a sentence such as (7.13) is meant to contain stacked relative clauses, it will usually have the surface structure given in (7.15)(i) below, and consequently, its deep structure will have the shape given in (7.15)(ii) below: (7.15)

(i) ha'jg geracaft 'et haftalban va(£e-) ganav 'et hayahalomim hu posea' yadua*. (ii)

ha'il^ racafi hafialban

ha'i% ganav hayahalomim

228 (The Object Markers and the Copulative Pronoun are absent from the deep structure because they are introduced by transformational rules.

As to the structure of S^, see note 21 in Chapter II.) Notice that the Relative Marker may be either repeated at the

beginning of each stacked relative clause or only at the beginning of the first one [see (7.15)(i) above].

The first case results from first

applying the relativization rule and then the rule for adjoining S^ and S^, while the second case is a case of the reverse process.

3.

Extraposition from an NP Even though word order is rather loose in Hebrew (cf. Chapter

Vl), certain word permutations, which are permissible in languages whose word order is quite strict, are not permissible in Hebrew. Such, for example, is the case of extraposition of a relative clause from its head NP. I refer to cases such as in (7.16) below which are grammatical in English, but their Hebrew literal translations are ungrammatical: (7.16)

(i)

I gave a gun to my brother which I had cleaned.

(ii)

He let the cats out which were meowing. 6

Thus in Hebrew there can be no movement of a relative clause without its head Noun, and no head Noun may be moved without the

229 relative clause which is dominated, by it.

Consider the following

examples as an illustration: (7.17)

(i)

hu natan la'aftiv 'et harove £ehu nika. 'He gave to his brother OM the gun that he cleaned.'

(ii)

la'aftiv hu natan 'et harove Mehu nika.

(iii)

'et harove Sehu nika hu natan la'aftiv.

(iv)

* hu natan 'et harove la'aftiv £ehu nika.

230 NOTES TO CHAPTER VII

1.

Cf. examples such as (4.59), (4.63), and (4.64) in Chapter IV.

2.

Usually there exists no pause whatsoever before a restrictive relative clause, but when there is a succession of relative clauses [e.g. as in sentence (7.1l)] there is usually a tendency to pause briefly before each relative clause.

3.

Such misinterpretation is not always avoided by the Gender/Number agreement.

At times it is avoided by selectional restrictions.

That is, if the Verbs in the relative clauses may serve as the Predicates of only a specific type of Subjects (e.g., +Human, +Animate, etc. ) and in only the relativized NP in the main clause meets these requirements, then the relative clauses are interpreted as stacked and not as successive. 4.

By claiming that the distinction is almost invisible, I mean that the only difference between the two types of sentences is a difference in the length of the pauses.

As was mentioned in note 2 above,

pauses exist in both types of sentences, and thus the differences are not between the absence vs. the presence of pauses, but rather between short vs. lengthy pauses.

This, of course, makes the

distinction quite difficult to discern. 5.

I intentionally avoided any periods in this sentence.

6.

The examples are extracted from Ross' Constraints, pp. 3-4.

A P P E N D I X

231

APPENDIX

The following is a list of the rules which play an important role in this study.

Rules which have been modified throughout the

discussion are stated here in their final form. The numbers to the right of each rule, in brackets, refer to the number of this rule, or to its intermediate stages in the text.

A.

Base Rules 1.

Predicate-Phrase

2.

VP - (V (Manner) )

3.

Predicate
4 5 4

3+Pro

Conditions :

[(3.16); ]predlcate

]g

5

(i)

3 does not contain Tense;

(ii)

e i t h e r 2 or 4 contains S or 2 and 4 are

[«Definite]. [(2.67)]

smixut-Construction Transformation [ [ X^NP^Poss 1 2 3

» ]NP 4

fi fi fi

4+312

P a r e n t h e t i c a l Clause Embedding [

[ X NP 1 2 1

[(2.12)]

Y ]g [ W NP Z ] g ] s 3 4 5 6

2+456

=>

fi fi fi

3

R e l a t i v i z a t i o n Transformation SD: SC:

(4.28)]

[ X NP # 1 2 3 1 2

fi

[

W 4

NP 5

[ (2.45); Z ] 6

RM+4i(a) [ +Pro ] ] 6

# fi

Y L 7 8 8

SC ( a ) - e i t h e r ( i ) i f 5 i s dominated by a Prep-Phrase node or by a coordinate NP-node; or ( i i )

if

4 . . . 6 i s dominated by S and does not contain Tense. SC (b) - elsewhere

234 Conditions :

6.

2 = 5

(i)

u

(11)

4...6

[(3.50); (4.46)]

Relative Clause Reduction X 1

NP 2

RM 3

Y 4

Aux 5

Predicate 6



0

fh

6

Conditions:

(i)

Z =*> 7 7

5 does not contain Tense and 2 £ mi/ma ;

(ii)

obligatory if 6 dominates NP or Adj and NP=i> 2... .7;

(iii)

if S =>2...7, applies optionallyonly if 6 dominates NP.

7.

X

imi ] ma I zman j makomj 2

0

RM

Y

3

4

3

4

Condition: 8.

[(4.47); (4.66)]

Reducing mi/ma/zman/makom =

obligatory if 2 = zman and 1 = k£-,

Agreement of Definiteness

r+N

i

|+DefJ

[+Adj] [ +N] 3 [3[+Def]

[(3.52)]

NP

235 Relative Markers Transformation SD:

NP 1

RM 2 S

[(5.10)]

Y 3 île- if 1 is mi/ma or if 1 = 0 and

(i)

dominated by Time. SC:

(ii)

'aSer if 1 = 0

Opt.(iii) ha- if 3 = V (iv)

10.

f'aSerl lie I

Ob.ject Marker Insertion [ X 1

V 2

NP 3

2

OM+3

J I 4 VP =>

Condition:

4 3 = [+Definite]

[ (2.68)]

B I B L I O G R A P H Y

236

BIBLIOGRAPHY

Annear, Sandra. English and. Mandarin Chinese: Definite and Indefinite Determiners and Modifying Clause Structure. Mimeographed. Ohio State University, 1965. .

Relative Clauses and Conjunctions. State University, 1967.

Mimeographed.

Bach, Emmon. An Introduction to Transformational Grammars. Holt, Rinehart and Winston, 1964. .

"Have and be in English Syntax."

Ohio New York:

Language 43, pp. 462-485.

, and Robert Harms (eds.). Universals in Linguistic Theory. New York: Holt, Rinehart and Winston, 1968. .

"Nouns and noun phrases," in Bach and Harms' Universals, pp. 91-122.

Baker, LeRoy. Definiteness and Indefiniteness in English. M.A. thesis, University of Illinois, 1965.

Unpublished

Bar-Adon, Aaron. "Analogy and analogic change as reflected in contemporary Hebrew." Proceedings of the Ninth International Congress of Linguists. Cambridge. Mass., 1962. The Hague: Mouton, 1964. .

"The Evolution of Modern Hebrew," in Acculturation and Integration. New York: American Histradrut Cultural Exchange, 1965.

Baruch, Y. "seder nose vanasu bamiìSpatim hapotafiim bamillot 'aìSer, ka'aSer, kase-, miìiSe-, lix£e-" [="The order of Subject and Predicate in subordinate clauses"]. Lasonenu La'am 19, pp. 195-206. Ben-David, Abba. la£on hammiqra 'o laiionfiaxamim[= Biblical Hebrew or Mishnaic Hebrew?]. Tel-Aviv: Machbarot Lesifrut, 1951. Chomsky, Noam. Morphophonemics of Modern Hebrew. Unpublished M.A. thesis, University of Pennsylvania, 1951. 237

238

.

Syntactic Structures.

The Hague:

Mouton, 1957.

. "A transformational approach to syntax," in A. A. Hill (ed.) Proceedings of the Third Texas Conference on Problems of Linguistic Analysis in English, 1958. Texas: The University of Texas, 1962. Reprinted in Fodor and Katz's The Structure, pp. 211-245. . Aspects of the Theory of Syntax. Press, 1965.

Cambridge:

The M.I.T.

, and Morris Halle. The Sound Pattern of English. Harper and Row, 1968. Cressey, William. "Relative adverbs in Spanish: analysis." Language 44, pp. 487-500.

New York:

a transformational

De Saussure, Ferdinand. Course in General Linguistics. Edited by Charles Bally and Albert Sechehaye (1915).Translated from the French by Wade Baskin. New York: Philosophical Library, 1959. Fodor, Jerry, and Jerrold Katz. Prentice-Hall, 1964.

The Structure of Language.

New Jersey:

Gesenius1 Hebrew Grammar. Edited and enlarged by E. Kautzsch (1909). Translated to English by A. E. Cowley. Oxford: Oxford University Press, 1910. Gladney, Grank. On Relative Clauses in Russian - I. University of Illinois, 1967. Hayon, Yehiel. Relative Clauses in Modern Hebrew. thesis. University of Texas, 1968.

Mimeographed.

Unpublished M.A.

Jacobs, Roderick, and Peter Rosenbaum. English Transformational Grammar. Massachusetts: Blaisdell Publishing Corp., 1968. Jespersen, Otto. The Philosophy of Grammar (1924). Norton and Co., 1965. Jouon, Paul. Grammaire de l'Hebreu Biblique. Pontificial, 1923.

Rome:

New York:

W. W.

Institut Biblique

Karttunen, Lauri. The Identity of Noun Phrases. Paper presented at the 42nd annual meeting of the Linguistic Society of America, Chicago, 1967.

239 .

Coreference and Discourse. Paper presented at the 43rd annual meeting of the Linguistic Society of America, New York, 1968.

Katz, J. J., and P. M. Postal. An Integrated Theory of Linguistic Description. Cambridge: M.I.T. Press, 1964. Koutsoudas, Andreas. Writing Transformational Grammars: tion. New York: McGraw-Hill, 1966.

An Introduc-

Kuroda, S.--Y. Generative Grammatical Studies in the Japanese Language. Ph.D. dissertation, M.I.T., 1965. .

"English relativization and certain related problems." Language 44, pp. 244-266.

Lakoff, George. sity.

Pronouns and Reference.

Mimeographed, Harvard Univer-

.

Counterparts, or the Problem of Reference in Transformational Grammar. Reprint of a paper presented at the summer meeting of the Linguistic Society of America under the title Semantics, Logic, and Opacity. Harvard University, 1968.

.

Selectional Restrictions and Beliefs about the World. Paper delivered at the 43rd annual meeting of the Linguistic Society of America. New York, 1968.

Langacker, Ronald. "Observations on French possessives." 44, pp. 51-75.

Language

Lees, Robert, and E. S. Klima. "Rules for English pronominalization." Language 39, pp. 17-29. McCawley, James. "Where do noun phrases come from?" To appear in Jacobs and Rosenbaum ' (eds.) Readings in English Transformational Grammars. Moore, T., T. Shopen, and R. Stockwell. "Relativization," in Working Papers of the English Syntax Conference. U.C.L.A.-Air Force, 1967. Oettlnger, Anthony (Principal Investigator). Mathematical Linguistics and Automatic Translation. Report NSF-17 of the Computational Laboratory. Cambridge: Harvard University, 1966.

240 O m a n , Uzzi. Nominal Compounds in Modern Literary Hebrew. dissertation, Hebrew University, 1964.

Ph.D.

Peretz, Yitzhak, mispat hazziqqa [= The Relative Clause]. Dvir, 1967.

Tel-Aviv:

Perlmutter, David. On the Article in English. University, 1968.

Mimeographed.

Postal, Paul. On Coreferential Complement Sub.ject Deletion. IBM Watson Research Center.

Brandeis New York:

.

"On so-called 'Pronouns' in English" in Monograph Series on Languages and Linguistics 19, 17th Annual Round Table, Georgetown University, 1966.

.

Cross-Over Phenomena, a Study in the Grammar of Coreference. New York: IBM Corporation, 1968.

Rosen, Chaim. ha'ivrit sellanu [= Our Hebrew]. 1956. .

'ivrit tova [= Good Hebrew].

Tel-Aviv:

Jerusalem:

Am Oved,

Kiryat Sefer, 1958.

Ross, John. "A proposed rule of tree-pruning," in Oettinger's Mathematical, pp. (iV-l) - (lV-18). .

"Relativization in extraposed clauses," in Oettinger's Mathematical, pp. (v-l) - (v-3).

.

Gapping and Order of Constituents. 1967.

.

Constraints on Variables in Syntax. M.I.T., 1967.

Mimeographed.

M.I.T.,

Ph.D. dissertation,

Rubinstein, Eliezer. hammi^pat halSMemani [= The Nominal Sentence]. Tel-Aviv: Hakibbutz Hameuchad, 1968. Segal, Moshe. diqduq la^on hammiSna [= Grammar of the Mishnaic Language]. Tel-Aviv: Dvir, 1936. Wolfe, P. and B. Partee. "Pronominalization," in Working Papers of the English Syntax Conference. U.C.L.A.-Air Force, 1967. Yoeli, Mordechai. 1961.

taftbir 'ivri [= Hebrew Syntax].

Tel-Aviv:

Yesodot,